Pipeline optimization using a novel hybrid algorithm combining front projection and the non-dominated sorting genetic algorithm-II (FP-NSGA-II)

CATEGORY: PIPELINES
2013 IEEE Congress on Evolutionary Computation (CEC), 20-23 June 2013, Page(s): 697 – 704, Cancun, Digital Object Identifier :10.1109/CEC.2013.6557636
Pipeline optimization using a novel hybrid algorithm combining front projection and the non-dominated sorting genetic algorithm-II (FP-NSGA-II)
Fettaka, S. ; Thibault, J.
Dept. of Chem. & Biol. Eng., Univ. of Ottawa, Ottawa, ON, Canada
Abstract
Authors describe a procedure for minimizing the pumping power, the number of pumping stations and the total pipeline mass required for a pipeline project using multiobjective optimization. They present two and three-objective optimization cases. The decision variables employed include the outer diameter, wall thickness, suction pressure and discharge pressure. They propose an innovative hybrid multiobjective optimization algorithm combining NSGA-II with a simple front prediction (FP-NSGA-II) to improve upon the performance NSGA-II. They apply the algorithm to a problem taken from the open literature.
The resulting Pareto domain is ranked using a cost function. Results suggest that FP-NSGA-II significantly improved convergence, spread and number of non-dominated solutions for the determination of the optimal design for a specified pipeline problem.
Authors explore two- and three-objective optimization cases to minimize the pumping power, the number of pumping stations and the pipeline mass requirements of a new pipeline project using a novel hybrid algorithm, the FP-NSGA-II. The algorithm is based on the Lamarckian approach. It utilizes a simple front projection module which improves upon the convergence and diversity of solutions obtained with existing NSGA-II. The main innovation of the algorithm is the use of a front projection operator following the fast non-dominated sorting procedure. The front projection operator computes the distance in the decision variable space between each solution in the first front and the nearest neighbor solution in the second front.
Full Text Source (Subscription or Fee): http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6557636