Value Stream Mapping is an important tool in the implementation of lean manufacturing. It identifies the waste in the system, paving the way for a successful lean implementation. VSM is a paper and pencil tool that captures the state of the system at the state it was drawn. Simulation can be combined with value stream mapping to give it power and flexibility in order to dynamically capture the state of the system. Component-based modeling divides a simulation model into a number of smaller simulation models each encapsulated in a component resulting in a set of simulation building blocks. These blocks can be used for the purpose of developing value stream maps as they are designed to be generic, reusable, and appear exactly like the traditional VSM icons. This paper introduces the Value Stream Map Simulator using ExtendSim (VSMSx) as a powerful tool designed to facilitate the implementation of lean manufacturing by simulating the value stream map. Compared to traditional value stream mapping, this tool outputs more quantitative information about the system under study and various scenarios allowing for better decision making, thus paving the way for successful lean implementation.

Download paper

Partnering with Indiana University Medical Group (IUMG), the author focused on outpatient care:

	Address the issue of missed appointments
	Analyze variability in patient flow
	Analyze performance of phone system

...to analyze and improve patient flow at an outpatient clinic of the Indiana University Medical Group. Queuing concepts were used to uncover sources of variability and to generate ideas to improve clinic operations that would mitigate the undesirable effect of variability. A process map, that matched the process at the clinic, was developed and validated. Data on task times was collected by observing the process with stopwatch or from historical records. A simulation model corresponding to the process map was developed, and the output was validated. Several ideas to modify clinic operations were tested on the validated simulation model. The overall result was an improvement in both the mean and the standard deviation of patient wait time, as well as higher utilization of physicians’ time. The clinic has implemented several of their recommendations and experienced improvements consistent with model predictions.

Download final presentation

Purdue University has partnered with an outpatient medical facility operating 20 clinics. Their first project involves testing changes in the registration process at a primary care facility. Process flow maps have been developed to document current processes. They are modeling how reducing variability and excessive waiting early in the visit reduces variability and waiting throughout the patient visit. Their second project specifically addresses the routing and staffing of phone calls at the same clinics. Purchased call management software provides best practices for call routing. Research is focused on the cost benefit trade off of staffing levels and providing recommendations for improving the routing of phone calls.

February 1, 2008 - Unfortunately my funding as a PhD student ran out at the time I received the Extend research grant. My professors accepted my work to date for my dissertation and I defended in August of 2007. Results of the project, a manuscript which is under review at IIE transactions, is confidential until it is approved for publication.

Thank you for the assistance throughout the program.

Many service providers offer customers the choice of either waiting in a line, or going offline and returning at a dynamically determined future time. The best known example is the FASTPASS® system at Disneyland. To operate such a system, the service provider must first make an up front decision on how to allocate service capacity between the two lines. Then, during system operation, he must dynamically provide estimates of the waiting times at both lines to each arriving customer. The estimation of offline waiting times is complicated by the fact that some offline customers do not return for service at their appointed time.

Managing Service Systems with an Offline Waiting Option and Customer Abandonment shows that when demand is large and service is fast, for any fixed capacity allocation decision, the two-dimensional process tracking the number of customers waiting inline and offline collapses to one dimension, and characterize the one-dimensional limit process as a reflected diffusion with linear drift. Next, we use the one-dimensional limit process to develop approximations for the steady-state distribution of the number of customers waiting inline and offline, the steady-state probability of abandonment from the offline queue, and to dynamically estimate inline and offline waits for each arriving customer. We conclude by considering a cost model, and optimize the up front capacity allocation decision.

Download complete report

November 21, 2006 - In an effort to improve the customer experience in service systems, we study a queueing system in which some customers immediately join the queue and some return at a later time for immediate service. We develop methodology to estimate real-time delays and return times at which immediate service can be guaranteed. One factor complicating the estimation procedure is that some customers promised immediate service will not return. We will use ExtendSim to build a simulation study verifying the accuracy of the proposed approximations.

The development of such methodology has applications for customer contact centers and hospital emergency rooms. In the customer contact center setting, callers may be given the option between immediate service and postponed service. In the hospital emergency room setting, patients in critical condition should receive immediate service, but patients in non-critical condition can be given an accurate “appointment” time later that day. ExtendSim’s rich discrete event simulation tool set will help us to refine our methodology for these applications contexts.

February 21, 2007 - We studied a queueing system in which customers have the choice to either join the queue immediately or to come back later. At a first step, the simulation we built works under the assumption that the waiting costs for the customers are equal for both lines and depend only on the ratio of service rate between the two queues. We run the simulation for possible values of this ratio and for different probabilities of abandonment of the offline queue (the abandonment time was assumed to be Exp(gamma)). We collected the results for the mean queue length of the two lines, the probability of abandonment and the queue ratio for the n-th system in a 95% confidence found after running the simulation enough times to generate over 1,000,000 arrivals. The results then were compared with our approximations that were proved to be quite accurate. (Click on the table to download the complete Excel table.)

As a next step, we are thinking of making our model more complicated by removing the assumption of equal waiting costs and by adding another similar queue. This model will be applicable in circumstances like entertainment parks were customers want to enjoy two rides and have to choose whether they will join the standard queues or they will take a fast pass to come back later.

June 2, 2007 - Unfortunately, the last quarter we didn't make any progress in our ExtendSim model because we were working on the mathematical background of the model trying to prove the computational results. We are currently working on an extension of the model; we study a queueing system in which customers have the choice to either join the queue immediately or to come back later, but now we allow the abandonment times to be generally distributed. We will soon have the ExtendSim model for this generalization and the simulation results.

August 26, 2007 - This quarter we worked on a generalized version of our model. More specifically, we worked on a single-server queueing model in which customers may choose between waiting for service in real-time, and returning for service at a dynamically specified future time point. Customers waiting for the delayed service option may not return for service. Customers arrive at the system at a rate lambda and are served with rate mu and their service follows a generalized processor sharing discipline. More details on the model can be found in "A Single-Server Queue with a Delayed Service
Option and Exponential Abandonment Times". Now we are working with the simulation model to look at the accuracy of the approximations by the convergence results suggested in Theorems 1-3 of the paper.

February 6, 2008 - Paper submitted in the Fall of 2007 and is being revised. A minor change was made to the model so are currently rerunning some of the simulations based on the new model. Results will be incorporated in the new version of the paper.

Variability in production is one of the largest factors that negatively impacts construction project performance. A common construction practice to protect production systems from variability is the use of buffers (Bf). Construction practitioners and researchers have proposed buffering approaches for different production situations, but these approaches have faced practical limitations in their application.

In Multiobjective Design of Work-In-Process Buffer for Scheduling Repetitive Building Projects, a multiobjective analytic model (MAM) is proposed to develop a graphical solution for the design of Work-In-Process (WIP) Bf in order to overcome these practical limitations to Bf application, being demonstrated through the scheduling of repetitive building projects. Multiobjective analytic modeling is based on Simulation–Optimization (SO) modeling and Pareto Fronts concepts. Simulation–Optimization framework uses Evolutionary Strategies (ES) as the optimization search approach, which allows for the design of optimum WIP Bf sizes by optimizing different project objectives (e.g., project cost, time and productivity). The framework is tested and validated on two repetitive building projects. The SO framework is then generalized through Pareto Front concepts, allowing for the development of the MAM as nomographs for practical use. The application advantages of the MAM are shown through a project scheduling example. Results demonstrate project performance improvements and a more efficient and practical design of WIP Bf. Additionally, production strategies based on WIP Bf and lean production principles in construction are discussed.

	Download final report (664 KB): "Work-In-Process Buffer Design Methodology for Scheduling Repetitive Building Projects" by Vicente González and Luis F. Alarcón
	Download Extend OR 6 model files (5.7 MB) You may view these models in ExtendSim or download the ExtendSim Demo.

April 18, 2005 - "The specific topic related to my PhD Research is buffer management in repetitive building projects. Like buffer I have been using work in process between dependent activities or production processes in building projects. The use of discrete simulation and the process interaction strategy is fundamental in my research to model the effect of different buffer sizes in the variability of production processes and the performance of projects construction (schedule and budget).For this reason, I have been working with ExtendSim and its powerful capacities to understand the effect of the buffers and to develop heuristic and analytic expressions at different levels of construction projects development to the decision making process."

February 16, 2006 – "During February, I'll be preparing an ISI paper and another one for a specific conference (International Group for Lean Construction, XIV Conference, Santiago, Chile, 2006)....so, I expect to finish these papers at the end of February or at the beginning of March, but these versions will be draft papers..the definitive paper should be ready in May or June..."

July 5, 2006 – Presented paper about WIP Buffer Design and Management at the 14th International Conference for Lean Construction, Santiago, Chile, July 2006.

December 29, 2006 - "My research is finishing. Next July (2007) I'll have my final exam about my research with the international advisors. About my models, I have developed a generic simulation modeling approach to size WIP Buffers in repetitive projects. By using Simulation-Optimization, I can find the optimum WIP Buffer size that minimizes the project schedule and maximizes the average productivity. During January and March I'll use neural networks to model the WIP Buffer size by using only a few parameters. These parameters are production rates, variability of cycle times, production amount, and numbers of processes."

January 27, 2008 - Research has been completed. It was developed using a methodology to manage uncertainty in repetitive building projects using Work-In-Process (WIP) buffers (Bf). The methodology has three hierarchical levels of application in production: strategic, tactical, and operational. At strategic level, a multiobjective analytic model (MAM) based on concepts of Pareto Front and simulation-optimization (SO) modeling was built. At the tactical level, SO modeling is directly used. At the operational level, a statistical model called Rational Commitment Model (RCM) was applied (in this level, simulation was not used). The methodology was applied in real and hypothetical cases demonstrating improvements in cost and productivity.

González, V., Alarcón, L.F. and Gazmuri, P. (2006). Design of WIP Buffers in Repetitive Projects: A Case Study. 14th International Conference for Lean Construction, Santiago, Chile, July 25th – 27th.

González, V. and Alarcón, L.F. (2005). Design and Management of WIP Buffers in Repetitive Projects (White Paper).

González, V., Rischmoller, L. and Alarcón, L.F. (2004). Management of Buffers in Repetitive Projects: Using Production Management Theory and IT Tools. PhD Summer School, 12th International Conference for Lean Construction, Helsinore, Denmark, August 5th – 8 th.

González, V., Rischmoller, L. and Alarcón, L.F. (2004). Design of Buffers in Repetitive Projects: Using Production Management Theory and IT Tools. 4th International Postgraduate Research Conference, University of Salford, Manchester, U.K., April 1st –2nd.

González, V. y Alarcón, L.F. (2003). Buffer de Programación: Schedule Buffers: A Complementary Strategy to Reduce the Variability in the Processes of Construction, Revista Ingeniería de Construcción Pontificia Universidad Católica de Chile, Volume 18 . Nº 2, pp 109 - 119, Mayo - Agosto 2003

July 8 to 9, 2010 - Paderborn, Germany
Application of Re-estimation in Re-planning of Software Product Releases
International Conference on Software Process (ICSP) 2010
Ahmed Al-Emran, Anas Jadallah, Elham Paikari, Dietmar Pfahl, Günther Ruh

Ahmed Al-Emran

Context -- Uncertainty is an unavoidable issue in software engineering and an important area of investigation. This paper studies the impact of uncertainty on total duration (i.e., make-span) for implementing all features in operational release planning.

Objective -- The uncertainty factors under investigation are:

	The number of new features arriving during release construction
	The estimated effort needed to implement features
	The availability of developers
	The productivity of developers

Method -- An integrated method is presented combining Monte-Carlo simulation (to model uncertainty in the operational release planning (ORP) process) with process simulation (to model the ORP process steps and their dependencies as well as an associated optimization heuristic representing an organization-specific staffing policy for make-span minimization). The method allows for evaluating the impact of uncertainty on make-span. The impact of uncertainty factors both in isolation and in combination are studied in three different pessimism levels through comparison with a baseline plan. Initial evaluation of the method is done by an explorative case study at Chartwell Technology Inc. to demonstrate its applicability and its usefulness.

Results -- The impact of uncertainty on release make-span increases – both in terms of magnitude and variance – with an increase of pessimism level as well as with an increase of the number of uncertainty factors. Among the four uncertainty factors, we found that the strongest impact stems from the number of new features arriving during release construction. We have also demonstrated that for any combination of uncertainty factors their combined (i.e., simultaneous) impact is bigger than the addition of their individual impacts.

Conclusion -- The added value of the presented method is that managers are able to study the impact of uncertainty on existing (i.e., baseline) operational release plans pro-actively.

Keywords: Operational release planning; Uncertainty; Impact analysis; Discrete-event simulation; Heuristic optimization; Explorative case study

The planning of releases is a key factor for the success or failure of incremental product development. The challenge here is deciding which features of a product are to be offered in which release (i.e., strategic planning) and how to carry out the feature development tasks within in each product release (i.e., operational planning). This planning process is extremely important since a bad release plan may cause late delivery of high-value features, unsatisfied customers, budget overrun, and decreased competitiveness. Existing research in release planning mainly takes place at strategic level. Thus, the research area is still in need of developing efficient and effective methods and tools that consider the operational feasibility of a proposed strategic release plan. The primary focus is allocating resources to feature development tasks (e.g., design, implementation, and test) with a “reasonable” release make-span. Three research objectives have been defined for the proposed research project in the area of operational release planning:

July 20, 2009 - We have recently submitted a journal article that studies the impact of uncertainty in operational release plans. We present a simulation-based method that considers planning parameters as the primary source of uncertainty. The method measures the impact of the planning parameters’ variations on total schedule time (make-span). More precisely, with our method product managers can analyze proactively the impact of both individual and combined variation of planning parameters using various levels of pessimism. We are going post the link to the article as soon it is accepted and published to the submitted journal.

February 26, 2009 - A re-planning method called H2W has been proposed to answer the "how?", "when?", and "what?" questions to re-plan of an existing product release strategy. At each of the re-planning iterations, H2W either provides a new improved plan or states that an improvement does not exist. As an instantiation of this method, an ExtendSim prototype has been developed as shown. The related research work has been published in International Conference on Software Process (ICSP) 2009.

We have further restructured the original research project during my comprehensive examination. At first, we thought we will be working on all of the parts in parallel. However, we revised our strategy and decided to work on the simulation part exclusively first and finish it completely. This project will culminate in two conference papers and one journal article which we would be happy to share with you upon their acceptance.

URL

Modelling of discrete operational events that impact the logistic support ramifications of unmanned aerial vehicle availability. Logistical studies of unmanned aerial vehicles (UAVs) are quite immature at this stage. This is due to the fact that most UAVs are brought into service under Advanced Technology Concept Demonstration (ATCD) programs and this method of introduction into service sees little effort put into precursory logistic planning - it is usually an after-thought. As such, while there has been a great deal of modeling of standard systems engineering logistic support, while Mr. Joyce has found none that he can identify in the literature pertaining to rapidly acquired aircraft, let alone unmanned aircraft systems.

March 30, 2010 - The project has been slightly delayed due to a stolen laptop. The project will be completed by September of 2010 and a project update will be posted in May of 2010.

October 2009 - Continuing to develop model (my work has been quite slow this month due to work commitments). I envisage having the first iteration of the model ready over the Christmas break and will commence validation of the model at this time.

September 2009 - Compiling data for use within model.

August 2009 - Overseas gathering data from a variety of sources (NASA, RAND, US Air Force, US Army, US Navy and contractors attending the UV Europe (Wales) and AUVSI North America (Washington DC) conferences).

July 2009 - Model development commencing with conceptual analysis and then beginning to map the ExtendSim blocks into their required functions.

June 22, 2009 - Unfortunately, progress in my project (UAV logistic support modelling) has been hampered over the last three months, but is getting underway. The plan from here is to develop the model over the next couple of weeks and then begin the analysis phase. I will be able to report much more in my progress report for the next quarter.

n/a

URL

This research work is part of collaboration between the University of Brasilia (DF, Brazil) and the University of Ottawa (ON, Canada). It aims to improve the quality of teaching materials of the former university with the know-how of the latter.

The teaching of Physiology - an important segment of Biology - strongly depends on dynamic simulations and experimentations. This need is specially augmented in e-learning environments where the quality of the didactic material is a factor. Most teachers of Physiology are unable to produce high-quality simulations because they are generally non-acquainted with code-based programming. To deal with this problem, this twofold project intends to first develop some lessons using block-based programming tools and, second, in the case of lectures, scientific papers and presentations in scientific congresses - to encourage average teachers to use those tools more frequently.

A broad review of the currently available block-based programming tools pointed to ExtendSim as one of the most suitable simulation software to meet our purpose. We plan to build lessons in biology and health sciences for students of the University of Brasilia. During the first term of the project, we will produce lessons for a course entitled "Introduction to Biostatistics". The point here is to overcome the natural resistance of those students vis-a-vis biostatistics, by stressing the use of 2-D simulations. Concurrently, this phase of the work will allow us to get used to ExtendSim.

During the second term, we intend to develop more complex 3D simulations to support classic physiology lessons. The point here is to help teachers and students to better handle the high dynamic and complex features of physiological systems.

January 4, 2010 - Due to health issues project completion date extended to July 2011.

July 2, 2009 - Name of the model: Average estimate
This first simulation aims to help students to better understand the differences between three popular statistics parameters used to estimate the average of a sample. The user is initially stimulated to change the distribution of a sample and to observe the plot of the new distribution. Then he may assess the new mean, median, and mode computed by the model. This simulation is being developed using some ExtendSim blocks such as sliders (to allow changes in the sample distribution), equations (to compute the parameters), and plot (to display the shape of the sample). I suppose that this simulation should be completed in about three weeks. Therefore, for the next quarter update, I would present the screen shots of the model. The development of a new simulation should be in progress as well.

March 31, 2009 - This is my first update. I've received my license approximately one month ago and since that I have been studying the manual, especially on those subjects concerning data managing and plotting tools. I am finishing the theoretical design of my first model, which will show up how parameters like mean and median change as the user changes the sample distribution.

VENEZIANO, W. H. ; ROCHA, A. F. ; GONCALVES, C. A. ; PENA, A. J. ; CARMO, J. C. do ; NASCIMENTO, F. A. O. ; RAINOLDI, A. . Confounding factors in water EMG recordings: an approach to a definitive standard. Medical & Biological Engineering & Computing, v. 44, p. 348-351, 2006.

GONCALVES, C. A. ; CARDOSO, I. P. ; CARVALHO, T. C ; FREIRE, V. D. . Action Potential Propagation/Propagate‹o do Potential de A‹o /(CD-ROM, Simulation Software)/Publisher: Editorial Universidade de Brasilia, DF, Brazil, 2005.

NOGUEIRA, Edna L’via ; CORREIA, Andressa da Costa ; SILVA, R. A. ; GONCALVES, C. A. ; MOTA, Yomara Lima. Electromyographic assessment of the effect of age over paravertebral muscles/O efeito do envelhecimento nos mœsculos paravertebrais lombares investigado pela eletromiografia. In: XI Congresso Brasileiro de Biomec‰nica, 2005, Jo‹o Pessoa. Anais do XI Congresso Brasileiro de Biomec‰nica, 2005. v. I. p. 1-4.

CARDOSO, I. P. ; OLIVEIRA, C. C. S. ; FREIRE, V. D. ; GONCALVES, C. A. . A methodology for building hypermedia courses in Physiology/Uma metodologia para desenvolver aulas em formato hiperm’dia para o ensino de fisiologia. In: FeSBE2004-XIX Reuni‹o da Federa�‹o de Sociedades de Biologia Experimental, 2004, çguas de Lind—ia, Brazil, 2004.

BRASIL-NETO, J. P. ; GONCALVES, C. A. ; GON‚ALVES, C. ; LIMA, R. R. F. ; PESSOA, V. F. . Development of a Computer-Based Method System for Studying Human Stereopsis:Contribution to the Study of Human Speed of Detection of Visual Depth. IEEE Computer Society, v. 1, p. 134-138, 1997.

URL

Although there has been extensive research and simulation conducted within the production and supply chain domain, in respect to the United States economy, research and simulation related to the services sector lags behind. This is somewhat paradoxical, since the service sector represents over 75% of the United States labor force and approximately 75% of the United States Gross Domestic Product. This research project is unique in that it completely focuses on the services sector, and more specifically targets the leisure and recreation industry.

This research and simulation project proposes to model, in a two phased approach, the service activities associated with a major ski resort in Utah by utilizing granular RFID data from the resort’s lift access system. The resort has many years of experience with RFID lift access systems, and is a notable industry leader. The RFID data will support the development of daily skier visit forecasts, frequency distributions and transition probabilities associated with the resort’s extensive lift system.

Phase 1
The first phase of this research project is to create a discrete event simulation model, of a major ski resort, to measure and possibly optimize the effects of a set of decision variables on a specific group of metrics such as inline and offline queuing, lift capacity upgrades, and real time queue information. Currently there are five proposed decision variables each having several experimental levels. Servicescape constructs and SERVQUAL measures will be evaluated.

Objective: The main goal is to make managerial recommendations of reasonable and appropriate alternatives related to demand and capacity management of the ski resort. The simulation will support a business case for capacity management recommendations. The management will market any implemented enhancements as competitive advantages to their customer base.

Phase 2
The second phase of this research project is to construct an agent based simulation model that replicates the behavior of several skier personas with as many as 5,000 agents active during one simulated day and then aggregate the results over a full ski season. An agent based simulation makes it possible to analyze the relationship between individual micro behaviors with the aggregate patterns of many agents that eventually emerges over a day or ski season. An agent based model is applicable to skier behavior because “skiers” are intelligent, act upon their own local information, have memory, and are goal oriented.

Objective: Due to this ski resort’s unique ability to passively gather RFID data on each skier during each day, the resort has the ability to leverage the data into numerous highly targeted customer relationship campaigns. The return on investment on CRM type projects are in general difficult to quantify, so most ski resorts in North America are not fully utilizing their available CRM system. This agent based simulation will support a business case for additional marketing resources.

July 14, 2010 - After an extensive period of data cleansing and scaling, I have been able to create interarrival schedules for each "process" in my proposed model. These "processes" of course, are the individual ski lifts at the resort. I built a database for the schedules and created a data feed into the model and ran the simulation for one day. I quickly noted that I had a major computer over utilization problem. After some model modifications to reduce CPU utilization, I still have a "hardware" problem. I have several options:

Currently, I am requesting a much more powerful computer with 8 GB of RAM for temporary loan from our computing staff.

While I am waiting during summer, I am validating more RFID data and also preparing for a quick trip back to the resort in August to obtain 2009/2010 data.

January 10, 2010 - I have recently returned from Utah were I was able to obtain complete RFID gate access data for the year 2008 - 2009. This was a major research goal accomplishment. The resort is in a growth phase and it took six months to arrange a meeting. I briefly met and talked with the president and then worked with the marketing director for almost a day downloading detailed data reports on skier days and lift rides. I will now build out the full resort model and add the interarrival times for each lift for each day of the season. The next steps are model verification and full factorial experimental model design.

August 19, 2009 - Over the last few months, I have been able to develop custom cumulative discrete distributions for skier arrival for each ski lift at the research site. A model for each lift has been created that includes balking behavior, a random batching process, entity persona attributes, a specific ride time, and an unbatching process. The current model has been constructed in a hierarchical design to force detail to lower levels of the model.

The next step is to travel to Utah and gather skier transition data between each lift, so a transition probability distribution can be created for each lift pair. This task will be by far the most time consuming and probably the most interesting, and in the end the most valuable to creating a valid ski resort discrete event model.

King, Michael A. “A Realistic Data Warehouse Project: An Integration of Microsoft Access® and Microsoft Excel® Advanced Features and Skills” Journal of Information Technology Education, The Informing Science Institute, Vol. 8, 2009

King, Michael A. “The Implementation of DSpace at the Intercontinental Hotels Group: A Knowledge Management Project Success.” Journal of Cases on Information Technology, The Idea Group, 10(2).

Conference Papers:
King, Michael A., “A Strategic Assessment of The Higher Education Industry: Applying the Porter’s Five Forces for Industry Analysis.” Proceedings of the annual Southeast Conference of the Decision Sciences Institute, Charleston, SC, February 2009.

King, Michael A., Zobel, Christopher. “Applying the R4 Framework of Resilience: Information Technology Disaster Risk Management at Northrop Grumman,” Proceedings of the annual Southeast Conference of the Decision Sciences Institute, Orlando, FL, February 2008.

Arnette, Andrew N., King, Michael A., “The Business Value of Geographical Information Systems In Business School Curriculum.” Proceedings of the annual Southeast Conference of the Decision Sciences Institute, Orlando, FL, February 2008.

King, Michael A., et. al. “Information Technology Project Retrospectives: The Northrop Grumman Sperry Marine Story.” Proceedings of the annual Southeast Conference of the Decision Sciences Institute, Savannah, GA, February 2007.

King, Michael A. “The Evaluation and Implementation of DSpace at the Intercontinental Hotels Group: A Project Portfolio Management Approach.” INFORMS Annual Conference, Pittsburg, PA, November 2006.

King, Michael A. “Is the Value of a Ph.D. Always Negative? A Case Study Comparing Real Option and Stochastic Discounted Cash Flow Valuation Techniques.” Decision Science Institute Annual Conference, November 2005.

URL

The project will be the design and operation of integrated sites under uncertainty and disruption.

We will consider both existing and newly integrated networks of production processes.

	For existing networks, we will examine modifications to either produce new products or increase current production capacity.
	For new networks, we will examine the overall design to meet reliability and capacity targets.

We will consider three major types of uncertainties:

Conceptually this is a very difficult problem because the uncertainties in the demands are continuous, typically represented by probability distribution functions, while the uncertainties in the availability and supplies are discrete in nature and correspond to major disruptions. To address this problem we intend to first investigate a probability measure that incorporates both the discrete and continuous uncertainties based on some of our previous work (Straub and Grossmann, 1990). We also intend to investigate the use of semi-analytical models for inventories in supply chains under disruptions (Snyder and Shen, 2006). We then intend to incorporate these elements in a multiperiod MILP model for optimizing the design and operation in the face of the three types of uncertainties. While our initial goal is to address design problems using somewhat more simplified models (eg linear mass balances for processes, neglecting changeovers in multiproduct plants), we intend to develop in a second phase more detailed models for operations that can be validated with plant data and used for day-to-day decision making of operating sites.

May 27, 2010 - During the last quarter, we continued working on combining specialized tools for mathematical optimization and discrete rate simulation in ExtendSim for the design of integrated manufacturing sites. In previous updates, we explained how simulations were run after using equation-oriented, mathematical programming software to obtain the optimal configuration of an integrated site. Recall that the model used in the equation-oriented software is an approximation of the model used in ExtendSim. We exploit state-of-the-art optimization technology for initially reducing the search space for the design of integrated sites. The idea of running the simulation after the optimization is to validate and fine-tune the design obtained in the optimization step using an approximate model.  

In the past months, we developed an idea to take advantage of ExtendSim in a new way. Namely, we used discrete event modeling tools embedded within the discrete rate model of an integrated site to construct samples of random events. The ExtendSim setup for representing the occurrence of random failures in the processing units of an integrated site was developed at the Dow Chemical Company and it’s summarized below. 

Unplanned failures are generated randomly from time-between-failures (TBF) and time-to-repair (TTR) distributions. Each failure mode is represented by a single item. For a unit with multiple trains (and same failure parameters), the item is duplicated so that there is one item per failure mode per train. A Queue Equation block is used to correctly model multiple failures. If a failure is generated when the unit is already down, that failure is held back until the unit is back in operation. If the unit is experiencing a rate cut, then the new failure is allowed to proceed, and the minimum rate of all active failures is used to determine the rate for that unit. 

The modeling technique just described was used to generate, for each simulation run, a list of exact times when failures occurred and the time they took to be repaired. This list was used to define failure scenarios in the equation-oriented optimization model. In this way we were able to obtain a closer approximation of the detailed model in ExtendSim and the equation-oriented mathematical programming model that we use for getting an initial design. 

We have begun to write a journal paper that will illustrate the way we use ExtendSim as part of our research on integrated site design.

February 2, 2010 - Last quarter I had to prepare for some conferences and wrote and corrected a research paper that we are about to submit to a Chemical Engineering Journal. Hence, I do not have any significant progress to report to you for the last quarter.

November 6, 2009 - In our last update we described how we used discrete rate models to validate the results obtained in the design optimization step. As a result of this procedure we obtained Pareto curves of minimum cost and maximum robustness against uncertainties. In that update we reported that the results predicted by the optimization model were in reasonable agreement with the simulation. However, we considered that the match between the mixed-integer optimization model and the discrete rate simulation could be improved. The results of our efforts during the last quarter show a better agreement between the mathematical programming model and the discrete rate simulation. For illustrating our results we used the following system: an integrated site made up of 9 plants that serve two external customers modeled as product sinks. We assume there is a given demand rate for the two products represented as the maximum total flow in the discrete event simulation. For confidentiality reasons we do not provide any more details on the process.

We developed a new mathematical programming model that corresponds to this integrated sites, with the objective of maximizing the probability of delivering the products at the demanded rate consistently. The optimization chooses the production units, their capacities, and the size of their storage tanks. One of the key ideas this last quarter was to make the objective function of this optimization similar to that of the network flow model embedded in the discrete rate simulation. The results of the optimal design predicted by the mixed-integer model are in very good agreement with the performance obtained by simulating the design using the discrete rate model.

Another key idea implemented during the last quarter was to add operational logic to the way that intermediate storage is managed. The optimal design of the integrated site installs extra plant capacity to keep the intermediate storage tanks from emptying. However operating at full plant capacity all the time disrupts the flow balance in the network. To avoid this effect, we added a valve between each plant and its storage tank. The valve allows the use of the full plant capacity only when the inventory falls below a set-point. The average inventory predicted by the mixed-integer optimization model is in very good agreement with the simulation results and the standard deviation is in the same order of magnitude.
Next quarter we will work on how to use ExtendSim to make improvements on the design obtained by mixed-integer optimization.

August 4, 2009 - The research project “Design and operation of integrated sites under uncertainty and disruption” represents a collaborative effort between Carnegie Mellon University and The Dow Chemical Company. The research problem can be stated as follows. We are given a set of finished products for which there is a demand that can be deterministic, or given by a specified probability distribution. The raw materials and the intermediate products involved in the integrated production site are known. The supply of raw material can also be deterministic, or given by a specified probability distribution. There are a predetermined number of steps involved in the transformation of raw materials to intermediate and finished products. Each of these steps is carried out by a different plant that can have multiple production units. Plants are connected through the flow of intermediate products between them. The network formed by the plants and their connections represents an integrated site. The plants in the integrated site are subject to random failures that result in corrective maintenance. As a result, plants experience some amount of down time during which production is decreased or stopped all together. Since plants are interconnected in the integrated site, a failure event will propagate downstream or upstream, forcing some other plants to decrease or stop their production. If these events are not considered while designing the integrated site, there is a risk that product demand will not be consistently met. The objective of this work is to develop a systematic method to determine the trade-off between capital investment and process robustness in the design of an integrated site subject to discrete and continuous uncertainties. The main design decisions considered are increases in process capacity, addition of parallel units, and addition of intermediate storage.

Discrete rate models are frequently used for simulating the design of integrated sites at Dow. At Carnegie Mellon we are proposing to leverage these modeling techniques as a part of an optimization & simulation approach for the design of such process networks. The optimization step is performed using rigorous mathematical programming techniques. However, in order to build an optimization model of practical size we are forced to make a set of approximations to the detailed behavior or the integrated site. Our goal is to use ExtendSim discrete rate modeling capabilities to correct and improve the initial design provided by the optimization formulation. For instance, information provided by the optimization technique, such as sensitivity around the optimal solution, could be used to design a set of simulation experiments in ExtendSim.

So far we have used ExtendSim discrete rate simulator to compare the actual performance of different designs obtained in the optimization step. As a result of this procedure we obtain Pareto curves of minimum cost and maximum robustness against uncertainties. We are working on developing a systematic procedure for using the full capabilities of ExtendSim, not only to validate, but also to correct and improve the optimal design obtained through the mathematical programming formulation

n/a

URL

Uncertainty is an inherent part of production systems. In construction processes, production variability emerges as one of the most typical representations of uncertainty. The negative impacts of variability over construction processes demands effective solutions to mitigate its effects on the accomplishment of projects. One of the tools to deal with variability in construction processes is the incorporation of buffers, which can help reducing the negative impacts of variability on projects. Despite the fact that buffering strategies have been developed for production environment in construction, there is no evidence of specific applications of these strategies to highway projects. Therefore, this study presents a new approach of buffering strategies applied to transportation construction projects.

Specifically, this research searches to study and analyze different types of buffers in order to figure out which are more relevant in road projects. Then, the buffers, which were chosen in the first part of this research, will be modeled with ExtendSim and, subsequently, validated through a case of study.

Full details on Mr. Forcael's proposed research can be downloaded here

n/a

The Conceptual Model - The model has been built based on the activities which are present in a transportation construction project. A schematic to represent the simulation model used for n linear sequential activities or processes was developed.

The Proposed Model - Based on the previously-mentioned conceptual model, ExtendSim was used to create a model to represent the most significant transportation construction activities or processes. The proposed model has three activities as an example (the full model includes about eight activities).

Capacity buffer and material buffer are simply modeled as queues which store the number of workers, machines and material available. However, the calculation of WIP is not directly obtained; therefore, it is necessary to calculate it using an alternate algorithm.

Preliminary Conclusions - This current research is now focused on testing the model. The next step will be to validate the model using three case studies, which outputs will be including within the next progress report.

Download March 31, 2010 progress report

James Woodward Saunders

URL

jim.saunder@pixie.co.za

URL

Kyle Naylor -- knaylor@calpoly.edu
Brandon Wolfe -- bwolfe@calpoly.edu

URL

February 27, 2010 - During the last couple of months, much progress has been done in the project.

Based on a set of flow charts developed at a smaller Norwegian hospital, a model has been created which consists of the patient treatment process from beginning to end. This main process will be used throughout the project as the basis for simulating the effects of security measures on a set of KPIs identified, all of which are used within the health sector in Norway.

Although the flow charts have been developed from the patient treatment process found at a different hospital than the one which the case study will be conducted on, we will use this process as our basis for the rest of the project. The reason for this is that the process is smaller, which makes it manageable given our short time frame available, but still contains the necessary elements in order to fulfill our requirements.

The progress of the case study is also advancing and the processes used in both the password scheme and the smart card solution have been identified, flow charts have been created and based on these, a model has been developed.

However, there is still work to be done in connecting the authentication model and the main model, as well as making some adjustments in order for the model to work as planned. When this work has been completed, our next phase (simulation runs) can begin before we can start analyzing the results.  

URL

This research is applicable to model the difficulties of a high volume
manufacturing environment with research findings focused on flexible work
cells. This consists of high volume high mix manufacturing with strict
process and quality requirements meeting international standards. This
requires modeling production for a mix of over 72 products from simple to
complex which impact cost and delivery rate. In detail the multivariates
include configuration of sub-components, products, testing, training, and
quality processes. Specifically this includes modeling the processes of:

Specifically this includes modeling the processes of:
	Demand forecasting of products and sub-assemblies.
	Product components and vendor logistics such as MRP.
	Subassembly: Build rate   Automated test rate   Items Inventory   Value
	Final Product: Build rate   Automated test rate   Items Inventory   Value   Labeling   Shipping
	Product tracking (barcode/RFID + database) through multiple processes including work in progress, automated testing and shipping.
	3D Modeling of specialized equipment & training required for production, quality and testing.

July 10, 2010 - Continuing development of test model for capacity planning and statistical analysis. Continuing changes in model based on results of PCB parallel PCB testing - completion is still on target for 2010.

March 8, 2010 - Continuing to develop test equipment model for capacity planning and statistical analysis. New Changes to the model for implementation of parallel PCB test are in implementation. The dual test model is requiring much more effort than originally anticipated, however completion is targeted for 2010.

July 23, 2009 - Continuing to develop test equipment model for capacity planning and statistical analysis. Delays continue with introduction of the new product line.

February 27, 2009 - Continuing to create a model for evaluation of test equipment, which has been delayed with the introduction of new product line. This will be used to determine factory capacity planning & statistical analysis for dual product test system.

February 20, 2008 - Project scope has changed slightly. Since I cannot use most of my previous employer's information/ process, I now plan to create a model for the multivariate's of products and process of my new employer. Plan to incorporate ExtendSim's 3D environment to evaluate specialized test equipment to determine factory capacity planning & statistical analysis. One specific case is our unique new test station concept where a single operator must use one PC with two LCD displays and manually operate two test adapters connected on this dual product test system.

November 15, 2007 - Continuing to model processes waiting on next Kaizen event to map external processes of quotation, purchasing, and engineering.

July 11, 2007 - Continuing to model manufacturing process based on Value Stream Mapping effort (Kaizen) to validate and improve competitive manufacturing methodologies with simulation.

March 29, 2007 - Research along with work related activities identified Kaizen for (continuous improvement) using operational excellence tools such as LEAN Value Stream Mapping to develop the framework for the ExtendSim simulation model. The goal is to model all processes for identification of improvement while eliminating waste and variability. A Kaizen event was completed to map all internal processes from receipt of raw materials through delivery for a specific product and customer (due to time constraints the external processes of quotation, purchasing, and engineering were excluded for a separate event). This LEAN exercise resulted in development of the current state and desired future state maps that provide the flow with batch and queue process for the internal simulation model that is currently under development.

December 29, 2006 - Research for previous work performed in this field is being performed using electronic databases to locate texts, journals, white papers and other publications. This will be used as a guide to compare, contrast or expand on the use of modeling to improve and create competitive manufacturing methodologies with simulation.

The research will help develop the framework for the simulation models. Working through the User's Guide tutorial to gain improved knowledge and skill of the ExtendSim tool. After completion will also review the creation of blocks as this may be necessary to create the final manufacturing solution libraries.

Bob Batt

URL

batt@wharton.upenn.edu

URL

“Determining Capacity in a Healthcare Organization with Changing Client Mix,” a presentation for the Decision Sciences Institute 2009 Annual Meeting.

URL

URL

For high-value workforces in service organizations such as call centers, scheduling rules rely increasingly on queueing system models to achieve optimal performance. Most of these models assume a homogeneous population of servers, or at least a static service capacity per service agent.

In our research we examine the challenge posed by fluctuating service capacity in service organizations. Our main line of inquiry involves building operations research models that can accommodate increasingly precise contemporary empirical studies of learning effects on the job. In "learning-by-doing," servers may increase their own service efficiency through experience; they may also decrease it through absence.

Today, large customer service providers such as call centers provide training and coaching to agents. But the benefits of "learning-by-doing" are not systematically optimized for in those providers' routing and staffing policies. We seek to quantify the benefits of learning-aware routing and learning-aware scheduling policies through discrete event simulation, queueing system models, and large-scale optimization programs. In particular, we seek policies that optimize worker development through "learning-by-doing" on the job, while at the same time maintaining a high level of service to customers.

March 3, 2010 - Our paper was provisionally accepted pending editors' requested revisions by a journal, and we're working on the revisions now. In the meantime I got a postdoctoral research position at SAP Labs. So I'm working on that paper with my coauthors in our free time. Not sure how the copyright rules work out, but if I can't send you the final copy for publication, I'll send you a customized version you can post with newer figures and a more readable text/less jargon.

July 6, 2009 - I graduated in June, and we're working to get these papers accepted in research journals. I just taught an undergrad lecture class of 172 students from March until June. You may download the submitted below.

February 27, 2009 - Successfully defended his dissertation in December, 2008. His committee requested some edits to his thesis, and he is finishing those up now, and is on schedule to turn in the completed manuscript at the end of March. Prototype modeling in ExtendSim has been very helpful to their team.

March 6, 2008 -Download Geoff's latest research and ongoing Extend model he has been using to generate results. Once he brings the model up to ExtendSim OR 7, he will be writing more object-oriented code inside equation blocks.

	Download submitted paper completed June of 2009 (7.4 MB): Routing to Develop Expertise in Customer Contact Centers -- A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Computer Engineering by Geoffrey S. Ryder
	Download in-progress report from March of 2008 (1.12 MB): Research Report: Knowledge Management in Call Centers—How Routing Rules Influence Expertise and Service Quality by Geoff Ryder
	Download Extend OR 6 model files (604 KB). You may view these models in ExtendSim or download the ExtendSim Demo.

Ryder, G. "Managing Changing Service Capacity Based on Agent Performance Data." INFORMS Annual Meeting, Service Industry III Session, Nov. 7, 2007.

Ryder, G. "How Learning and Forgetting Affect the Optimal Work Policy." INFORMS Annual Meeting, Management of Complex Service Systems Session, Nov. 4, 2007.

Ryder, G. and Ross, K. "Optimal Service Rules in the Presence of Learning and Forgetting." Sixteenth Annual Frontiers in Service Conference, San Francisco. October 4, 2007

"Optimal service policies under learning effects," International Journal of Services and Operations Management, Issue 6, Vol. 4, 2008. Accepted paper, to appear.

"Optimal service policies in the presence of learning and forgetting", by Kevin Ross and Geoff Ryder. Applied Probability Track, INFORMS Annual Conference, Pittsburg, PA, 2006.

"A probability collectives approach to weighted clustering algorithms for ad hoc networks," by Geoff Ryder and Kevin Ross. IASTED CCN Conference, Marina Del Rey, CA, October 2005.

URL

I am a professor at UNEFA University in Venezuela, at the Department of Post Graduate studies in Public Policy advising several post graduate students in the development of a model to analyze public resources allocation.

	Brief description of the problem/issues
	National oil companies (NOCs) face multiple challenges associated with a rapidly changing business marketplace, as well as challenges related to the development of local content. Plans developed for the development of local content often compete with the managerial urgency to complete projects on time to achieve the desired financial performance. Power point based plans to develop local content fail frequently, among other reasons, because of the lack of integration of such plans with the NOCs supply chain performance, and the lack of understanding of nonlinearities, delays, and capacity building issues related to the NOCs operations, priorities, and plans. A system dynamics based simulation model that links local content development strategy to the NOCs structure may help the definition of strategies to enhance the development of local services and manufacturing industries, the integration of local content in the core business, and consequently, help increase the impact of local companies on regional development, which is the ultimate objective of the related public policy.
	Question/s that will be answered by your research.
	Whether prioritization of capacity building projects over time, in combination with a clear understanding of the role of delays in decision making within NOCs, drives the difference between desired and real local content development, significantly.
	How research strategy will be designed to answer these questions.
	Having already designed the general structure model for a NOC (Exploration, Production, Storage, Transportation, Terminals, Refining, Commerce, Supply, and Finance) and having identified some of the entry points of local content providers in the supply chain, I am ready to pursue the development of a simulation model to accomplish the described task.
	How ExtendSim will be utilized in this project.
	Having tried other simulation programs extensively (Powersim, Vensim, I Think), and after having examined ExtendSim specifications, I find this simulation software appropriate to tackle the complexity associated with public policy formulation using simulation engines, which is the combination of the dynamic, discrete, and agents simulation paradigms, all present in ExtendSim; last, but not least, the user interface approach managed by ExtendSim increases the communication between non-technical policy makers and model designers by using animation.
	Project uniqueness
	Regarding the representation of the structure of NOCs using system dynamics, I have recently completed an effort for a major NOC; while on the side of local content policy, I designed the strategic plan for a NOCs local content support in 2008. Both efforts have given me an overview of the pros and cons of different strategies associated with the subject. I acknowledge, though, the following contributions as unique in the field: Dynamic Planning and Control Methodology: Monseo Park and Feniosky Pena Executing major projects through contractors using: Nicholas McKeena Different case studies on local content development strategy in Europe.
	Why the interest in the indicated subject?
	My interest is focused on this subject because I am a professor at the Public Policy Post Graduate School of the Armed Forces University in Venezuela, where I teach different subjects in Public Policy to Master's students.
	What will be learned from your research?
	The identification of policy levers to help improve the integration of local content companies in the core oil and gas business related to NOCs.
	What specific contribution would you like to give to the existing current state of knowledge in your subject area?
	The design of a formal policy framework for the development of local services and manufacturing industries related to National Oil Companies.

Full details on Mr. Chique's proposed research can be downloaded here

n/a

Aly Owida

URL

aly.owida@staff.aast.edu

Youness Eaidgah Torghabehi

URL

Anna Megarian, “Supply Chain Network Economics: Dynamics of Prices, Flows, and Profits”, Edward El gar Publishing, 2006

G. Tom as M, David J. Ketch en, Jr. Lowder, Garry L. Adams, Jeannette A. Mena, “Supply Chain Orientation and Balanced Scorecard Performance”, Journal of Managerial Issues, Number 4, 2008

Manish Govil, Jean-Marie Proth, “Supply Chain Design and Management: Strategic and Tactical Perspectives”, Academic Press, 2001

John D. Sterman, “Business Dynamics: Systems Thinking and Modeling for a Complex World”, Irwin/McGraw-Hill, 2000

David Simchi-Levi, Philip Kaminsky, Edith Simchi-Levi “Designing and Managing the Supply Chain: Concepts, Strategies, and Case Studies”, McGraw-Hill, 2008

A.E. Ellinger, “Improving marketing/logistics cross functional collaboration in the supply chain”, Industrial Marketing Management, Vol. 29, 2000

M.J. Meixell, V.B. Gargeya, "Global supply chain design: a literature review and critique", Transportation Research, Part E: Logistics and Transportation Review, V.B. 2005

Henry Livingston, “Avoiding Counterfeit Electronic Components”, IEEE Transactions on Components and Packaging Technologies, Volume 30, Issue 1, 2007

S. Bastia, “Next Generation Technologies to Combat Counterfeiting of Electronic Components”, IEEE Transactions on Components and Packaging Technologies, Volume 25, Issue 1, 2002

J. Stradley, D. Karraker, “The Electronic Part Supply Chain and Risks of Counterfeit Parts in Defense Applications”, IEEE Transactions on Components and Packaging Technologies, Volume 29, Issue 3, 2006

A. Gunasekaran, B. Kobu, “Performance Measures and Metrics in Logistics and Supply Chain Management: A Review of Recent Literature (1995-2004) for Research and Applications”, International Journal of Production Research, Vol. 45, No. 12, 2007

C. Shepherd, H. Günter, “Measuring Supply Chain Performance: Current Research and Future Directions”, International Journal of Productivity and Performance Management, Vol. 55, No. 3/4, 2006

A. Gunasekarana, C. Patelb, Ronald E. McGaugheyc, “A Framework For Supply Chain Performance Measurement”, International Journal of Production Economics, Volume 87, Issue 3, 2004

Dinu Corbu

URL

As representative for supporting this piece of research, there can be mentioned the article “Product warranty: modelling with 2D-renewal process” by Dinu Corbu, Stefanka Chukova, and Jason O’Sullivan, published in International Journal of Reliability and Safety 2008 - Vol. 2, No.3 pp. 209 - 220.