All the top dogs in the Monte Carlo Analysis spreadsheet universe have distribution-fitting capabilities. Their interfaces have common elements, of course, since they rely on (for the most part) the same PDFs in their arsenal of distribution-fitters. There are important differences, to be sure. It is hoped this comparison will illustrate pros and cons from a practical standpoint. Before going over our scorecard between Crystal Ball and ModelRisk, there is one more very important capability category begging for review: Correlation.

Let the battle begin anew. We continue our journey in uncertainty modeling, having understood how to fit distributions to data using Crystal Ball (CB). How does that experience compare to what ModelRisk (MR) has to offer?

Open the Duke 09_10 Scores spreadsheet with ModelRisk loaded in the Excel environment. We will first create the MR Objects representing the fitted PDFs. (Just as with the CB exercise, it is good practice to examine a variety of best-fitting distributions, rather than blindly accepting the top dog.) Then, in distinctly separate cells, we will create the VoseSimulate functions that behave as sampled values from the PDFs modeled by the MR Objects.

As with Crystal Ball, ModelRisk has the ability to fit distributions to historical data. The analyst looking to describe the variation of a Monte Carlo Analysis input can use "fitting" windows to select data and manipulate other options. How does the ModelRisk (MR) fitting experience stack up against the Crystal Ball (CB) methods and options? There are some important differences one should understand about MR before fitting PDFs to the Duke 09_10 Scores spreadsheet.

Are there discrete univariate probability distribution functions (PDFs) that can be used to simulate college basketball scores? Do we, as avid basketball observers, know enough to suggest one discrete PDF is better than another? In fitting distributions to data in your business problems, the analyst will be asking the same types of questions. If the analyst is not an expert on the inputs and their behavior, he or she should seek out a subject-matter expert (SME) who can provide insight. Putting experience and theoretical knowledge together this way is a best practice for distribution selection.

Let us assume we have a batch of historical data in a spreadsheet. Our mission-of-the-moment is to use this data and fit probability distributions that describe its past variability (or uncertainty). Consider using either Crystal Ball or ModelRisk to do this task. We offer free trials of both to registered users. If you register here, you can get yours too. Try fitting the same data using these two different packages. Let us know how and why one is better than the other. In demonstrating these capabilities, we gain first-hand experience on the usability and capabilities of the alternatives and which features compared have more priority. The best way to judge is to try them out for yourself.

Change is constant. Or so the saying goes. However, even change is ever-varying. So perhaps we should say: Change is constantly changing. As occupants of planet earth, we intuitively know this and yet strive to keep everything the same, at least those things that do well by us. Uncertainty derails the best of our plans, even uncertainties that we recognize up front.

Tolerance Analysis focuses on dimensional aspects of manufactured physical products and the process of determining appropriate tolerances (read: allowable variations) so that things fit together and work the way they are supposed to. When done properly in conjunction with known manufacturing capabilities, products don't feel sloppy nor inappropriately "tight" (i.e., higher operating efforts) to the customer. The manufacturer also minimizes the no-build scenario and spends less time (and money) in assembly, where workers are trying to force sloppy parts together. Defects are less frequent. There are a wealth of benefits too numerous to list but obvious nonetheless. Let us measure twice and cut once.

In the case of the one-way clutch example, the current MC quality prediction for system outputs provide us with approximately 3- and 6-sigma capabilities (Z-scores). What if a sigma score of three is not good enough? What does the design engineer do to the input standard deviations to comply with a 6 sigma directive?

How do Monte Carlo analysis results differ from those derived via WCA or RSS methodologies? Let us return to the one-way clutch example and provide a practical comparison in terms of a non-linear response. From the previous posts, we recall that there are two system outputs of interest: stop angle and spring gap. These outputs are described mathematically with response equations, as transfer functions of the inputs.