#TITLE_ALTERNATIVE#

• Main Author: HABIBI (NIM 25306010), LAFRAN
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/8449
• Institution: Institut Teknologi Bandung

Water distribution system is a very complex pipeline network. The complexity of the network causes problem in flow rate and pressure distribution which is related with the fulfillment of hydraulic criteria on water distribution system. That problem is well-known as network flow analysis. For solving that problem, it is necessary to set up a water distribution network model. The objective of this research is making a simple program for solving the problem of network flow analysis of water distribution system using Genetic Algorithm and Newton’s Method. In this research, steady-state model is set up from node (H) equation, by combining continuity equation and conservation of energy. As a hydraulic formula, this modelling uses formula of Hazen-Williams. Mathematically, the system model consists of a set of nonlinear simultaneous equations. This research produces a simple program using MATLAB which can be used to simulate network flow analysis of water distribution system. The result of simulation using the program shows that for simple pipeline network (pipeline network with 9 nodes and 12 links), Genetic Algorithm is able to work as solver by doing repeatedly simulation and set the paramater of Genetic Algorithm by adding the number of the population and maximum generation. Whereas for more complex pipeline network (pipeline network with 33 nodes and 40 links), Genetic Algorithm which stand alone as a solver may produce unsatisfying result. By combining Genetic Algorithm and Newton’s Method, the solution of water distribution system model becomes more powerful. The weakness of Genetic Algorithm, which is slow in the process, is covered by the power of Newton's Method, which has a quadratic convergence. Whereas, the weakness of Newton's Method, which need a good initial guess, is covered by the power of Genetic Algorithm, which can produce candidates of solution that close to the roots of the system model equation.