REAL-TIME SUPERVISION OF INDUSTRIAL WASTE- WATER TREATMENT PLANTS APPLIED TO THE SURFACE TREATMENT INDUSTRIES KONRAD SZAFNICKI 1 *, JACQUES BOURGOIS 1 , DIDIER GRAILLOT 1 , DOMINIQUE DI BENEDETTO 2 , PHILIPPE BREUIL 2 and JEAN-PIERRE POYET 2 1 Department of Environmental Engineering, Ecole Nationale SupeÂrieure des Mines de Saint-Etienne, 158 cours Fauriel, F-42023 Saint-Etienne Cedex 2, France and 2 Laboratory of Instrument Development and Chemical Analysis, Ecole Nationale SupeÂrieure des Mines de Saint-Etienne, 158, cours Fauriel, F-42023 Saint-Etienne Cedex 2, France (First received August 1997; accepted in revised form October 1997) AbstractÐThe project described in this paper consists of two main stages: the development of a dedi- cated instrument enabling continuous simultaneous measures of industrial pollutants (e.g. metal ions: Ni 2+ , Cu 2+ , Zn 2+ , ... ) and the design of a Real-time Expert System (RTES) which would provide bet- ter information and decision support so as to improve the supervision of waste-water treatment plants in real-time, applied to the surface treatment (electroplating) industries. The sensor is based on the prin- ciple of spectral absorption analysis. The real-time expert system is being developed using a commercial, industrially validated RTES-development software G2 2 (by Gensym Corp.). A mixed knowledge-base is being implemented, comprising both types of knowledge: quantitative (e.g. theoretical knowledge issued from chemistry) and qualitative (i.e. operators' `experience and know-how'). # 1998 Elsevier Science Ltd. All rights reserved Key wordsÐsupervision, real-time expert systems, UV-VIS spectrophotometry, metal ions analysis, industrial waste-water treatment, multivariable analysis INTRODUCTION Industrial activities, and particularly metal proces- sing activities (e.g. electroplating), use huge quan- tities of water during their manufacturing cycles. They thus generate signi®cant amounts of waste- water, which usually do not satisfy environmental regulations so as to be rejected directly to the receiving water. The regulations, however, tend to be more and more strict and limiting as far as admissible quan- tities of rejected pollutants (metal ions, cyanides, ... ) are concerned (e.g. in the new EEC-regu- lations). It is, therefore, of great interest, in order to be able to satisfy the present and future limi- tations, to determine and develop new, reliable means of supervision and treatment of industrial waste-waters (Serra et al., 1994). Moreover, Small and Medium-size Enterprises (SME) will certainly not feel very enthusiastic about the idea of installing huge numbers of additional sensors, if the idea is not properly justi®ed. The improvement of the treatment techniques thus implies two main directions of research eorts: . the development of sensors which would take into account both the quality constraints imposed by the present and future waste-water regulations and, on the other hand, the economical require- ments of the concerned enterprises; . an `intelligent' processing of information, pro- vided by the sensors already installed in situ. At present, most enterprises involved in the sur- face-treatments, and particularly SMEs, cannot aord a full-time agent for the supervision of the industrial waste-water treatment plants. On the other hand, operators can only note the problems (faults, pollutions, ... ) which have occurred, from the Central Monitoring Station, but still they can hardly prevent them, since there are no reliable industrial sensors available for the on-line simul- taneous monitoring of the amounts of dierent metal cations (particularly: ions of Cu, Ni, Zn, Fe, Cr, ... ) or cyanide ions in the rejected waste- waters. Moreover, the supervision agent (operator) rarely has at his/her disposal any computer support, which would provide an ecient aid in the analysis of the available information and the decision mak- ing. Modern Ultraviolet-Visible (UV-VIS) spectropho- tometers, associating both sophisticated hardware- ®bre optics, holographic ¯at®eld gratings, photo- diode arrays- and software-multivariate analysis Wat. Res. Vol. 32, No. 8, pp. 2480±2490, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/98 $19.00 + 0.00 PII: S0043-1354(97)00441-7 *Author to whom all correspondence should be addressed. [Fax: +33-4-77-42-66-66; E-mail: konrad@emse.fr]. 2480