Analysis of heat transfer fouling data using artificial neural networks
Zusammenfassung der Projektergebnisse
The formation of deposits on heat transfer surfaces is still the least understood phenomenon in heat exchanger design, causing severe problems in the design and operation of heat transfer equipment. Despite increased attention during the past decades, presently used design procedures involve massive uncertainties, while the recommended correlations and computer models can only be applied to a very limited number of highly idealised deposition processes. Part of this unsatisfactory situation must be attributed to the complexity of fouling, because of the large number of variables and mechanisms. However, some blame must also be laid on the inherent inadequacy of conventional regression methods to correlate experimental data with an ill-distributed parameter variation. One of these inadequacies is the lack of knowledge to justify whether the chosen regression equation is the best. In addition, the independent variables are usually selected based only on experimental observation or solely due to their availability. Introduction of redundant inputs or ignoring of important variables may severely reduce the reliability of any model. The proposed research project basically endeavoured 1. to analyse several comprehensive data banks with fouling data from numerous sources and applications, notably the unique, proprietary and hitherto unavailable HTRI (Heat Transfer Research Inc.) database, 2. and then to implement artificial neural networks (ANN) trained on these data bases to correlate/predict the fouling behaviour. Preliminary studies by the applicants demonstrate that the ANN architecture may be a promising tool for this purpose. Primary databases with the original experimental fouling data and the related operating conditions such as surface temperature, bulk temperature, fluid composition, flow velocity, flow geometry, etc. have firstly been converted into a working database consisting of an ensemble of possible dimensionless groups, e.g. Reynolds number, Prandtl number, Saturation Index, etc. One of the most important tasks was the selection of the fittest combination of dimensionless groups to be used as ANN (Artificial Neural Network) inputs to predict fouling resistances. In the present research project, the investigation then focussed on the ubiquitous and generic problem of scale formation from cooling water. The assembled data base contained fouling data for waters containing CaS04, CaCOs, BaSO4 etc., their mixtures, and cooling water containing various scaling inhibitors. In addition to the identification of key parameters and their effects on scale formation, the trained neural network has been used to generate with acceptable reliability a we 11-structured and well-distributed data base, which can then be used to curve-fit correlations for implementtation into computer codes.
Projektbezogene Publikationen (Auswahl)
-
An overview of fouling mechanisms, prediction and mitigation strategies in thermal desalination plants, proceedings of the llth Int. Water Tech. Conf. (IWTC2Q07), March 15-18, Sharm el-Sheikh, Egypt.
Malayeri, M.R., and Müller-Steinhagen, H.
-
Enhanced pool boiling heat transfer of refrigerants from plasma-spray coated surfaces, HeatSET-2007, Chambery, France, Vol. 1, pp. 287-294.
Malayeri, M.R., Schäfer, D., and Müller-Steinhagen, H.
-
Experimental investigation of crystallization fouling on grooved stainless steel surfaces during convective heat transfer: First results, Proceedings of the ECI Conf. on Heat Exchanger Fouling and Cleaning-VII, Tomar, Portugal.
Herz, A., Malayeri, M.R., and Müller-Steinhagen, H.
-
Fouling of roughened stainless steel surfaces during convective heat transfer to aqueous solutions, Proceedings of the 3r Int. Conf on Thermal Eng.: Theory and Applications, ICTEA_2007, May 21-23, Amman, Jordan, pp. 602-606.
Herz, A., Malayeri, M.R., and Müller-Stemhagen, H.
-
Initiation of CaSÜ4 scale formation on heat transfer surfaces under pool boiling conditions, Heat Transfer Eng,, Vol. 28, No. 3, pp. 240- 247.
Malayeri, M.R., and Müller-Steinhagen, H.
-
Intelligent discrimination model to identify influential parameters during crystallisation fouling, Proceedings of the ECI Conf. on Heat Exchanger Fouling and Cleaning-VII, Tomar, Portugal.
Malayeri, M.R., and Müller-Steinhagen, H.
-
Recent advances in heat exchanger fouling research,, Heat Transfer Eng., Vol. 28, No. 3, pp. 173-176.
Müller-Steinhagen, H., Malayeri, M.R., and Watkinson, A,P.
