Crystal Ball is most helpful in phases of quality improvement programs where uncertainty and variation have the greatest impact. In the Define phase, for example, cost savings forecasting, project cost estimation, and resource scheduling are all subject to uncertainty. These types of analyses are ideal targets for simulation.
In the Analyze and Improve phases, the focus is variation. Simulation can help reduce the variation around tolerances or process steps.
In any phase, defining model inputs (X’s) as a distribution of values yields a more realistic range of results. Then, the results (Y’s), defined as Crystal Ball forecasts, can be analyzed as demonstrated in the following chapters of this guide.
By default, the forecast certainty range is automatically set to the upper and lower specification limits (USL and LSL), if at least one of them is entered for the forecast. Then, you can tell at a glance the probability of output falling within them, 99.49% in the example in Figure 186, Piston Displacement Forecast Chart. Split View shows capability metrics beside the forecast chart.
Capability Metrics view indicates whether the forecast distribution is normal, to guide the interpretation of the capability metrics. For example, Z metrics are always displayed, although they are generally used only for normal distributions. The Z-total metric indicates the sigma level of quality. In this example, the level is 2.37 sigmas. For this forecast, 8,824 units per million will be outside the specification limits. The Cp and Cpk values are less than one, which suggests that the quality level is lower than 3 sigmas. To cycle among other views, such as Statistics and Percentiles, press Ctrl+Spacebar.
By default, Crystal Ball applies a 1.5 sigma Z-score shift value when calculating capability metrics. If the organization uses a different value or does not use a Z-score shift value, be sure to adjust this setting in the Capability Options panel. |
Crystal Ball’s sensitivity charts are also helpful to quality practitioners. They show which inputs, or assumptions, have the greatest impact on the selected forecast. For example, the following sensitivity chart (Figure 187, Piston Displacement Sensitivity Chart) shows that Connecting Rod Length has the greatest effect on the Piston Displacement forecast shown in the previous figure.
Crystal Ball’s data extraction and reporting capabilities are also helpful in quality analysis. For example, you can automatically extract capability metrics to a prominent location in the model for immediate access to current sigma levels and other important measures.
If you have OptQuest, it — along with the Crystal Ball Developer Kit and OptQuest Developer Kit — can add further power to the arsenal of quality improvement tools.
When you define certain cells as Crystal Ball decision variables, you can then optimize them with OptQuest and copy the new, optimal values back into the Crystal Ball model. For example, Using OptQuest to Optimize Quality and Cost, shows how you can use OptQuest to adjust several controllable inputs simultaneously to yield a set of optimal results.
Visual Basic for Applications (VBA) developers can use the Crystal Ball Developer Kit to support quality projects in a variety of creative ways. This model uses the Developer Kit to create a color-coded dashboard that highlights which process steps are lowest in quality and which are most important to the process output.
For other ways to apply Crystal Ball to the own quality improvement projects, review the examples in this guide and examine the other models supplied with Crystal Ball and posted online. For more information, see: