Chemical Engineering Science 59 (2004) 489–499 www.elsevier.com/locate/ces Fouling resistance modelling, identication and monitoring of a thermosiphon reboiler V. Grosls * , M. Kinnaert, Ph. Bogaerts, R. Hanus Service d’Automatique et d’Analyse des Syst emes, Universit e Libre de Bruxelles, CP 165/55, 50, Av. F.D. Roosevelt, Brussels B-1050, Belgium Received 9 January 2003; received in revised form 26 March 2003; accepted 16 October 2003 Abstract The aim of this paper is to describe a system for fouling monitoring of thermosiphon reboilers. This system is based on a static model of the fouling resistance. The latter has been constructed in three steps. First, a mathematical model of the fouling resistance based on the overall heat transfer coecient and on the inside lm heat transfer coecient is developed. Then, the overall heat transfer coecient is obtained from the mass and energy balances of the reboiler. In the last step, the inside lm heat transfer coecient is determined from correlations with some process variables. The resulting model is a function of measured quantities and of two unknown parameters. A procedure for identication of the unknown parameters from data selected so that hypotheses behind the model are fullled is presented. The eectiveness of the resulting monitoring system is illustrated by processing industrial data. ? 2003 Elsevier Ltd. All rights reserved. Keywords: Heat transfer; Fouling; Boiler; Modelling; Evaporation; Supervision 1. Introduction The fouling of a heat exchanger can be dened as the deposition of unwanted material on heat exchange surfaces. A consequence is the degradation of the performance of the reboiler due to the decrease of the heat transfer coecient. It implies economic and energetic losses and unexpected interruption of the process for cleaning. If the detection of fouling could be precocious, predictive maintenance of the reboiler could be achieved. The aim of this study is to build a fouling monitoring sys- tem for a thermosiphon reboiler. This system must be easy to use and to implement on-line and must take into account the industrial constraints. Indeed, the process knowledge is limited in practice (for instance, in the studied industrial plant, neither output temperature nor output pressure of the steam is measured). Moreover, the monitoring should not require any articial solicitation in the form of additional changes in the process-working mode. The heat transfer coecient or the pressure of the steam needed to evaporate the uid are not suitable to follow * Corresponding author. Tel.: +32-2650-2287/2613; fax: +32-2650-2677. E-mail address: valerie.grosls@ulb.ac.be (V. Grosls). fouling. Indeed, they are function of the process-working mode and give no absolute measure of fouling. In the literature, some eorts have been directed towards modelling and detecting fouling. On the one hand, models of the fouling resistance, dened as the deposit thickness di- vided by the thermal conductivity of the deposit, have been described (Himmelblau, 1978; Palen, 1986; Taborek et al., 1972a,b). However some of them (Palen, 1986) are dicult to use because of their complexity and of the introduction of variables that are hard to determine. The model proposed for cooling tower water fouling in (Taborek et al., 1972a,b) is based on deposition mechanisms, and requires the iden- tication of parameters from experimental fouling curves. Experimental determination of such curves is not feasible in most industrial environments, as is the case of our study, which severely limits the applicability of this approach. On the other hand, the static fouling resistance model for heat exchangers presented in Himmelblau (1978) is based on the overall heat transfer coecient obtained from an energy bal- ance, on the inside lm coecient and on the outside lm coecient, both dened as functions of the ow rate. In this work, the fouling resistance is computed by least squares given a few values of ow rates and temperatures. But the model is only valid for heat exchangers without boiling. Other methods to detect fouling are based on white or grey box models, but they are not suitable for our 0009-2509/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2003.10.011