On-line inference of tube-wall temperature in an industrial olefin pyrolysis plant

A major difficulty in the efficient functioning of industrial pyrolysis is the lack of suitable on-line tubewall temperature measurements. In this paper, an inferential scheme is developed to predict the tube-wall temperature, which is called skin temperature, in an industrial olefin pyrolysis plant. Based on a theoretically- derived model, the skin temperature is predicted from the available on-line measured process variables, such as gas temperature around the tube, fuel gas flowrate, and naphtha feed flowrate. Using the infrequently off-line measured skin temperature with the on-line measurements, adjustable parameters of the model are updated on-line according to the Kalman filter technique. The application of this scheme then is illustrated using operating data obtained from an industrial olefin pyrolysis plant. The results of the proposed scheme show that, by taking infrequent measurements of the skin temperature for model adaptation, the skin temperature can be predicted successfully for every tube in the pyrolysis over an extended time interval. Simulations suggest that as few as one or two manual measurements of the skin temperature, taken during the entire cracking operation, provide sufficient information to maintain the developed inferential system at a practical level.