Southern California Gas Company uses special ordered sets to model regulatory guidelines

Modeling rule-based decisions often conflicts with the mathematics of optimization because the rules may conditionally force actions that are not economically optimal. In order to simulate the reality of the public utility business, Southern California Gas Company (SoCalGas) includes in their operations planning models rule-based guidelines regulating the treatment of customers and suppliers. One approach to modeling such conditional guidelines is to use mixed integer programming (MIP); however, MIP is often avoided if possible because of its reputation for poor performance. SoCalGas uses an advanced mixed integer programming solution technique, special ordered sets (SOS) of type 2, to model the guidelines effectively and efficiently.