Natashia Boland and Open Pit Mining

I am attending the Australian Society for Operational Research meeting in Melbourne.  I had thought Australia always had wonderful weather, but it is a gray, rainy day here.

The conference was opened with a talk by Natashia Boland, currently at the University of Melbourne (but I hear she is moving) who spoke on her experiences in using integer programming in practice.

The first part of the talk was about modeling excavation in open pit mines.  The goal in this problem is to dig out a mine in the best possible way.  The most critical constraint is, of course, that you can’t dig under stuff that is not already dug away.  There are other constraints on how much ore can be processed in a year, and on other operational requirements.  Of course, the amount of money that is at stake is huge:  these mines generate  hundreds  of millions of dollars of income, and require huge investments (see the wikipedia entry on the Bingham Canyon Mine for one example:  the pictured example is from Russia).  So planning the excavation is an important decision, even if most of the savings are just the time-value of money (getting $100 million this year rather than next is quite a difference!).  The integer programs that Natashia and her colleagues get are very large, so they have had to develop specialized branching rules, along with aggregation approaches to the decisions.

Natashia’s second example had to do with shipping planning for a fertilizer company.  The most striking point she made is that her approach still leaves an 8% gap between the upper bound (feasible solution) and lower bound.  This bound may be because the lower bound is poor (no possible solution can reach that value) but even 1 or 2% of an operation that costs a few hundred million dollars to run is a big deal.

Natashia gives very good, clear, inspirational talks.  The thing I like best is her choice of problems.  They are non-standard, but still involve huge amounts of money.  Every percentage point she saves would be enough to support an enormous research effort in integer programming.  If only she were to get a fraction of the savings!