Evolutionary Algorithm for Minimization of Pumping Cost

Abstract

This paper deals with minimizing the total cost of pumping in a liquid pipeline. Previous experience with the most common solution procedures in pipeline optimization is discussed along with their strengths and weaknesses. The proposed method is an evolutionary algorithm with two distinct features: (1) The search is restricted to feasible region only; and (2) it utilizes a floating point decision variable rather than integer or binary as is the case with most other similar approaches. A numerical example is presented as a basis for verification of the proposed method and its comparison with the existing solver that utilizes the nonlinear Newtonian search. The proposed method provides promising improvements in terms of optimality when compared to the widespread gradient search methods because it does not involve evaluation of the gradient of the objective function. It also provides potential to improve the performance of previous evolutionary programs because it restricts the search to the feasible region, thus eliminating large overhead associated with generation and inspection of solutions that are infeasible. Comparison of the two solutions revealed improvement of the solution in favor of the proposed algorithm, which ranged up to 6% depending on the initial values of the decision variables in the Newtonian search. The proposed method was not sensitive to the starting value of the decision variables.