Return to Index
Operations Research Models and Methods
 
Computation Section
Hidden Iterations
  

Simulating many iterations on a worksheet requires many rows with perhaps complex functions in each. This will expand the Excel's memory requirements and the computational effort required for a simulation. The add-in provides a procedure that displays only a relatively few rows of the simulation, yet allows the sample size to be very large. We illustrate this feature with the built-in inventory model. After accepting the model parameters, the Simulation dialog appears as below. Note that we have indicated a sample size of 50, however, in the Show Sample region, the Specified Number option is selected. To the right, the Number to Show field has been given the value 25. This means that only 25 of the iterations of the sample will be shown on the worksheet, but a simulation run will include the entire sample size of 50. We use 25 and 50 in this example, but usually the sample size will be much greater than the number of iterations to be shown.

Pressing OK for this dialog, builds the worksheet for the inventory model. The image below shows the worksheet as it is initially constructed. Only the first 25 days are shown. The Sample Size and Show parameters at the top of the worksheet come from the dialog. The Sample Count in column I, which is currently 25, shows that the first 25 of the iterations are simulated directly on the worksheet.

 

Pressing the Simulation button in column A (not shown in the image) initiates the simulation run. The computer uses the display of 25 rows to perform the 50 iterations required for the sample. Because 50 divided by 25 is equal to 2, two 25 iteration replications are required. The image below shows the second replication. The second replication is a continuation of the first because the last four lines in the first replication are pasted into the initial rows region of the display. Only the values are pasted. The sample count cell in row I indicates that the simulation is complete.

 

The cumulative statistics for the entire sample are computed by the program and stored in the rows labeled Mean and Standard Deviation. The statistics shown in the example are for the entire 50 iterations.


 

Two other buttons are in column A. The Initialize button, returns the simulation model to its initial state, showing the first 25 iterations. The Simulate button repeats the simulation. If the Randomize Seeds option has been selected, each simulation will be given a new set of seeds.


 

Changing the sample size is accomplished by clicking the Change button. In the image below, we have changed the sample size and repeated the simulation. The display shows the last replication of the 40 replications of 25 iterations to simulate 1000 iterations. All the operations required for the ismulation are performed automatically, and only the first and last replications are shown. The simulation only used a few seconds on the author's computer.

 

 

The image shows a new feature. We have placed additional statistics in the Other Statistics row of the worksheet.

Cell
Formula Description
L12
=SUM(L19:L43)

This computes the sum of the reorder column. When the simulation is complete it shows the number of orders placed during the 1000 days.

N12 =MAX(N19:N43) This computes the maximum number in the backorder column. It is the maximum number if backorders at the end of the day during the entire simulated time.
O12 =MAX(O19:O43) This computes the maximum number in the inventory column. It is the maximum amount in inventory at the end of the day during the entire simulated time.

These statistics are computed with Excel statistical functions. There are many different types of statistics that might be of interest, but only a few are available with hidden iterations. They are: SUM, MAX and MIN.

The hidden iterations option of the add-in can be very useful to simulate large sample size without requiring a great number of cells on the worksheet. This facility does come with a price however. No longer is the simulation dynamic. This makes it more difficult to experiment with alternative parameter selections. For the example, it would be more difficult to find the optimum reorder point and lot size, because trial and error is much more easily accomplished with a dynamic simulation.

 


  
Return to Top

tree roots

Operations Research Models and Methods
Internet
by Paul A. Jensen
Copyright 2004 - All rights reserved

Next Page