Title: Siting Substations in an Electrical Network


	      Walter Murray and Vinayak Vishnu Shanbhag
	   Department of Management Science and Engineering
	              Stanford University
	       email: walter@leland.stanford.edu
	   web page: http://www.stanford.edu/~walter/


Abstract:
The problem of where to place new substations in an electrical network to 
meet anticipated increases in load is challenging. It can be shown to be 
equivalent to solving a large mixed-integer quadratic program. The difficulty 
arises because the number of integer variable may be as great as 10,000. 
However, as the size of the grid grows the density of the nodes on which 
substations are placed remains fairly constant, which implies any guess at 
the location for a specific substation has to transverse a similar number of 
nodes to get to the optimum location regardless of the size of the grid. This 
in turn suggests if there was a way of moving all substations simultaneously 
then the iteration count to solve the problem would be largely independent of 
the problem size. We describe an algorithm that makes such moves and report 
on its computational performance.