Related Topics and Links:

 Sensitivity Analysis Using Simulation Worksheets

Simulation Worksheets

Simulation worksheets allow you to vary values that are used when simulating a BlockSim RBD or a RENO flowchart. This allows you to investigate the effect of one or more settings on the simulation results.

Simulation worksheets consist of columns that can be defined as inputs or outputs. The input values are used during simulation, and they can be entered manually or transferred from a DOE experiment design. Values in output columns are determined through simulation. For example, you might design an experiment in a Weibull++/ALTA design folio and transfer all the factor level combinations to the simulation worksheet. Then you could simulate the experiment in BlockSim and transfer the output values back to the Weibull++/ALTA design folio to analyze the results. If, on the other hand, you do not want to test all combinations of all factor settings, you can enter the desired combinations of settings by hand.

In general, using a simulation worksheet will include the following steps:

Depending on what you are trying to achieve, some of the steps may not be necessary.

In this topic, we will consider an example in which we explore options for maintaining a spare part pool, with a goal of maximizing system availability while minimizing the cost of the pool. The factors that we will vary are the number of spare parts initially in the pool, the frequency with which the pool is restocked and the number of parts added to the pool at each restock.

Create the Variables and the Diagram in BlockSim/RENO

Build the diagram for the system in BlockSim/RENO. Use variables to represent the factors whose settings you want to investigate. Depending on the factors you are studying, you might use the variables directly (as field inputs for BlockSim blocks and resources, or in equations and functions in RENO) or via a dynamic model. In BlockSim diagrams, the following fields can be defined using a variable:

Any field that takes a model can use a dynamic model; it is up to you to decide if exploring various values for that field will yield useful results.

IMPORTANT: In RENO flowcharts, you will need to store the results of interest in variables. The When to Reset field for output variables must be set to Never Reset in order for the results to be available to the simulation worksheet. This ensures that the results are available to the simulation worksheet, rather than the variables being reset to their initial values.

Example

The spare part pool that is used by the tasks in our example is configured as shown next. The cost per item is specified as a constant model equal to $10 and the holding cost is specified as a constant model equal to $10 per month. Since the factors of interest (i.e., the initial stock level, the frequency of restocking and the number added per restock) will not be held constant, they are assigned to variables. These variables represent factor values that will be specified in the simulation worksheet.

Define the Factor Settings

For the factors of interest, you will decide on the settings to be used in simulation and the combinations of factor settings that should be tried (i.e., that will be used as inputs to the simulation).

This will ensure that the factor settings are tested in combinations appropriate to determine the effects of each factor and of factorial interactions.

After adding the new worksheet, you can enter the desired factor values manually, as shown next. (Note that the column names were changed by right-clicking the column header and choosing Rename Column on the shortcut menu.)

While you have the simulation worksheet open, it will be read-only in other applications. For instance, if you create the simulation worksheet in Weibull++/ALTA, it will be read-only in BlockSim and RENO until you do one of the following:

In either case, if you already have the simulation worksheet open in a read-only state, you will need to click Unlock and Edit on the control panel once you have released the lock in the original application.

Example

The experiment used in our example is a general full factorial design using the following settings:

The first fifteen rows of the simulation worksheet that is created in the transfer process are shown next.

Define the Simulation Worksheet Columns in BlockSim/RENO

Once the factor settings have been entered in the simulation worksheet, you will simulate the worksheet in BlockSim/RENO. The software will perform a set of simulations for each combination of factor settings in order to obtain response values. In BlockSim/RENO, open the simulation worksheet and click the Simulate icon on the control panel.

The Select Diagram window will appear, allowing you to select the diagram that will be used to simulate the response values (i.e., the data source) for the simulation worksheet.

Note: You can also click the Associated Data Sources icon to specify or change the source diagram.

The Worksheet Simulation window then appears, allowing you to specify the information contained in each column of the simulation worksheet.

Starting in Version 11, you can choose to assign the simulation end time to one of the input columns. This allows you to obtain results as a function of time.

Example

The Worksheet Simulation window used in our example is shown next.

The output columns were specified as follows:

Simulate the Worksheet in BlockSim/RENO

With all columns defined, you can perform the simulation in BlockSim/RENO.

Example

The simulation settings used in our example are shown next.

The first fifteen rows of the simulation worksheet, with results, are shown next.

Analyze the Results in Weibull++/ALTA

The results that were derived from the simulation can now be analyzed. To do this:

In the window that appears, select the design folio to transfer to and match up the columns in the design folio (factor/response columns) with the columns in the simulation worksheet (input/output columns), as shown next.

Note that all data will be transferred from the simulation worksheet, so if you have made changes to the input/factor columns in the simulation worksheet, those changes will be transferred to the design folio.

You can then perform any analysis you choose in Weibull++/ALTA, including optimization.

Example

The results of optimization for our example are shown next. A single optimal solution is found with Initial Stock Level = 5, Restock Time = 1148.693 and Number Added per Restock = 5. This gives a system availability of .906 and pool cost of 1173.71.

Verify the Results in BlockSim/RENO

Because BlockSim/RENO and Weibull++/ALTA rely on different methodologies to arrive at results, predictions made by Weibull+/ALTA will vary from the simulated results generated by BlockSim/RENO. Therefore, the optimal settings selected by Weibull++/ALTA must be simulated in BlockSim/RENO to ensure that the predicted results are reliable. To do this, return to BlockSim/RENO and, for each of the variables, set its initial value to the value given in the optimal solution and then simulate the diagram directly (i.e., not via the simulation worksheet) to make sure that the results are close to those estimated in Weibull++/ALTA.

Example

The simulation results from BlockSim are shown next. The results show a mean system availability of 0.904 and a total cost for the pool of $1212.55. These values are very close to the estimated values in Weibull++/ALTA.

 

© 1992-2018. HBM Prenscia Inc. ALL RIGHTS RESERVED.

E-mail Link