Today’s infrastructure is connected to many other infrastructure assets, systems, and networks that it depends on for normal day-to-day operations. These connections, or dependencies, may be geographically limited or span great distances (NIPP 2013). The many points of infrastructure connections, and their geographic distribution, make the infrastructure environment much more complex. The U.S. Department of Homeland Security (DHS) works to strengthen critical infrastructure security and resilience by generating greater understanding and action across a (largely) voluntary partnership landscape. This is achieved by working with private and public infrastructure stakeholders to resolve infrastructure security and resilience knowledge gaps, inform infrastructure risk management decisions, identify resilience-building opportunities and strategies, and improve information sharing among stakeholders through a collaborative partnership approach. This talk highlights the Department’s efforts to present a more comprehensive picture of security and resilience through a “system of systems” approach.

Academic methodological research is often done with the hope of creating mathematical methods that will be used in practice. At the same time, there is a significant gap between publishing papers and having the methods described actually be used in industry. In this talk, we offer advice for bridging this gap. We discuss challenges arising from a difference in focus between academic and industry research, and also an incomplete awareness within academia of the full context in which methods are deployed in industry. We then discuss strategies for overcoming these challenges, describing them using examples from the presenter's experiences as a data science manager at Uber working on Uber's carpooling product, UberPOOL, and as an academic developing Bayesian optimization algorithms for use at Yelp and the Bayesian optimization startup company SigOpt. This talk is aimed at academics who want their research to be used in industry, soon-to-graduate PhD students who are making a leap into an industry career, and practitioners interested in exploring ways to be more effective.

Many powerful statistical methods available for studying simulation output are under-appreciated and consequently under-used, because they are considered to be hard-to-understand, arcanely mathematical, and hard to implement. Such methods can invariably be implemented using data-driven resampling methods, making their underlying rationale quite transparent. There is little need for much formal mathematics, and what there is can be made visually obvious, with method and results explained and presented using figures and graphs, often with dynamic animation. This approach in studying simulation output will be illustrated by a number of examples drawn from simulation studies and real applications. A bonus of the approach is that it is quite easy to create one’s own ‘bespoke’ method of analysis tailored to a particular problem. Such examples will be presented and analyzed in ‘real-time’ in the talk itself, enabling the results to be immediately displayed.

An ambulance dispatcher decides which ambulances to allocate to new calls, and how to relocate ambulances in order to maintain a good coverage. Doing this, it is valuable to have information about the future expected response times in different parts of the area of responsibility, as well as the expected number of available ambulances. We present a simulation model that can be used to predict this, and compare the results to a naïve forecasting model. The results show that while it is difficult to accurately predict the future system state, the simulation based prediction manages this better than the naïve model.

We consider optimization problems with uncertain constraints that need to be satisfied probabilistically. When data are available, a common method to obtain feasible solutions for such problems is to impose sampled constraints, following the so-called scenario generation (SG) approach. However, when the data size is small, the sampled constraints may not support a guarantee on the feasibility of the obtained solution. This paper studies how to leverage parametric information and the power of Monte Carlo simulation to obtain feasible solutions even when the data are not sufficient to support the use of SG. Our approach makes use of a distributionally robust optimization formulation that informs the Monte Carlo sample size needed to achieve our guarantee.

Right Hospital – Right Time (RH-RT) is the conceptualization of the use of descriptive, predictive and prescriptive analytics with real-time data from Accident & Emergency (A&E)/Emergency Departments (ED) and centers for urgent care; its objective is to derive maximum value from wait time data by using data analytics techniques, and making them available to both patients and healthcare organizations. The paper presents an architecture for the implementation of RH-RT that is specific to the authors’ current work on a digital platform (NHSquicker) that makes available live waiting time from multiple centers of urgent care (e.g., A&E/ED, Minor Injury Units) in Devon and Cornwall. The focus of the paper is on the development of a Hybrid Systems Model (HSM) comprising of healthcare business intelligence, forecasting techniques and computer simulation. The contribution of the work is the conceptual RH-RT framework and its implementation architecture that relies on near real-time data from NHSquicker.

Water Distribution Networks are a particularly critical infrastructure for the high energy costs and frequent failures. Variable Speed Pumps have been introduced to improve the regulation of water pumps, a key for the overall infrastructure performance. This paper addresses the problem of analyzing the effect of the VSPs regulation on the pressure distribution of a WDN, which is highly correlated to leakages and energy costs. Due to the fact that water network behavior can only be simulated, we formulate the problem as a black box feasibility determination, which we solve with a novel stochastic partitioning algorithm, the Feasibility Set Approximation Probabilistic Branch and Bound, that extends the algorithm previously proposed by two of the authors. We use, as black box, EPANet, a widely adopted hydraulic simulator. The preliminary results, over theoretical functions as well as a water distribution network benchmark case, show the viability and advantages of the proposed approach.

At the 2001 Winter Simulation Conference, the progress of simulation in Ford Motor Company’s PowerTrain Manufacturing Engineering (PTME) department was documented in a paper that focused on the contributions of the department in changing the simulation environment at Ford. This paper reviews the progress, changes, and issues in the intervening years experienced by the UK-based PTME simulation team. It summarizes the development of a toolset from a model building capability to data analysis, to experimentation, and results analysis. It outlines how capabilities have expanded while maintaining quick delivery of results; it references the managements' changing attitude and how academic research has advanced simulation in PTME.

In emergency call centers (for police, firemen, ambulances) a single event can sometimes trigger many incoming calls in a short period of time. Several people may call to report the same fire or the same accident, for example. Such a sudden burst of incoming traffic can have a significant impact on the responsiveness of the call center for other events in the same period of time. We examine data from the SOS Alarm center in Sweden. We also build a stochastic model for the bursts. We show how to estimate the model parameters for each burst by maximum likelihood, how to model the multivariate distribution of those parameters using copulas, and how to simulate the burst process from this model. In our model, certain events trigger an arrival process of calls with a random time-varying rate over a finite period of time of random length.

Computer simulation experiments are commonly used as an inexpensive alternative to real-world experiments to form a metamodel that approximates the input-output relationship of the real-world experiment. While a user may want to understand the entire response surface, they may also want to focus on interesting regions of the design space, such as where the gradient is large. In this paper we present an algorithm that adaptively runs a simulation experiment that focuses on finding areas of the response surface with a large gradient while also gathering an understanding of the entire surface. We consider the scenario where small batches of points can be run simultaneously, such as with multi-core processors.

The current popular method for approximate simulation from the posterior distribution of the linear Bayesian LASSO is a Gibbs sampler. It is well-known that the output analysis of an MCMC sampler is difficult due to the complex dependence amongst the states of the underlying Markov chain. Practitioners can usually only assess the convergence of MCMC samplers using heuristics. In this paper we construct a method that yields an independent and identically distributed (iid) draws from the LASSO posterior. The advantage of such exact sampling over the MCMC sampling is that there are no difficulties with the output analysis of the exact sampler, because all the simulated states are independent. The proposed sampler works well when the dimension of the parameter space is not too large, and when it is too large to permit exact sampling, the proposed method can still be used to construct an approximate MCMC sampler.

Hybrid simulation, combining exact and approximate algorithms, provides an alternative to a completely stochastic simulation. However, one challenge for the efficient implementation of hybrid simulations is the additional overhead due to frequent switches between the two regimes. The amount of additional overhead considerably increases with the number of discrete events in the stochastic regime. However, reactions that take place rather frequently cannot completely be avoided due to the accuracy requirements. In this paper, we present an improved hybrid simulation method which takes advantage of the Hybrid Rejection-based Stochastic Simulation Algorithm (HRSSA), a variant of the hybrid simulation approach. To reduce the overhead on account of the switches from the stochastic to the deterministic regime, we analyse and record the dependencies of reactions as well as species between the stochastic and deterministic subnetworks. Comparing our technique with existing ones shows a clear improvement in terms of runtime, while preserving accuracy.

This paper considers the problem of ranking and selection with covariates and aims to identify a decision rule that stipulates the best alternative as a function of the observable covariates. We propose a general data-driven framework to accommodate (i) high-dimensional covariates and (ii) general (nonlinear) dependence between the mean performance of an alternative and the covariates. For both scenarios, we design new selection procedures and provide certain statistical guarantees, by leveraging the data-intensive environment and various statistical learning tools. The performances of our procedures are exhibited through simulation experiments.

This paper offers a process for understanding and analyzing the most effective interventions for eliminating human trafficking. The process consists of several methods for gathering, parsing and testing information about the Overseas Filipino Workers (OFW) economic system as it relates to trafficked persons. The methodology is comprised of a series of methods that include interviews, causal loop analysis, data collection, system dynamics simulation model, and scenario simulation runs. The last method, scenario simulation runs, tests the cause and effect relationships between government policies, overseas workers options, hiring companies, and the economy in which they all operate and interact. The importance of this process is that it is robust, repeatable, and efficient in a real-world setting.

The paper describes a prototype agent-based model used to predict the spontaneous settlements that can arise among internally displaced forced migrants in the Democratic Republic of the Congo. The internally displaced persons—in the real-world and the model—are constrained by geographic and social forces that dictate their ability to locate and reach organized camps run by humanitarian organizations or instead group with others to establish small temporary settlements. This research is of interest to humanitarian response stakeholders who try to locate self-settlements in order to administer humanitarian assistance and prevent further loss of human life.

This paper describes a prototype agent-based model used to explain why and how a norm of humanitarianism diffuses through a population. The model is constructed on norm diffusion theories as a foundation for developing explaining the emergence of Citizen Initiatives in a humanitarian and development context. We assume that in the model, some agents are already norm adopters (advocates), some have a humanitarian potential that can be activated with persuasion, while others will never adopt the norm of humanitarianism under any condition. In this model, we try to determine whether parameters such as agents’ values, thresholds for accepting alternative values, values degradation, and peer-pressure affect agents’ decision to become humanitarian activists.

Humanitarian aid is material or logistical assistance provided for humanitarian purposes, typically in response to humanitarian crises including natural disasters and man-made disaster. Humanitarian assistance requiring short response time windows in almost the whole world may be subject to long queues due to managing problems, e.g., the lack of control and/or inefficient infrastructure. This work tackles such challenge by proposing a low-cost planning and managing model and method based on a discrete-event simulation mirror connected through WEB tools to a near or far management level. The usual configuration of parallel servers (for instance, supported by local RFID monitoring) is implemented by a discrete-event simulation model that is validated by Jackson Networks (and vice versa). The results show a flexible model that may identify bottlenecks in advance in order to accommodate traffic flow variations.

Humanitarian logistics has recently gained increasing attention from both academics and practitioners. Although various research groups have addressed theoretical and technical developments in humanitarian logistics, only a limited number of those can actually be generalized, extended, accessed, and understood by non-technical practitioners. To tackle these challenges, we develop a simulation model for humanitarian logistics preparedness and a simulation-based serious game to raise awareness and provide accessibility on humanitarian logistics research to a wider audience. The simulation model aims to optimize the network configuration for prepositioning stocks of life-saving goods in Indonesia, while the game aims to provide a risk-free environment where players can craft various strategies to plan and deploy effective humanitarian operations.

Our research focuses on communications among a variety of organizations that coordinate their rapid responses to catastrophic disasters. Within the context of FEMA’s National Response Coordination Center, we constructed an agent-based simulation model of the inter-organizational communications happening via their Web-based Emergency Operations Center, email, phone calls, and face-to-face conversations as the support requests were addressed and fulfilled. We developed our model based on FEMA documentation, observations, interviews, and exercise data. In this paper we outline our model development process and provide details about our simulation model to highlight and address some of the particular challenges one faces when developing simulation models of disaster response activities. We describe what specific aspects of communication media and situational factors our model was developed to test, and also present the design and select results of our first research experiment using this model.

Even the best scientific minds cannot repeatedly produce desired results when working in sub-optimal systems. Complex enterprises are difficult to understand and manage. Cause-and-effect relationships are often separated in time and space, making real improvements challenging. To understand how complex systems work it is essential that we employ tools that accurately map and quantify the dynamics that drive results. Computer modeling and simulation (CMAS) is a valuable design, planning, management, and overall analytical decision-support tool to achieve effective and efficient results. CMAS could become a ubiquitous tool in the lengthy and complex environment of the clinical research (CR) enterprise. Without the comprehensive understanding gained when applying CMAS, organizations may continue to be overwhelmed by problems such as unnecessary bottlenecks, high costs, low productivity, and the inability of retaining critical staff. The approach explained here may complement or even replace traditional methods when organizations pursue greater enterprise capability.

The growing epidemic of drug shortages in the United States causes challenges for providers all across the critical health care infrastructure and demonstrates the lack of resiliency within drug delivery supply chains. With many drugs having no acceptable substitute, drug shortages directly translate to a public health and safety risk. One of the understudied elements driving this crisis is the role of human behavior and human decision making across the supply chain echelons. We propose an integrated simulation framework which allows for instantiating, testing, and improving supply chains when accounting for human components of the system. We demonstrate the potential insights that can be obtained with our method through several experiments in which supply chain decision makers account for how others might react to their decisions.

Right Hospital – Right Time (RH-RT) is the conceptualization of the use of descriptive, predictive and prescriptive analytics with real-time data from Accident & Emergency (A&E)/Emergency Departments (ED) and centers for urgent care; its objective is to derive maximum value from wait time data by using data analytics techniques, and making them available to both patients and healthcare organizations. The paper presents an architecture for the implementation of RH-RT that is specific to the authors’ current work on a digital platform (NHSquicker) that makes available live waiting time from multiple centers of urgent care (e.g., A&E/ED, Minor Injury Units) in Devon and Cornwall. The focus of the paper is on the development of a Hybrid Systems Model (HSM) comprising of healthcare business intelligence, forecasting techniques and computer simulation. The contribution of the work is the conceptual RH-RT framework and its implementation architecture that relies on near real-time data from NHSquicker.

In this paper, we build a system dynamics model to investigate the ongoing endeavor to transition from conventional non-renewable transportation systems to renewable ones. The model focuses on light to mid duty vehicles in the private transportation sector. The literature focuses on the environmental and economic aspects of such a transition. We adopt distributive justice as a new angle, define it as access to transportation, justify its relevance by considering it a vital need for people to actualize their capabilities in society and propose a measure to quantitatively measure it in our context. There are several layers of tradeoffs in policy appraisal, yet we are able to catalyze the transition to AFV’s while improving its sustainability. A policy that ensures such a harmonious behavior is one that focuses on the GHG emissions with little to no emphasis on the AFV quotas, while providing support for consumers to switch to AFV’s.

The issue of adequate urban bus routes is very important to general population life quality. It is shown that in Brazil, 25% of the citizens from urban areas uses buses as the main transportation method. In general, bus routes are created and modified without proper methodology or analysis. Discrete Event Simulation is a paradigm that can take into consideration the variability and dynamic nature of bus routes and, therefore, provide valuable insights especially when one is trying to propose a change in an existing bus network. On the other hand, a simulation model for general bus routes is not easy to develop. The present work proposes a generic simulation model for radial networks in order to take into account data which is normally available at buses networks and, then, apply this model to the case of Caieiras city bus network redesign.

For decades, users have been interacting with software through graphical user interfaces. This is also true for modeling and simulation (M&S) tools, a paradigm that affects how models are built and simulated. It also impacts how results are presented, 1) preventing non-experts from using M&S tools, and 2) excluding people who are atypical on a sensory spectrum. This paper focuses on the exploration of key factors contributing to designing M&S user experiences as an approach to mitigate existing thresholds and establish M&S as a means to improving people’s lives. We perform an empirical study with artists, who express creative skill and imagination in a visual form, creating in virtual environments. We report on results of a user experience survey completed after a creative session in virtual environments. Finally, we discuss how these results can benefit M&S tool development.

Simulation is experimentation with a model. The behavior of the model imitates some salient aspect of the behavior of the system under study and the user experiments with the model to infer this behavior. This general framework has proven a powerful adjunct to learning, problem solving, design, and control. In this tutorial, we focus principally on discrete-event simulation – its underlying concepts, structure, and application.

Discrete Event System Specification (DEVS) is a popular formalism for modeling complex dynamic systems using a discrete-event abstraction. At this abstraction level, a timed sequence of pertinent "events'' input to a system (or internal timeouts) cause instantaneous changes to the state of the system. Main advantages of DEVS are its rigorous formal definition and its support for modular composition. This tutorial introduces the Classic DEVS formalism in a bottom-up fashion, using a simple traffic light example. The syntax and operational semantics of atomic (i.e., non-hierarchical) models are introduced first. Coupled (i.e., hierarchical) models are introduced to structure and couple Atomic models. We continue to actual applications of DEVS, for example in performance analysis of queueing systems. All examples are presented with the tool PythonPDEVS, though this introduction is equally applicable to other DEVS tools. We conclude with further reading on DEVS theory, DEVS variants, and DEVS tools.

The Discrete Event System Specification (DEVS) formalism is a set of conventions for specifying discrete event simulation models. In this tutorial, we introduce the core concepts of DEVS. First, we introduce a set of informal requirements from which a formal specification is to be developed. Then, we present different modeling conventions at different levels of abstraction. The tutorial exploits the DEVS formalism’s support for modular model design. The concepts are discussed with an example of cyber-physical systems modeling and implementation, which can be used to understand the main concepts of the formalism.

Facilitated discrete event simulation offers an alternative mode of engagement with stakeholders (clients) in simulation projects. It is particularly beneficial when modeling systems with complex behavior, involving many stakeholders with plurality of opinions and objectives. PartiSim - short for Participative Simulation - is a facilitated modeling approach developed to support simulation projects through a framework, stakeholder-oriented tools, and manuals in facilitated workshops. This tutorial describes the PartiSim approach, available for analysts and simulation modelers to use. A PartiSim study includes six stages, four of which involve facilitated workshops. PartiSim has been developed and tested through working with health care organizations. It can, however, be applied to analyze operational problems in any other context within the services and manufacturing domains. This tutorial introduces PartiSim by describing the PartiSim framework and tools, some applications and example tools, a roadmap to adopting it and concludes with some tips for potential users.

Agent-based modeling (ABM) and simulation is an approach to modeling systems comprised of autonomous, interacting agents. Computational advances are making it possible to develop agent-based models in a variety of application areas, including areas where simulation has not been extensively applied. ABM applications range from supply chains, consumer goods markets, and financial markets, to predicting the spread of epidemics and understanding the factors responsible for the fall of ancient civilizations. Progress suggests that ABM could have far-reaching effects on the way that businesses use computer models to support decision-making and how researchers use models as electronic laboratories to identify promising research directions. Some contend that ABM “is a third way of doing science” and could augment traditional discovery methods for creating new knowledge. This brief tutorial introduces agent-based modeling by describing key concepts of ABM and addressing toolkits and methods for developing agent-based models.

Monte Carlo tree search (MCTS) is a general approach to solving game problems, playing a central role in Google DeepMind's AlphaZero and its predecessor AlphaGo, which famously defeated the (human) world Go champion Lee Sedol in 2016 and world #1 Go player Ke Jie in 2017. Starting from scratch without using any domain-specific knowledge (other than the game rules), AlphaZero defeated not only its ancestors in Go but also the best computer programs in chess (Stockfish) and shogi (Elmo). In this tutorial, we provide an introduction to MCTS, including a review of its history and relationship to a more general simulation-based algorithm for Markov decision processes (MDPs) published in a 2005 Operations Research article; a demonstration of the basic mechanics of the algorithms via decision trees and the game of tic-tac-toe; and its use in AlphaGo and AlphaZero.

Simulation models are integral to modern scientific research, national defense, industry and manufacturing, and in public policy debates. These models tend to be extremely complex, often with thousands of factors and many sources of uncertainty. To understand the impact of these factors and their interactions on model outcomes requires efficient, high-dimensional design of experiments. Unfortunately, all too often, many large-scale simulation models continue to be explored in ad hoc ways. This suggests that more simulation researchers and practitioners need to be aware of the power of designed experiments in order to get the most from their simulation studies. In this tutorial, we demonstrate the basic concepts important for designing and conducting simulation experiments, and provide references to other resources for those wishing to learn more. This tutorial (an update of previous WSC tutorials) will prepare you to make your next simulation study a simulation experiment.

How can you make your projects successful? How can you successfully create a “Digital Twin”? Modeling can certainly be fun, but it can also be quite challenging. You want your first and every project to be successful, so you can justify continued work. Unfortunately, a simulation project is much more than simply building a model - the skills required for success go well beyond knowing a particular simulation tool. A 30-year veteran who has done hundreds of successful projects shares important insights to enable project success. He also shares some cautions and tips to help avoid common traps leading to failure and frustration.

When a company wants to use business process simulation to support decision-making, a simulation study needs to be conducted. Using a clear stepwise method can avoid overlooking key activities during the simulation study and improves the study’s credibility towards decision-makers or project sponsors. In this tutorial, a method to perform a simulation study is proposed, which consists of nine steps and uses continuous assessment as a central feedback mechanism. It synthesizes existing methods proposed in literature and positions the actual construction of a simulation model in specialized software in a wider perspective. Given the method’s high-level nature, its operationalization is illustrated making use of a real-life simulation study at the emergency department of a hospital, showing that the relative weight of different steps in the method depends upon the project’s specificities.

In real-world systems, change and complex interactions are the norm. As a result, stakeholders face challenges and problems when they try to complete tasks and accomplish important activities in these systems. Linear and traditional analytic approaches fall short in helping us to understand problematic behavior in complex systems and in changing such behavior; they provide faulty and misguided recommendations. The system dynamics approach is based on feedback and control theory, and is well suited for tackling dynamic phenomena. This paper discusses the use and applications of system dynamics modeling, general lessons related to how and when to use this approach, and relevant tools.

Statecharts, introduced by David Harel in 1987, is a formalism used to specify the behaviour of timed, autonomous, and reactive systems using a discrete-event abstraction. It extends Timed Finite State Automata with depth, orthogonality, broadcast communication, and history. Its visual representation is based on higraphs, which combine graphs and Venn diagrams. Many tools offer visual editing, simulation, and code synthesis support for the Statechart formalism. Examples include STATEMATE, Rhapsody, Yakindu, and Stateflow, each implementing different variants of Harel's original semantics. This tutorial introduces modeling, simulation, and testing with Statecharts. As a running example, the behaviour of a traffic light, a simple timed, autonomous, and reactive system is modeled. We start from the basic concepts of states and transitions and explain the more advanced concepts of Statecharts by extending the example incrementally. We discuss several semantic variants, such as STATEMATE and Rhapsody. We use Yakindu to model the example system.

This tutorial is for analysts, consultants, and teachers of business modeling. No prior experience is needed. Simulating business challenges and plans has until recently been difficult and time-consuming, but things have changed! We will take you through an "agile" process that makes quantified, working simulations practical for non-experts to build, quickly and reliably. And, we need these tools! Spreadsheet-based methods just cannot handle the interdependencies, feedback, thresholds, and intangible factors that pervade all but the simplest cases. The resulting Living Business Models display with total transparency the factors that management recognizes, and the causality driving the performance outcomes they are interested in. Since everyone sees the same rigorous picture, they get a "joined-up" view that fully explains how everything has been changing. This allows them to explore likely future outcomes under alternative assumptions, decisions, and strategies.

We often simulate multiple variations of the same model – a simulation ensemble – to better understand intricate physical phenomena. The analysis of complex simulation ensembles represents a grand challenge which is approached by both computational and interactive, visual methods. We describe how modern visual analytics helps to analyze simulation ensemble data. A clever combination of computational and interactive methods supports the simulation expert to gain deeper insight into the data and into the physical phenomenon that is represented by the ensemble. An analysis environment that combines interactive visualization and computational analysis provides unique advantages for the exploration and analysis of complex ensemble data. It helps the domain expert to efficiently cope with analysis tasks, in particular when they are only partially defined. In this work, we describe the basics of interactive visual analysis, several approaches to interactive ensemble steering, and means for results quantification and analysis reproducibility.

Dynamical models of systems across many branches of science and engineering can be mathematically represented in terms of stochastic processes such as Markov chains, or deterministically through a system of difference or differential equations. Unfortunately, in all but special cases these models do not enjoy analytical solutions, hence one is left with computer-based approaches by means of stochastic simulators and numerical solvers. As a consequence, the computational cost increases with the dimensionality of the model under consideration, hindering our capability of dealing with complex large-scale models arising from accurate mechanistic descriptions of real-world systems. This paper offers an advanced tutorial on an array of recently developed algorithms that seek to tame the complexity of these models by aggregating their constituting systems of equations, leading to lower-dimensional systems that preserve the original dynamics in some appropriate, formal sense.

This tutorial provides an overview of guarantees on the probability of good selection (PGS), i.e., statistical guarantees on selecting – with high probability – an alternative whose expected performance is within a given tolerance of the best. We discuss why PGS guarantees are superior to more popular, related guarantees on the probability of correct selection (PCS) under the indifference-zone formulation. We review existing procedures that deliver PGS guarantees and assess several direct and indirect methods of proof. We compare the frequentist and Bayesian interpretations of PGS and highlight the differences in how procedures are designed to deliver PGS guarantees under the two frameworks.

Stochastic Gradient Descent (SGD), also known as stochastic approximation, refers to certain simple iterative structures used for solving stochastic optimization and root finding problems. The identifying feature of SGD is that, much like in gradient descent for deterministic optimization, each successive iterate in the recursion is determined by adding an appropriately scaled gradient estimate to the prior iterate. Owing to several factors, SGD has become the leading method to solve optimization problems arising within large-scale machine learning and "big data" contexts such as classification and regression. This tutorial covers the basics of SGD with an emphasis on modern developments. The tutorial starts with examples where SGD is applicable, and then details important flavors of SGD and reported complexity calculations.

We consider the problem of parametric verification, presenting a recent statistical method to perform parametric verification of linear time properties of stochastic models, estimating the satisfaction probability as a function of model or property parameters. The approach leverages Bayesian Machine Learning based on Gaussian Processes. Under mild conditions on continuity of parameters of the satisfaction probability, it can be shown that property satisfaction is a smooth function of such parameters. Gaussian Processes can effectively capture this smoothness and obtain more-accurate estimates of satisfaction probabilities by transferring information across the parameter space. We leveraged this approach to efficiently solve several tasks, like parameter synthesis, system design, counterexample generation, and requirement synthesis. In this tutorial, we will introduce the basic ideas of the approach and give an overview of the different applications.

CARMA (Collective Adaptive Resource-sharing Markovian Agents) is a process-algebra-based quantitative language developed for the modeling of collective adaptive systems. A CARMA model consists of an environment in which a collective of components with attribute stores interact via unicast and broadcast communication, providing a rich modeling formalism. The semantics of a CARMA model are given by a continuous-time Markov chain which can be simulated using the CARMA Eclipse Plug-in. Furthermore, statistical model checking can be applied to the trajectories generated through simulation using the MultiVeStA tool. This advanced tutorial will introduce some of the theory behind CARMA and MultiVeStA as well as demonstrate its application to collective adaptive system modeling.

Diverse methods for complex simulation experiments can contribute to developing and gaining insights into simulation models, for example simulation-based optimization, sensitivity analysis, or statistical model-checking. An effective tool for conducting simulation experiments must be highly flexible to support such a broad range of experimental methods. Furthermore, to facilitate reproducibility and communication of simulation experiments, an effective tool for simulation experimentation must yield experiment descriptions that are easily portable, executable, and human-readable. In this tutorial we introduce SESSL, a domain-specific language for setting up simulation experiments. SESSL is flexible and extensible, and experiment descriptions are executable, often succinct, and can be executed reproducibly across machines and operating systems. Based on a few examples, we demonstrate how SESSL can be leveraged to easily conduct complex simulation experiments while reusing existing software and methods, and how SESSL's capabilities can be extended and combined with arbitrary simulation software via bindings.

Data mining tools have been around for several decades, but the term "big data" has only recently captured widespread attention. Numerous success stories have been promulgated as organizations have sifted through massive volumes of data to find interesting patterns that are, in turn, transformed into actionable information. Yet a key drawback to the big data paradigm is that it relies on observational data, limiting the types of insights that can be gained. The simulation world is different. A "data farming" metaphor captures the notion of purposeful data generation from simulation models. Large-scale experiments let us grow the simulation output efficiently and effectively. We can use modern statistical and visual analytic methods to explore massive input spaces, uncover interesting features of complex simulation response surfaces, and explicitly identify cause-and-effect relationships. With this new mindset, we can achieve tremendous leaps in the breadth, depth, and timeliness of the insights yielded by simulation.

Diseases, like death and taxes, seem an inevitable part of life, with much of modern medicine acting to prevent, cure or manage them effectively. Modeling of diseases began over a century ago and has been one of the success stories of applied mathematics and, more recently, computer simulation. In this article, we describe modeling of both communicable and non-communicable diseases, and provide a review of the relevant literature. Two case studies are discussed, the first describing a modeling study of tuberculosis and HIV and the second capacity planning for stroke services. Our aim is to instigate a discussion of how modeling should be used in the future to answer the most up-to-date questions about global health, particularly those of disease prevention.

Exploratory modeling—as an approach for modeling under uncertainty—is based on the analysis of computational experiments representing many possible model responses in the face of uncertainty. Experiments are generated based on various design factors, such as the way uncertainties are defined and the techniques by which value sets are sampled from these uncertainties. The choice of the design of experiments can impact on the computational cost of experiments as well as has an effect on the results and the conclusions drawn from those results. Despite this significance, experimental design has not been adequately discussed in the exploratory modeling literature. This article investigates which dimensions and and what methods should be considered for an appropriate design of experiments in exploratory modeling. We conclude that there is a need to develop an analytical framework which can assist modelers to design experiments appropriately and to consider a wide range of model responses.

Computer simulation experiments are commonly used as an inexpensive alternative to real-world experiments to form a metamodel that approximates the input-output relationship of the real-world experiment. While a user may want to understand the entire response surface, they may also want to focus on interesting regions of the design space, such as where the gradient is large. In this paper we present an algorithm that adaptively runs a simulation experiment that focuses on finding areas of the response surface with a large gradient while also gathering an understanding of the entire surface. We consider the scenario where small batches of points can be run simultaneously, such as with multi-core processors.

A review of recently published papers demonstrates: simulation practitioners apply the standard methods of inferential and descriptive statistics for their reasoning with simulation generated samples without much critical reflection. Yet, simulation-generated samples differ in important aspects from empirical samples, for which the standard statistical methods have been developed. Simulation models do have inherent epistemic and computational limits for replication that do not exist with empirical data sets. Consequently, neither is simulation-based data generation the same as the collection of empirical data nor is the analysis of synthetic data equally beneficial as of empirical data. These differences are much more fundamental for computer simulation than the problems of specific techniques of inferential statistics which have been criticized recently. If simulation generated data is used for testing research hypotheses the core issue is not the method of statistical reasoning but the assurance of what might be called evidential content.

Monte Carlo techniques to radiation transport play a significant role in modeling complex astrophysical phenomena. In this paper, we design an application model (IMCSim) of an Implicit Monte Carlo (IMC) particle code using the Performance Prediction Toolkit (PPT), a discrete-event simulation-based modeling framework for predicting code performance on a large range of parallel platforms. We present validation results for IMCSim. We then use the fast parameter scanning that such a high-level loop-structure model of a complex code enables to predict optimal IMC parameter settings for interconnect latency hiding. We find that variations in interconnect bandwidth have a significant effect on optimal parameter values. Our results suggest potential value using IMCSim as a pre-step to substantial IMC runs to quickly identify optimal parameter values for the specific hardware platform on which IMC runs.

This paper considers parameter estimation of a reduced-order model (ROM) for building energy dynamics using the modulating function (MF) method. After briefly presenting a model of building dynamics, we recall the MF method and present a way of determining a modulating function directly from the available data instead of defining it a priori. Then, we look at the application of this method in a case study, where actual weather-based data (solar radiation and outdoor ambient temperature) is used. The results together with the advantages of the MF underline the potential of the proposed algorithm.

Real world multivariate data mostly contains correlation structure because generally some variables tend to have a similar behavior or some dependency structure. This is due to the nature of data generation process. The variables can contain cross-sectional correlation, correlation between variables at a given time stamp, and temporal correlation, correlation between various observations of a given variable at various timestamps. Also, there are datasets which contain both cross-sectional and temporal correlations. Modeling of such correlation structure in the data is important because it provides us the predictive power and affects the Machine Learning algorithms in various ways, due to relationships within variables. We propose a methodology to simulate a cyber activity dataset, containing both cross-sectional and temporal correlations, also called Panel dataset. The proposed methodology uses non-parametric approach to induce correlations among feature vectors and across time without disturbing the marginal distributions of features.

In emergency call centers (for police, firemen, ambulances) a single event can sometimes trigger many incoming calls in a short period of time. Several people may call to report the same fire or the same accident, for example. Such a sudden burst of incoming traffic can have a significant impact on the responsiveness of the call center for other events in the same period of time. We examine data from the SOS Alarm center in Sweden. We also build a stochastic model for the bursts. We show how to estimate the model parameters for each burst by maximum likelihood, how to model the multivariate distribution of those parameters using copulas, and how to simulate the burst process from this model. In our model, certain events trigger an arrival process of calls with a random time-varying rate over a finite period of time of random length.

We focus on a service system in which the customer arrivals are non-stationary and our goal is to determine a server staffing schedule that ensures that arriving customers do not experience long and/or unpredictable queue times. An airport ticket counter is an example of such a system. Passengers arrivals are nonstationary, yet arriving passengers do not wish to wait in long lines to check into their flights. Moreover, unpredictability is a significant issue in these environments as it often forces passengers to arrive earlier than necessary “just in case.” Unfortunately, we rarely know the precise form of the arrival process and must use observed samples to set the staffing policy. We show through a case study that simulation combined with a specialized input analysis tool can be used to determine good staffing policies in these environments.

Residential energy demand dynamics at household level can be studied through demographic, behavioral and physical characteristics of the household. In this paper, we develop an agent-based model using a bottom-up approach to build disaggregated energy demand estimates at the household level at an hourly interval. A household level analysis is made possible via the use of synthetic populations for the urban and rural areas of Virginia, USA. The energy consumption estimate is based on householders' demographics, their behaviors, and activities, ratings of appliances used in energy-related activities, space conditioning fuels, physical characteristics of the home, and weather conditions. Results from the simulation are then validated with actual demand curves from Rappahannock county in Virginia using dynamic time warping. The simulation results show that the model produces realistic energy demand profiles.

The Discrete Event System Specification (DEVS) is a general modeling formalism with sound semantics founded on a system theoretic basis. It can be used as a base for the development of specialized modeling formalisms. Usually, the extensions of DEVS expand the classes of systems models that can be represented in DEVS. However, with a growing number in new variants of DEVS and an increasing number of problems to be solved using discrete simulation techniques, it is necessary to define the relations among different approaches. This paper presents a conceptual modeling perspective applied to DEVS extensions that structure a framework over the traditional modeling and simulation approach. The framework provides a multilevel structure to analyze the features required for each extension type. Two main types of extensions are identified: variants and subclasses. In order to illustrate the proposed guidelines, the Routed DEVS formalism is presented as example of the subclass type.

Software-hardware co-design enabled with co-simulation is useful for building embedded computing systems. Indispensable to design is developing hardware and software simulation models at appropriate abstraction levels. Toward this goal, this paper presents a study of combined Register-Transfer-Level (RTL) and software system modeling. Specifically, composition of hardware and software models is proposed and a co-simulation environment to simulate the models is developed. The hardware and software parts of a prototypical Network on Chip (NoC) system are modeled and simulated. The hardware part is specified at RTL level using the DEVS Suite Simulator and the software part defined as a MATLAB script. A Functional Mock-up Interface (FMI) is developed for the DEVS-Suite Simulator to support hardware and software model coupling and co simulation. This study details modular development of hardware and software models executing on disparate environments instead of employing a monolithic modeling method supported with a monolithic simulation engine.

In designing a general modeling formalism for domain experts, a key challenge is to support a broad selection of their preferred paradigms yet minimize their exposure to complexity. With this aim, a formalism called Symmetric DEVS is proposed for specifying models that incorporate elements of discrete event simulation, dataflow programming, and agent-based modeling. Symmetric DEVS is based on the Discrete Event System Specification (DEVS) formalism, but differs in that atomic and composite nodes for discrete events are complemented with function and collection nodes for dataflow and agents. Like DEVS, nodes communicate over simulated time via message ports, but they also feature flow ports accommodating initialization and finalization operations. To minimize conceptual complexity, specifications are pared down to the essential elements and formulated to exhibit a high degree of symmetry. This paper defines the mathematical elements of Symmetric DEVS and presents an example of each of the four types of nodes.

We explore how to use conceptualization boards to familiarize students, teachers, parents, and the community in general with discrete-event simulations. We raise the question: how can we make simulations accessible to non-simulationists? We argue that with the use of tactile tools users can not only conceptualize but also create simulations. We developed a prototype board that captures discrete-event simulation’s blocks like arrivals, queues, processes and resources. The board applies the concept of gamification to engage users. The board is simple to replicate and can be created with accessible material or 3D printing capabilities. Further, we propose the board be used simultaneously with software applications to facilitate the transition from analogue to digital tools. Future work investigates 1) extended simulation, 2) universal access to simulation and 3) insight generation.

Domain-specific modeling languages (DSMLs) facilitate concise and succinct model descriptions. DSMLs are commonly realized by defining a custom grammar and executing models in an interpreter. This provides flexibility in language design as well as in the implementation of simulation algorithms. However, this type of implementation can lead to a negative impact on simulation performance in comparison to implementing models in general-purpose programming languages (GPL). To mitigate this problem, we propose using runtime code generation. This allows us to replace specific parts of a model at runtime by compiling generated GPL source code. In this paper, we demonstrate the potential benefit of this concept based on ML-Rules, a DSML for modeling and simulating biochemical reaction networks. Using code generation for arithmetic expressions in ML-Rules’ reaction rate functions, we were able to decrease the runtime by up to 40% in complex application scenarios.

Engineering workflow models are frequently used to optimize an engineering endeavor for some well-defined performance metrics, such as time-to-market or monetary costs. Static workflow analysis is often insufficient due to the complex interleavings of different activities and the interplay with limited resources. Simulation-based techniques provide a feasible alternative to static analysis. In this paper, we provide an automated translation from engineering workflow models to DEVS (Discrete Event System Specification) models, useful for simulation and subsequent performance evaluation. Our translation supports the vast majority of the essential workflow control patterns, previously identified by van der Aalst et al. Thanks to the use of simulation, our approach is able to deal with stochastically varying activity execution times and workflow decisions, potentially evolving between subsequent iterations of the workflow. Our approach is implemented in the Modelverse, where the mapping to and simulation of DEVS models remains completely hidden from the process modeler.

Middleware is required to support and interface multi-modal Dynamic Data Driven Application Systems (DDDAS) with back-end and other computing facilities. Middleware is also needed to support distributed simulations and emulations needed in earlier phases of system development. This work describes the Green Runtime Infrastructure (G-RTI), an energy-efficient client server based middleware developed to support distributed DDDAS simulation, emulation and deployment. G-RTI eases and accelerates the development and testing of multi-modal studies, testbeds and DDDAS systems. It serves as a platform for research in energy reduction techniques for middleware services. The services implemented by G-RTI are described and results of benchmarking studies are reported. Its application is demonstrated through a use-case for an end-to-end implementation of a connected vehicle application. G-RTI is open source.

Due to the evolution of programming languages, interpreted languages have gained widespread use in scientific and research computing. Interpreted languages excel at being portable, easy to use, and fast in prototyping than their ahead-of-time (AOT) counterparts, including C, C++, and Fortran. While traditionally considered as slow to execute, advancements in Just-in-Time (JIT) compilation techniques have significantly improved the execution speed of interpreted languages and in some cases outperformed AOT languages. In this paper, we explore some challenges and design strategies in developing a high performance parallel discrete event simulation engine, called Simian, written with interpreted languages with JIT capabilities, including Python, Lua, and Javascript. Our results show that Simian with JIT performs similarly to AOT simulators, such as MiniSSF and ROSS. We expect that with features like good performance, user-friendliness, and portability, the just-in-time parallel simulation will become a common choice for modeling and simulation in the near future.

Since simulation-based optimisation typically requires large numbers of runs to identify sufficiently good solutions, the costs in terms of time and hardware can be enormous. To avoid unnecessary simulation runs, surrogate models can be applied, which estimate the simulation output under a given parameter combination. Model preemption is a related technique that dynamically analyses the simulation state at runtime to identify runs unlikely to result in a high-quality solution and terminates such runs early. However, existing work on model preemption relies on model-specific termination rules. In this paper, we describe an architecture for simulation-based optimisation using model preemption based on estimations of the simulation output. In a case study, the approach is applied to the optimisation of traffic light timings in a traffic simulation. We show that within a given time and hardware budget, model preemption enables the identification of higher-quality solutions than those found through traditional simulation-based optimisation.

Dynamic Data Driven Applications Systems (DDDAS) is a paradigm where data is dynamically integrated into an executing application, and in reverse, the application dynamically steers the measurement process in a feedback control loop. Since its inception in 2000, the DDDAS concept has been successfully applied to a host of application areas. New technologies are emerging such as big data, the Internet of Things, and cloud/edge computing. With these trends DDDAS is poised to have large-scale impacts in areas such as smart cities, manufacturing, health care, and security, to name a few. Each author describes their views concerning the important research challenges facing the DDDAS paradigm and opportunities for impact in the years ahead.

Simulation interoperability is a recurring theme in simulation conferences and workshops for more than 20 years. With the IEEE Standards 1278 and 1516, two simulation interoperability standards were introduced, and both were adapted and implemented by the community. Nonetheless, the simulation community is still struggling with interoperability challenges that are not solved. Why is this the case? This paper gives an overview of the current approaches to simulation interoperability, including the standardized approaches as well as contributions of simulation formalism. It then addresses the mathematical foundations of simulation interoperability, including model theory. As a result, the need for the consistency in the representation of truth in all participating simulation systems emerges as the concept that needs to be addressed by interoperability solutions.

In this paper, we analyze the challenges in data acquisition for simulation models of production systems based on two cases from the robotics and aerospace industries. Unlike prior research, we focus not only on the challenges of data acquisition but also on how these challenges affect decisions in production systems. We examine this linkage using the concepts of strategic objectives, decision areas, and internal fit from operations management literature. Empirical findings show that for data acquisition to lead to improved production system performance it is necessary to develop standards. Standards should consider ownership of data by different functions within a manufacturing company, alignment of data to performance measurements, and the connection between data, information, and production decisions. Using these concepts, this paper proposes a set of guidelines that facilitate the standardization of data acquisition for simulation models in production systems. We conclude by discussing the managerial implications of our findings.

The latest innovations in Information and Communication Technologies (ICT) will change manufacturing and simulation forever. One of those changes involves the use of simulation-based cloud services and the integration of those services with operational problems on the shop floor. In this paper, we focus on the control problems, which we represent as a multilayer process graph. The nodes in this graph are control functions, which are cognitive in nature and can be executed by different agents. The links in the graph are the information objects needed by each agent to execute its assigned functions. We also describe a generic sets of control functions and the roles that simulation can play in executing those functions. This means that simulation is now part of an integrated process and can no longer be thought of as a stand-alone technology. In this paper, we discuss issues and propose approaches to addressing that integration.

An increasing number of approaches for specifying and executing simulation experiments emphasizes the desire to make this part of modeling and simulation studies explicit, and thus, also easier to replicate. We take this one step further by automatically generating simulation experiment specifications from documentations. Based on a template-based approach and documentations, we show how simulation experiment specifications can be generated and executed for experiments, such as statistical model checking and sensitivity analysis, and we identify crucial challenges.

Provenance information about a simulation model is information about people, artifacts, and processes that have contributed to its generation. It increases trust into the quality and validity of simulation models. Model documentation standards such as the ODD protocol have a similar goal, but are mostly concerned with “what has been generated”, and less with “how it has been generated”. Complementing ODD with provenance information offers a more structured approach to the “what” and fills the gap regarding the “how”. Thereby, simulation experiments play a crucial role, and are treated as first class artifacts, as are simulation models, data sources, and theories. The compliance to the Open Provenance Model allows using established tools for inferring the model’s origin. The approach is of particular value for models that are based on various data sources, theories, and earlier models, as we will show based on a model about migration from Senegal to Europe.

It is arguably difficult to reproduce the results of published work in Modeling & Simulation (M&S). Authors have certainly raised concerns about this issue and attempts by journals and conferences are being made to improve the situation. As part of a movement to tackle reproducibility in M&S, the Strengthening The Reporting of Empirical Simulation Studies (STRESS) reporting checklists were introduced in 2018. The STRESS guidelines aimed to improve knowledge management in industry and to maximize the chance that all important M&S details are included when writing up simulation research for publication. We extend this work by providing an applied example of using the STRESS-ABS checklist for documenting an Agent Based Simulation model. It is hoped that an applied example will both encourage and guide authors and practitioners to improve their reporting.

How important it is to our discipline that we can reproduce the results of Modeling & Simulation (M&S) research? How important is it to be able to (re)use the models, data, and methods described in simulation publications to reproduce published results? Is it really that important or are the lessons and experiences described in a paper enough for us to build on the work of others? At the 2016 Winter Simulation Conference, a panel considered opinions on reproducibility in discrete-event simulation. This article builds on these and asks if there really is a reproducibility crisis in M&S? A diverse range of views on the subject are presented including reflections on the reproducibility in terms of the art and science of simulation, the frustrations of poor reproducibility, perspectives from the industrial production & logistics community, the wider context of open science and artefact sharing, and the role of provenance beyond reproducibility.

Researches of AI planning in Real-Time Strategy (RTS) games have been widely applied to human behavior modeling and war simulation. Due to the fog-of-war, planning in RTS games need to be implemented under partially observable environment, which poses a big challenge for researchers. This paper focuses on extending Hierarchical Task Network (HTN) Planning in partially observable environment, and proposes a partially observable adversarial hierarchical task network planning with repairing algorithm named PO-AHTNR. By adding sensing action into HTN domain knowledge, a reconnaissance strategy and a history-based single belief state generation method are presented to obtain the best action. In order to verify the proposed algorithm, an empirical study based on µRTS game is carried out, and the performance of modified algorithm is compared to that of AHTNR and other state-of-the-art search algorithms developed for RTS games.

There has been significant growth in online social science experiments in order to understand behavior at-scale, with finer-grained data collection. Considerable work is required to perform data analytics for custom experiments. We also seek to perform repeated networked experiments and modeling in an iterative loop. In this work, we design and build four composable and extensible automated software pipelines for (1) data analytics; (2) model property inference; (3) model/simulation; and (4) results analysis and comparisons between experimental data and model predictions. To reason about experiments and models, we design a formal data model. Our data model is for scenarios where subjects can repeat actions (from a set) any number of times over the game duration. Because the types of interactions and action sets are flexible, this class of experiments is large. Two case studies, on collective identity and complex contagion, illustrate use of the system.

We design an experiment to evaluate behavior in a dynamic spatial game representing the incentives faced by drivers on a ridesharing platform while waiting to be matched with a rider. The design is unique in that it allows us to observe not only participants' choices, but also the considerations that went into those choices. The results of the experiment show that a large majority of player choices are consistent with myopic best responding - a myopic best response maximizes a player's flow payoff at the time of the decision but ignores strategic considerations regarding the future choices of opponents. Given this finding, we develop agent-based models of spatial competition built upon myopic agents. Myopic behavior in our model results in quite efficient outcomes, suggesting that ridesharing platforms may benefit from sharing with drivers the locations of other nearby drivers to allow them to compete spatially.

Smart Homes are currently one of the hottest topics in the area of Internet of Things or Augmented Living. In order to provide high-level intelligent solutions, algorithms for identifying which activities the inhabitants intend to perform are necessary. Sensor data plays here an essential role, for testing, for learning underlying rules, for classifying and connecting sensor patterns and to inhabitant activities, etc. However, only few and limited data sets are currently available. We present concepts and solutions for generating high-quality data using a flexible agent-based simulation tool. The basic idea is to integrate the simulation of a sensorized apartment with human behavior modelling based on constraint-based planning that produces a sequence of daily activities. The overall set-up is shown to generate data that exhibits the same relevant properties as data from a comparable real-world apartment.

Cities are complex, dynamic, evolving adaptive systems comprised of people as well as interconnected physical infrastructure. Simulation modeling can help us understand and shape the evolution of our cities. In this paper, we describe an agent-based simulation modeling framework applied to Chicago, called chiSIM (for the Chicago Social Interaction Model). Each person residing in Chicago is represented as an agent in chiSIM; all places where people can be located in Chicago also are represented. The model simulates the movements of people between locations on an hourly basis during the course of a typical day. Co-located agents engage in various kinds of social interactions, such as exchanging information, engaging in business transactions, or simply sharing physical proximity. We discuss technical approaches to largescale urban modeling including development of synthetic populations, efficiency gains through distributed processing, logging and analysis of simulation results, and visualization.

Fortifying facilities within a supply chain network can mitigate facility failures caused by disruptions. In this study we build an agent-based simulation model to study the r-interdiction median problem with fortification (RIMF), considering two types of facility disruptions: naturally-caused and human-caused disruptions. The objective of this study is to develop a simulation model that analyzes facility disruption and fortification as a repeated Stackelberg competition, where fortification decisions are made anticipating disruptions. The most important facilities - those with the largest demand coverage - are fortified, while the next most important facilities - those not fortified due to fortification resource limitations - are successfully disrupted. In addition, the study provides event-by-event updates of the fortifying, disrupting, and recovering processes, as well as how these events affect the total network cost over the course of the simulation; thus paves the way for a future study on how to make optimal fortification decisions.

The calibration of agent-based pedestrian simulation models requires empirical data. To avoid cost-intensive real-world experiments, human-in-the-loop simulations can be applied in which simulated pedestrians interact with human-controlled agents. However, the experiment results may be unrealistic if the human participants are presented with agents acting according to an uncalibrated model. We propose an incremental calibration approach that aims to address the circular dependency between the behaviour of human and simulated pedestrians. By incrementally adapting the parameters of the simulated agents to match the behaviour of the human participants, we aim to gradually approach a realistic interaction. We evaluate our approach using the simulation of the boarding procedure of a public transport vehicle in 2D and virtual reality experiments. The calibration results are compared with those gathered from a traditional non-incremental calibration. Our results indicate the feasibility of our approach and highlight the necessity for future research on efficient simulation model calibration.

Autonomous ridesharing systems (ARS) promise many societal and environmental benefits, including decreased accident rates, reduced energy consumption and pollutant emissions, and diminished land use for parking. To unleash ARS' potential, stakeholders must understand how the degree of passenger participation influences the ridesharing systems' efficiency. To date, however, a careful study that quantifies the impact of user participation on ARS' performance is missing. Here, we present the first simulation analysis to investigate how and to what extent user participation affects the efficiency of ARS. We demonstrate how specific configurations (e.g., fleet size, vehicle capacity, and the maximum waiting time) of a system can be identified to counter the performance loss due to users' uncoordinated behavior on ridesharing participation. Our results indicate that stakeholders of ARS should base decisions regarding system configurations on insights from data-driven simulations and make tradeoffs between system efficiency and price of anarchy for desired outcomes.

This study assesses feasibility aspects of using a real-time data-driven transportation simulation model to evaluate and visualize network performance indices to provide dynamic operational feedback in a real world environment, in a big data context. A hybrid traffic simulation model, consisting of a mix of preset and real-time data-driven intersections, is developed. The hybrid model represents a traffic corridor partially equipped with smart devices generating high velocity, high volume datasets with limited shelf-life. The model used in this study emulates seventeen consecutive intersections on a corridor. Signal controls and vehicle volumes at two of the intersections are driven by real-time data while the remaining intersections are driven by preset data. An optimized architecture is developed to enable control of the signals and the vehicle volumes using real-time data from in-field detectors, and real-time processing of the vehicle trajectories from the simulation output to generate travel-time, energy, and emissions performance indices.

This paper presents a novel agent-based simheuristic (ABSH) approach that combines simheuristic and multi-agent system to efficiently solve stochastic combinatorial optimization problems. In an ABSH approach, multiple agents cooperate in searching a near-optimal solution to a stochastic combinatorial optimization problem inside a vast space of feasible solutions. Each of these agents is a simheuristic algorithm integrating simulation within a metaheuristic optimization framework. Each agent follows a different pattern while exploring the solution space. However, all simheuristic agents cooperate in the search of a near-optimal solution by sharing critical information among them. The distributed nature of the multi-agent system makes it easy for ABSH to make use of parallel and distributed computing technology. This paper discusses the potential of this novel simulation-optimization approach and illustrates, with a computational experiment, the advantages that ABSH approaches offer over traditional simheuristic ones.

From the viewpoint of strong spatial cognition in graph problems, the shortest path search can be solved in one physical action using strings and pins that respectively represent graph edges and vertices. By pulling a start and an end pin, we can identify a series of stretched strings as the shortest path. We use agent-based models (ABMs) to translate this action into computational representations. Assuming that a set of strings and pins are hung on a wall with a start pin, agents are disseminated downward to a destination as gravity forces. We implemented three models: a discrete-event, an asynchronous, and an aggregated agent dissemination on top of the MASS (multi-agent spatial simulation) library. To address large-scale network environments, we blended HDFS into MASS so that a graph data set is read over a cluster system in parallel. This paper presents these ABM implementations and performance measurements over a cluster system.

Object Event Modeling and Simulation (OEM&S) is a general Discrete Event Simulation paradigm combining object-oriented modeling with the event scheduling paradigm. We show how to extend OEM&S by adding concepts of agent-based modeling and simulation, resulting in a framework that we call Agent/Object Event Modeling and Simulation (A/OEM&S). The main point for such an extension is to define agents as special objects, which are subject to general (physical) laws of causality captured in the form of event rules, and which have their own behavior allowing them to interact with their inanimate environment and with each other. Because agent behavior is decoupled from physical causality, an A/OE simulator consists of an environment simulator, which simulates the physical world (the objective states of material objects), and agent simulators, which simulate the internal (subjective) states of agents and their behaviors.

Disasters, whether natural or man-made, hinder the security and stability of societies worldwide. People in disaster prone areas live in constant fear, not knowing when and where disasters will strike. Therefore, it is critical to design emergency management systems and contingency plans to prepare for the worst. While a great number of plans have been proposed around the globe, the geographical properties and cultural differences between disaster sites significantly affect their efficiencies. This paper surveys existing literature on emergency management in China and summarizes different evacuation models, in particular, agent-based models and other models. We identify four major issues in the emergency management system in China and provide recommendations and guidelines for addressing these issues. To exemplify the implementation of these recommendations, we provide an example based on a statistical model and an agent-based simulation model to substantiate the development of the emergency management system in China.

Large-scale events are always vulnerable to natural disasters and man-made chaos which poses great threat to crowd safety. Such events need an appropriate evacuation plan to alleviate the risk of causalities. We propose a modeling framework for large-scale evacuation of pedestrians during emergency situation. Proposed framework presents optimal and safest path evacuation for a hypothetical large-scale crowd scenario. The main aim is to provide the safest and nearest evacuation path because during disastrous situations there is possibility of exit gate blockade and directions of evacuees may have to be changed at run time. For this purpose run time diversions are given to evacuees to ensure their quick and safest exit. In this work, different evacuation algorithms are implemented and compared to determine the optimal solution in terms of evacuation time and crowd safety. The recommended framework incorporates Anylogic simulation environment to design complex spatial environment for large-scale pedestrians as agents.

A means of egress from buildings is a critical aspect of building design and an important part of building fire regulations. However, elderly evacuees are often overlooked, being regarded as part of the average population, thereby ignoring the limitations elderly people may have. The computational egress model is a useful tool to evaluate postulated “what-if” scenarios, aiming to predict building egress performance under these designated scenarios. This paper first applies Multi-agent Based Simulation (MABS) supported by NetLogo to simulate the evacuation scenarios where evacuees are all elderly people, then statistical analysis is utilized to interpret results, and comparative analysis is conducted to offer some suggestions for egress design and crowd management.

Consider an agent-based social influence (belief adoption) network where agents share beliefs with neighbors using a linear regression model. One relevant question is: can aggregated, system-level belief also be fit by a linear regression model? Earlier work demonstrated several scenarios where system-level linearity of belief holds. This paper extends that research, varying model and simulation factors through experimental design. When linearity does not hold, we isolate the responsible factors. Finally, we investigate whether system-level linearity is as an absorbing state, that is, when system-level linearity is present at some time t, it continues to hold for all later times.

Social contagion is a process in which the behavior of a social group spreads across a population. We model and analyze this phenomenon with respect to fertility in an agent-based social simulation study. Our model integrates personal motivations for having children and social influences to allow for quantifying their respective impact on fertile behavior. We use a real-world dataset for model calibration and compare three hypothetical scenarios with the reference setting. Our results show that each of the social influences has a statistically significant impact on birth rates in the agent population. Moreover, we identify differences in the effects of social pressure and social support depending on a person’s age. This shows that our simulation is not only capable of realistically reflecting family development, but can also provide further insights into the functioning of social mechanisms in a controlled experiment setting.

Research works in collective problem-solving usually assume fixed communication structures and explore effects thereof. In contrast, in real settings, individuals may modify their set of connections in the search of information and feasible solutions. This paper illustrates how groups collectively search for solutions in a space under the presence of dynamic structures and individual-level learning. For that, we built an agent-based computational model. In our model, individuals (i) simultaneously conduct search of solutions over a complex space (i.e. a NK landscape), (ii) are initially connected to each other according to a given network configuration, (iii) are endowed with learning capabilities (through a reinforcement learning algorithm), and (iv) update (i.e. create or severe) their links to other agents according to such learning features. Results reveal conditions under which performance differences are obtained, considering variations in the number of agents, space complexity, agents' screening capabilities and reinforcement learning.

Working hours that are based on operating requirements may lead to conflicts between working and private times of the employees. In an attempt to reduce or even to avoid conflicts between employees’ and operating needs, a hybrid approach is presented here. This approach reduces, or even avoids, conflicts between employees’ needs and operational requirements. Starting from a preliminary shift schedule, an improved solution is developed through agent-based time swaps between the employees. The resulting solution is then dynamically evaluated using simulation. The simulation results show the advantages of the improved work schedules by means of personnel-oriented and logistics criteria. With the help of lexicographical preference ordering, an overall solution can be developed that improves the work-life balance of the employees and fulfills operating requirements. Finally, this hybrid approach is demonstrated using the intra-plant logistics department of a process automation devices manufacturer. Furthermore, the effects of diverse swapping heuristics are shown.

This paper focuses on constructing an abstract baseline model of agents, simulating buyers and sellers of health insurance, in the individual health insurance market. Each buyer assesses their expected medical expenses, surveys the insurance contracts available on the market, and decides each year whether to buy insurance, and which plan they will choose. Each seller, upon seeing the costs to provide medical care to its subscriber base, will update the premium price it offers to the market to remain profitable. The aim of this paper is two-fold – to produce a self-sustaining representation of the insurance market that qualitatively represents the health insurance market without any external policy implementations and to quantitatively replicate the large-scale metrics seen in the insurance market, most notably the price elasticity of demand for insurance.

We describe an agent-based conceptual model of a Blockchain system. Blockchain technology enables distributed, encrypted and secure logging of digital transactions in an add-only ledger record. We model and simulate an expanding Blockchain network, describing the participating agents, the transactions, and the verified add-only public ledger record achieving decentralized consensus. All of the essential details of the functioning of the Blockchain including the behaviors and decisions made by agents deciding to join as well as participate in the market are detailed in the prototype model. The aim of this paper is to illustrate the essential elements and functioning of a Blockchain system, implement a generalized simulation and a measure of Blockchain efficiency from an agent choice and energy cost perspective. Our preliminary results indicate that mining choice (transaction block to verify) coupled with proof of work incentives are critical for energy efficiency.

Zambia has one of the highest rates of cervical cancer in the world. To help policy makers make future intervention decisions regarding cervical cancer, we created a micro-simulation model to simulate cervical cancer and HIV in Zambia. Model calibration faced two major challenges: (1) Much of the available input data was on women in the United States, which do not allow calibration to align to the age-specific targets from Zambia; (2) Significant computational resources were not always available. We addressed issue one by creating age-specific calibration parameters to help better match specific targets. Issue two was addressed by using predictive models before conducting calibration simulations to discard parameter sets that are likely to produce poor results. This paper will demonstrate these two modeling strategies and show the dramatic effect they had on our ability to accurately calibrate to model targets.

Recent statistics predict an increasing demand for care services in elderly care as a result of demographic change. To forecast the amount, location, and type of emerging care demand, existing approaches like Microsimulation must be revised. These approaches lack consideration of the fact that every decision for a specific type of care is dependent on the individual care recipient’s situation and personal preferences. To close this gap, it seems reasonable to extend existing Microsimulation approaches by means of Agent-based Social Simulation. The goal of this work is the development of an agent-based model, that enables the simulation of individual decision-making processes. The presented model is based on sociodemographic data to take systemic properties and individual situations into account. Additionally, sociological actor types are used to implement individual preferences and the characteristics of care recipients. To evaluate the proposed approach, a case study using German census data is presented.

In this study, we propose a model that helps to analyze the behavior of chronic disease patients with a focus on heart failure patients based on their lifestyle. We consider how living conditions affect signs and symptoms of chronic disease and, accordingly, how these signs and symptoms affect chronic disease stability. We use an agent-based model, a state machine, and a fuzzy logic system to develop the model. Specifically, we model the required “living condition” parameters that can influence required medical variables. These variables determine the stability class of chronic disease. The model allows creation of virtual patients with a given pathology and analysis of their impact on quality of the Emergency Department. Analysis of the Emergency Department’s quality based on modeling and simulation enables us to develop new planning strategies for efficient care systems.

Designing complex systems with a specific emergent property is challenging even in perfect environmental conditions, and several additional challenges are introduced in contested, dynamic environments such as ongoing battles or military exercises. An example of a desired emergent property is disseminating information in a timely manner without causing significant network disruptions. While military communication strategies exist, their evaluation is performed in costly real-life deployments with limited exploratory capability. In this paper, we propose to achieve timely information dissemination through a rule-based expert system that adapts to network and operational context changes by change the priority of messages sent to the network. To analyse this emergent behavior, we present an evaluation simulation testbed that explores various operational conditions for the evaluation of communication strategies in a land-exercise focused scenario.

Modern service and humanoid robots are comprised of multiple computers distributed among the robots' hardware. During task execution, several software components are executed in parallel on the connected machines. Due to the complex control loops and communication requirements of robot tasks, a suitable assignment of software components to the available hardware units is necessary to achieve low reaction times. Currently, there is a lack of works on approaches to evaluate intra-robot communication. We propose a coupling between the robotics framework ArmarX and the network simulator OMNeT++ to support the evaluation and optimization of robot architectures. Our approach allows unmodified robot components to communicate across simulated network interconnects. In a case study, we examine the influence of different hardware assignments of software components on task execution times. We show that the timing information present in the simulation-based evaluations enables more efficient hardware assignments when compared to static graph partitioning.

The use of autonomous vehicles is a growing trend in the material handling and warehousing. Some challenges that face material handling include the navigation within a warehouse, precision localization and movement, and task selection decisions. In this paper, we address the issue of task selection. In particular, we develop a deep reinforcement learning methodology to enable a vehicle to select from among multiple tasks and move to the closest task in the context of material handling in a warehouse. To evaluate the deep reinforcement learning methodology, we conduct a simulation-based experiment to generate scenarios to first train and then test the capabilities of the method. The results of the experiment show that the method performs well under the given conditions.

Manufacturers need to quickly estimate cycle times for incoming orders for promising delivery dates. This can be achieved by using data analytics (DA) / machine learning (ML) approaches. Selecting the right DA/ML approach for an application is rather complex. Obtaining sufficient and right type of data for evaluating these approaches is a challenge. Simulation models can support this process by generating synthetic data. Simulation models can also be used to validate DA models by generating new data under varying conditions. This can help in the evaluation of alternative DA approaches across expected range of operational scenarios. This paper reports on use of simulation to select an approach to support the order promising function in manufacturing. Two DA approaches, Neural Networks and Gaussian Process Regression, are evaluated using data generated by a manufacturing simulation model. The applicability of the two approaches is discussed in the context of the selected application.

The complex interconnections between various critical infrastructure sectors make the system of systems (SoS) vulnerable to failures and highlight the importance of robustness and resilience. To this end, we first establish holistic probabilistic networks to model the interdependencies between infrastructure components. To capture the underlying failure and recovery dynamics, we further propose a Markov decision processes (MDP) model in which the response policy determines a long-term performance. To address the challenge of a large dimensionality, we exploit the sparsity of the network interconnections and solve an approximate linear program by the variable elimination, which leads to a distributed control policy under mild assumptions. Finally, we use a case study of the interdependent power and subway systems to corroborate the results and show that the optimal resilience resource planning and allocation can reduce the failure probability and mitigate the impact of failures caused by natural or artificial disasters.

Availability of large data sets and increased computing performance have contributed to many improvements in productivity and decision making. Simulation can exploit these by incorporating data mining capabilities, such as machine learning, in the modeling and analysis process. This paper demonstrates the integration of discrete event simulation with a deep learning resource, known as TensorFlow, to enable intelligent decision making in the form of smart processes. A bank credit approval process is modeled using these smart processes to evaluate customer credit worthiness based on 20 reported features. Comparison of three models is made where credit worthiness is (1) known, (2) randomly assigned, or (3) evaluated based on customer features. Additionally, the experiment compares results under conditions where the process is perturbed by an unexpected surge in customer arrivals. The presented models and results demonstrate the feasibility of enabling smart processes in discrete event simulation software and the improved decision-making fidelity.

Agent-based modeling and simulation (ABMS) has been applied to various domain problems. ABM consists of multiple agents and environments and focuses on their interactions. While such characteristics lead to its current popularity, they partially makes hard for the model calibration. Model calibration is generally performed by tuning model parameters, but in the ABM case, its interaction structure should be considered as well due to its huge influences to the results. To resolve this problem, this paper suggests a self-evolving ABMS framework. During the self-evolving process, ABM structure as well as parameters are explored and exploited for the model calibration. In particular, we adopt reconfigurable modeling in the proposed simulation framework, which is derived from Discrete Event System Specification (DEVS) formalism. This paper introduces a housing ABM, and the case study using this model shows that the accuracy of the model prediction was significantly increased through the self-evolving process.

When large-scale incidents occur on freeways, en-route traffic diversion is one effective strategy to reduce the impact of incident-induced congestion. In corridors with substantial commercial traffic, e.g., trucks, route diversion is complicated compared with non-commercial vehicular traffic, due to tough vehicle maneuvers along alternate routes, and higher safety risk. To address the issue of commercial vehicle diversions to alternate routes in response to large-scale incidents, this paper establishes a microscopic simulation methodology to analyze the impacts of various technologies on en-route diversion under large-scale traffic incidents in real-life corridors for single-unit and multi-unit trucks and passenger vehicles. Results show, in addition to incident duration and lane blockage, important factors as CAV, incident information availability, number of intersections, AADT, impact en-route truck diversions and hence the resulting delays. In future traffic operations practice, customizing incident information to truck and passenger vehicles separately is recommended.

Segmentation plays an important role within digitalized supply chain management in the pursuit of competitive advantages, especially in industries where complexity and volatility are present to high degrees. While research in this area is mainly focused on the development of segmentation concepts and the identification of influencing parameters, this paper focuses on the segmentation process and applies a hysteresis to provide stability and autonomous adaptability. Several novel concepts relating to the merger of hysteresis and supply chain segmentation are created, along with a discussion on the process and its requirements. In the end, the theoretical concept is validated by an industrial use case. The results substantiate the concept measured by bound capital and service level.

In this work, we tackle power consumption reduction of battery-dependent devices in a smart campus (including hospital) application. These devices are connected by networked systems which may be subject to fluctuation of the message delays that control essential equipment. We show through five case studies using discrete event simulation, that power consumption may be reduced using proper prioritization and balancing of the network emergency traffic. A predictor algorithm and a fuzzy logic controller was used to indicate the level upon which the system must switch off the load in order to reduce power consumption. The analysis of a case study shows that a considerable reduction in power consumption was achieved through the reduction of message delays and also due to the fuzzy control of AC and lighting equipment.

Cybersecurity assessment in automation requires a modeling and simulation framework to study the complex relationships between the cyber-based control mechanisms and the power systems. A real power system is not ideal for such an assessment due to potential disruption in operation. In this paper, we propose a simulation model of a virtual operational technology (OT) network in power system automation for cybersecurity assessment. The proposed simulation model is modularized with key components including power system and process, communication, automation, and enterprise network. We discuss Virtual OT Network (VOTNet), a specific instance of the model, in terms of simulated and emulated systems such as programmable logic controller, computing systems, control centers, and software. We evaluate VOTNet under two use cases: different attack scenarios and scalability vs. attack surface to study the impacts of cyber attacks. Particularly, unauthorized access, data manipulation in PLC, and denial of service in SCADA communication are demonstrated.

With the integration of renewable energy and advanced communication technologies, smart power grids can enhance the cost-efficiency and reliability of energy generation, transmission, and distribution. The communication network connecting numerous remotely distributed generators, devices, and controllers plays a vital role in the control of power grids. However, it is vulnerable to cyber attacks. In particular, because of its frequency and lasting impact, the Distributed Denial of Service (DDoS) attack poses an important threat to smart power grids. This paper presents a simulation-based stochastic programming approach to guide the unit commitment and economic dispatch decisions, which accounts for the prediction uncertainty of wind power and the impact from DDoS attacks. The case study demonstrates that the proposed approach can lead to a more cost-efficient and reliable operational decision guidance for smart power grids.

Booms are extensively used for confining oil spilled at sea and for carrying it away from sensitive areas or toward a gathering point. The success of booming operations depends on effective boom deployment, which, in turn, is conditioned by the state of the sea – wind, tides, currents, and waves – and its effects on the spill spreading shape and speed. Planning the boom deployment beforehand could help overcome problems derived from sea condition. Besides, the employment of Unmanned Surface Vehicles (USVs) for towing spill recovery booms represents an interesting alternative to the use of ordinary vessels, and it could lead to a fully automatic oil spill recovery system. A basic tool has been developed in Python to simulate the deployment of booms towed by USVs. The aim is to explore simple trial scenarios, working out information that helps to prototype real USV-boom systems and to plan automatic deployments.

In this paper, a novel approach is proposed to design a bank of unknown input observers(UIOs) for aero-engine distribute control system(DCS) with multi-sensor fault. First, considering the system with multi-sensor fault which doesn’t satisfy the observer matching condition, a UIO is proposed based on the disturbance decoupling principle of unknown input observer(UIO) and control theory. The parameter matrixes of UIO are solved in terms of linear matrix inequality(LMI) toolbox. Second, the fault detection and isolation scheme is proposed for DCS with multi-sensor fault and the UIOs are designed for a bank of fault detection and isolation sets which are obtained by reorganizing the control system. The residual signal generated by each observer is robust to the sensor fault in the matching set, and sensitive to the sensor fault out of the matching set. Finally, the residual signals were simulated and the results show the effectiveness of the proposed method.

Electric taxi fleet simulation calls for a lot of uneasily accessible data. We learned from historical data in order to predict electric vehicle consumption and charging time. Although simulation was the main goal, the developed models will be used by the dispatching team of the company. Moreover, the data analysis process allowed us to discover flaws in the infrastructure (e.g., defective stations and defective batteries).

This article proposes a discrete event simulation model of an organization that maintains computer hosts and incurs several millions of dollars in maintenance and incident response costs. The common maintenance policy is referred to as “out-of-sight is out-of-mind” (OSOM) because the majority of hosts are absent from scans and ignored. Hosts are “dark” (absent) because they are not accessible (turned off or with restricted permissions). The proposed model is used to compare OSOM with alternatives including improved analytics that make dark host vulnerabilities visible. Findings clarify the apparent benefits of OSOM unless indirect costs for intrusions or improved policies are applied. Also, benefits from using Windows operating systems and improved policies are clarified including millions in expected savings (vs. Linux).

A key challenge for manufacturers today is efficiently producing and delivering products on time. Issues include demand for customized products, changes in orders, and equipment status change, complicating the decision-making process. A real-time digital representation of the manufacturing operation would help address these challenges. Recent technology advancements of smart sensors, IoT, and cloud computing make it possible to realize a “digital twin” of a manufacturing system or process. Digital twins or surrogates are data-driven virtual representations that replicate, connect, and synchronize the operation of a manufacturing system or process. They utilize dynamically collected data to track system behaviors, analyze performance, and help make decisions without interrupting production. In this paper, we define digital surrogate, explore their relationships to simulation, digital thread, artificial intelligence, and IoT. We identify the technology and standard requirements and challenges for implementing digital surrogates. A production planning case is used to exemplify the digital surrogate concept.

The representation of complex Cyber-Physical Systems usually requires the ability to combine models described in different modeling formalisms and involving several numerical methods. While ordinary differential equations (ODEs) are a common formalism for representing continuous systems, their solution cannot be efficiently and accurately accomplished by a single numerical integrator. A plethora of methods have been developed to tackle several issues including efficiency, accuracy, and stiffness. Given the variety of methods, a common representation to enable their interoperability is still a major research challenge. In this paper, we introduce the Hybrid Flow System Specification formalism (HyFlow) as a general framework for describing numerical integrators in a modular form. Each numerical method is represented as a particular model in the formalism. Co-simulation is then enabled, since all HyFlow models can be composed by design. In this paper, we show examples demonstrating the co-simulation of geometric, pulse, and exponential numerical integrators.

Designing hybrid controllers for cyber-physical systems raises the need to interact with embedded platforms, robotic applications being a paradigmatic example. This can become a difficult, time consuming and error-prone task for non-specialists as it demands for background on low-level software/hardware interfaces often falling beyond the scope of control designers. We propose a simulation-driven methodology and tool for designing hybrid controllers based on a model continuity approach. The simulation model of a controller should evolve transparently from a desktop-based mocking up environment until its final embedded target without the need of intermediate adaptations. DEVS-over-ROS relies on the DEVS framework for robust modeling and real-time simulation of hybrid controllers, and on the ROS middleware for flexible abstraction of software/hardware interfaces for sensors and actuators. We successfully tested DoveR in a case study where a custom-made crafted robotic system is built concurrently with the design of its controller.

This paper describes the design, implementation, and early experiences with a novel agent-based simulator of online media streams, developed under DARPA's SocialSim Program to extract and predict trends in information dissemination on online media. A hallmark of the simulator is its self-configuring property. Instead of requiring initial set-up, the input to the simulator constitutes data traces collected from the medium to be simulated. The simulator automatically learns from the data such elements as the number of agents involved, the number of objects involved, and the rate of introduction of new agents and objects. It also develops behavior models of simulated agents and objects, and their dependencies. These models are then used to run simulations allowing for future extrapolations and "what if" analyses. Results are presented on using this system to simulate GitHub transactions. They show good performance in terms of both simulation accuracy and overhead.

Timeliness is a fundamental property of Cyber-Physical Systems that has been intensively investigated within the scope of real-time and critical systems. The advent of the Internet of Things, however, brings an intense communication flow between devices and the Internet. In this scenario, modeling time requirements in terms of data, rather than the tasks that manipulate them, may be advantageous as a data-centric design can promptly encompass other first-order requirements, such as geolocation, security, and trustworthiness. In this paper, we propose a strategy to design complex CPSs by modeling their data using the SmartData construct, which, besides encompassing means to handle the aforementioned requirements, also defines the concept of data expiry to guide scheduling decisions. With SmartData, local tasks are scheduled to produce the freshest data and to manipulate them before expiration. Likewise, network packets are scheduled prioritizing data whose expiry is close. We validate the strategy through simulations using Castalia.

In this paper we propose a different approach for interacting and analyzing agent-based models. The approach relies on creating empathy and understanding between physical agents in the physical world (people) and artificial agents in the simulated world (simulated agents). We propose a simulated empathy framework (SEF) in which artificial agents communicate directly with physical agents through verbal channels and social media. We argue that artificial agents should focus on the communication aspects between these two worlds, the ability to tell their story in a compelling way, and to read between the lines of physical agents speech. We present an implementation of the SEF and discuss challenges associated with implementing the framework in an artificial society.

Visitor expectations continually evolve as new forms of technology mediate ever more personalized interactions, not only within the museum, but also virtually around the physical environment. Designing visitor journey experiences that support visitor heterogeneity are complex. Consequently, it is difficult for museum management and collection managers to respond effectively in their design of heritage experiences. Understanding human behavior at scale is challenging, often explored in other disciplines by simulating generic process models and scenarios. Creating experiences for cultural heritage persona remains a challenge because no clear methodology currently exists. This paper proposes a methodological framework supporting museum and collection managers in their design and simulation of heritage experiences. Clear classification of visitors and journeys is undertaken at the outset. Journey mapping is used as a modeling primitive for both agent and system dynamic modeling.

Co-simulation requires the ability to represent systems in a modular form, while guaranteeing that simulation algorithms rely exclusively on model external interface. These conditions enable models to be composed without exposing their internal state, a requisite needed for the co-simulation of cyber-physical systems. In this paper we provide a modular representation of geometric integrators, a type of integrator essential to the simulation of 2nd-order energy preserving systems. These integrators offer an alternative to the conventional decomposition of systems into 1st-order Ordinary Differential Equations (ODEs). This latter approach, although commonly used in nowadays M&S software, is not acceptable when long simulation runs are needed. Geometric integrators are represented in the Hybrid Systems Specification (HYFLOW), a modeling formalism to represent hybrid modular dynamic topology systems. We show that HYFLOW enables the composition of geometrical solvers, allowing the co-simulation of complex 2nd-order energy preserving systems.

Particle tracking in physical systems is a well known simulation challenge in many domains. In particular, High Energy Physics (HEP) demand efficient simulations of charged particles moving throughout complex detector geometries in a magnetic field. Quantized State Systems (QSS) is a modern family of hybrid numerical methods that provides attractive performance features for these problems. Its state-of-the-art implementation is the general-purpose QSS Solver toolkit. Meanwhile, Geant4 is the most widely used platform for computational particle physics, embedding vast amounts of physics domain knowledge. Yet, Geant4 relies rigidly on classic discrete time numerical methods. In this work we present a robust co-simulation technique to apply QSS in the simulation of HEP experiments, thus leveraging the best of both toolkits. We obtained speedups of up to three times in synthetic, yet representative scenarios, and a competitive performance in a difficult benchmark modeled after the Compact Muon Solenoid (CMS) particle detector at CERN.

Acquisition planning involves decisions to be made regarding the number of assets to be acquired initially and the type and timing of replacement and upgrade actions to maintain performance measures efficiently. Acquisition planning is challenging for high-valued assets because of considerable uncertainties in their long-term life cycle. This article proposes an approach to determine which acquisition strategy—i.e. what initial number of assets, what number of new acquisitions, and in what time throughout a long-term planning period—can robustly fulfil multiple performance objectives in the face of plausible future scenarios. The article incorporates robust optimization for the treatment of uncertainty inside the simulation multi-objective optimization process where the robustness of different acquisition strategies in future scenarios is analyzed by running many simulations. A fleet management system is used as an illustrative hypothetical example. The results show an adaptation map of robust acquisition strategies over the life cycle of the fleet.

Hybrid simulation, combining exact and approximate algorithms, provides an alternative to a completely stochastic simulation. However, one challenge for the efficient implementation of hybrid simulations is the additional overhead due to frequent switches between the two regimes. The amount of additional overhead considerably increases with the number of discrete events in the stochastic regime. However, reactions that take place rather frequently cannot completely be avoided due to the accuracy requirements. In this paper, we present an improved hybrid simulation method which takes advantage of the Hybrid Rejection-based Stochastic Simulation Algorithm (HRSSA), a variant of the hybrid simulation approach. To reduce the overhead on account of the switches from the stochastic to the deterministic regime, we analyse and record the dependencies of reactions as well as species between the stochastic and deterministic subnetworks. Comparing our technique with existing ones shows a clear improvement in terms of runtime, while preserving accuracy.

Symbiotic simulation is one of Industry 4.0 technologies that enables interaction between a physical system and the simulation model that represents it as its digital twin. Symbiotic simulation is designed to support decision making at the operational levels by making use of real- or near real- time data that is generated by the physical system, which is used as an input to the simulation model. From the modeling perspective, a symbiotic simulation system comprises a hybrid systems model that combines simulation, optimization and machine learning models as well as a data acquisition module and an actuator. The actuator is needed when the symbiotic simulation system is designed to directly control the physical system without human intervention. This paper reviews the components of a symbiotic simulation system from the perspective of hybrid systems modeling and highlights research questions needed to advance symbiotic simulation study.

Recent advances in information technology systems have enabled organizations to store tremendous amounts of business process data. Process mining offers a range of algorithms and methods to analyze and extract metadata for these processes. This paper presents a novel approach to the hybridization of process mining techniques with business process modelling and simulation methods. We present a generic automated end-to-end simulation framework that produces unbiased simulation models using system event logs. A conceptual model and various meta-data are derived from the logs and used to generate the simulation model. We demonstrate the efficacy of our framework using a business process event log, achieving reduction in waiting times using resource reallocation. The intrinsic idea behind our framework is to enable managers to develop simulation models for their business in a simple way using actual business process event logs and to support the investigation of possible scenarios to improve their business performance.

The study of complex systems usually requires hybrid simulations because they have components that are continuous in nature and other that are discrete. It has been proved that DEVS is s a common denominator to combine different Modeling and Simulation methodologies (such as Petri Nets, Cellular Automata, Modelica etc.). We present a cellular model solution to solve PDEs as an extension of classical numerical methods combined with the Cell-DEVS formalism. We explain how to use Cell-DEVS to solve PDEs, focusing on two examples: the PDEs of a Heat Diffusion Process solved with the Method of Lines, and the Shallow Water Equations solved using the Lax-Wendroff method. We will discuss the advantages of solving PDEs with Cell-DEVS, in particular its integration with other hybrid models.

This paper promotes a unique system dynamics-discrete event simulation hybrid modelling framework. The way the hybrid model is developed is intended to simplify the modelling process and make the framework flexible to a variety of situations. In the current study, the framework is used to investigate the success possibility of introducing within-shift job rotation in the plant and its optimal frequency. The intention is to reduce worker exhaustion and by so doing increase productivity and manufacturing throughput. The improvement decisions generated from the simulation results could not have been achieved using any other means. The hybrid modelling framework presents a coherent, straightforward standardized approach that includes the main tasks in SD-DES hybrid modelling. A recommendation for future research is to test the framework's adaptability to a variety of settings.

There is a lack of integration of facility planning techniques in the design and layout of healthcare facilities. Computational models can be used to evaluate minimal distances or cost functions. Discrete event simulations can be used to model the stochastic nature of operations to check the impact on specific performance measures. Visualization can be used to immerse decision makers in the future environment to aid model validity, communication, and understanding. In this paper, we discuss three techniques: mathematical optimization, discrete event simulation, and virtual visualization, and present a framework for using them in an integrated hybrid simulation modeling approach in the healthcare planning process. We present objectives for the use of these techniques throughout the lifecycle in one unit of a healthcare facility. These techniques, while described in a healthcare context, have implications for other domains where uncertain and latent processes are components of the layout decision making process.

Dynamic Data Driven Applications Systems (DDDAS) is a systems design framework that focuses on integrating high-dimensional physical model simulations, run-time measurements, statistical methods, and computation architectures. One of the foremost DDDAS successes was scientific theory assessment of natural disasters such as wild fire monitoring and volcanic plume detection. Monitoring the atmosphere with DDDAS principles has evolved into domain methods for space awareness, unmanned aerial vehicle (UAV) design, and biomedical applications. Recent efforts reflect the digital age of information management architecture design such as multimedia analysis, power grid control, and biohealth concerns. Underlying a majority of DDDAS developments are advances in sensor design, information filtering, and computational systems. The paper provides a motivation, explanation, and literature review of DDDAS.

Hybrid Simulation (HS) is the combined application of simulation approaches like SD, DES and ABS in the model implementation stage of a simulation study. Its objective is to better represent the system under scrutiny. Hybrid Systems Modelling (HSM), on the other hand, is the combined application of simulation with methods and techniques from disciplines such as Applied Computing, Computer Science, Engineering and the wider OR. HSM can be applied to multiple stages of a simulation study. In this paper, we present a classification of HS and extend it to include HSM approaches which use simulation with other OR techniques. The paper contributes to the debate on what constitutes HS and offers a unifying conceptual representation for mixing simulation approaches with HSM methods and techniques.

Discrete event simulation is an established methodology for investigating the dynamic behavior of complex manufacturing and logistics systems. In addition to traditional simulation studies, the concept of data farming and knowledge discovery in simulation data is a current research topic that consist of broad scale experimentation and data mining assisted analysis of massive simulation output data. While most of the current research aims to investigate key drivers of production performance, in this paper we propose a methodology for investigating productive efficiency. We therefore developed a concept of combining our existing approach of data farming and visual analytics with data envelopment analysis (DEA), which is used to investigate efficiency in operations research and economics. With this combination of concepts, we are able to determine key factors and interactions that drive productive efficiency in the modeled manufacturing system, but also to identify the most productive settings.

The development of care pathways is increasingly becoming an instrumental artefact towards improving the quality of care and cutting costs. This paper presents a framework that incorporates Simulation Modeling along with Machine Learning (ML) for the purpose of designing pathways and evaluating the return on investment of implementation. The study goes through a use case in relation to elderly healthcare in Ireland, with a particular focus on the hip-fracture care scheme. Initially, unsupervised ML is utilized to extract knowledge from the Irish Hip Fracture Database. Data clustering is specifically applied to learn potential insights pertaining to patient characteristics, care-related factors, and outcomes. Subsequently, the data-driven knowledge is utilized within the process of simulation model development. Generally, the framework is conceived to provide a systematic approach for developing healthcare policies that help optimize the quality and cost of care.

Hospital planning teams are always concerned with optimizing staffing and scheduling decisions in order to improve hospital performance, patient experience, and staff satisfaction. A multi-method approach including data analytics, modeling and simulation, machine learning, and optimization is proposed to provide a framework for smart and applicable solutions for staffing and shift scheduling. Factors regarding patients, staff, and hospitals are considered in the decision. This framework is piloted using the Emergency Department(ED) of a leading university hospital in Dublin. The optimized base staffing patterns and shift schedules actively contributed to solving ED overcrowding problem and reduced the average waiting time for patients by 43% compared to the current waiting time of discharged patients. The reduction was achieved by optimizing the staffing level and then determining the shift schedule that minimized the understaffing and overstaffing of the personnel need to meet patient demand.

This study examines how changes in the age structure of Polish population inhabiting an administrative region, the Wrocław Region (WR), will affect the demand for healthcare services during the period 2016–2017. We constructed a hybrid simulation model that operates on two submodels. The system dynamics submodel simulates the evolution of population using an aging chain approach to forecast the number of individuals belonging to the respective age-gender cohorts. On the other hand, the discrete event submodel forecasts the number of patients in and number of treatments generated by the respective age-groups. We found that the effect of the WR population growth on the total healthcare demand will be small; however, it will have a larger impact on the number of elderly patients and the volume of more expensive services delivered in WR hospitals.

This paper provides an overview of a hybrid, discrete event simulation (DES) agent based model (ABM), simulation model of the overdue pregnancy outpatient clinic at the Obstetrics department of Akershus University Hospital, Norway. The model is being developed in collaboration with clinic staff. The purpose of the model is to better plan resources (e.g. staffing) to improve patient flow at the outpatient clinic given the uncertainty associated with demand. The uncertainty is due to an increase in the size of the hospital’s catchment area, changes to overdue pregnancy guidelines in Norway and that women can give birth before their appointments. The ABM model component represents the human parts of the system, the women and the clinic staff. The DES component represents the outpatient clinic’s physical location and processes/pathways that operate within it. The technicalities of the model are presented along with some illustrative results.

The last 10 years have witnessed a marked upsurge of attention on Hybrid Simulation (HS). The majority of authors define HS as a joint modelling approach which includes two or more simulation approaches (mainly Discrete Event Simulation, System Dynamics and Agent Based Simulation). Whilst some may argue that HS has been in existence for more than 5 decades, the recent rise tended to be more problem driven rather than technical experimentation. Winter Simulation Conference (WSC) 2015, 2016, 2017 have witnessed 3 panels on the purpose, history and definition of HS, respectively. This paper reports on a comprehensive review conducted by the panelists on HS and its applications. The aim of the paper is to move the debate forward by exploring potential platforms for developing concrete avenues for research on HS in conjunction with the modelling life cycle.

In this paper we present a discrete event simulation model of a cruise terminal for decision support and strategic planning. The main objective is to provide an analysis tool for systems with high capacity constraints, heterogeneous subsystems, time-dependent demand and stochastic passenger transfer and travel times. Cruise terminals are connected by bus to a transport interchange or intermobility center where passengers are transferred to taxis. All terminals are affected by capacity constraints such as the number of bus platforms for passenger pick-up and drop-off and the limited queuing capacity for taxis and passengers. Passenger arrivals are time-dependent and cruise-type-dependent. The purpose of this model is to support the strategic decision-making process through what-if-scenarios. Decisions involve determining the required number of bus platforms, taxi buffer sizes and passenger distribution policies, among other factors. The results identified and proposed the solution of bottlenecks and enabled a sensitivity analysis to be performed.

Car rental companies have to operate the fleet of vehicles considering a cost effective fleet sizing and rebalancing of the vehicles. Additionally, they have to ensure a good vehicle availability at the rental stations. We present a simulation model combined with optimization and queueing network analysis for mobility-on-demand networks with one-way car sharing. The system is modeled as a closed queueing network. On the one hand, this allows an estimation of the vehicle availability; on the other hand, constraints for the optimization model can be derived from the mathematical equations of the queuing network model. To optimize the total revenue, the optimization model considers price incentives for customer trips and cost of empty trips resulting in a mixed-integer linear programming problem. The simulation uses the optimization component for an optimized operation of the mobility-on-demand networks and computes the optimal fleet size, the cost of empty runs, and the vehicle availability.

The car rental business has experienced recent growth under a fiercely competitive market. This paper describes a generic discrete event simulation tool for studying the complex nature of a car rental company’s operations. To construct the tool, we conducted several interviews with experts to first establish the proper requirements, after which we collected relevant data for populating the simulation model, thereby determining the best strategies for car class pricing and upgrade offers that would result in the highest revenue. A realistic case study was conducted to assess the tool and gain a better understanding of the underlying system. Finally, simulation experiments regarding rental prices and car upgrades were designed, and the system was run, achieving a 0.5% error for various performance criteria. For the given case, we found that two upgrade offers and setting prices slightly higher than the average customer budget provides the best strategy.

We study the problem of constructing confidence intervals (CIs) for simulation outputs in the presence of input uncertainty, where the constructed CIs capture both the statistical noises from the simulation replications and the input data. We present a simple technique based on sectioning input data that provides exact asymptotic confidence guarantees. Unlike some existing approaches, our technique bypasses the need to consistently estimate variances that could be computationally demanding. It can be flexibly applied to dependent data and to both parametric and nonparametric input models.

Given a set of point-process event times with constant dispersion ratio, we are interested in estimating the rate function by grouping the event times into count data from equal-width time intervals. We group in order to smooth the resulting piecewise-constant rate function using one of our two existing methods: I-SMOOTH and MNO-PQRS. Using the mean integrated squared error (MISE) for piecewise-constant rate functions, we create two estimators; minimizing the estimated MISE function yields the chosen number of intervals. The MISE function provides insights into the optimal number of intervals as a function of the rate-function shape and expected number of events. Across several examples, our two number-of-intervals estimators perform well and similarly; nevertheless, one dominates in terms of realized MISE value.

Rather than the standard practice of selecting a single "best-fit" distribution from a candidate set, frequentist model averaging (FMA) forms a mixture distribution that is a weighted average of the candidate distributions with the weights tuned by cross-validation. In previous work we showed theoretically and empirically that FMA in the probability space leads to higher fidelity input distributions. In this paper we show that FMA can also be implemented in the quantile space, leading to fits that emphasize tail behavior. We also describe an R package for FMA that is easy to use and available to download.

It has become increasingly important to assimilate “online data” that arrive sequentially in time for real-time decision. Input uncertainty quantification in stochastic simulation has been developed extensively for batch data that are available all at once, but little has been studied for online data. In this paper, we propose a computationally efficient method to incorporate online data in real time for input uncertainty quantification of parametric models. We show finite-sample bounds and asymptotic convergence for the proposed method, and demonstrate its performance on a simple numerical example.

Despite the great advances in modeling and simulation techniques, modelers and researchers acknowledge that models are simplified representations of reality and, hence, are subject to uncertainty and errors. Although models are inevitably uncertain, they can still be a valuable decision-support tool if the users are informed about the uncertainty in the results. The importance of model uncertainty identification and quantification becomes clear in this context, but there are numerous challenges that remain. In this work, an uncertainty analysis framework is proposed for simulation models. This framework comprises of the steps that must be performed to analyze the uncertainty in simulation models. Next, an application of the framework is discussed where entropy is used as a possible measure of input-uncertainty. By using this framework, stakeholders can be better advised regarding the applicability and uncertainty of the simulation model, which will lead to an appropriate adjustment of expectations on the model results.

In stochastic simulation, input uncertainty refers to the output variability arising from the statistical noise in specifying the input models. This uncertainty can be measured by a variance contribution in the output, which is estimated commonly via the bootstrap. However, due to the convolution of the simulation noise and the input noise, the computation effort required in the existing bootstrap schemes are typically substantial. This paper investigates a subsampling framework as a computation saver. We demonstrate the strengths of our subsampled bootstrap in terms of theoretical computation requirements, and substantiate them with numerical illustrations.

The moment-independent sensitivity analysis technique introduced by E. Borgonovo has gained increasing attention to characterize the uncertainty of complex systems and optimize their reliability. The estimation of corresponding indices is a challenging task. This paper aims at presenting a new estimation scheme valid for dependent model inputs. This scheme is build on the copula representation of indices and uses maximum entropy methods to estimate this copula. Accuracy of the proposed method is evaluated through numerical simulations and is compared to two competitive methods, namely an importance sampling based approach and a second one which uses the Nataf transformation.

Metamodel-based bootstrap methods for characterizing input model uncertainty have disadvantages for settings where there are a large number of input distributions, or when using empirical distributions to drive the simulation. Early direct bootstrapping of empirical distributions did not take into account the distinction between intrinsic and extrinsic variation in the resampled quantities. When the intrinsic uncertainty is large, the result is overcoverage of the bootstrap percentile intervals. We explore ways of accounting for both sources in direct bootstrap characterization of input model uncertainty, and study the impact on confidence interval (CI) coverage. Four new bootstrap-based CIs for the expected simulation output under the unknown true distribution are proposed, basic shrinkage CI, percentile shrinkage CI, basic hierarchical bootstrap CI, and percentile hierarchical bootstrap CI, and their empirical performances are demonstrated using an example.

We study sample average approximations under adaptive importance sampling in which the sample densities may depend on previous random samples. Based on a generic uniform law of large numbers, we establish uniform convergence of the sample average approximation to the function being approximated. In the optimization context, we obtain convergence of the optimal value and optimal solutions of the sample average approximation.

We propose an algorithm for simulating bipartite or directed graphs with given degree sequences, motivated by the study of financial networks with partial information. Our algorithm sequentially computes certain “maximum entropy” matrices, and uses the entries of these matrices to assign probabilities to edges between nodes. We prove the correctness of the algorithm, showing that it always returns a valid graph and that it generates all valid graphs with positive probability. We illustrate the algorithm in an example of an inter-bank network.

In a previous article, we studied a then-new class of standardized time series (STS) estimators for the asymptotic variance parameter of a stationary simulation output process. Those estimators invoke the well-known reflection principle of Brownian motion on the suitably standardized original output process to compute several "reflected" realizations of the STS, each of which is based on a single reflection point. We then calculated variance- and mean-squared-error-optimal linear combinations of the estimators formed from the singly reflected realizations. The current paper repeats the exercise except that we now examine the efficacy of employing multiple reflection points on each reflected realization of the STS. This scheme provides additional flexibility that can be exploited to produce estimators that are superior to their single-reflection-point predecessors with respect to mean-squared error. We illustrate the enhanced performance of the multiply reflected estimators via exact calculations and Monte Carlo experiments.

In this paper, European option prices are computed analytically, as well as simulated, for underlying asset price models with stochastic volatility and jump discontinuities. The analytical price is derived using the Fast Fourier Transform method developed in previous literature, which enables prices to be computed quickly. This model is compared with the Black-Scholes model, and the results suggest that this model addresses a known issue with the Black-Scholes model, the under and over valuations of short maturity options. The analytical solution is also used to investigate effective control variates in Monte Carlo simulations. Simulation experiments indicate that the random motion of the asset price serves as an effective control variate.

We study the behavior of a splitting method for sampling from a given distribution conditional on the occurrence of a rare event. The method returns a random-sized sample of points such that unconditionally on the sample size, each point is distributed exactly according to the original distribution conditional on the rare event. For a cost function which is nonzero only when the rare event occurs, the method provides an unbiased estimator of the expected cost, but if we select at random one of the returned points, its distribution differs in general from the exact conditional distribution given the rare event. However, we prove that if we repeat the algorithm $n$ times and select one of the returned points at random, the distribution of the selected point converges to the exact one in total variation when $n$ increases.

The current popular method for approximate simulation from the posterior distribution of the linear Bayesian LASSO is a Gibbs sampler. It is well-known that the output analysis of an MCMC sampler is difficult due to the complex dependence amongst the states of the underlying Markov chain. Practitioners can usually only assess the convergence of MCMC samplers using heuristics. In this paper we construct a method that yields an independent and identically distributed (iid) draws from the LASSO posterior. The advantage of such exact sampling over the MCMC sampling is that there are no difficulties with the output analysis of the exact sampler, because all the simulated states are independent. The proposed sampler works well when the dimension of the parameter space is not too large, and when it is too large to permit exact sampling, the proposed method can still be used to construct an approximate MCMC sampler.

Approximate Bayesian computation is a popular methodology for simulation-based parameter inference in scenarios where the likelihood function is either analytically intractable or computationally infeasible. The likelihood of simulator parameters fitting given data is approximated by iteratively simulating samples that are generated according to a specified prior distribution. The convergence speed and the quality of inference are highly sensitive to the choice of hyperparameters such as the chosen summary statistic, the value of the acceptance threshold and the distance function. The choice is typically left to the domain expert as summary statistics vary across disciplines and threshold values are problem-specific. This work explores automated hyperparameter optimization for approximate Bayesian computation using Bayesian optimization as an alternative to time consuming manual selection. The problem setting assumes availability of a fast low-fidelity simulator, which is used during the optimization process. The optimized hyperparameters are then used to perform inference using the high-fidelity simulator.

We study the problem of designing good importance sampling (IS) schemes to simulate the probability that a sophisticated predictor, built for instance from an off-the-shelf machine learning model, gives a prediction that exceeds a large threshold. This problem is motivated as a step towards building good learning algorithms that takes into account the extremal risks of the prediction. We provide a framework to design IS for two common machine learning models, namely random forest and a basic neural network. Our approach utilizes some available mathematical programming formulations to optimize over these models and a simple "cutting plane" idea to look for dominant points under Gaussian input distributions.

A commonly used approach to analyzing stochastic differential equations (SDEs) relies on performing Monte Carlo simulation with a discrete-time counterpart. In this paper we study the impact of such a time-discretization when assessing the stationary tail distribution. For a family of semi-implicit Euler discretization schemes with time-step h>0, we quantify the relative error due to the discretization, as a function of h and the exceedance level x. By studying the existence of certain (polynomial and exponential) moments, using a sequence of prototypical examples, we demonstrate that this error may tend to 0 or infinity. The results show that the original shape of the tail can be heavily affected by the discretization. The cases studied indicate that one has to be very careful when estimating the stationary tail distribution using Euler discretization schemes.

Economic capital (EC) is a risk measure that has been used by financial firms to help determine capital levels to hold to protect (with high probability) against large unexpected losses of credit portfolios. Given a stochastic model for a portfolio's loss over a given time horizon, the EC is defined as the difference between a quantile and the mean of the loss distribution. We describe Monte Carlo methods for estimating the EC. We apply measure-specific importance sampling to separately estimate the two components of the EC, which can lead to much smaller variance than when estimating both terms simultaneously. We provide Bahadur-type representations for our estimators of the EC, which we further exploit to establish central limit theorems and asymptotically valid confidence intervals. We present numerical results for a simple model to demonstrate the effectiveness of our approaches.

We develop simulation estimators of risk measures associated with the distribution of the hitting time to a rarely visited set of states of a regenerative process. In various settings, the distribution of the hitting time divided by its expectation converges weakly to an exponential as the rare set becomes rarer. This motivates approximating the hitting-time distribution by an exponential whose mean is the expected hitting time. As the mean is unknown, we estimate it via simulation. We then obtain estimators of a quantile and conditional tail expectation of the hitting time by computing these values for the exponential approximation calibrated with the estimated mean. Similarly, the distribution of the sum of lengths of cycles before the one hitting the rare set is often well-approximated by an exponential, and we analogously exploit this to estimate the two risk measures of the hitting time. Numerical results demonstrate the effectiveness of our estimators.

We survey recent developments concerning Sequest and Sequem, two simulation-based sequential procedures for estimating steady-state quantiles. These procedures deliver improved point and confidence-interval (CI) estimators of a selected steady-state quantile, where the CI approximately satisfies user-specified requirements on the CI's coverage probability and its absolute or relative precision. Sequest estimates a nonextreme quantile (i.e., its order is between 0.05 and 0.95) based on the methods of batching and sectioning. Sequem estimates extreme quantiles using a combination of batching, sectioning, and the maximum transformation. Two test problems show both the advantages and the limitations of these procedures. Based on the lessons learned in designing, justifying, implementing, and stress-testing Sequest and Sequem, we discuss future challenges in advancing the theory, algorithmic development, software implementation, performance evaluation, and practical application of improved procedures for steady-state quantile estimation.

Importance sampling has been extensively studied as a variance reduction tool in rare-event simulation. The design and efficiency of this method often relies on structural knowledge about the target problem in hand. In this paper, we consider a simple rare-event setting, driven by Gaussian input variates, where there is no direct information about the system of interest, or where the system of interest can only be evaluated through a "black-box" model. This setting mimics, on a basic level, situations where the system of interest could be too complex to analyze as in the conventional importance sampling literature. We investigate an approach based on two-stage sampling, where the first stage learns the rare-event set of interest, and the second stage uses a change of measure that exploits the first-stage finding. We present some guarantees and discuss comparisons of this scheme with more naive approaches.

For complex stochastic systems, simulation can be used to study the system inherent risk behaviors characterized by a sequence of percentiles. In this paper, we develop a Bayesian framework to quantify the overall estimation uncertainty of percentile responses. Suppose that the input parametric families are known. The input model estimation uncertainty is quantified by posterior samples of input parameters. Then, a distributional metamodel is introduced to simultaneously model the percentile response surfaces, which can efficiently propagate the input uncertainty to outputs. Our Bayesian framework can deliver credible intervals for percentiles, and a variance decomposition is further derived to estimate the contributions of input and simulation uncertainties. The empirical studies indicate that our approach has promising performance for system risk analysis.

Metamodeling has been a topic of longstanding interest in stochastic simulation because of the usefulness of metamodels for optimization, sensitivity, and real- or near-real-time decision making. Experiment design is the foundation of classical metamodeling: an effective experiment design uncovers the spatial relationships among the design/decision variables and the simulation response; therefore, more design points, providing better coverage of space, is almost always better. However, metamodeling based on likelihood ratios (LRs) turns the design question on its head: each design point provides an unbiased prediction of the response at any other location in space, but perhaps with such inflated variance as to be counterproductive. Thus, the question becomes more which design points to employ for prediction and less where to place them. In this paper we take the first comprehensive look at LR metamodeling, categorizing both the various types of LR metamodels and the contexts in which they might be employed.

When a Gaussian Markov random field (GMRF) is used as a metamodel of an unknown response surface for a discrete optimization via simulation (DOvS) problem, the hyperparameters of the GMRF are estimated based on a few initial design points in a large feasible solution space. Although the maximum likelihood estimators (MLEs) are most commonly adopted to estimate these hyperparameters, its computation time increases polynomially in the size of the feasible solution space. We introduce new generalized method of moments (GMM) estimators of the hyperparameters of GMRFs and their initial sampling schemes, and show they are consistent under some conditions. Unlike MLEs, the computation time for these GMM estimators does not depend on the size of the feasible solution space. We show empirically that the GMM estimators have smaller biases and standard errors than MLE for a wide range of initial simulation budget while requiring orders of magnitude smaller computation time.

The finite-difference method has been commonly used in stochastic derivative estimation when an unbiased derivative estimator is unavailable or costly. The efficiency of this method relies on the choice of a perturbation parameter, which needs to be calibrated based on the number of simulation replications. We study the setting where such an a priori planning of simulation runs is difficult, which could arise due to the variability of runtime for complex simulation models or interruptions. We show how a simple recursive weighting scheme on simulation outputs can recover, in an online fashion, the optimal asymptotic bias-variance tradeoff achieved by the conventional scheme where the replication size is known in advance.

Multistart stochastic gradient descent methods are widely used for gradient-based stochastic global optimization. While these methods are effective relative to other approaches for these challenging problems, they seem to waste computational resources: when several starts are run to convergence at the same local optimum, all but one fail to produce useful information; when a start converges to a local optimum worse than an incumbent solution, it also fails to produce useful information. For problems with a one-dimensional input, we propose a rule for allocating computational effort across starts, Most Likely to Succeed (MLS), which allocates more resources to the most promising starts. This allocation rule is based on a novel Gaussian-Process-based statistical model (SGD-GP) for a start's limiting objective value. Unlike previously proposed statistical models, ours agrees with known convergence rates for SGD. Numerical results show our approach outperforms equal allocation of effort across starts and a machine learning method.

We study and compare the efficiencies of stochastic approximation (SA) and sample average approximation (SAA) for stochastic optimization problems when the decision variables are inside the generating probability distributions or when importance sampling can be applied (as in the case of simulation optimization with Monte Carlo samples). We explain how SA is statistically more efficient than SAA in such contexts, a behavior different from conventional situations where SAA is usually held as the statistically optimal procedure. We support our claim with a theoretical minimax framework and a weak duality argument. We also demonstrate our theoretical findings with some simple simulation examples.

We consider a single stage stochastic program without recourse with a strictly convex loss function. We assume a compact decision space and grid it with a finite set of points. In addition, we assume that the decision maker can generate samples of the stochastic variable independently at each grid point and form a sample average approximation (SAA) of the stochastic program. Our objective in this paper is to characterize an asymptotically optimal linear sample allocation rule, given a fixed sampling budget, which maximizes the decay rate of probability of making false decision.

We propose a simulation-based prediction framework which can quantify the prediction uncertainty of system future response and further guide operational decisions for complex stochastic systems. Specifically, by exploring the underlying generative process of real-world data streams, we first develop a nonparametric input model which can capture the important properties, including non-stationarity, skewness, component-wise and time dependence. It can improve the prediction accuracy, and the posterior predictive distribution can quantify the prediction uncertainty accounting for both input and stochastic uncertainties. Then, we propose the simulation-based prediction framework which can efficiently search for the optimal operational decisions hedging against the prediction uncertainty and minimizing the expected cost occurring in the planning horizon. The empirical study demonstrates that our approach has promising performance.

In the ranking-and-selection field, many current existing fully sequential procedures are developed under the indifference-zone (IZ) formulation which assumes an optimality gap between the best alternative and the others. In this paper, by modifying one classical fully sequential procedure, Paulson's procedure, we devise a new type of fully sequential procedure which can provide the probably approximately correct (PAC) selection guarantee. The procedure is able to select an alternative within a certain distance from the best one with the desired probability even when the IZ assumption is violated. In order to provide this selection guarantee, we lose some efficiency on the procedure. However, we show, both theoretically and numerically, that this efficiency loss is limited and acceptable when the number of alternatives is large.

We review static and dynamic optimization formulations for simulation allocation and selection procedures and revisit several sampling approaches under a single umbrella. We conduct some new simulation experiments to illustrate where the static optimization approach may be inadequate to capture the dynamic sampling decisions and show how many existing sampling procedures ignore certain important considerations.

We boost the performance of the Optimal Computing Budget Allocation (OCBA) algorithm, a widely used and studied algorithm for Ranking and Selection (as known as Best Arm Identification) under a fixed budget. The proposed fully sequential algorithms, OCBA+ and OCBAR, are shown to have better performance both theoretically and numerically. Surprisingly, we reveal that in a two-design setting, a constant initial sample size in a family of OCBA-type algorithms (including the original OCBA) only amounts to a sub-exponential or even polynomial convergence rate of the probability of false selection (PFS). In contrast, our algorithms are guaranteed to converge exponentially fast, as is shown by a finite-sample bound on the PFS.

This paper considers the problem of ranking and selection with covariates and aims to identify a decision rule that stipulates the best alternative as a function of the observable covariates. We propose a general data-driven framework to accommodate (i) high-dimensional covariates and (ii) general (nonlinear) dependence between the mean performance of an alternative and the covariates. For both scenarios, we design new selection procedures and provide certain statistical guarantees, by leveraging the data-intensive environment and various statistical learning tools. The performances of our procedures are exhibited through simulation experiments.

Water Distribution Networks are a particularly critical infrastructure for the high energy costs and frequent failures. Variable Speed Pumps have been introduced to improve the regulation of water pumps, a key for the overall infrastructure performance. This paper addresses the problem of analyzing the effect of the VSPs regulation on the pressure distribution of a WDN, which is highly correlated to leakages and energy costs. Due to the fact that water network behavior can only be simulated, we formulate the problem as a black box feasibility determination, which we solve with a novel stochastic partitioning algorithm, the Feasibility Set Approximation Probabilistic Branch and Bound, that extends the algorithm previously proposed by two of the authors. We use, as black box, EPANet, a widely adopted hydraulic simulator. The preliminary results, over theoretical functions as well as a water distribution network benchmark case, show the viability and advantages of the proposed approach.

We present a study on simulation-based optimization for the Viennese subway system. The underlying discrete event simulation model has several stochastic elements like time-dependent demand and turning maneuver times, direction-dependent vehicle travel and passenger travel as well as transfer times. Passenger creation is a Poisson process which uses hourly origin-destination-matrices based on mobile phone data. The number of waiting passengers on platforms and within vehicles are subject to capacity restrictions. As a microscopic element, passenger distribution along platforms and within vehicles is considered. There are trade-offs between service quality (e.g. waiting time) and costs (e.g. fleet mileage). This bi-objective optimization problem is transformed into a single-objective one by normalization and scalarization. The goal is to find optimal time-dependent headways. Computational experience is gained from 48 test instances which are based on real-world data. Several population-based evolutionary algorithms were applied. The covariance matrix adaptation evolution strategy (CMA-ES) performed best.

Modern advancements in computing architectures have been accompanied by new emergent paradigms to run Parallel Discrete Event Simulation models efficiently. Indeed, many new paradigms to effectively use the available underlying hardware have been proposed in the literature. Among these, the Share-Everything paradigm tackles massively-parallel shared-memory machines, in order to support speculative simulation by taking into account the limits and benefits related to this family of architectures. Previous results have shown how this paradigm outperforms traditional speculative strategies (such as data-separated Time Warp systems) whenever the granularity of executed events is small. In this paper, we show performance implications of this simulation-engine organization when the simulation models have a variable granularity. To this end, we have selected a traffic model, tailored for smart cities-oriented simulation. Our assessment illustrates the effects of the various tuning parameters related to the approach, opening to a higher understanding of this innovative paradigm.

We propose a framework and specific algorithms for screening a large (perhaps countably infinite) space of feasible solutions to generate a subset containing the optimal solution with high confidence. We attain this goal even when only a small fraction of the feasible solutions are simulated. To accomplish it we exploit structural information about the space of functions within which the true objective function lies, and then assesses how compatible optimality is for each feasible solution with respect to the observed simulation outputs and the assumed function space. The result is a set of plausible optima. This approach can be viewed as a way to avoid slow simulation by leveraging fast optimization. Explicit formulations of the general approach are provided when the space of functions is either Lipschitz or convex. We establish both small- and large-sample properties of the approach, and provide two numerical examples.

Parallelizing Bayesian Optimization has recently attracted a lot of attention. The challenge is usually to estimate the effect multiple new samples will have on the posterior distribution of the objective function, and the combinatorial explosion of the possible sample locations. In this paper, we show that at least for multi-task Bayesian Optimization, parallelization is straightforward because the benefit of samples is independent as long as they are sufficiently far apart in the task space. We propose a simple penalization approach proportional to correlation based on the kernel and demonstrate that for the problem settings we considered, the efficacy of our parallel multi-task Bayesian Optimization algorithm is close to the sequential version, while being able to exploit parallel computation to speed up optimization. Depending on the setup considered, we observe efficiencies of parallelization between 69% and 100%.

This paper studies integrated random search algorithms for continuous optimization-via-simulation (COvS) problems. We first tailor the Gaussian process-based search (GPS) algorithm to handle COvS problems. We then analyze the potential sampling issue of the GPS algorithm and propose to construct a desirable Gaussian mixture model (GMM) which is amenable for efficient sampling and at the same time also maintains the desirable property of exploitation and exploration trade-off. Then, we propose a Gaussian mixture model-based random search (GMRS) algorithm. We build global convergence of both the tailored GPS algorithm and the GMRS algorithm for COvS problems. Finally, we carry out some numerical studies to illustrate the performance of the GMRS algorithm.

Improving production line performance and identifying bottlenecks using simulation-based optimization has been shown to be an effective approach. Nevertheless, for larger production systems which are consisted of multiple production lines, using simulation-based optimization can be too computationally expensive, due to the complexity of the models. Previous research has shown promising techniques for aggregating production line data into computationally efficient modules, which enables the simulation of higher-level systems, i.e., factories. This paper shows how a real-world factory flow can be optimized by applying the previously mentioned aggregation techniques in combination with multi-objective optimization using an experimental approach. The particular case studied in this paper reveals potential reductions of storage levels by over 30%, lead time reductions by 67%, and batch sizes reduced by more than 50% while maintaining the delivery precision of the industrial system.

Building construction comprises interaction and interdependence among processes. Discrete-event simulation (DES) is widely applied to model these processes interaction. To find optimal construction plans, optimization technique is usually integrated with DES. However, present simulation-optimization integrated method directly invokes simulation model within optimization algorithms, which is found significantly computationally expensive. This study proposes a machine learning based construction simulation and optimization integrated method. After trained by DES, the machine learning model accelerates simulation-optimization integration by nearly real-time providing fitness evaluation within optimization. This method was implemented into a real construction project for construction time-cost-environment optimization. Results show that proposed machine learning based method significantly reduce computing time compared with original simulation-optimization integration. Less than 1% of construction cost and time improvement were miss, while greenhouse gas emissions obtained same performance. The new method could be a more effective DES and optimization integration approach for practical engineering application.

Simulation optimization is often conducted by applying optimization heuristics (e.g., genetic algorithms) whereby the simulation model delivers the objective function value for the respective parameter set. For real world simulation models, their evaluation time is a crucial constraint. This holds especially for material requirements planning (MRP) parameter optimization of real production systems with many products, because of an extensive search space. Approximating the objective function values by surrogate models can be applied to reduce the search space. Based on a real world production system simulation model, the performance of different regression models to identify simple surrogate models for fast objective function approximation is evaluated in this paper. Specifically, a focus is put on the relationship between the MRP parameters: lot-size and planned lead time, and the performance indicators: inventory and tardiness costs. The paper evaluates a set of simple regression models and compares their objective function fit.

Green simulation is the reuse of past simulation outputs to enhance the efficiency of current and future simulation experiments. One natural application of green simulation is in the context of simulation optimization, wherein outputs from past iterations in a search can be reused in subsequent iterations. In this article, we draw attention to challenges that arise when green simulation likelihood ratio estimators are naively employed in simulation optimization. In particular, we show that for searches that identify new designs based on past outputs, outputs in different iterations are conditionally dependent, violating one of the assumptions for the validity of the likelihood ratio estimator. As a result, green simulation likelihood ratio estimators of the objective and gradient can become biased. We demonstrate how this conditional dependence and bias can adversely affect the behavior of gradient-based optimization algorithms.

We study a statistical method to estimate the optimality gap, as an assessment of the quality, of a given solution for a stochastic optimization using limited data. Our approach is based on bootstrap aggregating the resampled optimal values of sample average approximation (SAA), by connecting these SAA values with the classical notion of symmetric statistics. We discuss how this approach works on general stochastic optimization problems and is statistically more efficient than some previous methods. We substantiate our findings with several numerical experiments.

We consider optimization problems with uncertain constraints that need to be satisfied probabilistically. When data are available, a common method to obtain feasible solutions for such problems is to impose sampled constraints, following the so-called scenario generation (SG) approach. However, when the data size is small, the sampled constraints may not support a guarantee on the feasibility of the obtained solution. This paper studies how to leverage parametric information and the power of Monte Carlo simulation to obtain feasible solutions even when the data are not sufficient to support the use of SG. Our approach makes use of a distributionally robust optimization formulation that informs the Monte Carlo sample size needed to achieve our guarantee.

Many service centers offer a “call-back” option, in which customers entering the queue are informed of the anticipated (on-line) wait and can choose to wait either online or off-line till an agent contacts them. We show that such a policy has the potential of both improving service performance and server utilization, by balancing the load between overloaded and underloaded periods. However, our analysis suggests that companies need not offer that service at all times, and that the delay guarantees proposed should be planned according to the anticipated load throughout the day. In order to optimize the operation of such a system, we develop an Iterative Simulation Algorithm to determine what delay guarantees the company should offer in a time-varying environment. Those guarantees depend on service level targets the company wishes to provide and the delay sensitivity of the online vs. off-line customers.

Worldwide Electrical Power Systems (EPSs) are faced with tremendous challenges because of the reduction of greenhouse gas emissions and the increasing number of renewables. EPS analysis can help to show future developments in an uncertain environment and is an important task for the assessment of greenhouse gas emissions. In order to perform such a complex analysis of future EPSs, a huge number of input parameters is needed. Moreover, technical and also economical processes have to be considered. Thereby, one major task is the modeling of electricity markets. In this paper, we present an approach for the modeling of the German EPS including electricity markets using hybrid simulation and mathematical optimization. We contribute an object-oriented electricity market model which can be utilized to study different exchange mechanisms and behavior patterns of generation unit operators. Simulation results show market results for different generation unit operators and realistic market prices.

This paper deals with the Emergency Department (ED) patient flow management problem. After triage, where acuity of patient’s illness is assessed, each patient waits for treatment. Different care requires disparate resources, leading to different care-paths in the ED. During treatment patients can be in a diversity of treatment stages (waiting for the first consultation, doing clinical tests, waiting for a second consultation, etc.). Therefore, the selection of the next patient to be seen by a physician is not trivial, especially when there are different quality goals to be attained simultaneously. This research investigates disciplines based on accumulation of priority while waiting (APQ). Pure priority disciplines are independent of the quality goals set for the ED while the new ones can be optimized to achieve those goals as much as possible. The optimal APQ policies are obtained by using simulation-based optimization. Both types of policies are tested in a real case.

Although the mean is a widely-used performance measure for stochastic simulations, the quantiles have become very attractive to measure the variability and the risk of the simulated systems. Optimizing high quantiles, which is typically of interest for decision makers, can be challenging with limited budget as it may cost a large number of simulation runs to obtain estimators for high quantiles with acceptable accuracy. This restricts the application of some traditional Monte Carlo approaches. In this work, we propose a multi-level metamodel (co-kriging) based algorithm to optimize high quantiles more efficiently. Utilizing non-decreasing properties of quantile functions, we first search on some cheaper and informative lower quantiles, and then search the higher quantiles with a focus on the promising regions identified by lower quantiles. This results in less wastage of budget in non-promising regions. A numerical study is presented to show the power of this algorithm.

Multi-fidelity simulation optimization is an emerging area looking at the use of low-fidelity (computationally cheap but inaccurate) models to optimize high-fidelity (expensive and accurate) models. In this context, low-fidelity models exhibit a mismatch to high-fidelity models whose values can be point-wise obtained by querying an expensive simulator. Herein, an efficient multi-fidelity algorithm is proposed for continuous global optimization. The algorithm is made up of an additive model that consolidates low-fidelity and bias (mismatch) predictions. Two sampling criteria with different use of the cumulated high and low-fidelity information are introduced as well as a cheap certificate guiding the decision on whether to sample from the expensive simulator. The performance of proposed algorithms is evaluated using a state of the art stochastic search benchmark algorithm. The results show that the proposed methods can beat the benchmark with improved accuracy, while essentially maintaining the same performance in terms of number of expensive simulations.

We introduce Direct Gradient Augmented Response Surface Methodology (DiGARSM), a new sequential first-order method for optimizing a stochastic function based on Response Surface Methodology (RSM). In this approach, gradients of the objective function with respect to the desired parameters are utilized in addition to response measurements. We establish convergence of the proposed method. We compare methods that use only response information, and those that use only gradient information with the proposed approach, which uses both. Moreover, we conduct numerical simulations to illustrate the effectiveness of the proposed method.

The airline industry faces many causes of disruption. To minimise financial and reputational impact, the airline must adapt its schedules. Due to the complexity of the environment, simulation is a natural modelling approach. However, the large solution space, time constraints and system constraints make the search for revised schedules difficult. This paper presents a method for the aircraft recovery problem that uses multi-fidelity modelling including a trust region simulation optimisation algorithm to mitigate the computational costs of using high-fidelity simulations with its benefits for providing good estimates of the true performance.

In many practical problems, simulation models are used to support complex decision-making processes comparing hundreds or thousands of solutions. These problems typically have a key objective but the final decision may be dependent on other factors, which cannot be incorporated into the simulation model. In such cases, decision-makers may request a short list of `good' solutions, which work well for the main objective and satisfy one or more chance constraints. While fully sequential ranking and selection procedures can be effective at solving these problems, surveys of experimentation practice suggest that they are under-utilized, potentially due to difficulties automating commercial software. We develop an approach with just two stages of replications. The approach, which has been designed to cope with the use of common random numbers, draws on ideas from indifference zones and makes use of bootstrapping to find a subset of high quality solutions. A Python implementation is freely available.

When we have sufficient computational resources to treat a simulation optimization problem as a ranking & selection (R&S) problem, then it can be "solved." R&S is exhaustive search---all feasible solutions are simulated---with meaningful statistical error control. High-performance parallel computing promises to extend the R&S limit to even larger problems, but parallelizing R&S procedures in a way that maintains statistical validity while achieving substantial speed-up is difficult. In this paper we introduce an entirely new framework for R&S called Parallel Adaptive Survivor Selection (PASS) that is specifically engineered to exploit parallel computing environments for solving simulation optimization problems with a very large number of feasible solutions.

We consider the multiobjective simulation optimization problem, where we seek to find the non-dominated set of designs evaluated using noisy simulation evaluations, in the context of numerically expensive simulators. We propose a metamodel-based scalarization approach built upon the famous ParEGO algorithm. Our approach mainly differentiates from ParEGO and similar algorithms in that we use stochastic kriging, which explicitly characterizes both the extrinsic uncertainty of the unknown response surface, and the intrinsic uncertainty inherent in a stochastic simulation. We additionally integrate the Multiobjective Optimal Computing Budget Allocation ranking and selection procedure in view of maximizing the probability of selecting systems with the true best expected performance. We evaluate the performance of the algorithm using standard benchmark test functions for multiobjective optimizers, perturbed by heterogeneous noise. The experimental results show that the proposed method outperforms its deterministic counterpart based on well-known quality indicators and the fraction of the true Pareto set found.

As the IEEE standard for distributed simulation, High-Level Architecture (HLA) defines the data distribution management (DDM) service to filter unnecessary communication among federates. In DDM, interest matching algorithm is essential for the message filtering. Among existing interest matching algorithms, sort-based algorithms have been proven to be the most efficient method in most scenarios. However, existing sort-based algorithms still have some drawbacks, the overhead of sorting the bounds can be further reduced and a portion of unnecessary bit or matching operations can be eliminated. In this paper, we propose a binary search enhanced sort-based interest matching algorithm (BSSIM). Based on two exclusive judging conditions for region overlap, sorting overhead can be remarkably reduced. Moreover, unnecessary bit or matching operations can be eliminated by binary searches. Experimental results show that BSSIM algorithm outperforms existing sort-based algorithms, and considerable performance improvement can be achieved.

It is known that incorporating gradient information can significantly enhance the prediction accuracy of stochastic kriging. However, such an enhancement cannot be scaled trivially to high-dimensional design space, since one needs to invert a large covariance matrix that captures the spatial correlations between the responses and the gradient estimates at the design points. Not only is the inversion computationally inefficient, but also numerically unstable since the covariance matrix is often ill-conditioned. We address the scalability issue via a novel approach without resorting to matrix approximations. By virtue of the so-called Markovian covariance functions, the associated covariance matrix can be invertible analytically, thereby improving both the efficiency and stability dramatically. Numerical experiments demonstrate that the proposed approach can handle large-scale problems where prior methods fail completely.

Large solution space is one of the main features of simulation-optimization problems. Reducing the cardinality of the set of alternatives is a key point for increasing the efficiency of simulation-optimization methods. In this work, a new cutting approach is proposed for this purpose. The approach exploits the Benders Decomposition framework that can be effectively applied when the simulation-optimization problems are represented using Discrete Event Optimization models. Benders Decomposition subproblems represent the simulation components, hence, cuts can be easily generated observing the values of the variables while a system alternative is simulated, without solving any subproblem. The cut generation procedure is proposed to approximately solve the Server Allocation Problem in a tandem queueing system. Results on randomly generated instances show its effectiveness in decreasing the computational effort by reducing the solution space.

Hierarchical problem decomposition methods are widely used in optimization when the scale of the problem is large. The master problem is hierarchically decomposed to several sub-problems and the detail level of the sub-problems increases during the optimization from bottom to top. When simulation is used to estimate unknown functions, models with different detail are used at each level. However, the simulation outputs used to solve the sub-problems of a hierarchy level are not used anymore at higher levels. An approach is proposed in this paper to reuse these experiment data to improve the efficiency of the simulation-optimization algorithm. A multi-fidelity surrogate model is built in each sub-problem to guide the search of the optimum. The performance of the approach is numerically assessed with the goal of understanding its potentialities and the effect of algorithm parameters over optimization results.

Continuous descent operation (CDO) is fuel-saving noise-abating arrival procedures that are used at the Kansai International Airport (KIX). However, CDO can only be performed at night so as not to disturb airflow during heavy daytime traffic. Altitude window has been proposed as a solution to expand the CDO operations. The operational conditions surrounding the arrival route, such as the percentages of aircraft types and the wind conditions, should be adequately considered while designing the altitude window. This study describes the simulation techniques that are used to formulate this design. A fast-time simulation (FTS) is exhibited to calculate the maximum possible range of the vertical-CDO trajectory. A performance-review method for the altitude window is also depicted using an FTS. Exemplary simulations are conducted using the actual KIX data. Finally, the application of a review method for the altitude window is discussed to develop the total design process.

The following paper presents an innovative approach for dealing with complex capacity problems in aviation. We introduce a sliding window framework composed by an optimization method with a simulation component. By applying this framework in diverse problems that are dependent on time it is possible to find feasible and close-to-reality solutions in shorter time than the ones that could be achieved by evaluating the problem in the complete time-horizon. The framework can be applied to solve diverse problems in aviation or similar industries. We exemplify the approach with a model of Paris Charles de Gaulle Airport in France.

This study successfully implements flight specific delay costs in an air traffic simulation with a multi-criteria trajectory optimization and exemplifies a coupling of turnaround and trajectory optimization of historical real flights. Therein, delay costs and detour costs for the reduction of contrail formation are individually calculated for each flight and considered in a flight specific multi-criteria trajectory optimization with the air traffic simulation environment TOMATO. Detours in the optimized trajectories are mainly caused by the intent of avoiding contrail formation. With this case study, the historical flight plan could be stretched and departures and arrivals could be more homogeneously distributed during the analyzed three hours while at the same time ecological costs could be saved by 15 per cent. Therewith the promising potential of System Wide Information Management between airports, airlines, air traffic control and customers could once again be shown.

Capacity analysis for systems with time varying constraints is still an open problem in Operations Research due to the non-stationarity of the problem domain. This is particularly true for Defence manpower supply which is subject to frequent temporal policy and resource changes. As such, the problem cannot be completely covered with a single overriding simulation or optimisation solution, but, rather, better described using piecewise interplay between simulation and optimisation. This paper describes such an approach for a flexible, interactive capacity analysis simulator with an embedded integer linear programming (ILP) optimiser.

The air corridor is an airspace concept to optimize operational efficiency of flights inside it. However, it inevitably becomes a huge obstacle for aircraft flying outside it. In this study, the cost-benefit analysis is performed in order to clarify the feasibility and the conditions at which the air corridor installation becomes beneficial. The increases of the operational cost of the flights outside the air corridor is analyzed through the comparison between the actual trajectories and numerical simulations of the trajectories to avoid the air corridor. The reduction of the operational cost by flying along the optimum trajectories inside the air corridor is also analyzed through comparison with actual trajectories. It has been clarified that even a small number of flights operated inside the air corridor achieves total operational cost reduction. It is further concluded that the air corridor should be installed as early as onboard self-separation capability becomes available.

The cost efficient management of spare parts for low-volume high-tech equipment is inherently difficult. In this on-going study, we seek to improve the OEM’s spare parts inventory management by incorporating the condition information from a large number of distributed working units in the field. For that purpose, the condition information relayed by sensors is put in context with usage parameters, preventive replacement policies, customer plans, and current economic indicators to create an aggregate forecast and inventory ordering policy. This requires a synthesis of the state of the art knowledge from multiple research streams. In this paper, we outline a simulation environment of the maintenance management of a jet engine program over its life cycle, and provide preliminary results highlighting several modules for future research to improve the performance of spare part inventory policies and assess the value of health monitoring.

During arrival and departure procedures at an airport, aircraft generate noise. High levels of noise during these terminal procedures can have a negative impact on the communities located near the airport. We assess the impact of the noise generated by a standard instrument departure of a twin-engine narrow body mid-range aircraft over the communities nearby Amsterdam Airport Schiphol, the Netherlands. During the departure procedure, we consider a stochastic flight trajectory that is subject to lateral position errors. We estimate, using Monte Carlo simulation, the distribution of the sound exposure level and of the number of awakenings generated by a departure. We also identify the residential areas where the number of awakenings are overestimated or underestimated, when comparing a stochastic and a deterministic departure approach. Lastly, we approximate the distribution of the noise level for a generic aircraft departure, which can be further employed for further optimization of departure procedures.

Reliable and predictable ground operations are essential for 4D aircraft trajectories. Uncertainties in the airborne phase have significantly less impact on flight punctuality than deviations in aircraft ground operations. The ground trajectory of an aircraft primarily consists of the handling processes at the stand, defined as the aircraft turnaround, which are mainly controlled by operational experts. Only the aircraft boarding, which is on the critical path of the turnaround, is driven by the passengers’ experience and willingness or ability to follow the proposed procedures. We propose a machine learning approach predict the boarding time. A validated boarding simulation provides data input for a recurrent neural network approach (discrete time series of boarding progress). In particular we use a Long Short-Term Memory model to learn the characteristic passenger behaviors over time.

Airport retail revenue has long been recognized as a critical revenue stream to ensure an airport’s financial sustainability and stability. However, there is a lack of simulation model on how airport terminal could be better designed to facilitate this vital revenue stream. This paper presents a conceptual agent-based simulation model on passengers shopping behavior in the airport context. This model attempts to investigate the relationship between terminal design and retail performance through different scenarios studies. Results show that finger pier terminal shape can have a negative impact on retail revenue if shops are decentralized. Terminal with centralized shopping areas also performed better than a terminal with decentralized shopping area. Future research directions were proposed at the end to improve the existing simulation model with the aim of making it an essential evaluation tool for future terminal design.

Booms are extensively used for confining oil spilled at sea and for carrying it away from sensitive areas or toward a gathering point. The success of booming operations depends on effective boom deployment, which, in turn, is conditioned by the state of the sea – wind, tides, currents, and waves – and its effects on the spill spreading shape and speed. Planning the boom deployment beforehand could help overcome problems derived from sea condition. Besides, the employment of Unmanned Surface Vehicles (USVs) for towing spill recovery booms represents an interesting alternative to the use of ordinary vessels, and it could lead to a fully automatic oil spill recovery system. A basic tool has been developed in Python to simulate the deployment of booms towed by USVs. The aim is to explore simple trial scenarios, working out information that helps to prototype real USV-boom systems and to plan automatic deployments.

In this paper, a novel approach is proposed to design a bank of unknown input observers(UIOs) for aero-engine distribute control system(DCS) with multi-sensor fault. First, considering the system with multi-sensor fault which doesn’t satisfy the observer matching condition, a UIO is proposed based on the disturbance decoupling principle of unknown input observer(UIO) and control theory. The parameter matrixes of UIO are solved in terms of linear matrix inequality(LMI) toolbox. Second, the fault detection and isolation scheme is proposed for DCS with multi-sensor fault and the UIOs are designed for a bank of fault detection and isolation sets which are obtained by reorganizing the control system. The residual signal generated by each observer is robust to the sensor fault in the matching set, and sensitive to the sensor fault out of the matching set. Finally, the residual signals were simulated and the results show the effectiveness of the proposed method.

Electric taxi fleet simulation calls for a lot of uneasily accessible data. We learned from historical data in order to predict electric vehicle consumption and charging time. Although simulation was the main goal, the developed models will be used by the dispatching team of the company. Moreover, the data analysis process allowed us to discover flaws in the infrastructure (e.g., defective stations and defective batteries).

A simulation model representing the dynamics of a sow farm is presented in contrast with other modelling approaches. To highlight relevant aspects of the model a real application for comparing two different groups of sows is considered. The main contribution of the model is that it is flexible to represent different sow herd management alternatives through an extensive number of input parameters. This features allow to measure the productivity of sows under different set of parameters and performing virtual experimentation. The example presented here is related to estimate economic performances of two phenotypes of sows from a real experiment. Phenotypes consist of sows resistant to a common disease (porcine reproductive and respiratory syndrome) with respect another group sensitive to the same reproductive disease. Furthermore, the implementation in Extend performs efficiently additional calculations to enlarge the understanding of the productive behavior in each case.

With an increasing demand to manage biological resources even when the supporting systems undergo significant changes there is great need for ecological models which take into account all system levels relevant to the resource. However, the large ecological models rising to this challenge require expertise from, at times, quite distinct branches of science. Therefore, rather than tasking a single individual or working group to expand models beyond their area of expertise, fusing smaller submodels into large complex ones appears as a natural way forward. We propose to ensure the semantic validity of the fused models by carefully interlinking composition and validation steps exploiting earlier simulation experiments. A simulation model of the cod in the Baltic sea will elucidate the approach as the different aspects of respiration, energy budgets and behavior are fused and validated as a whole after being modeled and validated individually.

Sustainability is related with environmental, social and economic variables. Each one of these areas is, by itself, complex due to the huge number of factors that one must analyze. Because of the combination of the levels with the factors, and the needed replications, an exponential growth in the number of executions appears. In this paper we describe a methodology that helps us to deal with this complexity applying three key concepts, formal representation of simulation models, optimization algorithms and high-performance computing. We present an infrastructure named NECADA that supports the methodology. This approach can be applied to a building refurbishment or to define optimal parameters in new buildings. The specialists work with the conceptual model, and from it with the system; following the method, they will be able to find optimal scenarios using a selection of build-in heuristics that can be applied for the problem resolution.

Cyclists riding in polluted urban environments may be exposed to unhealthy conditions. Therefore, the design process of bikeway routes should take into account the exposure of cyclists to air pollutants. Riding a bicycle is a common way to commute in Bogotà, a highly polluted city. Quinto Centenario is a 25-km bikeway, to be built in the coming years. This study aims at producing estimates for Quinto Centenario users exposure to particulate matter. We simulated the movement of bikers along the planned route, and we used air quality data collected by a monitoring network to estimate the pollution concentration bikers will be exposed to. Bikeway traffic estimates were obtained from official city surveys, which we analyzed to determine origin/destination matrices for bike trips and trip duration distributions. The output of simulation is captured by a spreadsheet that computes bikers' exposure for any path along the bikeway.

Renewable energy production is increasing, and the cost of power storage is decreasing, enabling users to exercise control over their power consumption from the electricity grid. To achieve partial or complete energy independence, consumers need to be able to calculate the energy storage required to cover periods when renewable power is not being produced. To address this need, a model is introduced which determines energy storage characteristics from a single simulation of an energy production system. From this model, the optimal storage size necessary for a required level of grid-independence can be calculated quickly, to a high degree of accuracy. By using existing data, no user estimation of demand parameters is required, making this model accessible to non-specialists. The optimal battery size for a domestic photovoltaic - battery system under varying demand and production scenarios is analysed and the accuracy and computational performance of the model assessed.

Pakistan is rapidly becoming a water stressed country, thus affecting people’s well-being. Authorities are faced with making drastic water conservation policies toward achieving effective management of available water resources and efficient water supply delivery coupled with responsible demand side management. Due to the lack of modern water metering in Pakistan, water consumption is not being accurately monitored. To achieve this goal, we propose a hybrid modeling and simulation framework, consisting of: (i) Agent-Based Modeling (ABM) paradigm that deals with the behavior and characteristics of individuals and (ii) System Dynamics(SD) paradigm that accounts for water flow dynamics. Our approach provides dual-resolution expressiveness suitable for replicating real-world urban infrastructure scenarios. The key objective of the research is to assist authorities to understand and forecast short-term and long-term water consumption through examining varying patterns of water consumption in different climates and thus improving demand side water usage dynamically subject to water supply availability.

In this work, we tackle power consumption reduction of battery-dependent devices in a smart campus (including hospital) application. These devices are connected by networked systems which may be subject to fluctuation of the message delays that control essential equipment. We show through five case studies using discrete event simulation, that power consumption may be reduced using proper prioritization and balancing of the network emergency traffic. A predictor algorithm and a fuzzy logic controller was used to indicate the level upon which the system must switch off the load in order to reduce power consumption. The analysis of a case study shows that a considerable reduction in power consumption was achieved through the reduction of message delays and also due to the fuzzy control of AC and lighting equipment.

Cybersecurity assessment in automation requires a modeling and simulation framework to study the complex relationships between the cyber-based control mechanisms and the power systems. A real power system is not ideal for such an assessment due to potential disruption in operation. In this paper, we propose a simulation model of a virtual operational technology (OT) network in power system automation for cybersecurity assessment. The proposed simulation model is modularized with key components including power system and process, communication, automation, and enterprise network. We discuss Virtual OT Network (VOTNet), a specific instance of the model, in terms of simulated and emulated systems such as programmable logic controller, computing systems, control centers, and software. We evaluate VOTNet under two use cases: different attack scenarios and scalability vs. attack surface to study the impacts of cyber attacks. Particularly, unauthorized access, data manipulation in PLC, and denial of service in SCADA communication are demonstrated.

With the integration of renewable energy and advanced communication technologies, smart power grids can enhance the cost-efficiency and reliability of energy generation, transmission, and distribution. The communication network connecting numerous remotely distributed generators, devices, and controllers plays a vital role in the control of power grids. However, it is vulnerable to cyber attacks. In particular, because of its frequency and lasting impact, the Distributed Denial of Service (DDoS) attack poses an important threat to smart power grids. This paper presents a simulation-based stochastic programming approach to guide the unit commitment and economic dispatch decisions, which accounts for the prediction uncertainty of wind power and the impact from DDoS attacks. The case study demonstrates that the proposed approach can lead to a more cost-efficient and reliable operational decision guidance for smart power grids.

Traditionally, discrete event simulation is associated with logistic system analysis due to its role as an operations research method. In this article, we consider a multiple case study on the alternative use of simulation for operations management education. When stressing the role of the student or trainee instead of operations—as in operations research—alternative demands on model set up and use arise. Identifying such demands and ways to deal with them extends simulation methodology by showing how to exploit existing simulation knowledge and tools for pedagogic purposes. Our multiple case study evaluates four simulation-based games reflecting operations systems in health care and manufacturing. Respective games target learning activities of a different nature. Outcomes of the evaluation provide insights on modelling requirements, and guidelines and good practices that are supportive in meeting these.

Gaming simulations (games) are increasingly becoming the tool of choice for modeling and understanding the complexity of today’s systems. This increased popularity has consequently revealed the weaknesses of games in several areas. These limitations range from inconsistencies on the game design to the unexploited explicit and tacit knowledge that games invoke. This paper focuses on games that do not aim at generalizing the produced knowledge but, instead, at understanding how a system works within a specific context. The first step of the analysis is identifying these limitations based on an extensive literature review. Based on this, different directions that could mitigate or even fully address these limitations are proposed. The paper concludes with a focused research agenda.

The recent growth of sophisticated digital gaming technologies has spawned an $8.1B industry around using these games for pedagogical purposes. Though Digital Game-Based Learning Systems have been adopted by industries ranging from military to medical applications, these systems continue to rely on traditional measures of explicit interactions to gauge player performance which can be subject to guessing and other factors unrelated to actual performance. This study presents a novel implicit eye-tracking based metric for digital game-based learning environments. The proposed metric introduces a weighted eye-tracking measure of traditional in-game scoring to consider the mental schema of a player’s decision making. In order to validate the efficacy of this metric, we conducted an experiment with 25 participants playing a game designed to evaluate Chinese cultural competency and communication. This experiment showed strong correlation between the novel eye-tracking performance metric and traditional measures of in-game performance.

Researches of AI planning in Real-Time Strategy (RTS) games have been widely applied to human behavior modeling and war simulation. Due to the fog-of-war, planning in RTS games need to be implemented under partially observable environment, which poses a big challenge for researchers. This paper focuses on extending Hierarchical Task Network (HTN) Planning in partially observable environment, and proposes a partially observable adversarial hierarchical task network planning with repairing algorithm named PO-AHTNR. By adding sensing action into HTN domain knowledge, a reconnaissance strategy and a history-based single belief state generation method are presented to obtain the best action. In order to verify the proposed algorithm, an empirical study based on µRTS game is carried out, and the performance of modified algorithm is compared to that of AHTNR and other state-of-the-art search algorithms developed for RTS games.

There has been significant growth in online social science experiments in order to understand behavior at-scale, with finer-grained data collection. Considerable work is required to perform data analytics for custom experiments. We also seek to perform repeated networked experiments and modeling in an iterative loop. In this work, we design and build four composable and extensible automated software pipelines for (1) data analytics; (2) model property inference; (3) model/simulation; and (4) results analysis and comparisons between experimental data and model predictions. To reason about experiments and models, we design a formal data model. Our data model is for scenarios where subjects can repeat actions (from a set) any number of times over the game duration. Because the types of interactions and action sets are flexible, this class of experiments is large. Two case studies, on collective identity and complex contagion, illustrate use of the system.

We design an experiment to evaluate behavior in a dynamic spatial game representing the incentives faced by drivers on a ridesharing platform while waiting to be matched with a rider. The design is unique in that it allows us to observe not only participants' choices, but also the considerations that went into those choices. The results of the experiment show that a large majority of player choices are consistent with myopic best responding - a myopic best response maximizes a player's flow payoff at the time of the decision but ignores strategic considerations regarding the future choices of opponents. Given this finding, we develop agent-based models of spatial competition built upon myopic agents. Myopic behavior in our model results in quite efficient outcomes, suggesting that ridesharing platforms may benefit from sharing with drivers the locations of other nearby drivers to allow them to compete spatially.

Recent work has illustrated how Bayesian approaches to discrete simulation optimization can be applied to clinical trial design. These approaches balance the expected cost of running the trial with the expected economic benefits of adopting one of the treatments, based on the information which accumulates during the trial. Some work in this space has presented a model of a fully sequential trial, but with simplifying assumptions; other work has incorporated some special features of pragmatic clinical trials into a one-stage (that is, non-sequential) framework. This paper shows how simulation optimization ideas can be used to model fully sequential sampling for trials which contain these features.

Improving bioanalytical procedures (assays) to measure levels of immunologically relevant proteins such as haptoglobin requires minimizing the uncertainty associated with the measurement process. In this work, we consider the question of optimally selecting the concentration levels of calibrators (solutions with known concentrations of haptoglobin) used to calibrate the haptoglobin immunoassay. Specifically, we aim to select calibrator concentration levels such that the net measurement uncertainty at medical decision points is minimized. We accomplish this by constructing a Monte Carlo simulation of the haptoglobin immunoassay that estimates the measurement uncertainty associated with the assay, and then formulating the problem of optimally selecting calibrators to minimize the assay uncertainty at medical decision points as a discrete simulation optimization problem. We demonstrate the application of the NSGS and KN procedures to solve this problem. This work represents a first step towards the utilization of simulation optimization in the optimal design of clinical assays.

Sepsis, the body's inflammatory response to infection, is a serious complication and leading cause of in-hospital mortality. Timely intervention is both difficult and critical since sepsis can rapidly worsen to organ dysfunction and septic shock. However, the lack of a gold-standard definition renders the lines between these transitions unclear, complicating medical decision-making. Using electronic health records, we build a simulation framework to study the evolution of dynamic physiological and cellular responses to therapeutic interventions in septic patients. Since sepsis trajectories can manifest differently depending on patient characteristics, we incorporate patient heterogeneity through comorbidity, age, race, and gender. Under therapeutic interventions recommended by the Centers for Medicare and Medicaid Services (CMS), we illustrate the framework on patient trajectories using the CMS criteria for sepsis definition. The framework is designed to support the comparison and quantification of the impact of clinical definitions and recommended interventions on the timely identification of sepsis states.

A modelling and simulation framework for the performance evaluation of emergency units and intelligent control in the case of major crisis is proposed in this paper. Taking into account new challenges related to the optimization of the patient care pathway in the emergency unit in the context of major crises, we propose a flexible tool that can be used by health-care practitioners as a decision aid in various situations under the shape of a digital twin of the emergency unit. The modelling framework is based a modular model with specific representation of care pathways and resources activities. Various arrival processes are modelled in order to comply with the periodic variations observed in data history as well as exceptional massive arrivals that may be related to major crisis. A practical experiment is provided in order to determine the best available leverages to optimize the operations of the system.

Hospital-wide strategies for reducing mental health (MH) boarding in the emergency departments (ED) have focused on improving the onsite patient flow and resource management. A gap in literature is still residing on the lack of consideration for the MH transition process from the ED into external community and hospital inpatient psychiatric settings, as well as the external bed capacity limitations. Discrete event simulation was used to understand the mental health patient flow within the ED and to inpatient settings. It provided a forecast of system changes when three scenarios were tested to determine the percentage increase in beds necessary to reduce MH ED boarding time. These alternative approaches illustrate the applicability of the model as a decision support tool for evaluating solutions to MH ED boarding.

A major challenge of an emergency department (ED) is managing the growing patient volume without diminishing quality of care nor increasing costs. Therefore, this case study aims at minimizing the average patient’s length-of-stay (LOS) subject to a given staffing budget. It is based on data of a multidisciplinary ED in Germany with an annual volume of 50,000 patients. We analyze the impact of process modifications, staff pooling, and optimized staffing levels on the LOS. In order to optimize staffing levels, we combine optimization models with discrete-event simulation. The simulation model is built on treatment processes of different medical specialties that were mapped by time-motion studies. Results show that avoiding boarding time and pooling of nurses originally assigned to internal medicine or neurology are particularly promising. These lead to improvements of 15% and 14% in avg. LOS, respectively. Optimizing staff allocation results in reductions of up to 9% in avg. LOS.

The Ambulance Location and Dispatching Problem (ALDP) identifies the location of the available ambulances and the best dispatching policy to minimize the response times to answer the calls. However, the uncertain nature of the emergency calls makes it impossible to know in advance if the ambulance identified by the dispatching policy is available or not upon a call arrival. Thus, the probability that a vehicle is busy when a call arises, denoted as busy fraction, is usually considered in the literature. Probabilities can be estimated in several manners, but simulation seems to be well suited for this purpose. In this work, we propose four Recursive Optimization-Simulation Approaches to estimate the ALDP busy fraction, and we apply them to a set of realistic instances. Numerical results confirm that the most sophisticated and computing demanding approaches offer a better performance.

An ambulance dispatcher decides which ambulances to allocate to new calls, and how to relocate ambulances in order to maintain a good coverage. Doing this, it is valuable to have information about the future expected response times in different parts of the area of responsibility, as well as the expected number of available ambulances. We present a simulation model that can be used to predict this, and compare the results to a naïve forecasting model. The results show that while it is difficult to accurately predict the future system state, the simulation based prediction manages this better than the naïve model.

Emergency Medical Service is one of the most important health care services as it plays a vital role in saving people’s lives and reducing the rate of mortality and morbidity. A peak of emergency demand can determine overcrowding at the emergency department. In this operative context, a challenge is the definition of proper real-time dispatching, routing and redeployment policies (DRRP) in such a way to maximize the number of emergency requests served within a time threshold, and to minimize the waiting times. The contribution of this paper is twofold. The former is a simulation model capable to deal with the real-time management of the ambulances, and to generate new ad hoc instances. The latter is a set of simple online algorithms to implement several DRRP. An extensive comparison among different DRRP is also provided.

The facilities management department of a U.S. health system has experienced a 42.28% turnover rate within the last year. Despite the human resources team’s work to mitigate turnover, insufficient attention has been given to burnout within the department. There is a need to explore alternative frameworks to investigate the issue that take into account the complexities and impact of staff burnout on the health system. This paper starts by developing an understanding of the root causes and behavior of burnout within a department and scrutinizing the complexity of how departmental decisions impact the health system as a whole. This is followed by a simulation model based on the data collected from managers at the health system. We conclude with a discussion of the insights that were derived from both causal loops as well as that from the results of the Vensim simulation. Finally, plans for future work are outlined.

The growing global diabetes epidemic is a serious public health problem. We developed a system dynamic model to study the cost-effectiveness of different diabetes treatment options. According to existing literature, we estimated dynamic costs and changes of hemoglobin A1c levels of two first-line monotherapies and a hypothetical innovation therapy for glycemic control over a 15-year horizon. Incremental cost-effectiveness ratios were expressed as dollars per HbA1c decrement from perspectives of the patient, insurance-payer, and society. Simulation results showed that better adherence with a more expensive and efficacious drug results in better control of HbA1c and cost-saving in the long-run. The results also showed that the cost-effectiveness ratio varied with patients’ pre-determined out-of-pocket payment for health expenditure. The higher the rate of their out-of-pocket payment for extra health care expenditure to their household income, the more cost-effective it is for the innovative drug from the perspectives of the patient, insurance-payer, and society.

Complex patients, limited specialists, and high demand create long referral queues ill-equipped to deliver timely quality care. One approach to reduce queue length is for specialists to review referrals through an electronic consultation (e-consult) and determine if 1) the patient requires a specialty appointment or 2) the primary care physician can continue care with the specialist’s recommendations. As inappropriate patients are removed from the queue, it is hypothesized that lead times will decrease and the system be more efficient at delivering the right care to the right patient at the right time. A discrete-event simulation was built to estimate the impact e-consults have on Rheumatology clinics at Duke Health; specifically considering lead time, queue length, and specialists’ workload. This is an adaptable tool with visualizations for which any specialty clinic, each with its own complex challenges, could estimate the impact e-consults have on delivering quality care to referred patients.

We describe the development and application of a model that simulates the impact of CommunityRx, an information-based health intervention, on the utilization of community-based resources. The model includes a synthetic population of agents matching the sociodemographic characteristics of the South Side of Chicago, along with their activities and behaviors. We simulate the information-based intervention and model agent decision-making about using community resources to maintain health, based on a dynamic dosing of information about community resources, gained through interactions and experience. Through in silico experiments, our model aims to demonstrate the flow and spread of information from primary agents to others in the community, and through these dynamic interactions, the impact of an individual-level information intervention on resource utilization.

Often large health care organizations will have multiple satellite clinics in addition to the primary flagship location. The value of having multiple clinics includes the ability to better serve patients by improving accessibility while also increasing efficiencies from common structures and policies. Unlike most simulation models of health care clinics, which are uniquely designed for one specific location and are not easily adapted without significant effort, we develop one flexible discrete-event simulation model to represent two orthopedic centers. In addition to demonstrating the validity of employing the same model for two clinics, a graphical user interface is constructed to enable the clinic managers to modify simulation parameters, minimizing the need for future model adaptations. The value of the simulation model is verified by evaluating the effects of new policies impacting the role of mid-level providers and results show that the best policy for improving timeliness of care varies among clinics.

Fragmentation is a common problem in the United States healthcare delivery system. For patients with complex medical conditions requiring multiple providers, this problem is particularly perplexing. One solution being developed, called an Integrated Practice Unit (IPU), involves a co-located multidisciplinary team of providers addressing a patient’s medical condition over a full care cycle. In this paper, we use simulation to design a WorkLife IPU that treats medical issues arising in the workplace from minor illnesses to acute non-urgent injuries. More specifically, we determine a patient appointment schedule and the number of resources required to achieve patient throughput, clinic closing time, and patient waiting time targets set by the IPU management.

The treatment of patients in outpatient healthcare clinics is a continually growing trend as technology improves and recovery benefits are recognized. In this paper, we focus on the operational aspects of clinics that will impact clinic design decisions relative to productivity, efficiency, and quality of patient care. We develop a generalized Healthcare Clinic Design Simulator (HCD-Sim) to study the dynamic system behavior and to analyze alternative outpatient healthcare clinic designs. Our simulation framework is designed to have a data-driven structure that is able to represent a large class of outpatient healthcare clinics through the specification of clinic data relative to patient flows, work flows, and resource requirements. We describe our methodology and demonstrate the capabilities of experimentation utilizing the simulation framework.

Advances in facility design and technology are leading to novel uses of healthcare delivery space including flexible and shared clinical space increasing collaboration and communication in multidisciplinary care settings. With new outpatient care delivery facilities and plans emerging, the need to evaluate how flexible and shared resources are assigned arises. We present a discrete-event simulation model based on the daily operations of an outpatient clinic where multiple specialties share resources including support staff, exam rooms, and ancillary providers. Being a new clinical space, the primary objective is to design resource assignment and staffing policies which maximize the use of the new facility. While resource utilization is the primary objective, the simulation model results and analysis incorporate other competing performance criteria including patient waiting, provider idleness, and clinic length of day. Results are presented based on a multidisciplinary clinic at the University of Minnesota.

Since 2015 researchers in Austrian health-care research project DEXHELPP (Decision Support for Health Policy and Planning) benefit from having access to a validated generic agent-based population model (GEPOC ABM) of Austria’s population. This simulation model delivers a valid virtual image of Austria’s population and is also able to make feasible prognoses. During the last years the model has been extended, remodeled and applied to several use-cases. We were able to add aspects like vaccination strategies, treatment pathways or spread of infectious diseases which underlines the flexibility of the implementation. Yet, a number of challenges have been identified, being the basis to contribute to the general discussion of population models. We will discuss evolving challenges according performance issues and present a newly implemented time-update approach. Thereafter we will discuss different parametrization concepts when adding a disease model. Finally we will present how we integrated GIS information based on Delauney Triangulation.

Respite care is a new service to decrease burnout risk of caregivers. Hospitalization related to caregivers burnout are costly and should be avoided. Pre-identification of caregivers with severe burnout is crucial to better manage respite care services through smart admission policies and health resources management. In this article we propose a mixed machine learning and agent-based simulation for respite care evaluation taking into account smart admission policies. Results show that neural networks approach demonstrate best results for burnout prediction and allows a significant decrease of undesirable hospitalizations when used as decision aid for admission control.

Since the referral system of the Chinese hierarchical healthcare delivery system is still developing, patients can go directly to the specialists for all outpatient care, which results in a significant imbalance of patient flow. To balance the patient flow, the general hospital (GH) and the community healthcare center (CHC) form a healthcare alliance. We propose a two-stage game-theoretic approach to study the operations of the healthcare alliance and employ simulation to analyze the revenue sharing in the alliance. In the first-stage game, two providers negotiate fixed proration rates to share the revenue from referral patients. In the second-stage game, the GH makes the capacity allocation decision and the CHC decides referral rates to maximize their own revenues. We first analyze the Nash equilibrium in the second-stage game through simulation, and then back to the simulation of the first-stage game to investigate the revenue sharing rule’s feasibility and efficiency.

During an outpatient chemotherapy infusion visit, patients typically have blood work done, see their oncologist in the clinic, wait for the pharmacy to prepare their drugs, and receive their infusion. The time variability at each of these steps can introduce delays, which not only negatively impact the patient but propagate through the system to negatively impact other patients and staff as well. One major opportunity to reduce patient delays is by pre-mixing (i.e. making drugs before the patient arrives for their infusion appointment) at the pharmacy. This, however, requires careful consideration of the trade-off between time savings versus the potential cost of wasting a drug if the patients are deemed ineligible for treatment on the day of their appointment. We present a discrete-event simulation model to predict the effectiveness of various make-ahead drug policies utilizing data from our collaborators at the University of Michigan Rogel Cancer Center (UMRCC).

Recent health care reform debates have triggered substantial discussion on how best to improve access to insurance. Colorectal cancer (CRC) is an example of a largely preventable condition, if access to and use of healthcare is increased. Early and ongoing screening and intervention can identify and remove polyps before they become cancerous. We present the development of an individual-based discrete-event simulation model to estimate the impact of insurance expansion scenarios on CRC screening, incidence, mortality, and costs. A national repeated cross-sectional survey was used to estimate which individuals obtained insurance in North Carolina (NC) after the Affordable Care Act (ACA). The potential impact of expanding the state’s Medicaid program is tested and compared to no insurance reform and the ACA without Medicaid expansion. The model integrates a census-based synthetic population, national data, claims based statistical models, and a natural history module in which simulated polyps and cancer progress.

Mass casualty incidents still cause a huge amount of deaths and injuries in the twenty-first century. Research on these events is challenging due to inherent ethical and logistical difficulties. Computer simulation models can overcome these difficulties, and offer evidence on which to base policy and decisions. In this paper a discrete event simulation model is described, designed to analyze prehospital policies and commonly made decisions. We studied an airplane crash scenario and analyzed mortality as a primary and treatment and transport times as secondary outcome measures. We implemented resource dispatching, search and rescue at the disaster site, triage, treatment and evacuation of victims to healthcare facilities. Overall we conclude that for this scenario - where treatment capacity is sufficient - the best outcome for victims can be achieved by the combination of triage and quick distribution of victims to regional hospitals.

The Standard Unified Modeling, Mapping, and Integration Toolkit (SUMMIT) is an innovative simulation platform that enables users to discover/reuse models and execute case studies to analyze emergency events. SUMMIT accelerates “what-if” analysis of disaster scenarios and improves the planning and execution of exercises. The key features of SUMMIT are in reuse of models that have been integrated into the model library, an interface that helps non-programmers to compose scenarios, ability to rapidly test and compare disaster scenarios from one location to another. SUMMIT contains a number of case files that constitute synthetic data for analysis of response and recovery operations, trends, impacts, and risks. SUMMIT supports the National Preparedness System as described in the Presidential Policy Directive 8 (PPD-8) and has been transitioned into an operational environment for use in training, exercise, threat/hazard risk assessment, resilience, and other preparedness activities.

The 1999 Columbine attack changed police response to the active shooter incidents (ASI) by the public and first responder’s tactics and training. With FBI data suggesting ASI events increasing, this study offers an AnyLogic models to understand mitigation actions such as Run.Hide.Fight. Our model represents a general densely populated area, such as public transportation terminal or indoor arena. Model agents include civilians, police, and shooter agents interact with the following parameters: civilian evacuation time, the response of police, firearm discharge by the shooter and police. The casualty rates vary from 85 to 1 causalities when the shooter’s rate of discharge was 1 to 60 seconds, respectively. The model as developed was shown to provide a method to evaluate and compare actions such as adequacy of training, introduction of technology into public buildings and the general design of public spaces to reduce the impact of ASI events.

Modern shipping ports require computer systems to accommodate an increasing number of port calls, larger vessel sizes, and tighter supply chains. Disruptions to assets on these networks have the potential to propagate to other critical infrastructures at great economic cost. Such disruptions may be introduced intentionally by adversaries that include nation states, organized crime, hacktivists, and insiders. Area Maritime Security Committees (AMSCs) must develop security plans to minimize disruptions’ impact. This paper explores one way to couple a simulation of the flow of commodities through a shipping port with an optimization that minimizes the cost of disruptions to the port transportation system. Our intent is to enable stakeholders to run what-if scenarios, to understand the impact and effect of cyber-physical disruptions, and to optimally mitigate their effect. This research, based on ongoing fieldwork with Port Everglades and the USCG, hopes to improve security policies that integrate cyber and physical effects.

Critical infrastructures such as the electric grid, oil refineries, telecommunications, transportation, water, emergency services, etc., all depend on some cyber infrastructure. Risk analysis of a critical infrastructure to compromised cyber infrastructure is therefore very important, but has certain challenges. This paper examines the problem of integrating analysis of a cyber network with the impacts possible on the critical infrastructures that are possible through cyber access. We find that adding the infrastructure impact cost considerably increases the complexity of the assessment problem.

Government agencies face the challenge of selecting and prioritizing research and development (R&D) investments to address security threats in a dynamic threat landscape. This challenge is compounded by the fact that proposed projects often span technology readiness levels (TRL), cost, and developmental timeframes. The following paper describes a probabilistic, risk-based method that assesses the future expected value of potential investments, incorporating uncertainty from the threat environment - both current and projected - and uncertainty in technology development. The developed method facilitates hedging strategies and redirection of investments at future decision points, based on shifts in threat and reduction of uncertainty; it also captures discounting of achieved value over time to balance projects that have a smaller but faster realization of risk reduction with longer-term solutions that have a large reduction. Ultimately, this approach to R&D investments produces clear, quantitative results for a balanced portfolio that maximizes expected risk reduction.

The reduction of police officers in UK forces in recent years has reduced the number of staff in police custody, hence it is imperative that staff, and other resources, are utilized appropriately to optimize the custody process. This work aims to develop a simulation model that looks at resource utilisation in police custody. Resource availability can be modelled and tested using discrete event simulation. Developing an accurate conceptual model is a key stage to link the real-world problem and the simulation model. There is minimal literature presenting an accurate police custody conceptual model and hence, that is the focus of this paper. The key sources of information for constructing this model, previous literature, custody record analysis and a custody suite visit are discussed. A final conceptual model is presented with a discussion of how this will be converted into a simulation.

This paper focuses on process analysis and computational simulation to address public health management in case of disaster response. By adopting a Business Process Management perspective, we perform organization modeling and analysis as a management tool by comparing results from agent-based and discrete event simulations. We focus on the consequences of a mass tragedy, exploiting real data from an Emergency Department after a crowd disaster where people were stamped as a consequence of mass panic. Our models consider activities with corresponding durations, resources as well as patients arrivals based on real data. Finally, once models were validated by managers, simulations can be used to provide suggestions as well as to propose different set of responses to disaster stress in emergency management.

The concept of Smart Cities is introduced to deal with the problems and challenges that come along with the growth of urbanization. Among all the challenges cities face, one fundamental concern is the safety of its citizens, especially during large sporting events. To test how smart capabilities can affect the evacuation process, an agent-based model was developed of the Georgia Tech football stadium. The model was developed to evaluate possible social behaviors during an evacuation, as well as produce the metrics of interest. To better visualize the results, a user interface was developed that includes a scenario evaluator and a decision making tool.

Horizontal collaboration (HC) is an innovative strategy aimed at reducing distances traveled through better resource utilization and consolidated product flows. Especially concerning urban freight movements, often summarized in the concept of City Logistics, cooperation between actors on the same supply chain level can be an effective concept to ensure efficiency and sustainability. The aim of this paper is to evaluate possible business opportunities for collaborative freight operation via discrete event simulation. Certain key performance indicators (KPIs) addressing both economic and environmental aspects are used in order to present and evaluate the execution of delivery tours in the city center of Athens and its outskirts. The primary data used for this evaluation are real-life data from Greek third party logistics (3PL) operators and the results show the current impact of two collaboration scenarios involving an Urban Consolidation Center.

The negative effects of urban freight transports, such as air quality problems, road congestion, and noise emissions lead in many cities to major difficulties. A widely studied measure to reduce these negative effects are Urban Consolidation Centers (UCCs), which aim to bundle freight flows to reduce the number of urban freight transports. However, many projects showed that the additional costs of UCCs often made it unattractive for carriers to participate in such schemes. This paper presents an agent-based simulation to assess the impact of urban access regulations on the cost-attractiveness of UCCs for carriers. A case study inspired by the Frankfurt Rhine-Main area is presented to compare deliveries of a group of carriers with and without a Urban Consolidation Center under various urban access scenarios. The simulation shows that regulations increase the cost-attractiveness of UCCs for carriers to varying degrees while increasing the overall traffic volume.

Probe vehicle (or connected car) data are becoming an important source of real-time travel information for a variety of intelligent transportation system applications. Since traditional sensors have significant installation and maintenance costs, technological companies are interested in traffic data from these alternative detection techniques for computing traffic-aware shortest routes. This paper analyzes and evaluates the use of data provided by probe vehicles in two reactive navigation strategies and how this affects a set of city and driver key performance indicators. The case study adopts a microscopic simulation approach to emulate real-size fleets of probe cars providing positions and speed data. The paper presents and discusses the modeling approach and the obtained results after conducting an experimental design for a Barcelona district scenario. Moreover, a simulation-based framework is introduced for simplifying the analysis of simulation results and easily visualizing origin-destination paths for the proposed driver segments (experts, regular, and tourists).

This paper presents a joint simulation project in the area of railway fleet maintenance management. A simulation model was developed to understand and visualize the complex interaction in a railway system comprising rolling stock, depots, and maintenance guidelines. In many cases, requirements on such a system come from different sides, such as fleet operation (timetables and availability), maintenance engineering (maintenance regime), and depot management (depot restrictions). The paper describes the domain-specific challenge, the model, the implemented scheduling algorithm, and resulting insights. It, furthermore, describes how the model can be used for a variety of different use cases all along a railway project: from sizing to forecast and performance analysis and from initial tender theory check to operational risk analysis. Two real-world use cases are presented: West Coast Main Line and TransPennine Express, both in the United Kingdom.

Decision makers (DMs) for logistics networks (LNWs) have the complex task of maintaining their networks in good conditions while internal and external demands are changing. Therefore, the DMs need to identify promising actions in order to adapt to the LNW’s changing state, e.g., increasing the stock level of stock keeping units (SKUs). The authors have developed a logistics assistance system (LAS) that automatically alters the LNW’s model, for improving it under changing conditions, by applying actions and evaluating their effects on the LNW’s performance. Promising actions are suggested to the DM. As the LNW grows in size, the number of potential actions increases and therefore, the response time of the LAS increases as well under the additional computational burden. In this paper, the authors describe a novel concept for reducing the number of actions by grouping similar actions together, leading to faster convergence and shorter response time of the LAS.

Continuously maintaining a logistics network (LNW) in good condition is a challenging task for decision makers. For purposes of improving an LNW’s performance, promising actions need to be identified, such as the centralization of a stock keeping unit (SKU). In order to support the decision maker, the authors have developed a logistics assistance system (LAS) based on discrete-event simulation. With an increasing size of the LNW, the response time of such an LAS increases exponentially. In this paper, the authors present an approach for utilizing domain-specific information to guide the search for promising actions and, therefore, reduce the LAS’s response time. The given examples show that the LAS’s response time can be decreased. For example, the approach reduces the number of iterations needed by an evolutionary algorithm to converge.

Port planning and management has become a difficult task due to the increasing number of elements involved in port operations, their nature (random in some cases), and the relationships between them. Generally, traditional methods involving empirical formulas or queuing theory can be useful though only for simple cases. In the case of complex systems, the problem needs to be approached from a holistic point of view, and more advanced methodologies should be considered. In such cases, simulation may be the most appropriate solution, especially nowadays, when computation and data management are increasingly more efficient. A methodology for port planning and management, based on Monte Carlo simulation, is presented in this paper. A new software application, OptiPort, has been developed based on this methodology, proposed for use by port managers and planners. The methodology and the software have been validated with real data from the Port of Algeciras.

Maritime logistics plays an important role in the global trading scene with over 80% of global trade by volume and more than 70% of the trade value being handled by vessels and seaports worldwide. Today, the maritime industry is facing both new challenges and opportunities. Amongst the existing review papers, an in-depth and systematic summary on the integration of simulation and optimization is lacking. To fill the gap, this paper reviews dozens of papers on the integration of simulation and optimization for maritime logistics since 2010. Five modes of integration are classified according to how the two techniques interact with each other. Lastly, the paper introduces new challenges and opportunities in the maritime industry, and how the integration of simulation and optimization can help to boost the development of the next generation maritime systems.

Effective traffic management can help port operators gain a competitive edge in service level and efficient use of limited resources. One critical aspect of traffic management is gate operations management, ensuring a good customer experience to logistic carriers and considering the impact of congestion in and around the port. In this paper, we describe the design and implementation of a decision support tool to help gate operators plan for future scenarios with fluctuating demand and limited resources. We propose a simulation-optimization framework which incorporates theoretical results from queuing theory to approximate complex multi-lane multi-server systems. Our major contribution in this paper is the demonstration that the proposed design, when coupled with real data, can indeed help port operators improve their performances. To provide concrete real-world evidence that such technology has benefits, we have tested the system operationally since December 2017 and present the results and analysis in this paper.

This paper focuses on the influence of different order placement behavior of users on the allocation of common resources inside a multi-user warehouse. Furthermore, the interdependencies between one user’s resource usage on other users’ order processing time is investigated. For this objective, an agent-based simulation model has been developed, depicting a rectangular warehouse with two users and one order picker. Results show that different order placement behavior and resource usage of one user have a strong influence on order processing times of other users. Furthermore, by simulating uneven order placement by one user, it can be shown that peaks in order demand influence other user’s order processing times with a delay of up to two hours after the peak occurred. Thus, the results highlight the need for coordinated order placement of partners inside a multi-user warehouse.

The Container Loading Problem (CLP) is an optimization challenge where the constraint of dynamic stability plays a significant role. The evaluation of dynamic stability requires the use of dynamic simulations that are carried out either with dedicated simulation software that produces very small errors at the expense of simulation speed, or real-time physics engines that complete simulations in a very short time at the cost of repeatability. One such engine, PhysX, is evaluated to determine the feasibility of its integration with the open source application PackageCargo. A simulation tool based on PhysX is proposed and compared with the dynamic simulation environment of Autodesk Inventor to verify its reliability. The simulation tool presents a dynamically accurate representation of the physical phenomena experienced by cargo during transportation, making it a viable option for the evaluation of dynamic stability in solutions to the CLP.

This paper presents a scalable and flexible agent-based simulation (ABS) model for the planning and control of AGVs at automated container terminals (ACTs). A comprehensive and generic multi-agent system is presented to effectively manage the container handling process between the quay and stacks of ACTs. Specifically, we design an agent-based Traffic Manager, which serves as a control application layer for all processes related to the horizontal transport of containers using AGVs. Our ABS model provides a playground to easily test out the effectiveness of different types and layouts of ACTs.

New market developments force warehouses to handle a large number of orders within short time windows. Although research optimizing storage assignment and picker routing is extensive, there remains a gap between academic research and practice. Real-life issues, such as picker blocking, high-level storage locations, precedence constraints, and safety constraints, have been considered insufficiently while optimizing order picking planning problems. This paper goes beyond the current academic state-of-the-art by showing the effects of these real-life issues on order picking efficiency and explaining the importance of incorporating these real-life issues while planning order picking operations. Multiple horizontal and vertical storage assignment policies, as well as multiple routing policies are simulated with the aim of reducing travel and waiting times. The results of a full factorial ANOVA are used to formulate managerial guidelines to increase order picking efficiency in narrow-aisle systems to address the new market developments resulting in enhanced customer service.

Shuttle Vehicle-type Mini-load automated storage and retrieval systems (SVM-AS/RSs) allow for rapid storage and retrieval, enhancing the buffering function of flexible storage and sorting operations. The systems considered in this study consist of lightweight shuttle vehicles installed at each storage level, storing and retrieving lifters, layer conveyors connecting lifters and shuttle vehicles, and incoming and outgoing aisle conveyors. First, a method is demonstrated whereby it can be determined whether lifters or shuttle vehicles are the bottlenecks in a designated system. It is then shown how simulation can be used to precisely analyze the performance of different layouts, taking storage-location allocation rules and operation priorities into consideration. This work shows that key performance indicators derived from the results of such a simulation analysis are valuable tools for the selection of the most effective and economical set of specifications for an SVM-AS/RS under given conditions of operation priorities.

Industrial laundries in Germany face a high pressure to increase their efficiency due to an ongoing market concentration. While many processes in an industrial laundry are already highly automated and optimized, order picking is still done manually. As RFID proliferates in industrial laundries, goods become traceable and identifiable. This enables the effective use of automated picking systems. Automation systems in the laundry industry face the challenge of pliable (form unstable) goods. This paper shows how the planning and dimensioning of an automated Laundry Order Consolidation System (LOCSys) can be supported by simulation modelling, already beginning in the early conception stage of the system.

Few problems in Operations Research are regarded as highly as the Vehicle Routing Problem (VRP). Its relevance within management and industrial settings has led to the variants of this problem being widely studied by the scientific community. With the aim of solving the VRP with stochastic demands we analyze an extension of the classical GRASP metaheuristic. This work hybridizes a biased-randomized GRASP (BR-GRASP) with a two-stage Monte Carlo simulation which has the ability to attain robust and competitive solutions. In the first stage, a promising set of local optimum solutions is identified based on a short simulation evaluation. In the second stage, the promising solutions are tested for reliability using a larger number of simulation runs. The most reliable solution is the final solution. Experiment results are provided that demonstrate that the proposed integrated algorithm leads to higher quality solutions than the equivalent approach without such an integration.

Multi-constrained Vehicle Routing Problems are gaining steadily in importance. Especially, the dynamic version of the problem has become more emphasis due to modern service requirements, such as short-term or express delivery. With a growing number of dedicated solution approaches for these problems, we investigate a simulation-based supervised learning approach to determine the suitability of a particular algorithm from a set of algorithms for a given dynamic problem instance based on a variety of its characteristics. This decision is known as the Algorithm Selection Problem. We explore the performance space for Greedy and Re-planning algorithms for different dynamic problem instances by simulation and an evolutionary algorithm. For the algorithm selection we test several problem features in combination with two supervised machine learning techniques. The applicability of our approach is demonstrated in a use case for autonomous algorithm selection for Dynamic Vehicle Routing Problem instances.

In real-life logistics, distribution plans might be affected by weather conditions (rain, snow, and fog), since they might have a significant effect on traveling times and, therefore, on total distribution costs. In this paper, the distribution problem is modeled as a multi-depot vehicle routing problem with stochastic traveling times. These traveling times are not only stochastic in nature but the specific probability distribution used to model them depends on the particular weather conditions on the delivery day. In order to solve the aforementioned problem, a simheuristic approach combining simulation within a biased-randomized heuristic framework is proposed. As the computational experiments will show, our simulation-optimization algorithm is able to provide high-quality solutions to this NP-hard problem in short computing times even for large-scale instances. From a managerial perspective, such a tool can be very useful in practical applications since it helps to increase the efficiency of the logistics and transportation operations.

In the context of smart cities, unmanned aerial vehicles (UAVs) offer an alternative way of gathering data and delivering products. On the one hand, in congested urban areas UAVs might represent a faster way of performing some operations than employing road vehicles. On the other hand, they are constrained by driving-range limitations. This paper copes with a version of the well-known Team Orienteering Problem in which a fleet of UAVs has to visit a series of customers. We assume that the rewarding quantity that each UAV receives by visiting a customer is a random variable, and that the service time at each customer depends on the collected reward. The goal is to find the optimal set of customers that must be visited by each UAV without violating the driving-range constraint. A simheuristic algorithm is proposed as a solving approach, which is then validated via a series of computational experiments.

Supply network design is an important strategic operations management problem. The paper discusses the use of simulation-based optimization for solving this task in a performant manner. A three-tier hierarchical approach is presented: On the first tier, a multi-population genetic algorithm decides about the active nodes of the network and assigns product groups to nodes. On the second tier, the production capacities and sourcing paths are determined. On the third tier, a simulation model calculates an objective function value that describes the quality of the network design. While the genetic algorithm acts as a central control instance, the evaluation of the individuals on the second and third tier are parallelized in a computer network. In the course of the optimization process, partial solutions are saved and reused in order to reduce computational cost. The approach is tested on a real-world example of a European supply network.

Optimization problems arising in real-life transportation and logistics need to consider uncertainty conditions (e.g., stochastic travel times, etc.). Simulation is employed in the analysis of complex systems under such non-deterministic environments. However, simulation is not an optimization tool, so it needs to be combined with optimization methods whenever the goal is to: (i) maximize the system performance using limited resources; or (ii) minimize its operations cost while guaranteeing a given quality of service. When the underlying optimization problem is NP-hard, metaheuristics are required to solve large-scale instances in reasonable computing times. Simheuristics extend metaheuristics by adding a simulation layer that allows the optimization component to deal with scenarios under uncertainty. This paper reviews both initial as well as recent applications of simheuristics, mainly in the area of logistics and transportation. The paper also discusses current trends and open research lines in this field.

This paper describes the design and application of a model simulating the passenger flows on the high-speed rail line Rail Baltica. The proposed model allows one to define passenger flow service indicators, according to data on the behavior of potential passengers that take the train in different localities along Rail Baltica. If corresponding baseline data is provided and the duration of the longest trip will not exceed one day, the model can be used to study processes on other railways with a similar structure. The developed model simulates movements and accumulation of passengers at stations and in trains on four sections connecting five cities along Rail Baltica. Besides that, the model allows for studying particular scenarios related to large sports or other events that will take place near one or several cities through which the railway passes.

By outsourcing major aircraft systems to Tier 1 suppliers, original equipment manufacturers depend heavily on their supply chain to meet the growing demand for aircrafts. However, capacity constraints upstream of the Tier 1 suppliers increase the risk of schedule disruption. Discrete event simulation is commonly applied to analyze capacity constraints in manufacturing systems while analytical models assess financial investment scenarios for capital equipment. This paper demonstrates a combined simulation and analytical modeling approach to simulate the operational and financial implications of capacity constraints in aerospace supply chains. A three-tier supply chain is modeled with the option of investing to remove a capacity constraint in a sub-tier supplier. The results demonstrate how a supplier’s capacity investment decisions and the increasing demand for aircrafts can affect their cash flow and delivery schedule adherence.

Snowplow operations are critical for public safety and economic success in countries where difficult driving conditions occur in winter. Specifically, the salt supply ensuring good driving conditions is a crucial factor. In this paper, the strategic supply chain design of a winter service provider in Austria is investigated. Two research directions on the influence of bigger and fewer salt silos per depot and the logistic costs for a unique summer salt purchasing strategy are addressed applying two independent solution approaches. On the same data basis, a simulation model is developed and a mixed integer linear problem is applied to answer the respective research questions. The first study shows that the current depot availability is quite good but that bigger and fewer salt silos per depot could be a risk. Finally, the second study shows the logistic costs for the unique summer salt purchasing strategy and the optimal salt-warehouse locations.

This paper describes the simulation modeling of bulk petroleum supply chains through an object-oriented simulation library. The paper focuses on the conceptual modeling that is useful for resource and transportation capacity analysis during contingency and war gaming analysis. In addition, the modeling provides the capability to understand how surges in demand will affect inventory service. The methods, insights, and capabilities developed within this research facilitate the analysis of bulk petroleum military supply chains and also permit the analysis of the resilience of commercial bulk petroleum supply chains under conditions of disruption.

Medical device companies have a growing interest in additive manufacturing (AM) for orthopedic implant production. AM provides an opportunity for economically feasible customized implants to supplement the current traditionally-manufactured, standard-sized prosthesis. In this paper, a simulation model is developed to analyze the supply chain of custom hip stem production using AM. A test case is implemented into the simulation model to determine the supply chain’s responses in key performance indices, such as production lead time of a uniquely customized prosthesis and resource utilization, to resource level variations based on current state of the practice for additive and subtractive manufacturing. The method of this research was to use a simulation focused on measuring the performance of the supply chain for hip stem implants in order to determine supply chain resource requirements.

A main characteristic in the field of supply chain simulation is that typically several independent organizations are involved. However, most simulation studies only consider a very limited number of different organizations. One main reason is that suppliers usually do not want to publish sensitive data, such as resource capacities or cost rates to their customer. In this paper we present an overall architecture to tackle this problem of multi-organizational simulation studies storing all relevant data at an independent third party, which also carries out all experiments and distributes the results as a cloud service. In order to define scenarios, different rights to use the provided data can be granted to other participants. The necessary user-specific encryption of datasets needs to be established on the underlying data base structure. In this paper we focus on three approaches for storing and encrypting data along with a detailed comparison in terms of performance.

We consider a pharmaceutical supply chain where the manufacturer sources a customized product with unique attributes from a set of unreliable suppliers. We model the likelihood of a supplier to successfully deliver the product via Bayesian logistic regression and use simulation to obtain the posterior distribution of the unknown parameters of this model. We study the role of so-called input-model uncertainty in estimating the likelihood of the supply failure, which is the probability that none of the suppliers in a given supplier portfolio can successfully deliver the product. We investigate how the input-model uncertainty changes with respect to the characteristics of the historical data on the past realizations of the supplier performances and the product attributes.

We consider a discrete-event stochastic simulation that represents a generic manufacturing process flow and assume the availability of limited knowledge about part inter-arrival times and machine processing times. We investigate how the state-of-the-art simulation methodologies can be utilized to propagate input inter-arrival-time and service-time uncertainty through such a manufacturing simulation. We quantify the impact of limited knowledge on steady-state manufacturing-line performance measures such as utilization, lead time, inventory, and throughput. First, we conduct experiments for single-stage manufacturing systems where we additionally study the impact of yield loss uncertainty on annual throughput predictions. Then, we switch our focus to multi-stage manufacturing systems and investigate whether there may be situations in which it becomes difficult to correctly identify the system bottleneck. We conclude with the identification of the manufacturing system stages that contribute most to the variability in lead time and inventory predictions.

Inspired by an application in the field of on-demand public transportation, we perform a Monte Carlo simulation study on the probability distribution of the length of Traveling-Salesman-Problem (TSP) tours between small numbers of random locations. We consider a fixed convex region, where we generate a fixed number of random locations from a known probability distribution and find the corresponding euclidean TSP tour for them. We simulate this process extensively and perform both quantitative and qualitative analyses of the resulting experimental distribution for the TSP tour length. We show that, under certain assumptions on the shape of the region and the probability distribution of locations, the length of the TSP tour is well-approximated by a normal distribution, even for as few as five locations. Furthermore, we propose experimental models for estimating the mean and standard deviation of the tour length.

In this paper we present a discrete event simulation model of a cruise terminal for decision support and strategic planning. The main objective is to provide an analysis tool for systems with high capacity constraints, heterogeneous subsystems, time-dependent demand and stochastic passenger transfer and travel times. Cruise terminals are connected by bus to a transport interchange or intermobility center where passengers are transferred to taxis. All terminals are affected by capacity constraints such as the number of bus platforms for passenger pick-up and drop-off and the limited queuing capacity for taxis and passengers. Passenger arrivals are time-dependent and cruise-type-dependent. The purpose of this model is to support the strategic decision-making process through what-if-scenarios. Decisions involve determining the required number of bus platforms, taxi buffer sizes and passenger distribution policies, among other factors. The results identified and proposed the solution of bottlenecks and enabled a sensitivity analysis to be performed.

Car rental companies have to operate the fleet of vehicles considering a cost effective fleet sizing and rebalancing of the vehicles. Additionally, they have to ensure a good vehicle availability at the rental stations. We present a simulation model combined with optimization and queueing network analysis for mobility-on-demand networks with one-way car sharing. The system is modeled as a closed queueing network. On the one hand, this allows an estimation of the vehicle availability; on the other hand, constraints for the optimization model can be derived from the mathematical equations of the queuing network model. To optimize the total revenue, the optimization model considers price incentives for customer trips and cost of empty trips resulting in a mixed-integer linear programming problem. The simulation uses the optimization component for an optimized operation of the mobility-on-demand networks and computes the optimal fleet size, the cost of empty runs, and the vehicle availability.

The car rental business has experienced recent growth under a fiercely competitive market. This paper describes a generic discrete event simulation tool for studying the complex nature of a car rental company’s operations. To construct the tool, we conducted several interviews with experts to first establish the proper requirements, after which we collected relevant data for populating the simulation model, thereby determining the best strategies for car class pricing and upgrade offers that would result in the highest revenue. A realistic case study was conducted to assess the tool and gain a better understanding of the underlying system. Finally, simulation experiments regarding rental prices and car upgrades were designed, and the system was run, achieving a 0.5% error for various performance criteria. For the given case, we found that two upgrade offers and setting prices slightly higher than the average customer budget provides the best strategy.

The setup process, Numerical Control (NC) program configuration and the linked configuration of point of origins, workpiece position, tool ranges require high computational effort that include multiple simulation runs during the work preparation process. This contribution describes an automatic setup optimization process, including validation of position parameters using a virtual tooling machine as simulation model. In the first step, the developed simulation-based optimization approach minimizes the production time while the collision information and NC program validation are provided by the simulation. In the next step, a cluster method is applied to avoid a high number of single simulation runs, but the validation effort is still high. In order to address this point the developed system offers a selection and data reconciliation procedure using supervised learning methods to determine feasible workpiece positions.

Kanban systems constitute a widely used pull control for inventory management in production systems. As a result of an increasingly volatile and individualized customer demand, Kanban systems have to be reconfigured dynamically to achieve minimal inventory levels while maintaining a stable production. This paper investigates the application of an incremental online learning platform called XELOPRO to optimize inventory levels using the current state of the production system, while including contextual information, e.g., time-related information. As the platform uses an incremental support vector machine to update its models during runtime without the need to store and reevaluate large amounts of historical data, it constitutes a suitable tool for a decentralized inventory management. Results show a good performance with drastic decreases in inventory levels compared to static configurations and a higher reliability compared to a dynamic application of standard Kanban rules.

The paper describes a methodology for layout and intralogistics redesign using simulation. This methodology is composed of five stages: Topography, Product & Process, Linkages, Simulation, and Results. Key foundations are defined as PFEP (Plan for Every Part), Workstation as basic object for simulation, high level script language, cyclic processes, and 3D core simulation. The simulation model is built based on an earlier-prepared AutoCad drawing as background and on data in Excel about the Bill of Materials, PFEP, Container List, Assignment, and Operations. The simulation model is built using the developed template, which significantly shortens the modeling time.

Today, delivery of industrial maintenance services is planned under uncertainty, as, for example, the true task duration is unknown. Thus, providers rely on estimates at the stage of planning. During operations, the true task duration becomes known which results in a schedule update. This work evaluates the effects of information uncertainty on business-relevant performance indicators in industrial maintenance. We present a simulation model to reflect the evolution of information and the schedule revision during operations. Within a simulation experiment, we measure the business-relevant impact of different uncertainty levels on operations. Indeed, the findings of our research confirm the common understanding that reduced uncertainty at the stage of planning has a positive effect on service delivery. However, our research also shows that in order to achieve business-relevant impact in service delivery quality, the reduction of uncertainty at the stage of planning needs to reach a certain threshold.

Most revenue management publications dealing with substitutable products in a manufacturing context have focused on pricing issues. They consider that substitution is a customer's decision which occurs as a response to product price differences. In our study, substitution is considered as the firm's policy. We focus on the extension of the revenue management to practical applications in manufacturing and we are motivated by the Canadian softwood lumber case where product substitution is a common demand fulfillment practice. We aim, first, to propose a generic consumption model integrating both capacity control and product substitution decisions and second, to evaluate, using a rolling horizon simulation, the performance of this integrated model in different settings compared to common demand fulfillment approaches. In addition to practical implications, our study contributes to the existing demand fulfillment literature since we simulate different consumption models integrated with a Sales and Operations Planning model.

A big challenge for manufacturers today is to create a flexible and efficient production system. One way of managing this challenge is to establish a virtual factory, a virtual model of the production unit. Working smarter by using the advantages that digitalization implies enables production of personalized products at increasing speed. This paper explores how to implement such a concept by stepwise increasing the maturity of the virtual factory. Evaluated at a large-scale Swedish manufacturer, local needs and enabling technologies benchmarked at industry leaders have been identified and strategically mapped to their corresponding maturity step. This paper shows that the implementation of a virtual factory relies on standardized work procedures, ensuring its use as a decision aid throughout the company. Implementing a virtual factory in this manner will facilitate user-driven development and more accurate decision making, generating support for efficient production systems.

Industry 4.0, a part of the German high tech strategy, prefers self-organization in production over central production planning for the sake of greater flexibility, faster response to disruptions and to deviations, and less effort. Current planning systems usually plan centrally. We developed a universal self-organizing production and empirically compared its performance to a centrally planned production. The self-organizing production is based upon agents. For better comprehensibility, we additionally implemented central planning. The results of self-organization in production are promising in relation to central planning; especially when disruptions and deviations occur.

The aim of this contribution is to provide a modeling approach for multi-project manufacturing. Existing approaches for capacity planning and detailed scheduling in those complex environments are using separate models that are linked by instruction and feedback slopes. In contrast, we include multiple levels hierarchically into one simulation model. Thus, the possibilities to propagate restrictions such as precedence dependencies and starting times from gross to detailed levels in a consistent manner are established. The approach is implemented in Java, extending the modeling capabilities of a simulation-based optimization framework. A real-life application to project-oriented aircraft maintenance is presented to highlight the practicality and efficiency of the integrated model for simulation-based capacity planning using work packages as well as detailed scheduling using activities of work plans.

Simulation methodologies are developed to model the joint optimization of preventive maintenance and spare part inventory for a specific industrial plant under different production configurations. First, spare part provision for a single-line system is considered, with the assumption that the demand is driven by maintenance requirements. The results indicate that a periodic review policy with replenishment as frequent as inspection is cost-optimal. Second, the joint optimization model for a multi-line (parallel) system is developed. It is found that a just-in-time review policy with inspection as frequent as replenishment produces the lowest cost policy. In this latter case, an implication of the proposed methodology is that, where mathematical modeling is intractable, or the use of certain assumptions make them impractical, simulation modeling is an appropriate solution tool. Under both production settings, the long-run average cost per unit time is used as the optimality criterion for the comparison of several policies.

In recent years, environmental aspects became one of the key interests in manufacturing. Accordingly, simulation studies had to include factors like energy or emissions. This paper aims to provide a comprehensive introduction to the state of the art in modeling of energy and emissions in simulation of manufacturing systems. We review existing literature to develop a landscape of common approaches and best practices. Typical goals and objectives of the reviewed simulation projects are summarized. Furthermore, we will evaluate the structure and life cycle phases of the examined manufacturing systems and look into the requirements and implementation of respective simulation studies. Finally, we will discuss open questions and future trends in this field of research.

The increasing reliance on volatile renewable energy sources in the European Union raises questions regarding the future mechanisms of the energy markets. Energy-intensive production industries are particularly expected to take a more active role by shaping their energy demand according to the availability of wind and sun. Hence, they will need to align their production processes with external energy market signals. This paper presents an application example for a Siemens Plant Simulation extension that makes holistic material flow and energy flow studies of factories possible. The so-called eniBRIC class library provides functionalities for investigating the flow of energy between infrastructure and production equipment. Since its latest update, it can also be used to model renewable energy sources as well as energy storages. A case study of the E³-Research Factory showcases the features of eniBRIC and provides an outlook on future research in the field of energy-flexible production.

Human-Robot Interaction (HRI) systems were recently introduced in the manufacturing industry. Robots acting autonomously and working in a direct collaboration with humans create a hazard potential, which is influenced by human factors in complex working situations. In this paper, a concept of a full-scope simulator for HRI applications is introduced, followed by a pilot installation of the simulator. Different environmental influences can be created and repeatably reproduced in an experimental environment within the simulator. The HRI full-scope simulator works completely autonomous and enables comprehensible and repeatable ergonomic proband experiments for an optimal design of HRI systems with regard to human factors and allows for statistical conclusions, e.g., about the human situational awareness.

At the 2001 Winter Simulation Conference, the progress of simulation in Ford Motor Company’s PowerTrain Manufacturing Engineering (PTME) department was documented in a paper that focused on the contributions of the department in changing the simulation environment at Ford. This paper reviews the progress, changes, and issues in the intervening years experienced by the UK-based PTME simulation team. It summarizes the development of a toolset from a model building capability to data analysis, to experimentation, and results analysis. It outlines how capabilities have expanded while maintaining quick delivery of results; it references the managements' changing attitude and how academic research has advanced simulation in PTME.

In the last years, the increase of data availability together with enhanced resource flexibility shed light on the possibility to develop planning and control methods with real-time inputs. Literature is rich of approaches to simulate, to quickly evaluate system performances, and to take decisions based on optimization criteria. Further, simulation has been identified as one of the pillars for the Industry 4.0 revolution. However, the lack of a generally recognized approach and methodology to deal with real-time decision-making through simulation is evident. Simulation approaches can and should play a central role in industry for the years to come. This position paper analyses the current research context with a brief state of the art on existing approaches, includes considerations about the issues for implementing Real-Time Simulation (RTS) concepts and their current state of development. Finally, it outlines research directions for the simulation community.

Real-time batching in job shops decides 1) whether to start processing a batch or to wait for more jobs joining the batch, and 2) which batch should be processed first. It is addressed as a sequential decision-making problem and formalized based on Markov decision processes. By adding a dummy batch, which means no batches are selected and all batches wait for additional jobs, the first decision-making is generalized to the second. A simulation-based neural fitted Q learning is introduced to solve the Markov decision processes and build a decision maker. The well-trained decision maker decides which batch in a batch list should be processed first at each decision epoch. The experiment results show that the proposed approach outperforms some other decision rules.

Modern companies have to face challenging configuration issues in their manufacturing chains. One of these challenges is related to the integrated allocation and scheduling of resources such as machines, workers, energy, etc. These integrated optimization problems are difficult to solve, but they can be even more challenging when real-life uncertainty is considered. In this paper, we study an integrated allocation and scheduling optimization problem with stochastic processing times. A simheuristic algorithm is proposed in order to effectively solve this integrated and stochastic problem. Our approach relies on the hybridization of simulation with a metaheuristic to deal with the stochastic version of the allocation-scheduling problem. A series of numerical experiments contribute to illustrate the efficiency of our methodology as well as their potential applications in real-life enterprise settings.

The highly competitive automobile market requires automotive companies to become efficient by continuously improving their production systems. This paper presents a case study where simulation-based optimization (SBO) was employed as a step within a Value Stream Mapping event. The aim of the study was to promote the use of SBO to strengthen the continuous improvement work of the company. The paper presents all the key steps performed in the study, including the challenges faced and a reflection on how to introduce SBO as a powerful tool within the lean continuous improvement standards.

The main focus of the research presented in this paper is to propose new methods for filtering and cleaning large-scale production log data by applying statistical learning models. Successful application of the methods in consideration of a production optimization and a simulation-based prediction framework for decision support is presented through an industrial case study. Key parameters analysed in the computational experiments are fluctuating reject rates that make capacity estimations on a shift basis difficult to cope with. The most relevant features of simulation-based workload estimation are extracted from products’ final test log, which process has the greatest impact on the variance of workload parameters.

Modern semiconductor foundry business involves dozens of different products sharing the same production line. With each of them having distinct flows with potentially different mask layers, process steps and dedicated tools, it is not easy to get a good overview. Furthermore, analyzing simulation results is difficult because the flow of material on the shopfloor inside the simulation is difficult to understand. We describe a dynamic approach of automated, interactive, aggregated flow visualization based on nodes and edges graphs, augmented by current state and simulation data like waiting and upcoming material, tool states and line holds. This allows for a profound analysis of the current state of the production line and the material flow anticipated by the simulation considering all constraints.

We consider the dynamic price and lead time quotation problem in the practical context of the semiconductor industry. Our model considers an inventory decoupled supply chain and accounts for a limited capacity, stochastic demand and processing times and quote-sensitive customers. We focus on performance evaluation under two decision making strategies. The first is lead time based pricing (LTBP). It follows a sequential approach where the firm decides first on the lead time quote (manufacturing) and then quotes the price under the given lead time (marketing). The second strategy suggests determining the lead time and the price quotes simultaneously. From the practical view-point, it is interesting to first understand the system performance under LTBP and then look for the ways to realize it. Based on our numerical results, we elaborate on the effect of LTBP on the key performance indicators and discuss conditions for close performance to a simultaneous decision strategy.

In the semiconductor manufacturing literature, production planning models mainly aim at minimizing production, inventory and backlog costs. Solving these models may lead to a poor utilization of the production capacity when there are not enough demands. In this paper, after presenting a first generic linear programming model with fixed lead times when total costs are minimized, a model where productivity is maximized is introduced. Then, a model where the total profit is maximized is proposed, which considers the net present value of the profit through an actualization rate. These models are then compared using a data set from the literature. The numerical results show that, although the model with productivity maximization is performing as expected, the model with profit maximization is more relevant since it also helps to increase productivity. The impact of the actualization rate is analyzed, and also the limitations of the production of some products.

Due to the significant rise in environmental awareness of companies and customers for the past few years, research on how to optimize business with respect to carbon dioxide (CO2) emission has gained more attention and importance. This paper investigates how flexibility can be an enabler for CO2 reduction over a global production network especially in a capital intensive and high volatile market like the semiconductor one. We tested this hypothesis with discrete-event simulation experiments based on a case study obtained from a semiconductor company. The study indicates that global supply chains (SCs), like those in the semiconductor industry, should be equipped with a certain level of flexibility to cope with demand volatility if the CO2 burden due to transportation is low compared to those due to manufacturing. This flexibility provides ecological benefits to companies in reducing the carbon footprint of their products.

We discuss a bicriteria scheduling problem for parallel identical batch processing machines in semiconductor wafer fabrication facilities (wafer fabs). Only jobs that belong to the same family can be batched together. The performance measures of interest are the total weighted completion time and the electricity power cost. Unequal release dates of the jobs are taken into account. The jobs can have non-identical sizes. We provide a Mixed Integer Linear Programming (MILP) formulation for the general setting. Moreover, we analyze the special case where all jobs have the same size, the maximum batch size is an integer multiple of this job size, and all jobs are available at time zero. We prove certain properties of Pareto-optimal schedules for this special case. These properties lead to a MILP formulation that is more tractable than the one for the general setting. We perform computational experiments with the εconstraint method for both formulations.

In this work, we propose a re-enactment simulation-based optimization method to determine the minimal total buffer capacity in an assembly line required to meet a target throughput. A distinguishing feature is the use of real-time event traces, in a fast fluid flow simulation model. Employing real-time event traces avoids the necessity to make restrictive modeling assumptions. The fluid simulation is combined with a multi start search algorithm. To demonstrate its effectiveness, the method is applied to a real-world use case in lead frame based semiconductor back-end manufacturing. This use case considers an assembly line consisting of six machines, for which the proposed method determines optimal buffer size configurations within several minutes of computational time.

Process and data are equally important for business process management. Data is especially relevant in the context of automated business processes, and process controlling. In the context of knowledge-intensive domain, data modified by engineers and used by a fully automatic manufacturing system can lead to unpredictable errors. In this paper, a method permitting the discovery of truly enforced business rules used by the process is proposed. These rules will be highlighted using data from the model of the studied process. The proposed methodology is mainly based on a data mining approach. The proposed method has been tested on data from semiconductor industry, in which processes are known to be complex, and the number of business rules is known to be important. The results show that the method is efficient in assisting engineers in process errors detection and practical process improvement.

Vendor managed inventory (VMI) is a widespread supply chain collaboration model. To avoid product shortage and excessive stocks, various measurement approaches defining minimum and maximum inventory levels have been designed. None of the known approaches take up the aspect of responsibility regarding the status of the inventory levels. This is especially important in industries facing long process times, short product life cycles and volatile demand patterns, such as the semiconductor industry and their downstream customers. Within this work, a VMI classification checklist is established to build a basis for a common understanding of this collaboration model. This is followed by a generic VMI measurement approach including a process to assign responsibilities for poor performance. The novel process distinguishes from other measures, as it is of a bilateral nature and parameters can be fitted uniquely to the needs of different VMI partnerships.

The Infineon Dresden 300mm fab is the first wafer fab in the world to run high volume production of power technology products. This requires the coordination of multiple production facilities in the supply chain. Output from one facility is fed as an input to the subsequent facilities in a push or pull operation mode, depending on technology lines. In a push operation mode, lots transition from one facility to another in a continuous manner, while in the pull operation mode, lots are stored and continuation of the lots is triggered by customer order to the supply chain. The latter case triggers the need of synchronizing material flows from one facility to another, and thus requiring an accurate material forecast for better planning and execution. In this paper, we discuss the modelling issues and solutions associated with this challenge, and the use cases of the simulation forecast.

The introduction of a new product into a semiconductor manufacturing facility can have significant adverse effects on both current and new products. We combine a simulation model of a simplified production system and a learning model to optimize releases of old and new products using simulation optimization. Our results show that the simulation optimization model yields significant improvements in performance over simple alternatives, and provides insights into the structure of optimal release policies.

Production planning models face the fundamental problem of capturing the nonlinear relation between resource workload and cycle times. One approach to this has been the use of nonlinear clearing functions that represent the expected output of a production resource in a planning period as a function of the expected workload in that period. Recently an alternative data driven approach that represents the behavior of the system using a set of system states and their corresponding output levels has been proposed. We compare the clearing function based approach to the data driven approach using a simulation model of a scaled-down semiconductor wafer fabrication facility and discuss the strengths and weaknesses of the two approaches.

In this paper, we address the question of how to maximize output over the ramp-up period of a semiconductor fab. We postulate that by keeping a margin between the constraint capacity and the ramp starts cumulative output can be increased over a desired planning horizon (e.g. first year of production). A mathematical framework of the problem is offered and then used to prove our hypothesis. We demonstrate the benefit via a numerical example and a set of simulation experiments.

Advanced transport scheduling for conveyor-based automated material handling systems (AMHS) in semiconductor fabricating facilities (wafer fabs) can reduce transport times and thus cycle times. Commonly, transport operations of arriving wafer lots at conveyor junctions, such as rotary tables, are sequenced ad-hoc by myopic first-come-first-serve policies. In contrast, improved transport schedules for the transport operations can be produced ahead of the time in conjunction with the overall production schedule for process operations. More precisely, such a transport schedule can either be sequentially constructed by fixing one transport after another, or it can be optimized by simultaneously fixing several conflicting transports. Hence, two conceptually different transport scheduling methods, which both avoid congestions by enforcing a no-wait constraint, are compared with special regard to transport-related variability. Furthermore, three different AMHS models that exhibit the typical interbay-intrabay layout are used for computational experiments.

We discuss a qualification management problem arising in wafer fabs. The steppers need to be qualified to process lots of different families. A qualification time window is associated with each stepper and family. The time window can be reinitialized as needed and can be extended by on-time processing of lots from qualified families. In this paper, we propose a hierarchical approach for this problem. The base-level is a dispatching strategy that takes into account qualification decisions. The medial-level consists of a mixed integer linear programming formulation for making qualification decisions, whereas the top-level is a linear programming (LP) formulation that computes target quantities for the steppers for a planning window taking fab-wide objectives into account. We present results of simulation experiments where the hierarchical approach is applied in a rolling horizon manner. The results demonstrate that the LP-based approach outperforms a heuristic to determine target quantities.

This research is motivated by the expensive costs of scraps because of timelink misses in a semiconductor manufacturing line with tool downs. A timelink is a time constraint between defined process steps. This paper presents a mixed integer programming model (MIP) and an constraint program (CP) with downscaled time constraints. With the assistance of the survival analysis, a safety value will be computed and included as a constant in the MIP and as a dynamic expression in the CP, to downscale the allowed time between two specific operations. The MIP and CP models are tested on a realistic production area example with different problem sizes. The quality of the solution and the performance of these two approaches are compared with each other. The test results show that the CP Model outperforms the MIP and finds much earlier usable schedules for large problem sizes.

Cluster tools are closed mini production environments that are especially used in the semiconductor industry. They consist of multiple processing chambers. In contrast to single processing tools they can handle multiple lots with different process characteristics at the same time. The major challenge for scheduling cluster tools is the hard to predict process times of even comparable lots as they strongly depend on a changing product mix over the tool. In this paper a mathematical model for the external scheduling of a cluster tool workstation is presented. The goal is to minimize the weighted cycle time by accounting for the changing cycle times of each lot. A MIP model assigns lots to tools of a workstation and determines for each lot the chambers, which are used during the processing steps inside the cluster tool. Finally the developed method is compared with a simulation that uses dispatching strategies.

Most manufacturing systems in which jobs arrive dynamically and their processing times have variations use dispatching rules to obtain production schedules. In LCD manufacturing, several dispatching rules that reflect the knowledge of the fab operator have been developed and prioritized to select a unique job for processing on a machine. However, engineers rank the dispatching rules based on their experiences without any systematic analysis method. Hence, there is a great need for a tool that can analyze how the order of dispatching rules affect the key performance indicators (KPIs) of schedules. Therefore, we provide a framework for the performance analysis of dispatching rules so that engineers can examine the KPIs for a given order of dispatching rules and find the best order of dispatching rules.

In this paper, we study the problem of considering the internal behavior of complex machines when solving complex job-shop scheduling problems encountered in semiconductor manufacturing. The scheduling problem in the diffusion area is presented, and the complex structures and behaviors of the different machine types in this area are described. Previous related research is reviewed, and our approach to consider the complexity of these machines with batching constraints when scheduling the diffusion area is described. The main part of the paper presents and discusses numerical experiments on industrial instances to show the benefit of choosing a suitable modeling for complex machines.

This presentation will discuss requirements for reaching the long-standing goal of designing a reference model for planning and control functions in semiconductor supply chains. A recently proposed simulation testbed for semiconductor supply chains will be described as an intermediate step towards reaching this goal. Some applications of the testbed will be presented. The discussion of the testbed will be related to recent initiatives for replicated computational results and in more general terms to reproducibility of scientific results and open research efforts.

For the past decades, Virtual Metrology (VM) has been widely studied and covered in the literature for semiconductor industries where cycle time is a critical aspect and the elimination of non-productive metrology measurements is expected to significantly contribute to its reduction. A wide variety of approaches has been proposed but not effectively implemented. An ideal VM model should be able to provide accurate predictions and to reveal the hidden relationship among production/process factors. For this aim, we employ the Gaussian Bayesian Network (GBN) to investigate the implicit relationship not only between the metrology and the control factors but also among the production/process parameters. Instead of working purely as a black-box data-driven methodology, GBN enables the flexibility to integrate domain knowledge through the corresponding connected graph. The effectiveness of proposed approach is validated using real industrial data from a Chemical-Mechanical Polishing (CMP) process.

Achieving desired yields in advanced manufacturing industries, as is the case for semiconductor fabrication, requires perfectly matching the performance of parallel machines/chambers at all production steps. Moreover, in the high-mix/low-volume manufacturing environment of the IC makers, the need for highly precise production processes makes the task of matching identical machines/chambers more and more complicated and challenging. In this paper, a methodical approach to deal with machine/chamber mismatching using all available data such as sensor data, product measurements, and maintenance log is proposed and validated with real practices. In fact, after detecting and identifying the key variables source of variance in the process, a modeling step known as Virtual Metrology is performed to quantify accurately the recipe adjustments i.e., R2R control, which will match as much as possible both sensor data and metrology measurements between similar machines/chambers.

This paper proposes a performance indicator based on the Bayesian theory. This indicator is used for assessing the behaviour of Run-to-Run controllers. The indicator is calculated by analyzing four main points: the output/target error, the output dispersion, the out of tolerance (oot) rate and the industrial risk. The proposed Bayesian method has been tested on the Run-to-Run loops of a semiconductor manufacturing foundry.

The semiconductor sector is undergoing one of the fastest market growths. Demand is increasing and market forecasts are optimistic. New markets are emerging and product portfolios are broadening significantly. Dynamic supply chains are developing with increasing number of customers, products, suppliers and manufacturing partnerships. Up to now due to modeling complexity and computation time constraints, disjoint systems are used for local supply chain control and optimization. For efficient control, these complex semiconductor supply chains require a global approach for simulation and optimization. This is the motivation for this conceptual paper. In the conceptual part this paper introduces and discusses novel model aggregation approaches and model validity improvements with novel hierarchical modeling concepts. In the modeling part this paper incorporates that into an aggregated simulation model approach as well as a novel hierarchical interface concept for coupling of disaggregated analytical and simulation models to systematically improve overall model validity at Bosch.

Semiconductor manufacturing systems, with several hundreds of different production steps, reentrant loops, cleanroom conditions and a job shop organization, are probably the most complex production systems and so is the manufacturing data structure. This high complexity requires precise planning, modelling and simulation to improve manufacturing transparencies and avoid planning inefficiencies. Although many modelling approaches for semiconductor production systems have been published, much less work has been conducted to define a common data model that supports the industrial complexity. This paper presents a common conceptual data model for semiconductor supply networks, developed in close collaboration by three European semiconductor manufacturers. Described semiconductor manufacturing use cases motivate the proposed conceptual data model. This model includes master entities, defining the structure of semiconductor supply networks with common definitions of key elements, and tracing entities, describing time dependent changes and events of the system. Expected benefits of the conceptual data model are also discussed.

The development of Industrial Cyber-Physical Systems (ICPS) requires new ways of collaboration between ICPS vendors and companies which provide them with parts and components. As ICPS extend existing systems, e.g. machines or healthcare equipment, by electronic components, new supply chains will evolve between companies which were not connected before. Hence, traditional methods for collaboration will not fit for them anymore. We propose a new approach for collaborations in ICPS development. It relies on technologies from the Semantic Web and consists of a web platform where ICPS vendors and vendors of electrical, electronical, information and communication technology components can virtually meet. Component vendors can describe their products, ICPS element vendors their product ideas and requirements. To match requirements and offers, semantic descriptions of both are necessary. Then, matching of requirements and offers will be realized by semantic reasoners. They have to find RDF graph chains which link requirement and offer descriptions.

In the global, volatile semiconductor market classical capacity planning approaches need to be improved to stay competitive. Resource bottlenecks constrain production, limit the availability of finished product which can be promised to the customer and cause different utilization levels. Bottlenecks may vary periodically due to changes in supply and demand. One bottleneck can have an impact on a whole global production network. How can bottlenecks be identified and allocated in the capacity plan of global enterprises with alternative production routes and daily updated, integrated planning across sites? This paper reviews how capacity planning is covered in the literature, focusing on the semiconductor industry. Local vs. global capacity planning in global networks is identified as an area where there is limited research available so far. A case study of a semiconductor company is used to show the relevance of that area and what solutions have been found by practitioners so far.

Infineon Villach wafer fab facility has a unique characteristic of producing products of different wafer sizes with shared resources, both operators and tools. Daily adaptation of resource allocation to manage changing workloads is done dynamically and imposes challenges in providing high accuracy simulation forecast. In this paper, we discuss the modelling approach that we have taken to represent this dynamic feature, where resources are adjusted in simulation based on WIP situation. The concept is similar to acquiring KANBAN prior to running a production lot. KANBAN quantity is kept at a minimum number in low WIP situation, while the quantity is kept at maximum number at high WIP situation. These values were adjusted by studying the historical moves observed for a period of 3 months. Using this approach, we managed to improve the forecast quality by 7 percentage points in moves and 64 percentage points in WIP.

When solution approaches are developed to facilitate decision processes in real manufacturing systems, fundamental information constraints may apply. The data required by theoretically efficient decision support tools may not be available or accessible; as a consequence, the possibility of implementing the solutions developed is compromised. In this paper, assignment strategies subjected to information constraints and previously developed for a particular wet station are generalized to wet stations characterized by different tool configurations and production recipes. These strategies have been designed to be integrated with data management systems and make use of real time data to minimize the stations' average cycle time. The experiments run aim at supporting the hypothesis that progressive incorporation of details on the system status in assignment strategies enhances the strategies' performance. The relevance of data-driven simulation-based decision support is demonstrated with reference to wet stations operating in a real semiconductor manufacturing facility.

Recent advances in simulation optimization (SO) research and explosive growth in computing power have made it possible to optimize complex manufacturing system problems. Semiconductor manufacturing is known as one of the most complex manufacturing systems. Based on a review of literature in the field of semiconductor manufacturing operational and planning problems, there is little reference to SO methods, an approach that has many advantages over other solution approaches. In this paper, we first distinguish between different users of SO then consider different approaches of SO applied to semiconductor and other manufacturing problems. The article then describes the main operational and planning issues in semiconductor manufacturing drawing actively from a Bosch fab, which could be addressed using SO. Finally, we attempt to provide insights on how SO can be applied to these problems.

An efficient scheduling of bottleneck areas within the semiconductor manufactory gets more and more important. Due to the high complexity within the manufacturing area it is currently not possible to optimize the whole (or even a bigger part of the) factory at once. So mostly only work center specific optimization approaches are investigated. Typically scheduling problems only deal with two dimensions – jobs and equipment. But in some areas of semiconductor manufactory also a third dimension have to be considered – a limited secondary resource. In this paper a Constraint Programming model for such limited secondary resource problems is presented. Thereby the scheduling model also deals with setup matrices for the first and also secondary resource. The modeling of this CP model is shown in detail and the results are compared to a discrete event simulation. The test data for the first tests are orientated on real production data.

Photolithography plays a key role in semiconductor manufacturing systems. In this paper, we address the capacity allocation problem in the photolithography area (CAPPA) subject to machine dedication and tool capability constraints. After proposing the mathematical model of the considered problem, we present a new genetic algorithm named RGA which was derived from a psychological concept called Reference Group in society. Finally, to evaluate the efficiency of the algorithm, we solve a real case study problem from a semiconductor manufacturing company in Ireland and compare the results with one of the genetic algorithms proposed in the literature. Results show the effectiveness and efficiency of RGA to solve CAPPA in a reasonable time.

The roots of today’s military modeling and simulation approaches date back to 1938 when OR emerged after a disappointing exercise conducted by the RAF to test the effectiveness of the newly developed radar. It is fair to say that OR support in WWII was decisive in winning the Battle of Britain in 1940 and the Battle of the Atlantic in 1943. After WWII, military OR was re-awakened by NATO when the Cold War began by facilitating the build-up of national military OR institutions to support defense planners and militaries in sustaining a NATO force structure capable of deterring a Soviet aggression. During the decade of cooperation with Russian analysts after the end of the Cold War we found out that, based on the results of war games and battle simulations, Soviet leaders concluded that the risk of not meeting the operational objectives of a successful attack on NATO was too high. Given Putin’s revisionist policy, NATO’s problem today is how to re-establish deterrence in an ever more complex environment characterized by cyber threats and hybrid warfare. Hopefully, modeling and simulation will again help stabilizing the situation.

Although representations of many elements of hybrid conflict can be identified in the literature, its more holistic view needs more research. The use of Modeling and Simulation (M&S) to represent past and emerging hybrid conflicts can aid in better understanding of their factors and deceptive mechanisms leading to the accumulative effects. Gained understanding could help in preventing, mitigating, and winning hybrid conflicts. Because hybrid conflict is not an entirely new phenomenon this paper offers a few historical examples followed by an attempt to clarify relevant to hybrid conflicts terms. Causal loop diagram (CLD) is used to represent theoretical model of hybrid conflict, while Dynamic Bayesian Network (DBN) demonstrates its implementation. The final section discusses a coordinated defense approach against hybrid threats and needs for a better hybrid conflict representation.

This article develops a data-driven forecast of repair parts for the M109A6 Paladin self-propelled 155 mm howitzer, and this forecast drives a discrete-event simulation to assess requirements for Additive Manufacturing (AM) to be a feasible part of the U.S. Army’s expeditionary supply chain. Actual part demand from the initial invasion of Iraq in 2003 during Operation Iraqi Freedom (OIF) feeds a sample-path-based forecasting method to obtain part demand for each scenario. A simulation of a conceptualized deployed Army 3D-printing facility integrated into the supply chain evaluates the performance and feasibility of the different operational policies. Results indicate current technology could support one battery (or smaller unit) for parts below 100 cubic inches while keeping performance comparable with OIF. These results are incorporated in realistic recommendations for how the Army can potentially improve its supply chain practices with this progressive technology.

High expectations to leverage Modeling and Simulation (M&S) capabilities to support operational testing, systems analysis and user/operator training is motivated by a number of factors. As the complexities of new operations, such as multi-domain operations and operations in highly contested area operations, can no longer be tested and trained in live environments, leveraging M&S capabilities efficiently is paramount. In this paper, we use the differentiation of combat simulation requirements in terms of functional (e.g., model fidelity) and non-functional (e.g., real-time, distributed) characteristics to define a trade space to highlight challenges in relation to simulation purpose. 5th generation weapon systems require lots of data with high fidelity, depending on the purpose of the simulation. As functional and non-functional characteristics are interdependent, the design of supporting simulation systems must be well aligned with the purpose when developed, and is constraining possible purposes to be supported once it is developed.

This paper highlights the accomplishments and shared vision between the International Council on Systems Engineering (INCOSE) and the Modeling and Simulation community (represented by the Society for Modeling and Simulation, International (SCS), and Simulation Interoperability Standards Organization (SISO, among others). We describe convergence between the model-based systems engineering initiative of the INCOSE community and the model-based simulation developments of the SCS community. The goal is not only to highlight the outstanding accomplishments of our time, but also to emphasize the parallels and relationships. The paper is intended to enhance communications and facilitate the outreach already in motion. Modeling and Simulation (M&S) represents a core capability and need for addressing today’s complex and grand challenges. We suggest a collaboration of INCOSE and SCS, as leaders in the systems and M&S communities, to solve these challenges complicated by multi-dimensional, hierarchical, and uncertain Big Data and propelled by exascale computational platforms.

We tackle some of the challenges of optimal recruitment strategies of pilots and other aircrew for Royal Australian Navy aviation. Aside from financial costs, too many students in the training system can create bottlenecks and harm morale, while too few poses a risk to delivering operational capability. We propose a stochastic programming approach that can determine the minimum number of students to recruit and how to distribute those students amongst the training system, in order to meet a capability target with a prescribed probability. This approach is parallelizable and reasonably computationally efficient, and utilizes multiple randomized trials of a greedy method to seek a global minimum. Our work provides insights into the relationship between minimization of entry recruitment numbers and minimization of total student numbers in the system, as well as the relationship between risk constraints and convergence to local minima.

Mission engineering is a recently proposed concept that needs practicable means to implement. Integrating a new system into technical and organizational architectures is a critical part of this initiative. Additionally, it requires an in-depth understanding of the system’s operational employment. This paper describes the development of the integration and operations support system, an innovative use of scenario methodologies, computer simulation, and experimentation to shape a strategic analysis framework for mission engineering. A scenario from an actual computer aided exercise, Cobra Gold 2018, is the backdrop of the support system. The Joint Theater Level Simulation drives the operational vignettes for the training audience in this multinational exercise. Our approach is to automate the exercise, thereby creating an experimentation environment for a knowledgeable team to form credible, analytically derived insights about a new system. We generalize this approach and provide a use-case via a planned study on future naval capabilities.

The problem of determining optimal naval gun firing patterns for engagement of manoeuvring surface targets using traditional simulation approaches is computationally intensive, particularly for large salvo sizes. A simplified modelling technique based on representing warhead effects using Gaussian function approximations calibrated from more detailed modelling is reported here. The simplified model permits the parameter space defining lay-down of rounds in a firing pattern to be searched so as to determine optimal patterns that maximise salvo probability of kill. The method employs Newton’s method to formulate a system of equations defining local extrema, which are then solved using Gaussian elimination. These extrema are then searched to obtain the pattern that maximises salvo kill probability. This paper presents the underlying theory and gives initial results obtained using the model calibrated for an illustrative example from a more detailed model.

The design and analysis of simulation experiments and the analysis of simulated alternatives are critically important tasks when employing combat simulations in support of Army modernization. Effective sensitivity and ranking analyses enable insight to be gained on the marginal contributions of sub-systems to overall operational effectiveness, as well as comparatively assessing competing system alternatives. This paper makes some unifying discoveries related to sensitivity analysis via linear multiple regression, and discusses some empirical findings concerning ranking analysis via score-based, partition-based and consensus-based methods. Finally, an investigation into the combined ranking sensitivity problem is also attempted.

This work is motivated by the need for the Australian Defence Force to produce the right number of trained aircrew in the right place at the right time. This necessitates the development of optimal recruitment strategies while sustaining squadron capability within some risk tolerance. The challenge is that Defence Aircrew training environments typically have highly variable failure rates and relatively small numbers of students. We investigate three receding horizon strategies, each of which use inflated notional targets with some deterministic assumptions to mitigate risk. The first strategy back-fills expected demand given fixed targets; the second strategy dynamically chooses targets using Monte Carlo simulations; and the third strategy incorporates Integer Linear Programming for partial solutions. We show that the first two strategies scale well and maintain steady states, and that the second strategy successfully incorporates the risk tolerance, resulting in an efficient and highly scalable strategy for the recruitment problem.

Virtual testbeds are essential tools for evaluation of cyber-security problems and cyber-security solutions in embedded control systems such as those that control the power grid. However, without explicit control, virtual testbed behavior is not repeatable, which limits our ability to replay a testbed experiment to re-create a surprising event, or even isolate and fix a bug. In this paper we consider extensions to TimeKeeper emulation framework aimed at improving its repeatability. One approach steps through assembly language instructions, the other asks emulated applications to report their own progress through virtual time. Empirical data demonstrates the potential of our proposals.

As network technologies undergo an exponential growth in terms of bandwidth and topology complexity, the gap is worsened between the performance of network simulation techniques and real network scenarios. Fluid-flow models for network dynamics are a well know option for reducing simulation overhead while offering useful averaged approximations of network metrics. Yet, the methods and tools established in the packet-level simulation community are alien to those used in continuous system modeling by means of differential equations. This hinders the synergy between specialists in both techniques. In this work, we present a novel modeling methodology and simulation tool to unify the experience of designing network simulation models both with fluid-level and packet-level techniques under a single modular and hierarchical formal framework. We verified the efficacy of our approach both in terms of simulation speedups and modeling simplicity for canonical network simulation scenarios.

Scalability of parallel discrete-event simulation (PDES) systems is key to their use in modeling complex networks at high fidelity. In particular, intranode scalability is important due to the prevalence of many-core systems, but MPI communication between cores on the same node is known to have drawbacks (e.g., software overheads). We have extended the ROSS optimistic PDES framework to create memory pools shared by MPI processes on the same node in order to reduce on-node MPI overhead. We perform experiments to compare the performance of shared memory ROSS with pure MPI ROSS on two different systems. For the experiments, we use several models that exhibit a variety of characteristics to understand the conditions where shared memory can benefit the simulation. In general, higher remote event rates means that simulations are more likely to benefit from using shared memory, but this may also be due in part to improved rollback behavior.

Many service systems provide queue length information to customers to aid their decisions of what queue to join. One example is at Walt Disney World (WDW), where waiting times are posted to customers via an app. However, it has been observed that the real waiting times are not posted in the app. In fact, WDW rounds the waiting time up to nearest five minute interval. In this paper, we build a simulation model to study the impact of rounding this information and when the information is lagged. We show that rounding or delaying information can result in oscillations in the queue length process. Moreover, increasing the rounding parameter or the delay in information causes oscillations to increase. We also demonstrate that our queueing model can mimic the observed dynamics in the data seen in WDW. Thus, we show the importance of understanding the impact of rounding or delaying information.

Validation and Verification (V&V) is one of the most important aspects of the simulation modeling process. During the three stages of V&V (conceptual validation, verification, and operational validation), it is possible to identify inconsistencies in the model’s development and, therefore, increase the trust that the model is an adequate representation of reality. The aim of this paper is to demonstrate practical problems and benefits while applying V&V techniques, even if takes a lot of project time, as an indispensable step in guiding the modeler to solving any missed issue. Shortcomings and benefits of V&V techniques application were demonstrated through a case study at a Children’s hospital Call Center.

Defining efficient decentralized load balancing schemes exhibiting low memory and communication costs is an important ongoing topic. In particular, characterizing critical regimes where a system optimizes resource usage is uncharted territory. We consider here dynamic balancing schemes in a set of processor sharing servers with finite capacity. Guided by recent results for insensitive load balancing schemes, we applied a modeling and simulation strategy to characterize systematically and extensively several classes of balancing policies under various statistical conditions. We found that there is a class of efficient policies for which a common critical regime can be identified and interpreted as a generalization of the Halfin-Whitt-Jagerman regime for one-server systems. We also study the gap between full and partial information systems, and analyze the performance sensitivity to jobs' size distribution. This study is motivated by the network architecture in the ATLAS experiment at CERN, where load balancing plays a key role.

Realizing continuous operations of autonomous plants subject to finite specialist crew resources for maintenance and repair is vital to achieving productivity and cost-effectiveness in construction operations. This paper presents a practical Monte Carlo simulation-based method to develop autonomous plants operations and maintenance programs. To balance the cost of plant production loss against the cost of hiring maintenance crews, we define a cost function which factors in production output value, resource utilization efficiency and direct cost in connection with both autonomous plants and maintenance crews. An illustration case of planning maintenance crew resources in operating autonomous crushing plants at a quarry site is used to shed light on required input data, simulation processing, and output analysis. The case also has increasing relevance to the construction industry in the near future in terms of planning the operation of a fleet of autonomous equipment in site operations.

In typical construction projects, general contractors devote significant resources to coordinate the work of multiple subcontractors. Subcontractors and general contractors interests are frequently misaligned, and they strive for local optima. In particular, subcontractors balance the demands of different projects as they search to deploy their resources most productively. The problems are exacerbated where customization, such as tailoring of apartments by clients, disrupts workflow. Separation of standard (structure, public spaces) from customized work (apartment interiors) may drastically improve the production system. Separation represents a fundamental change to industry practice. Evaluating separation requires experimentation on a multi-project scale; agent-based simulation is the only practical research method. Simulations of single project systems showed that Lean interventions improve productivity, cash flow and project duration. Recent work with agent-based simulation with two projects and eight subcontractors has shown that the standard/customized work separation provides additional advantages. Future work will extend the simulations to multiple projects.

Customized engineering and the build-up of unique products are complex tasks, where project management contains lots of uncertainties. Simulation techniques could help to evaluate and achieve improved and more robust plans during project management, but are typically not applied in industry, especially at SMEs (small and medium-sized enterprises). This paper covers some ideas of the joint research project SimCast of the Universities of Kassel and the University of Applied Sciences Zwickau. It aims at the development of a method for duration estimation of a project task during project planning. Based on the researched state-of-the-art, requirements and a planning process are described as well as a draft of the current technical infrastructure of the intended modular prototype. First plug-ins are implemented and already show possible benefits for the project management process. The paper describes possible scenarios for the use of simulation techniques in this setting, based on gained experience.

Construction industry has one of the highest rates of bodily injuries including serious Work-related Musculoskeletal Disorders (WMSDs). The Occupational Safety and Health Administration (OSHA) identifies force level as a risk indicator associated with WMSDs. Without direct measurement, quantifying the force exerted during a given task performed by workers is difficult, if not impossible. Therefore, an indirect and non-intrusive way of identifying excessive force applied during physical tasks can effectively reduce the risk of WMSD injuries. In this research, a series of physical activities involving pushing and pulling are simulated by the research team in laboratory-scale experiments. The exerted force is measured using a work simulator tool and accelerometer data is collected from a smartphone sensor affixed on the working arm. Artificial Neural Network is trained with the accelerometer data and the force levels. Testing results indicate that the trained model can predict the force level with over 87.5% accuracy.

Ground conditions remain an uncertain factor in tunneling projects, complicating the ability of practitioners to reliably estimate project productivity and, in turn, duration. This study proposes a Bayesian-based approach to incorporate real-time project data into simulation-based ground prediction models to improve prediction accuracy. Changes in ground conditions are modeled using a Hidden Markov Model, which is updated with actual project data using the Baum-Welch algorithm. The prediction model is then incorporated in Simphony.NET to enhance simulation of tunneling construction operations. A case study conducted in Edmonton, Canada, demonstrates that the proposed approach is capable of incorporating real-time data in a manner that resulted in enhanced duration prediction accuracy.

This paper introduces an extension of the well-established Resource-Constrained Project Scheduling Problem (RCPSP) to apply it to maintenance problems of highly frequented infrastructure. A major complication of the observed scheduling problem is that the infrastructure is only temporarily available for maintenance and repair work during the shift of the workforce. A multi-criteria evolutionary algorithm with a novel problem representation is introduced which is capable of revising technician-task allocations whereas the duration of the task may be stochastic. The main objective is to develop shift plans which maximize the utilization rate of technicians due to a minimization of waiting times caused by the use of the infrastructure through other actors. The results of the already implemented core algorithm for an actor-oriented model illustrate a fast convergence towards an optimal work allocation within a team as well as an efficient sequence of tasks.

Building energy use is significantly influenced by building occupants or users. The integration of a robust occupant behavior model that captures energy-related behaviors and a building energy model will have the potential to improve energy simulation performance, as current virtual model of building lacks dynamic and practical occupant information input. Agent-based Modeling (ABM) has been successfully applied to model interactions between occupants and building components, but most of the models were developed on a simulation basis without actual data involvement. To address on this issue, this paper proposes an approach to modeling occupant behaviors in office buildings via the design of a novel ABM and relevant data collection for model testing and validation. A case study is conducted to investigate the performance of the model. The results show the applicability of the ABM and provide a feasible direction for tuning ABM for the purpose of building energy simulation improvement.

A holistic approach for the simulation of evacuations from buildings in cases of fire and toxic gas spread is developed within the German project iSiGG to achieve high reliability in fire safety planning. Its essence is in the mutual interaction of the domains of crowd simulation, pollutant gas spread simulation (CFD) and Building Information Modeling (BIM), embedded in a coherent IT system. The conceptual basis of this system is provided by a dynamic multimodel ensuring interoperability of all system components and supplying simulation tasks with the necessary building and environmental data. More importantly, it allows to take into account various possible changes of the state of building elements, which may be caused by inhabitants or by the building control systems and can lead to strong changes in the simulation models. The simulations themselves are coupled on numerical level through a shared Voxel Model in a co-simulation approach.

Smart buildings are complex cyber-physical systems that typically involve a high degree of human interaction. This interaction, if controlled and utilized adequately, can provide a significant hint in fine-tuning and optimizing systems, as well as in diagnosing problems in systems in a timely manner. In this paper we define a Zoning Framework for enhanced and, potentially, optimal automation of buildings that utilizes the potential of human interaction and its multifaceted nature in optimizing on the various performance metrics of buildings. We, furthermore, present a workflow in which the framework can be utilized to optimize data collection processes, and, thus, aid in meeting performance goals of smart buildings.

Most management problems addressed by simulation studies can be characterized as complex and difficult to analyze. Simplification is instrumental in creating and employing simulation models that are useful – by focusing on those system elements that matter, and feasible – by reducing study efforts. Although simplification is considered a fundamental modelling activity, simulation educational support for mastering associated modelling skills is limited. Main textbooks either do not address this topic or tend to restrict their guidance to a few rules of thumb. This reflects how the topic of simulation model simplification is underdeveloped, despite the field of simulation being around 50 years of age. The purpose of this panel is to initiate a discussion about the way we teach simulation model simplification with the view to identifying improvements. This paper addresses the motivation for the panel, and presents panelists’ “positions” about the way forward for education on simulation model simplification.

Discrete Event Simulation (DES) is a powerful tool for decision making. It has been widely used in different application fields. There are many works investigating the simulation modeling process, but little is found in the literature that considers how to manage a DES project. In this respect, this paper has the objective of proposing a specific methodology for the management of DES projects based on PMBOK®. This was considered to be the methodology that most fits with the requirements of DES. In order to test the approach, PMBOK® was applied in a real simulation project. The methodology was then evaluated by the simulation analysts who provided their feedback on the study.

Business process simulation can support the analysis and improvement of business processes and, hence, is a valuable technique to teach to students. Consequently, introductory simulation courses are included in a multitude of study programs. Besides providing students with theoretical knowledge and getting them acquainted with simulation software, it is also important to let them experience the complexity involved in conducting a simulation study in practice. In this respect, this paper outlines the inclusion of a real-life simulation study in an introductory simulation course. Besides the content of the simulation study, the instructional design, student feedback, company feedback and challenges perceived by the instructor are outlined. Student feedback shows that, even though the project was perceived as highly challenging, conducting the real-life simulation study provided extensive learning opportunities.

An academic/industry alliance can improve the preparedness of students to enter the workforce. When the alliance is centered around a laboratory providing students real world research and development experience, both partners benefit from the intellectual engagement. This paper discusses the benefits of a collaborative laboratory being developed in support of the modeling and simulation engineering program at Old Dominion University. The laboratory focuses on simulation-based test and evaluation (T&E) of autonomous vehicles, an important economic and social problem. Economically, driverless cars and drone delivery systems are the wave of the future. Socially there are issues of public acceptance of these systems as evident by public reaction to the infrequent, but highly public, accidents involving driverless cars. The laboratory allows academia and industry to address these important problems by improving the T&E of these systems while also improving student skills, giving undergraduate and graduate students an opportunity to participate in research.

In many curricula and degree programs, simulation courses are not required, but these tools and techniques could be beneficial to students preparing for a variety of careers. The current paper describes two examples of simulation education embedded into broader course topics from a development perspective. The examples, from courses generally described in this paper as Cloud Computing and Big Data, offer a recommended approach for exposing students to practical uses of simulation. In the first example, we use simulation techniques to develop a web service emulation response database within a cloud computing environment for software testing. In the second example, simulation techniques provide an approach to generate data sets for learning data analytics techniques.

Gamification and experiential learning are increasingly used in education as they create an immersive environment to stimulate students and promote deeper learning. In industrial engineering education, computer simulations and digital games are commonly used to teach technical skills in supply chain management and production planning. Used alongside other teaching methods, they allow students to apply theories learnt and reflect on the impact of their decisions. Other “hands-on” games can also foster the development of professional skills such as leadership, teamwork and communication. Focusing on serious games and game-based learning in industrial engineering, this paper reviews examples to discuss games’ benefits and drawbacks as educational tools. Finally, the author suggests ways for game developer to consider how game aspects align with learning outcomes in the cognitive and affective domains.

In this paper, we share our approach to teaching conceptual model development and simulation modeling at a Department of Computer Science in the UK. We explore the challenges and opportunities we face teaching static and dynamic modeling to the 2nd and 3rd year undergraduate students. While our Business Computing students tend not to be as technically savvy as our Computer Science students which at times limits the technical complexity of what we can cover in our courses, we designed the Business Analysis and Process Modeling course as a tool to empower students to analyze and discuss business problems in-depth. The paper reports on the design and findings of the undergraduate teaching of modeling and simulation content. We finalize the paper by providing some concluding remarks and future directions.

This paper describes a framework for ethical analysis of the practice of computer Modeling & Simulation (M&S). Each of the authors presents a computational model as a case study and offers an ethical analysis by applying the philosophical, scientific, and practical components of the framework. Each author also provides a constructive response to the other case studies. The paper concludes with a summary of guidelines for using this ethical framework when preparing, executing, and analyzing M&S activities. Our hope is that this collaborative engagement will encourage others to join a rich and ongoing conversation about the ethics of M&S.

We discuss the application of simulation to estimate a nominal, or target, processing times for work stations on a serial assembly line. The expectation is that having different processing times per sta-tion per product will increase the throughput of the line, compared to having a constant time for all stations. A demonstration case at ABB Robotics in Sweden will be presented. This is a small part in the “Process Automation for Discrete Manufacturing Excellence” project (PADME) involving five manufacturing industry partners and four research organizations, that aim at adapting Industrie 4.0 strategies and existing state-of-the-art technologies into new configurations, serving as a framework that can be used by similar industries.

Internet of Things (IoT) has been attracting much attention due to its economic impacts and high expectations for drastically changing the competitive domain in various industries in the recent year. Digital Twin (DT) has been proposed as a tool for collecting and synchronizing real-world information in real time on the cyber side. However, it is still at the conceptual stage and only a few studies have specifically discussed methods for their construction and implementation. In this study, a framework for constructing DT in an IoT-aided manufacturing environment was proposed. As a case study, we constructed and implemented a DT-oriented simulation model for a sensor-equipped manufacturing system to verify the effectiveness of the framework.

More and more simulation applications need high performance computing to deliver the results from experimentation in a timely manner. Cloud computing presents an attractive cost-effective alternative to using a local computing cluster. Normally a user would decide how many cloud computing resources to hire, provision them and then use them for experimentation. However, it may be the case that the user has paid for many instances that were not used. We have proposed the Microservice-based Cloud Application-level Dynamic Orchestrator (MiCADO) to automatically orchestrate and scale cloud computing applications. This article describes the architecture of a version of MiCADO that has been adapted for simulation experimentation.

This paper describes how simulation can be an effective tool for investigating the dynamic behavior of complex manufacturing systems. Flexible Manufacturing System (FMS) aims to provide optimal machine flexibility, process flexibility, and product flexibility. However, releasing flexibility in FMS is challenging due to the necessary complexity of coordinating all subsystems while simultaneously scheduling multiple product families. In this paper, an automotive industry case study is outlined, and a Discrete Event Simulation (DES) model has been developed. The model sufficiently represents all primary subsystems to capture their dynamic interactions, providing a bird’s eye view of all critical activities within an FMS. A series of manufacturing problems and their corresponding hypothetical solutions have been explored and evaluated in a number of simulation scenarios. The results demonstrate that releasing FMS flexibility can result in significant performance improvements whilst also removing barriers (such as operation sequencing or line balancing) associated with dedicated manufacturing systems.

Simulation modelling has long been used to support operations strategy decisions in the mining industry. Recently, the commodity cycle and push for lower costs has driven efforts to improve operations efficiency. The use of simulation has typically been challenging in the operational time horizon due to the difficulty of initializing the system state and the sensitivity of the results to initial conditions. However, a recent explosion in data availability has made it feasible to know, in real time, the location of each piece of equipment in the fleet, what it is carrying and where it is going. This makes it possible to simulate and predict production performance within a shift and to allow testing of what-if scenarios to improve operations efficiency. In this case study we describe the approach taken, the application of simulation for short-term forecasting and the challenges faced implementing this for a global mining company.

Simulation models can benefit from the latest 3D laser scanning technology, as showed in the current projects SUMMIT and ECOPRODIGI. Using point cloud data of the physical environment generated from 3D laser scanning, we can emulate a spot welding process as-is in an immersive virtual reality environment that looks like the real facility. Point cloud data can also be used to evaluate volumetric filling of cargo space, as visualized in this demo using a 3D laser scanned RoRo ferry. The point cloud data can also help generate accurate models of the hulls for larger ships that can be used to simulate the flow of water, allowing engineers to predict the effect of changes to the hull and traveling conditions to reduce emissions and fuel consumption.

Timely installation of fittings is a vital element of successful construction of offshore structures. In particular, pipe spools take up a large part of installation work of the entire fittings. Most of the pipe spools are fabricated and supplied by partner companies. Whereas the procurement and supply of raw materials are carried out by Samsung Heavy Industries (SHI). Therefore, it is necessary to not only manage the timely receipt of raw materials, but also check a bill of materials (BOM). Inventory and delivery schedule are considered to forecast a possibility of outfitting fabrication. This paper discusses an example of the application of an algorithm that supports schedule confirmation and decision making when forecasting a schedule.

Production plans are based on fair assumptions of process performance and all operation parameters are taken as averages. There are a number of events that happen in any manufacturing setup during the course of production like periodic delivery of raw materials or changeovers on a machine. The interaction between these events is non-linear and cannot be easily visualized. As a result of which most of the production plans in any company have only a limited realization. This paper provides an example of how simulation using Anylogic has been applied in one such plant scenario to visualize the plan outcome.

A worldwide leading company in the cosmetic industry was dealing with great challenges regarding adapting its positioning strategy to the dynamically changing behaviors of the market. The company needed to decide where to invest its marketing budget to optimize its revenue and was using different traditional marketing mix models without any success. A marketing mix model was developed using agent-based modeling to predict the market’s reaction to a given distribution of a certain budget among the different touchpoints available. This innovative model uses market information and consumer information collected from surveys to estimate the company’s sales, level of awareness and level of consideration given its distributed investment. The tool was implemented as part of the marketing plan decision making process, providing the ability to test different scenarios and generate quantitative analysis of its results.

This study focuses on simulating the evacuation of a neighborhood of Montréal, Québec, Canada in the event of flooding. Two strategies are simulated, and it is concluded that a significant reduction in the average evacuation time per family can be obtained by appropriate scheduling of the evacuation buses.

Training instructors are using advanced, interactive use-of-force simulators to produce unique training experiences for law enforcement officers. While today’s simulators are sophisticated with scenarios that make them invaluable tools in law enforcement training, much of the training doesn’t take full advantage of simulator capabilities when used in the context of a larger training environment. Based on the professional experiences of the author, an immersive concept that would require trainees to not just react to the video on the simulator, but to work the scenario within the context of a larger environment would be extremely beneficial. A training venue has been developed with the goal of revolutionizing law enforcement training in Wisconsin. Using simulation scenario-based learning and training offers many benefits to law enforcement officers and advances realism in law enforcement training.

With the ever-increasing importance of simulation in supporting decision making, strategic planning and operations within a fast paced business environment, traditional simulation has lagged behind, constrained as it is by long lead times for model update and validation. Symbiotic simulation has emerged as a solution, enabling a timely response to abrupt changes in the physical system. In this presentation, we introduce the implementation of symbiotic simulation as part of an overhaul of Ford Power Train Manufacturing Engineering throughput simulation methodology and discuss challenges, opportunities and lessons learned.

S-Flow, or Simulation-Flow, is the name of the methodology for utilizing discrete-event simulation (DES) that has evolved from over 20 years of use in automotive manufacturing. This process has been refined to address typical pitfalls of applying DES and to ensure acceptance by non-experts. The methodology walks the simulation engineer through stages of project planning, model planning, data analysis, model development, model verification and validation, scenario testing, results analysis, and project reflections. Each stage is explored in depth to show the potential pitfalls that can be avoided. Complete and consistent usage of the methodology promotes awareness and approval of DES analysis as well as the opportunity to further improve the S-Flow methodology.

This paper use a case study to investigate the process analysis methods using value stream mapping and discrete event simulation in early phases change decision making. The case study discuss the practical situation for the modeler. The output of each tool and the combination is evaluated with the modelling effort in mind. The result shows that it may be preferable to analyse issues like static bottlenecks and average capacity with value stream mapping while for dynamic issues discrete event simulation may be necessary. For industrial application purposes it is proposed to build the models simultaneously and to take time to do the proper static analysis before doing the dynamic analysis in order to understand the dynamic results.

In recent years, automated guided vehicles (AGV) have been widely utilized in automated container terminals for material transfer. AGV can navigate in a facility via communicating with transponders buried beneath the surface of the ground. However, little research has been done on the placement configurations of the transponders which have significance in facility layout, bringing in flexibility and not hindering productivity. In this scope, this study provides a simulation model to evaluate the effectiveness of different transponder placement configurations and helps company decide the best transponder layout.

Nowadays, in the fast fashion industry, high delivery frequencies and small lead times are key fac-tors for market competitiveness. To achieve this goal, many companies are evaluating the possibil-ity of opening urban hubs close to the retailers to guarantee more than one replenishment at the re-tailers each single day. In this paper, we investigate the impact on the service level, inventory in-vestment and transportation costs of adding such hubs in an Italian fast fashion supply chain.

As UK and the EU are in a critical phase of Brexit negotiations, it is important for the transportation and logistics industry to comprehend the full extent of the consequences of Post-Brexit scenarios. In this case, a simulation-based scenario mapping model is developed in collaboration with an Irish food logistics and transportation company to test for the short-term effects of Brexit on company’s performance. O’Toole Transport founded in 1996 with one refrigerated vehicle operating from Gal-way. Now, the company has become the first logistics company in Ireland to be awarded the prestigious BRC accreditation for the cross-docking and distribution of chilled and frozen food.

In this case study, we are considering the use of Discrete Event Simulation (DES) and Agent Based Simulation (ABS) with our Flexible Healthcare Simulation Optimization (FHSO) algorithm in Emergency Department (EMD) following the Emergency Severity Index (ESI) with a lot of patient tracks, to optimize staff assignment, and better determinate the number of resources that could share tasks between the EMD and surgery room taking into consideration the indirect tasks executed by the resources. The main purpose of this study was understand the capacity of the EMD, patient’s cycle time or Length of Stay (LOS) and staff assignment for different demands of patients using simulation & optimization inside FlexSim due to its multi method modeling capability (DES and ABS) and high flexibility. The project was handled inside the major Latin America’s hospital.

Creutzfeldt-Jakob disease (CJD) is a fatal disease caused by prions. Prions on surgical instruments are not completely deactivated and subsequent patients may become infected resulting in surgically transmitted CJD (stCJD). Work was undertaken to assess the cost-effectiveness of strategies to reduce the probability of stCJD. Strategies evaluated included: no changes; ensuring that instruments were kept moist; prohibiting instrument migration between sets; and employing single-use instruments. A stochastic mathematical model was constructed with key model parameters populated via elicitation. The model was calibrated to match the possible number of observed stCJD cases between 2005 and 2018, noting that stCJD cases could be misdiagnosed. Results indicated that keeping instruments moist reduced the number of stCJD cases and saved money. Further measures (prohibiting set migration, or single-use instruments) reduced the risk of stCJD cases, compared with keeping instrument moist, but at considerable cost and poor value for money.