Soft-sensing of key process variables in a biomass combustion plant

Using physical instrumentation and measuring network to monitor a large set of process variables is of a crucial importance in any industrial plant. However, endowing the process with more sophisticated instrumentation will not only increase the investment capital in the plant, but also the maintenance planning and scheduling time. Furthermore, some process variables that are of relevance cannot be measured. A cost-effective way to overcome such limitations is by using the soft-sensing methodology. In this work, a virtual sensor is developed for a biomass heat recovery power plant to predict multifarious key process variables that will help in estimating the calorific value of biomass solid fuel. For this purpose, the process measurements, obtained from the existing physical instrumentation, are leveraged by using an analytic model, which is based on biomass combustion stoichiometry. Finally, the concept is validated by comparing the predicted steam amount obtained from the soft-sensor against the measured one.