WSC 2005 Final Abstracts

Predicting the Effect of Process Time Variability on Annual Production in a Cell Culture Manufacturing Facility
Stephen Hall (Bristol-Myers Squibb)

Abstract:
Microsoft Excel and Visual Basic for Applications were used to model process time variability in a cell culture facility. A deterministic simulation had already established that the plant had sufficient buffer, media, water and CIP (clean-in-place) capacity. The Excel model was built to answer the question, “How many batches per year can be expected, as a minimum, knowing that random failures occur and that processing time variability results in a failed batch if a queue exceeds a set value?” Aspects to be presented: a) simplifying assumptions that drilled into the question of interest, b) use of VBA to create the multi-batch per year simulation, c) drawing timelines on the screen that speak to the customer without requiring simulation jargon, d) design of modeling experiments that simulated 250 years of operation with a variety of assumptions, and e) use of statistical functions to obtain confidence intervals and probability data.

TARGET II: The First Discrete Event Simulation of Diabetes and Its Complications
J Jaime Caro, Alexandra Ward, and Jörgen Möller (Caro Research)

Abstract:
A model showing the pharmaco-economic effects of different diabetes-treatments in 1000 patients by cloning the patients once their characteristics are assigned.The Arena model works jointly with an Excel-shell and looks at the effects on the individual patient and on the aggregated economic implications over typically 20 years.

Scaling and Using Large Pandemic Agent-based Models
Steven Naron (IBM)

Abstract:
The NIH NIGMS MIDAS grant is focused on building global agent-based models for studying world-wide epidemics. As the chief architect of the MIDAS project, I am part of a team providing informatics support to our researchers from Hopkins, Emory, and Virginia Tech. Because of the number of agents and complex social networks we are dealing with, our models tend to require very large amounts of computational resources. We are working with a combination of high level modeling tools, automated means of scaling models and balancing work across grids and clusters, and sophisticated tools for managing complex simulation experiments. A goal is to integrate these functions into a software workbench that minimizes many of the distractions on the researchers in order to allow them to stay focused in a rapid iterative research and development mode.

Simulating Coated Paper Production and Distribution
Gail Kenny (Rockwell Automation) and Tim Rech (Stora Enso)

Abstract:
Stora Enso, domiciled in Finland, is an integrated paper, packaging and forest products company. In North America, Stora Enso is a leading producer of coated and supercalendered papers for the printing and publishing industries, and a premier producer of specialty papers. Stora Enso's North American operation continually examines its supply chain and the functions involved in sourcing, producing, and delivering its products. Because of the complexity of its manufacturing process, a strategic planning tool was needed to help management determine the appropriate improvements to the existing process. The primary objective was to balance capacity, customer service, and inventory levels on a continuing basis. The project scope encompassed two Stora Enso North American paper mills, a large sheeting operation, and up to five distribution centers. This presentation describes how Arena Simulation Software was used to develop this decision tool.

Evaluating Increasing Capacity of Steel Hot Rolling Line to 4.2 million Ton per Year
Mehdi Ashrafi Nasrabadi (Mobarakeh Steel Complex) and Soheil Mardani (Simaron Pardaz Co.)

Abstract:
There is complex logistic operation to transfer slabs which are being produced with continuous casting machines to hot rolling in MobarakehSteelComplex. First the slabs should get cooled and then there will be quality control and scarf operation is needed for some of them. Then slabs get transferred via conveyor to warehouse & then transferred to 4 furnaces. The conveyor is bottleneck. The simulation model proved if a parallel conveyor gets used to increase transferring rate because next warehouse will get bottleneck, it will not increase production. Then Hot Charge technology gets evaluated which there is no need to cool the slabs & then again making them hot. So the slabs will be transferred to furnace area with out losing temperature & considerable energy will be saved. Simulation has been used to evaluate what kind of facility is needed & how to assigning the furnaces to Hot Charge & cooled slabs.

Assembly Line Simulation for a Rear Suspension Cradle
Charles Stuart (American Axle & Manufacturing / Lawrence Technological University)

Abstract:
The manufacture of automotive frames and subframes is a complex undertaking, requiring many disparate components and subassemblies to be melded into a functional whole. The competitive environment in manufacturing in general, and the automotive industry in particular, require that both capital and labor be employed efficiently. The throughput constraint for the manufacturer is the assembly line. These lines are generally multi-station automated machines, built to purpose by outside suppliers. They usually consist of a mixture of automated and manual gas-metal arc welding fixtures, buffers, positioning fixtures, error-proofing fixtures, robots, and indexing equipment. This case study presents an analysis of a proposed assembly process for a product designated as the X cradle. Since the assembly line has not been deployed, an existing line was used as a surrogate to verify modeling assumptions, time to failure, and time to repair information. The objective of the analysis was to examine efficiencies and throughput for the proposed line.

The Lean Transformation in Engine Overhaul
Mark Orth and Franklin Young (United Airlines)

Abstract:
United Services, the MRO division of United Airlines, is transforming its operations through Lean. Eliminating waste and reducing repair cycle time make possible for additional work from outside customers and incremental revenues. One of the most dramatic transformations is in the engine overhaul area. A flow line similar to an auto assembly line is being built for the Pratt & Whitney 4000 engines flown on the B747-400 and B777. To quantify its economics, United’s Industrial Engineering developed a simulation model of the proposed line. This model gave key insights without pouring an ounce of concrete or cutting any metal.

Business Driven Simulation of a Postal Letter Sorting Facility in Oslo, Norway.
Knud Erik Wichmann and Niels Erik Larsen (PA Consulting Group)

Abstract:
Using advanced 3D simulation models (incorporating physical layout and logic control rules ) to verify that the mechanical system can handle a certain flow of parcels, products whatever are well known. Yet, many systems do not perform quite as expected even after a simulation study has been conducted? We will argue that key to successfully modelling the new business, production or material handling system is to ensure that all constraints and fluctuations within actual operation, and variations in external business demands are reflected such that scenarios in this regard can be tested. This mean that modellers should be adopting a philosophy with more emphasis on specific customer insights and analysis – a “customer business pull” rather than a “vendor solution push” in the approach to modelling.

A Simulation Based Approach to Evaluating The Impact Of Port Operation Improvements On The Pacific Northwest Rail Network
Amy M Brown and Jim Sawyer (Automation Associates, Inc) and Blair Garcia (Transystems)

Abstract:
This presentation reviews modeling transportation systems which include port operations and the extended railroad network used by port related traffic. The case study presented encompasses the port systems and rail networks of the Pacific Northwest. A new approach for unloading and loading ships at port is introduced in this presentation. Traditionally, a ship is completely unloaded and then loaded. The new approach greatly reduces turn around time by allowing the crane to make an unload move followed immediately by a load move, reducing the number of empty moves made by the crane. Shipments are then transferred away from the port and sorted at an inland interface center. Simulation modeling is used to determine if the current railroad infrastructure can accommodate the increased throughput of improved port operations and aid decision makers in determining the ideal location for inland interface centers.

Simulation Support for a New Approach to Multimodal Transportation Handling
Markus Klug and Shabnam Michèle Tauböck (ARC Seibersdorf research GmbH), Helmut-Klaus Schimany (ÖBB Holding AG), Dietmar Schratt (Rail Cargo Austria AG) and Günther Kaluza (ICN – Intermodal Corridor Network GmbH)

Abstract:
New approaches for multimodal transportation solutions are necessary to increase competitiveness, efficiency and speed of container transportation on railways. Especially the modal changes between train and truck are subject to be investigated. A new innovative approach, the “Innovative Handling Terminal (IUT)” addresses the problem. Putting road and train track side by side together with dividing handling and storing into two task-specific systems was the winning approach of a EU-founded project on new methods. Because of huge investing efforts in the development, the whole system was primarily built in the simulation model, with emphasis on flexibility and visualization. The model was realized with the possibility to compare different layouts (different sizes, different, number of handling units, etc.) of terminals just by changing parameters from an external data source. The simulation model together with the animation turned out to be unavoidable support for the whole development process, and is still continued to be extended and used.

Using Simulation to Evaluate Emerging Transportation Concepts
Beth Carpenter Kulick (Automation Associates, Inc)

Abstract:
Recent cargo growth projections for the Port of Los Angeles/Long Beach have the container traffic tripling over the next 10 to 15 years. The majority of these containers will pass through the LA basin via truck or rail and onto the rest of the country. To accommodate the anticipated growth in trade, an inland port has been proposed to be built in the desert region of Southern California. There are numerous physical challenges associated with the transportation infrastructure to move containers from the Port to the inland region and there are public pressures to remove trucks from highways and onto rail to reduce congestion, pollution, and improve safety. Simulation offers the ability to demonstrate and evaluate transportation concepts and is being used to assess the capacity of regional rail infrastructure and to explore the potential capabilities of emerging freight transportation concepts such as a high-speed magnetic levitation (MagLev) technology.

Driving Large Call Center Simulations Using OLAP Data Cubes
Pam Laney Markt (Progressive Insurance)

Abstract:
Simulating large call centers can require weeks or months. Additional time is spent on detailed analysis of input data to drive the models. Once a model is complete, the input data can be obsolete, requiring additional analysis and re-validation. This hinders the ability to provide accurate, timely models. This process is greatly improved by the use of OLAP data cubes to provide input data. Most simulation software packages have the functionality to read data from Excel workbooks. Data cubes provide data in a standard Excel pivot table format that can be easily updated to current data. Using OLAP cubes can significantly reduce the time to bring a simulation model to market, by streamlining the access to input data. Cubes also provide a quick, easy method to update data to the most current information. This ultimately provides a simulation analyst with a means to produce accurate, relevant and timely simulation solutions.

Simulating a Virtual Customer Service Center
Daniel S. Riley (IBM Global Services)

Abstract:
In today’s global economy, it may be cost efficient to spread customer service center work across multiple cities, countries or continents. The obstacle to enabling these efficiencies is sizing how much work to send to each location and how many agents it would require to handle that work. Further complications include the potential desire for one-way sharing and prioritization of centers to handle calls. Traditional call center models are inadequate to accurately size virtual customer service centers, as sub-optimization may be counterproductive to the whole. Using simulation, this gap can be bridged, allowing a truly optimal staffing solution for a call center that may have agents taking the same set of calls in New York, Hong Kong and anywhere in-between.

Process Management for Federal Agencies
Ed Stephan (CACI International)

Abstract:
This case study examines the variation in the key attributes of discrete event simulation when dynamic modeling is applied to a variety of business processes within the Federal Civil Government. The key simulation attributes that will be compared and contrasted include: the rate of arrival of work entities, the degree to which work is subdivided, the frequency with which work is batched, activity duration, the degree of correlation between successive work entity arrivals, and perspectives on simulation results. The business process application areas examined include personnel recruitment, help desk operation, call center operation, logistics, image capture, and image processing.

Probabilistic Simulation Analysis of Risk Potential in Cost Management Programs
Irina Warthen and Kris Arvind (CGN & Associates)

Abstract:
Today’s business leaders recognize that advanced decision making capabilities using simulation techniques can directly improve their organization’s performance. In this case study, CGN will describe how we helped a large equipment manufacturer identify areas of potential risk within their portfolio of cost management initiatives for a new product line. In order to define the interdependencies between multiple cost initiatives, their relevant business processes and their subsequent business drivers, a sophisticated understanding of prevailing market conditions was necessary. The case study will explain how the simulation model was designed to analyze the variation of individual business drivers such as steel prices, currencies, labor rates, machine reliability etc., and then predict the potential for success or failure of the overall cost management program using probability distributions. Finally, the paper will demonstrate how intelligent input and output reports can provide a degree of scalability to such simulation models and thus, allow for incorporation of future initiatives.

Simulation Model for Analyzing Effects of Employment Tax Incentive Programs
Robert Yerex (Unicru)

Abstract:
In the United States, a variety of federal, state and local tax incentive programs exist for the purpose of encouraging private sector corporations to employee certain targeted classes of disadvantaged workers. This study demonstrates the use of OR modeling and simulation approaches in the analysis of the federal Work Opportunity Tax Credit (WOTC) program in terms of its impact, and the potential for increased value to corporations seeking to optimize their participation in the program.

Process Simulation as a Key to Uncovering True Organizational Value
Kris Arvind (CGN & Associates)

Abstract:
As organizations struggle to remain competitive in a global sourcing environment, business leaders are challenged to increase overall yield and increase product/process quality, while lowering total costs and cycle time. CGN will describe how we helped a division of a large equipment manufacturer model and simulate their organizational processes that identified key levers critical to maintaining competitive advantage. Traditional process analyses determine inefficiencies and non-value add by conducting a linear pass-through the process. Simulation affords business leaders the capability to build a complex model of their business that simulates reality. In this case study, we will illustrate how CGN’s approach: Identified key business parameters and formulated a model that provided a hypothesis-based approach to solution building; Visualized multiple organizational processes simultaneously, resulting in a quantifiable solution set matrixed against KPI value drivers; and Provided guidance regarding initiatives to achieve an optimal organizational / process structure.

Using Discrete-Element Simulation to Evaluate Laboratory Instrumentation Needs
Valerie G. Caryer Cook (DaimlerChrysler and Lawrence Technological University )

Abstract:
Strategic planning for laboratory operations capital investment involves predicting the usage of equipment used in daily operations. Development laboratories are particularly difficult to model equipment usage, due to the large variation in the types of projects or tests conducted and complex equipment needs. Discrete-element simulation can be used to effectively model the use of equipment inventory in this highly variable environment.This presentation demonstrates the use of a WITNESS discrete-element simulation model to optimize capital investment in data acquisition equipment for an NVH laboratory. The model is used to evaluate the effect of increasing investment in data acquisition chassis units versus component cards to maximize the efficiency of the instrumentation process. The model was also adjusted to reflect future laboratory demand conditions to increase the accuracy of the prediction. The study resulted in a recommendation of the most cost-efficient combination of chassis units and component cards for the laboratory.

Validation of a Time-of-Supply Inventory Policy Through Simulation
Benjamin Martin and James Francis (Sara Lee Intimates & Hosiery)

Abstract:
A popular safety-inventory policy utilized in the apparel industry is time-of-supply. Time-of-supply specifies inventory held to protect against forecast error and supply disruptions in terms of safety-time rather than safety-stock. Determining the proper safety-time is critical to ensure customer service. This case study presents a method for determining safety-time for raw materials in an apparel manufacturing operation. Additionally, the results of a simulation study that validated the time-of-supply inventory policy are presented.

Using Monte Carlo Simulation to Choose Among Alternative Statistical Definitions
David M. Northcutt (IBM Global Services)

Abstract:
We frequently need to come to agreement with our customers about the formulas that will be used to calculate various statistics relating to the measurement of contractual terms and conditions. In many cases there is only one generally accepted operational definition, but in other cases, such as computing percentiles, there are multiple definitions that can be considered. This presentation looks at work I have done using Monte Carlo simulation to compare two different formulas for estimating a population percentile from small samples; the application is the analysis of benchmarking data. This work compares two different percentile calculations by comparing the performance of both approaches under multiple probability models with respect to estimation of the 10th, 25th, 50th, 75th, and 90th percentiles. In addition, a non-parametric confidence interval generation method for each of these percentiles was also examined to determine how well it performed relative to its stated 95% confidence level.

Simulating Customer Ownership Patterns for Technology Goods in the Mature Market Phase
Chris Robson (Parametric Marketing LLC)

Abstract:
This presentation shows how we have applied System Dynamics modeling tools and techniques to the problem of marketing mature technology products. There is a large body of work dating back to the early 50s on how technology products are initially adopted, but many companies are now struggling with products that are well into the maturity phase -- their customer base has saturated and sales come from replacement and competitive shift. We will present how simulating the installed base using SD techniques has helped our F500 clients understand the effects of upgrades, loyalty, ownership patterns and installed-base marketing as they try to capture the maximum value from their customers in this difficult market phase.

When Things Go Wrong
Darrell Starks (Rockwell Automation)

Abstract:
When performing simulation projects, all analysts or project managers are taught to follow the simulation process: Define, Formulate, Verify/Validate, Analyze, and Recommend. However, few of us are taught what to do when all does not go well. Most projects have at least one bump in the road which if not smoothed out as soon as possible can become a crater that can swallow the project and all team members. This presentation gives examples of when things went wrong and what was done to correct these issues.