Accounting for stream variability in retrofit problems using Monte Carlo simulation

Abstract

The aim of this paper is to use Monte Carlo simulation to analyse the effect of stream data variation on the economic performance of retrofit designs. The input stream data is fitted to a distribution which can be sampled from to calculate the outputs of the Heat Exchanger Network model, stochastically. A simple four-stream problem is used to demonstrate the method, comparing two retrofit designs that each reduce the hot utility by 700 kW. Monte Carlo simulation analyses how the similar designs behave differently under the variable conditions and the results show that time-average analysis can not only overestimate the Total Retrofit Profit but can also favour a different design to that favoured by the Monte Carlo results. Of the two compared designs, the Design 2 was selected with an annual Total Retrofit Profit of NZD 215,553, requiring one new exchanger and additional area on an existing one. The Monte Carlo analysis also shed light on controllability and feasibility issues in the Heat Exchanger Network.