Work around

[Former Member]

Hi gurus,

We have an extremely large model with about 500,000 discrete variables. We have found that most of them are from Transport Lot Size profile as we need to deliver lot sizes multiples.

Does anyone know a work around to this issue?... we think the best idea is to delete the discrete restriction, but the issue is how to deliver this lot size multiples? .

Any idea would be appreciated.

-Italo

[Former Member · Accepted solution]

Hi there

I agree that this is a large discretisation problem, however I think the solution will depend on your exact requirements.

My first question would be, do you need the final result coming from the Optimiser? By this, I mean would it be okay fo rthe Optimiser to run without discrete lot sizes and suggest the ideal quantity. Then using Deployment Heuristic to round the final values to required Lot Size? This will cause a few stock issues where there is insufficient stocks to round all values, this is where your particular requirements kick in as to the answer.

We have a similar issue and think that the manual corrections following the Deployment Heuristic would be of a manageable size. By running this way we can have a Linear Optimisation run followed by a Deployment Heuristic running in far less time than a Discrete Optimisation model (by a fair few hours too).

Hope this helps and Good Luck.

Ian

[Former Member · Accepted solution]

Thanks Ian,

I think this fit to our needs.

-Italo