engleski

Design and Multi Objective Optimization of Aspirating Smoke Detector Pipeline

This work presents multi-objective optimization of early fire detection system which uses pipeline to transport air sample from sampling hole to the analysing module. First, 3D computational fluid dynamics simulation was conducted which has shown that the smoke transport will not always take place in the centre of the pipe and that one dimensional analysis is not able to provide information in which layer smoke transport will take place. To bind the calculation on the safe side, velocity from slower layer near the pipe wall was taken as an input for calculation of the transport time. Another reason for choosing this approach is that the 1D calculation doesn’t include diffusion phenomena, so in the case of pure convective transport, especially in the laminar or sublaminar flow regime, the sample will have lower velocity near the pipe wall, compared to the mean velocity. It has also been shown that mathematical model of the smoke detection chamber filling is valid for stationary or quasistationary transport respectively, which corresponds to the case when large part of transport pipeline is filled with smoke. By employing the multi-objective optimization approach balancing of specific transport time and volume flow at the samplings points’ location was achieved. Pipeline configuration was also varied in order to provide better mixing of samples at junction points. Four optimization problems have been set, with four different objective functions, and with the same inequality boundaries. For the first objective function Euclidean norm of differences was used, and for the other three objective functions the infinity norm of differences was used. From all optimization problems that were solved, best solution gives fourth problem where transport time balancing and volume flow balancing were combined together.

Multi-objective optimization ; Objective function ; Norm ; Smoke detectors ; Chemical species transport ; CFD ; Transport time ; Pipeline ; Sampling

nije evidentirano