The Tragedy of Simulation Modeling

Don’t get accused of stealing company resources. Five techniques to reduce simulation modeling time and increase quality.

1. Make use of accessible data
   • Business leaders have screamed their dissatisfaction with the time it takes to update the details of a model. Shaking his head in disgust, one manager said, “I just want updated information entered in the model. Why does it take so long?” Many modelers are figuratively digging ditches with “spoons” by manually entering data into their models. Instead, a simulation modeler could have used a “track-hoe” to update an entire ProcessModel  simulation model with a single click. That’s right — one click prepares data, enters it into ProcessModel and runs the model. This means that complex model changes error proofed and entered in just seconds. Learn more about data importing
2. Simplify Changes to a Model
   • With the creation of a custom interface, a model can be easily used by anyone. One innovative company turned a model over to a group of their salesmen. Remember, while salesmen have one of the most stressful and difficult jobs in any company, they tend not to be as technically skilled. Despite that, these salesmen were able to use the model to predict the results of various product configuration changes. It worked because the only thing the salesmen saw was a custom interface with simple entries and a button that said “Run.” The whole process took only minutes, but allowed them to make dramatic changes to their system configurations with confidence. Learn more about custom interfaces
3. Use a Throw Away Model
   • Highly trained Six Sigma Black Belts collected detailed information about every step of the process before starting the model building process. They recited their internal mantra “garbage in, garbage out.” But effort and time spent on the collection of irrelevant and non-critical information is an enormous waste. Collecting detailed process information before the needs of the system/process/simulation are understood is like selecting a teenager’s wardrobe before knowing the youths gender.A better process is to use a model to help understand the information needed to describe the system, the same way a student makes an outline before drafting a paper. I build a quick model, or a ‘throw away’ model,  so that I understand the process requirements and timing. This throw away model allows me two major advantages: first, I can eliminate collection of some elements entirely because they won’t be used by the model; second, I can use rough estimates to determine that some elements won’t matter even if I am off by 100%. I can usually even expand the ‘throw away model’ into the final model. Rough estimate of distributions allow the model to be tested for sensitivity. Some elements don’t effect of the model while others will dramatically change the outcome. By focusing detailed data collection of only on the areas that will matter, the amount of time collecting data can be slashed. This single change in project procedure allowed this process team to cut projects times by more than 45%.
4. Build in the Proper Sequence
   • I used to build process models from the first step to the last, putting in detail all along the way. A brilliant modeler finally showed me how to cut my modeling time by 60 percent. His method is simple to follow and makes modeling more like a game. Here is the basic outline to for a fast Model Building Sequence. This method of building models allowed ProcessModel to build a model for NASA/Johnson Space Center, that was voted Process Improvement of the Year. The simulation was built using the procedure described above in less than 4 hours! That wasn’t only to build the model, but included gathering data, getting buy-in and presenting to management. This procedure works and it will save you a ton of time (usually about 2/3rds).
5. A Little Bit goes a Long Way
   • It is easy to get overwhelmed with the details of most systems. Often, a simplified model will show the problem clearly without sacrificing accuracy. Sometimes adding more detail decreases the accuracy of a model. That’s right, some detail hurts the accuracy more than it helps. Detail spawns more detail, and before you know it, a model is difficult to build, change and understand. The question that should always be asked is, “will this detail add significant accuracy or capability to the model?” If the answer is no, eliminate the detail.

These five principles will help you speed the process of model development, make model changes easier and help you get to the “success” faster. The real tragedy is much greater than the inefficiency of a modeler. It even goes deeper than the money that a company will lose — projects get delayed. By far the biggest tragedy is loss of quick, compounding successes. Success begets success and leads to quick change in the organization. Take advantage of these quick successes by using the principles outlined above.

Would you like to speak to one of our senior consultants about any of the items discussed in this article? Fill out the information here and we will contact you to arrange a time. Or, you can just call us at 801-356-7165 X172.