Abstract

This paper presents a robust soft sensor for estimating heat exchanger fouling in combustion–thermal power plant context. The approach is data–driven and focuses on identifying the effect of plant load changes to fouling estimation. Proposed method is applied to real process measurements and results are presented. The method consists of two blocks; a static energy balance calculation block for “traditional” fouling indicator calculation and a dynamic subspace identification block for finding sootblowing– and load level dynamics components of the static fouling indicator signal. Results from applying the proposed method to real plant data show that load level dynamics can be decoupled from static fouling estimate.