Solution of a Large-Scale Forest Scheduling Problem by Linear Programming Decomposition

The linear programming decomposition solution of a large-scale forest scheduling problem involving the cutting of 1166 woodlands over the next 24 years is discussed. The original problem was formulated as a linear programming problem which involved about 28,000 variables and 1200 restraint equations. However, a problem of this size is too large for a computer solution to be achieved by usual linear programming simplex solution methods. The decomposition approach treats a large linear programming problem as a series o! smaller linear programming subproblems whenever the structure of problem permits. Although the decomposition procedure was here terminated after the solution of 930 such subproblems, the gain in yield was extremely small alter 240 iterations such that three hours of IBM 7044 computation would have provided excellent operating results.