Tag Archives: discrete event simulation

Teaching Simulation to Graduate Students Using ProModel Products and Real-World Problems

by

ProModel Guest Blogger:  Larry Fulton, Ph.D. & MSStat – Assistant Professor of Health Organization Management at Texas Tech University Rawls College of Business.  After serving 25 years in military medicine, Dr. Fulton began a second career in teaching and research.

Larry Fulton, Ph.D. & MSStat

Larry Fulton, Ph.D. & MSStat

Teaching introductory Monte Carlo, Discrete Event, and Continuous simulation to business graduate students requires at least two components beyond a good set of reference materials:   realistic or real-world problems and an excellent modeling platform allowing for relatively rapid development.  In the case discussed here, the real-world scenarios derived from interests and background of the professor and students (portfolio analysis, sustainability, and military medicine), while ProModel products addressed the platform requirements. Each of the case study  scenarios served to underscore various simulation building elements, while ProModel supported rapid  product development for a 14-week, lab-intensive course that included some  reviews of probability, statistics, queuing, and                                                    stochastic processes.

Scenario 1:  Monte Carlo Simulation (Portfolio Analysis)

Business students generally have an affinity for portfolio analysis, and I do as well. Using ProModel  features, one of the earliest student projects involves fitting univariate distributions to return rates to several funds and simulating results of investment decisions of various time horizons.  Students discuss methods that might account for covariance as well as autoregressive components in these simulations.  While developing the simulations, students also determine sample size requirements to bracket mean return on investment within a specified margin of error and confidence interval and use random numbers seeds.

Scenario 2:  Continuous Simulation using Rainwater Harvesting

Students in this course are generally from a semi-arid region (Central Texas), which has significant water shortages (so much so that desalinization is being considered.)  I rely 100% on rainwater harvesting for my home water supply, so extending this to each student’s particular home location is trivial. The “Tank Submodule” provides an easy mechanism for developing the simulations.  Students develop conceptual models of rainwater mechanism as well as flowcharts.  They gather rainfall data from the National Oceanic and Atmospheric Administration regarding rainfall and evaluate various roof sizes (capture space), demand figures based on occupants, and tank sizes. They also learn about the importance of order statistics (the distribution of the minimum in the tank) versus measures of central tendency that often dominate discussions of simulation. Finally, they incorporate tools and techniques to improve and assess V&V.

­­­Scenario 3:  Discrete Event Simulation using Military Scenarios

While serving as the Chief of Operations Research Branch for the Center for Army Medical Department Strategic Studies, I encouraged the use of MedModel for multiple DES projects.  The team built strategic models (resource constrained and unconstrained) for analyzing medical requirements for strategic operations. These same models serve as a basis for a team-based MedModel student capstone project.  The primary entity for these models was the patient with attributes of severity, injury type, and evacuation type.  The primary processes involved collection, treatment and evacuation. Resources included ground ambulances, air ambulances, medics, intensive care units, and operating rooms.   Locations were geographic locations throughout the entire of Afghanistan.  Evacuation paths were built, and treatment logic (triage, ground evacuation, air evacuation, etc.) provided the flow.

Bottom Line:  The ProModel products are outstanding for use in both teaching and industry.

 Larry Fulton Bio:

http://www.depts.ttu.edu/rawlsbusiness/people/faculty/hom/larry-fulton/

ProModel Customized Solutions: Hogistics

by

There are times when the best solution is a customized solution. For those situations ProModel has a highly skilled and agile development team ready to work with you to develop the one of a kind predictive analytic solution that meets your needs.

We have created unique custom predictive analytic technology applications and training programs for the Army, Navy, and Government Agencies, as well as companies in the Aerospace & Defense Manufacturing, Pharmaceutical, Healthcare and Services Industries.

Here is just one of our latest custom solutions:

Hogistics, created in partnership with Zoetis, a global animal health company that delivers medicines and vaccines, complemented by diagnostic products and genetic tests and supported by a range of services.

Zoetis, in conjunction with ProModel, created an analytics tool to help farmers predict barn and system pig growth, mortality, and sales volume over time. This helps them achieve an optimal pound per pig as prescribed by the packing companies to which they sell.

Pork production is a highly variable process due to many factors including, genetics, environment, and disease. Other events, such as vaccine protocols, weather and individual animal care can also have an effect. Hogistics will help swine producers with the following:

  • Marketing weight and weight distribution projections weeks ahead of time
  • Data to optimize animals per sale
  • Scenario analysis capabilities to evaluate animal heath, vaccines, husbandry education
  • Pig placement, marketing and transportation schedule data
  • An inventory management tool for placement and planning of multiple pig flows

If you are interested in Hogistics view the Hogistics page on Zoetis website

Contact saleshelp@promodel.com to find out how ProModel can help your organization make better decisions – faster.

Teaching Process Management Using ProModel

by

ProModel Guest Blogger:  Scott Metlen, Ph.D. – Business Department Head and Associate Professor at University of Idaho

Scott Metlen, Ph.D.

Scott Metlen, Ph.D.

Understanding process management, the design, implementation, management and control, and continuous improvement of the enterprise wide set of an organizations processes is the key to well deployed strategies. It was not until Tim Cook made Apple’s total set of processes world class including all supply chain linked processes (Brownlee, 2012) that Apple hit its amazing climb to become the world’s highest valued company; even though the company had cutting edge products before his arrival. Gaining effective understanding of process management is not easy due to the strategic variability inherent in the portfolio of products that companies sell, and in markets they service. This strategic variability (Rajan, 2011) in turn drives variability in many processes that an organization uses to operate. For instance, different markets require different marketing plans supported by different processes.  Order processes often vary by product and target market. Employee skill sets differ by product requiring different hiring and training processes. Different products, whether it be services or goods that have a slight variation require, at the very least, an adjustment to the production process. Adding to, and often caused by the variability just mentioned, are multiple process steps, each with different duration times and human resource skills.  Depending on what product is currently being produced, process steps, process step order and duration time, interdependency between the process steps, and business rules all vary. Where a product is in its life cycle will drive the experience curve, again creating variation across products. In addition, the numerous interfaces with other processes all vary depending on the product being produced. All of these sources of variability can make process management hard to do, teach, and learn. One tool that helps with process management in the face of variance is discrete event simulation and one of the best software suites to use is ProModel. ProModel is a flexible program with excellent product support from the company.

Effective process management is a multi-step process. The first step of process management is to determine the process flow while at the same time determining the value and non-value added process steps. Included in the process flow diagram for each step are the duration times by product and resources needed at each step, and product routes. Also needed at this time are business rules governing the process such as working hours, safety envelopes, quality control, queueing rules, and many others. Capturing this complex interrelated system begins by visiting the process and talking with the process owner and operators. Drawing the diagram and listing other information is a good second step, but actually building and operating the process is when a person truly understands the process and its complexities.  Of course many of the processes we want to improve are already built and are in use. In most cases, students will not be able to do either of these. However, building a verified and validated simulation model is a good proxy for doing the real thing, as the model will never validate against the actual process output unless all of the complexity is included or represented in the model. In the ‘Systems and Simulation’ course at the University of Idaho students first learn fundamentals of process management including lean terms and tools. Then they are given the opportunity to visit a company in the third week of class as a member of a team to conduct a process improvement project. In this visit students meet the process owner and operators. If the process is a production process, they walk the floor and discuss the process and the delta between expected and actual output. If the process is an information flow process, such as much of an order process, the students discuss the process and, again, the delta between expected and realized output. Over the next six weeks students take the preliminary data and begin to build a simulation model of the current state of the process. During this time period students discover that they do not have all the data and information they need to replicate the actual process. In many cases they do not have the data and/or information because the company does not have that information or how the model is operated is not the same as designed. Students then have to contact the process owner and operators throughout the six weeks to determine the actual business rules used and/or make informed assumptions to complete their model.

Once the model has been validated and the students have a deep understanding of the process, students start modeling process changes that will eliminate waste in the system, increase output, and decrease cost. Examples of methods used to improve the process include changing business rules, adding strategically placed buffers and resources, and reallocating resources. To determine the most effective way to improve the process, a cost benefit analysis in the form of an NPV analysis is completed. The students use the distribution of outputs from the original model to generate appropriate output and then compare that output to output pulled from the distributions of each improvement scenario. This comparison is then used to determine a 95% confidence interval for the NPV and the probability of the NPV being zero or less. Finally, several weeks before the semester is finished, students travel to the company to present their findings and recommendations.

Student learning on these projects is multifaceted. Learning how to use ProModel is the level that the students are most aware of during the semester, as it takes much of their time. However, by the end of the semester they talk about improving their ability to manage processes, work in teams, deal with ambiguity, manage multiple projects, present to high level managers, and maintain steady communication with project owners.

Utilizing external projects and discrete event simulation to teach process management has been used in the College of Business and Economics at the University of Idaho for the past six years. As a result, the Production and Operation area has grown from 40 to 150 students and from five to 20 projects per semester. More importantly, students who complete this course are being sought out and hired by firms based on the transformational learning and skill sets students acquired through the program.

References:

Rajan Suri. Beyond Lean: It’s About Time. 2011 Technical Report, Center for Quick Response Manufacturing, University of Wisconsin-Madison.

Brownlee, John. Apples’s Secret Weapon 06/13/2012. http://www.cnn.com/2012/06/12/opinion/brownlee-apple-secret/index.html?hpt=hp_t2. 12/301/2014.

Scott Metlen Bio:

http://www.uidaho.edu/cbe/business/scottmetlen