 |
Download a pdf copy of this help file here |
|
Download a pdf copy of this help file here |
Simulation Requirements define rules on the acceptable values for stochastic
variables within the model. They are checked at the end of a simulation
run when a new set of decision variables has being tested and the simulated
values for the variable in question can be analyzed. Simulation requirements
are probabilistic in nature, i.e. they define a requirement for some statistical
attribute of the variable in question.
For example, a requirement might be that the average delivery time is less than two weeks, or that the standard deviation of profit is less than 21%, or the expected (mean) profit is greater than 12%, or that there is more than a 95% probability of finishing the project on time.
Clicking the Add button in the Simulation Requirements tab opens a dialog for you to select the spreadsheet location where the simulation requirement is to be placed. This is then entered into the Address field and can be edited later by double-clicking this field.
Address is the cell position of the simulation requirement (constraint) that is to be monitored. This cell should already have a formula entered for the Optimizer to work with
Name is a text field that identifies the constraint
Type provides a dropdown list of constraint options: Min; Max; Equals or Between
Value1 specifies the value of the constraint. When Type is Equals, the constraint is that the formula in the cell is precisely equal to Value1; when Type is Min the constraint is that the formula is greater than or equal to Value1; and if Type is Max the constraint is that the formula is less than or equal to Value1. Value1 will nearly always be a fixed value: if it is linked to a spreadsheet cell, ensure that the cell referred to is not generating random values.
Value2 field is only available if Type is Between, in which case the constraint is that the formula in the cell lies between Value1 and Value2. Value2 must be > Value1. ModelRisk will generate an error message if not and will switch the constraint off, showing it to be invalid in the dialog box.
Enabled is a tick box switching the constraint on and off. In the Excel formula, this appears as TRUE or FALSE respectively.
Entering the information as described above will add the requirement to the model. In the example above, where the first variable ‘Component 3 in the cell is constrained to have a 10 percentile of 0, the following formula now appears:
= [formula] + VoseOptRequirementMin("Component3",VoseOptPercentile(0.1),0,TRUE)
where [formula] is the equation that was in the cell before the constraint was added.
ModelRisk incorporates three functions for defining decision constraints:
VoseOptRequirementMin(Name, Statistic, Value, Enabled)
VoseOptRequirementMax(Name, Statistic, Value, Enabled)
VoseOptRequirementEquals(Name, Statistic, Value, Enabled)
Options for the function parameters are:
Name: a text field identifying the decision variable to the user
Statistic “mean”; “median”; a value between zero and one for the Percentile option; “min”; “max”; “stdev” (standard deviation); “variance”; “range” (the difference between the minimum and maximum observed values); “skewness”; “kurtosis”; “CofV” (coefficient of variation); and “FinValue” (the final value of a simulation, used if one builds a model that simulates across iterations in some way).
Value: any numerical value
Enabled: a Boolean parameter taking either TRUE (or equivalently 1) when the decision constraint is to be used, or FALSE (or equivalently 0) when the decision constraint is not to be used.
Example model