Cost Optimization of Nanofiltration with Fouling by Natural Organic Matter

Abstract

An artificial neural network and genetic algorithm routine has been developed for predicting and optimizing membrane system performance. The model predicted system behavior in response to operating conditions of applied pressure and crossflow velocity. Artificial neural networks accurately modeled mechanisms involved in fouling of membranes by natural organic matter. The model correctly predicted the effects of calcium within the solution in exacerbating fouling, binding of the divalent calcium ions to the natural organic matter macromolecules, and the formation of complexes. The model also correctly predicted the role of increased pressure in inducing fouling and the reverse scenario of mitigating fouling with increased crossflow velocity. The model was applied to membrane plant design for determining cost-effective operations. The genetic algorithm routine searched the predictions of the system model to determine the optimal operating conditions. Fouling conditions induced by the presence of calcium resulted in escalating costs with increases in calcium concentration. Membrane-related cost components were shown to be a significant cost factor that is sensitive to operating conditions and represents a prime target for optimization.