WSC 2001 Final Abstracts

Modeling Methodology Track

Monday 10:30:00 AM 12:00:00 PM
Object-Oriented Paradigm

Chair: Tayfur Altiok (Rutgers University)

Component-Oriented Simulation Architecture: Toward Interoperability and Interchangeability
Gilbert Chen (Rensselear Polytechnic Institute) and Boleslaw K. Szymanski (Rensselaer Polytechnic Institute)

In this paper we investigate two issues at the kernel of simulation reusability: interoperability and interchangeability. Their implications on the simulation technology are discussed. Based on our previous work on simulation component oriented world view and simulation component classification, the Component-ORiented Simulation Architecture (CORSA) is devised to address both issues. The ideas and considerations which motivated us in developing CORSA are discussed. The design and implementation of a prototype is also described briefly. A sequential PCS simulation has been developed using CORSA. This exercise demonstrated several advantages of the component-based approach: flexibility, extensibility as well as reusability. Experimental results show that the component-based approach is only slightly slower than the monolithic approach, whose complexity quickly grows to nearly unsurmountable proportions with the growth of complexity of the simulated system.

A Capacity Planning Tool for the Tuxedo Middleware Used in Transaction Processing Systems
Tayfur Altiok (Rutgers University), Wei Xiong (BEA Systems Inc.) and Mesut Gunduc (BEA Sys. Inc.)

In this paper, we present a brief overview of Tuxedo middleware system and introduce an object-oriented computer simulation template developed for the purpose of capacity planning and performance analysis of Tuxedo application environments. Arena/Siman simulation software is chosen and a CP_Tool template specific to Tuxedo environment is developed. The template consists of a number of modules representing client and server nodes, network nodes and other critical components of the system. Any Tuxedo environment can be created using the modules from the CP_Tool. The paper discusses the tool and its capabilities.

A Framework for Distributed Simulation Optimization
Bjrn Gehlsen and Bernd Page (University of Hamburg)

The system presented bridges the gap between three different research areas: discrete event simulation, heuristic optimization methods and distributed systems technology. Its goal is to provide a framework which supports an efficient implementation of simulation optimization projects, including heuristic optimum seeking procedures and parallel execution of experiments. It is written completely in Java and only uses components that are publicly available, including software libraries from academic institutions or the Java API from Sun Microsystems.

Monday 1:30:00 PM 3:00:00 PM
Extreme Modeling

Chair: Lee Schruben (University of California Berkeley)

Modeling Design Development in Unpredictable Environments
Nuno Gil and Iris D. Tommelein (University of California, Berkeley) and Robert Kirkendall (Industrial Design Corporation)

This paper presents a process simulation model representative for design development of a building system in an unpredictable environment. Unpredictability means that design criteria are prone to change as design development unfolds. The model was implemented with a discrete-event simulation engine based on event graphs. Events capture moments when tasks start or end, or changes that cancel future scheduled events and schedule new design iterations. Between conceptualization and concept development, we assume that managers can impose a time lag so as to minimize rework of concept development tasks due to upstream changes of design criteria. Simulation illustrates the effects of adopting different postponement strategies. The results show that postponing the start of concept development consistently reduces the average resources spent in concept development and increases process reliability, but it augments the average design duration. The judicious choice of a postponement lag can thus yield gains in terms of cost versus time.

Resource Graphs for Modeling Large-Scale, Highly Congested Systems
Paul Hyden (Clemson University) and Lee Schruben and Theresa Roeder (University of California at Berkeley)

Simulations often execute too slowly to be effective tools for decision-making. In particular, this problem has been found in semiconductor manufacturing where conventional job-driven simulation models explicitly track each lot of wafers as it progresses through the system. While a job-driven simulation model offers some advantages, they inherently execute slowly. This paper explicitly defines resource-driven modeling. Here jobs are implicitly tracked through their resource usage. Resource-driven simulations typically run much faster than job-driven simulations. This speed-up is insensitive to congestion and is most dramatic when the system is highly congested and therefore most interesting to the analyst. There can also be a significant reduction in memory footprint. However, there is a potential tradeoff in information loss.

Simulating Biotech Manufacturing Operations: Issues and Complexities
Prasad V. Saraph (Bayer Corporation)

The Biotech industry is still an emerging application area for simulation techniques. This paper describes the hierarchical discrete event simulation efforts at Bayer Corporation's Berkeley facility that manufactures second generation recombinant DNA technology based drug, Kogenate-FS. The facility consists of multiple manufacturing areas housing state-of-the-art biotech processes. The main simulation issues included discretization of continuous activities, building appropriate level of detail in the models and conceptualizing biotech operations for simulation. Complexities arose from spread of manufacturing operations, sharing of common utilities, limited lifespan of product and materials in-between stages coupled with limited storage capacities, regulatory constraints, industry-specific quality requirements and varying shift schedules, production capacities and batch sizes across stages. Even though the simulation efforts are not complete, the simulation models developed so far have saved Bayer substantial amount of money and have offered forward visibility for various strategic decisions over the last two years.

Agent-Based Simulation and Greenhouse Gas Emissions Trading
Hideyuki Mizuta (IBM Japan, Ltd.) and Yoshiki Yamagata (National Institute for Environmental Studies)

The need for new theoretical and experimental approaches to understand dynamic and heterogeneous behavior in complex economic and social systems is increasing recently. An approach using the agent-based simulation and the artificial market on the computer system is considered to be an effective approach. The computational simulation with dynamically interacting heterogeneous agents is expected to reproduce complex phenomena in economics, and helps us to experiment various controlling methods, to evaluate systematic designs, and to extract the fundamental elements which produce the interesting phenomena for future analytical works. In the previous works, we investigated the stability of a virtual commodities market and the aggregated behavior of the dynamic online auctions with heterogeneous agents. In this paper, we will introduce a simple framework to develop agent-based simulations systematically and consider an application of the agent-based simulation for a dynamical model of the international greenhouse gas emissions trading.

Monday 3:30:00 PM 5:00:00 PM
Panel: Simulation Environment

Chair: Voratas Kachitvichyanukul (Asian Institute of Technology)

Simulation Environment for the New Millennium (Panel)
Voratas Kachitvichyanukul (Asian Institute of Technology), James O. Henriksen (Wolverine Software Corporation), C. Dennis Pegden (Rockwell Software Inc.), Ricki G. Ingalls (Oklahoma State University) and Bruce W. Schmeiser (Purdue University)

A panel discussion session of the past and present of the simulation environment. Issues related to research and development, methodology, software will be discussed by distinguished panel members.

Tuesday 8:30:00 AM 10:00:00 AM
Supply Chain Modeling

Chair: Jerry Evans (University of Louisville)

A Real Options Design for Product Outsourcing
Harriet Black Nembhard and Leyuan Shi (University of Wisconsin-Madison) and Mehmet Aktan (Ataturk University)

We develop a financial model to assess the option value of outsourcing. We value the real options associated with outsourcing an item using Monte Carlo simulation. This valuation gives decision makers a way to choose the appropriate outsourcing strategy based on an integrated view of market dynamics. A simulation example is used to demonstrate the application of real options to value outsourcing. The simulation program code was written in JavaScript so that the valuation task would be accessible to other users because of its web enabled feature.

Supply Chain Agent Decision Aid System (SCADAS)
Anurag Gupta and Larry Whitman (Wichita State University) and Ramesh K. Agarwal (National Institute for Aviation Research)

Supply chain decisions are improved with access to global information. However, supply chain partners are frequently hesitant to provide full access to all the information within an enterprise. A mechanism to make decisions based on global information without complete access to that information is required for improved supply chain decision making. Mobile agents can support this requirement and these are the programs that can be initiated on a single host and then migrate from host to host over a network. At each host, a process can be spawned which will provide a "black-box" view into that host's information. This provides access to necessary information, while maintaining privacy for company sensitive information. This paper will discuss mobile agents and how they are useful in designing and managing the supply chain. The Supply Chain Agent Decision Aid System (SCADAS) is presented as tool to provide the flexibility of mobile agents while protecting company sensitive information.

Production Scheduling Validity in High Level Supply Chain Models
David J. Parsons and Richard A. Phelps (Simulation Dynamics)

Although they focus on the big picture, high level supply chain models cannot gloss over the capacity of production nodes to meet production allocations. Capacity is not simply a reflection of equipment production rates. Short runs drive down utilization by increasing total time lost to changeovers. Multistage plants require coordination of capacities at the several production stages. In short, production capacity is crucially affected by the way production runs are scheduled through plants. Modeling actual scheduling practice is often unrealistic, since methods vary from plant to plant, and involve a blend between planned schedules and on-the-fly adjustments. This paper suggests that there is a range of approaches to modeling production scheduling. In the modeling of supply chains, modeling alternatives must be assessed in terms of cost of development and implementation versus validity.

Tuesday 10:30:00 AM 12:00:00 PM
Panel: GPSS 40th Anniversary

Chair: Tom Schriber (University of Michigan)

GPSS Turns 40: Selected Perspectives
Thomas J. Schriber (University of Michigan School of Business), Peter Lorenz (Otto von Guericke University of Magdeburg), Springer Cox (Minuteman Software), Julian Reitman (University of Connecticut - Stamford), James O. Henriksen (Wolverine Software Corporation) and Ingolf Sthl (Stockholm School of Economics)

GPSS (General Purpose Simulation System) is celebrating its 40th birthday this year. We recognize this notable birthday by assembling a panel of discussants consisting of some of the folks who have contributed significantly to GPSS and its use over the years. The panelists are Springer Cox (GPSS/PC and GPSS World), Jim Henriksen (GPSS/H and Proof Animation), Peter Lorenz (promoter of GPSS in Europe and on the Web), Julian Reitman (principal in early interactive use and accommodation for large-scale simulations), and Ingolf Sthl (micro-GPSS for Windows and on the Web), with Tom Schriber (author of the "Red Book") as moderator. Each panelist has contributed written perspectives describing aspects of his involvement with GPSS. A Geoffrey Gordon memoriam is included in the paper. (Geoffrey Gordon, who conceived and evolved the idea for GPSS and brought about its IBM implementations, died in 1989.)

GPSS 40 Years of Development
Ingolf Sthl (Stockholm School of Economics)

This year GPSS celebrates its 40th birthday. This paper reports on the development during these 40 years, starting with the first version developed by Gordon at IBM in 1961, and the following development of GPSS II, GPSS III, GPSS/360 and GPSS V, all IBM products. A major section is devoted to GPSS/H, which has dominated the GPSS scene during the last years. There is one section on the GPSSR family of GPSS versions and one on GPSS/PC and GPSS World. There are also many GPSS systems, projects and ideas of a mainly academic nature. A great number of GPSS textbooks are noted. The concluding section discusses the reasons for the popularity of GPSS.

Tuesday 1:30:00 PM 3:00:00 PM
Verification and Validation

Chair: Michael Metz (Innovative Management Concepts, Inc.)

Automated Object-Flow Testing of Dynamic Process Interaction Models
Levent Yilmaz (Trident Systems Incorporated)

This paper deals with the assessment of accuracy of simulation models from the perspective of dynamic object flows. Dynamic objects (also called temporary entities or transactions) move physically or logically from one model component to another and represent entities such as aircraft, data packet, passenger, and vehicle. Accurate flow (movement) of thousands or millions of dynamic objects within a complex simulation model significantly affects the overall model validity. We present a new automated testing technique for assessing the accuracy of dynamic object flows. The permissible sequence and precedence of dynamic object flows are specified using the context-free grammar formalism. The specification accuracy is assessed using a variety of verification and validation techniques. The executable model is instrumented and dynamic object flow trace data is generated. The trace data is automatically compared with respect to the specification and each dynamic object movement traced during model execution is automatically verified.

Verifying and Validating a Simulation Model
Anbin Hu, Ye San, and Zicai Wang (Harbin Institute of Technology)

This paper presents the verification and validation (V&V) of simulation model with the emphasis on the possible modification. Based on the analysis, a new framework is proposed, and new terms are defined. An example is employed to demonstrate how the framework and terms related are used in verifying and validating an existing model.

Verification of Object-Oriented Simulation Designs
Michael L. Metz (Innovative Management Concepts, Inc.) and Jack Jordan (BMH Associates, Inc.)

This paper discusses the verification process for object-oriented simulation high-level and detailed designs based on the authors experience with the Joint Warfare System (JWARS). There is an overview of the JWARS simulation, the software development process, and the design artifacts. The paper describes how the JWARS V&V Team developed a tailored process and method for verification of the high level design and the detailed design and attempted to determine and document the completeness of the design. Also the V&V Team's verification experts attempted to identify the linkage and traceability of the simulation from the pre-design artifacts to the design and from the design to the implemented code. Included is a discussion of how JWARS uses the IBM VisualAge and UML Designer tools and how the Verification Agent was able to use them to support the verification process.

[ Return to Top | Return to WSC '01 Program ]