Null completions and taking averages

Null completions and taking averages

Postby andrew_student » Sun Feb 28, 2010 5:32 pm

2 of my work exit points sometimes never actually end up with any work items. One of them gets no work items in a run something like 80% of the time and the other about 15% (both only guesstimates). The average time taken by work items ending at these work exit points is needed but when I performed trials it was coming out way bellow expected.

I have found out why though; for each run where there are no work items which finish here the system records an average of zero, it then averages these averages over all the trials.

I tested this hypothesis by throwing in one extra work item per run requiring the same amount of time as the expected average and feeding it directly into the work exit point. The result came out as I had expected, thus proving the hypothesis and the range of values was also as expected.

However is there a more elegant way of solving this sort of problem?

Return to “SIMUL8 Cafe”