AnyLogic parameter variation output to Excel efficiency

Question

My model is running a parameter variation experiment for an external optimization problem. I have to export the output of each experiment replication to Excel. I was using a function that is triggered at On destroy: in the main agent and then using the following code for each output value:

ExcelResults.setCellValue(planned_leadtime_1, varExcelSheetName, ActRow, ActColumn++);

Currently, there are around 90 output values for each replication. Unfortunately, this method makes runtime very slow.

Trying to solve the issue, I stored the results of each replication in an ArrayList within the parameter variation experiment in After simulation run:. Now I would like to transfer this ArrayList directly to Excel in After iteration: and reset the ArrayList.

Is it possible to transfer the ArrayList directly to Excel without iterating over every value in the list by using setCellValue()? Or is there another computationally efficient way to transfer the output of the parameter variation experiment to Excel?

Answer 1

To solve the issue I stored the value of each replication in an ArrayList experimentsResults in the parameter variation experiment. After each iteration, the ArrayList representing an iteration of multiple replications is transferred to a CSV file.

To store the data in an ArrayList a java class was created to represent all the data that needs to be transferred to the list. Then the code below is used in a function that is being triggered in After simulation run:

ReplicationData row = new ReplicationData();
    row.experimentDescription = experimentDescription;
    row.varCurrentIteration = varCurrentIteration;
experimentResults.add(row);

To write the ArrayList to CSV I used this function that is triggered in After iteration:

for(int j=0; j<experimentResults.size(); j++)
{

    String resultstring=""; 
    
    resultstring = experimentResults.get(j).experimentDescription+";"+
        experimentResults.get(j).varCurrentIteration+";"+       
    csvResults.println(resultstring);
}
csvResults.close();

The main thing that I was not aware of, was to place the Text File from the connectivity pallet directly in the experiment. This improved the writing speed a lot. After the experiment, the CSV file can be easily converted to Excel.