After you collect this information, you are ready to run a simulation.
If you're not sure which distribution to choose and you have historical data from the system, choose Use data to decide. Then, browse to the CSV file that contains data. Be sure that the data represents expected future performance. Companion automatically finds the best-fitting distribution. For more information, see "Choose a distribution" in this topic.
When you have a complex or a large simulation, you can create groups to define the model by function. For example, you might want to describe different actions or various parts' behavior within the simulation. With groups, you can categorize inputs and outputs to help you manage and organize your simulation.
In Monte Carlo simulations, it is typical for simulated responses to violate the assumption of normality. Therefore, Companion uses a nonparametric method to calculate capability in the simulation tool because it works for both normal and nonnormal data. The nonparametric method calculates the spread of the output distribution using the observed 0.135 and 99.865 percentiles of the simulated data, which is analogous to +/-3 sigma in a normal distribution.
Because there are no subgroups and no concept of long term and short term variation in the simulation context, Cpk and Ppk values are equivalent in the Companion Monte Carlo simulation. You can choose which label to display in Companion's simulation results. Choose and select the label you prefer.
Based on the spread in the data and the specification limits you set in the model, Companion calculates ppl and ppu to find the corresponding Ppk. As with typical capability calculations, Companion takes the distances between the process "center" and each spec and divides by the spread of the distribution to get ppl and ppu. The smallest value is chosen to represent the capability.
Companion displays the results of the simulation, how your results compare to generally accepted values, and guidance for next steps.
Each time you repeat the simulation, the results will vary because the simulation is based on randomly selected values for the inputs.
After you analyze the results, you may want to return to the model and change inputs or outputs, and then rerun it. This allows you to test a number of "what if" scenarios allowing you to gain insight into the behavior of your system and make better decisions.