Operations Research topics: SAS/OR,
SAS Optimization, and SAS Simulation Studio

ATM Cash Managment Problem Formulation

Frequent Contributor
Posts: 75

ATM Cash Managment Problem Formulation


I am exploring the ATM cash managment optimization problem as described in the following link:


In the MINLP section it says that "The predictive model fit is defined by the following data for each ATM $a$ on each day $d$: $c_{ad}, c^ x_{ad}, c^ y_{ad}, c^ z_{ad}, c^{u}_{ad}$"

What do the data express - what do they mean in business terms?

Also as i understand the problem has three variables xa, ya and ua. The xa and ya are binary and the ua is continuous, what do they express in business terms?

Thanks in advance,


SAS Employee
Posts: 49

Re: ATM Cash Managment Problem Formulation

Posted in reply to andreas_zaras

At a very high level, these variables represent a weighted fit model for the replenishment (of cash) schedule at each ATM. The idea is to try to match the predicted cash (subject to some side constraints) as closely as possible given the possible replenishment policies. The details of the replenishment schedules (and the fit model) are proprietary and were provided to us by the client who created the model as a proof of concept for their business problem. The goal of the documentation example is to show how we can go from an MINLP formulation to an approximate MILP formulation that has nice decmoposable structure and solves well using the DECOMP feature.

Super Contributor
Posts: 358

Re: ATM Cash Managment Problem Formulation

Posted in reply to andreas_zaras


Unfortunately, the link you posted is not working so I can't see the original problem.

I have had some experience in this area because I created ATM cash management programs and

predictive models for a major bank about 15 years ago - we were a little ahead of the game

at the time.

If I can access the link, I may be able to help here.

Posts: 8,094

Re: ATM Cash Managment Problem Formulation

Frequent Contributor
Posts: 75

Re: ATM Cash Managment Problem Formulation

Ask a Question
Discussion stats
  • 4 replies
  • 1 like
  • 4 in conversation