Assessment of overall heat transfer coefficient models to predict the performance of laboratory-scale jacketed batch reactors

Heat transfer models for agitated, jacketed, laboratory-scale batch reactors are required to predict process temperature profiles with great accuracy for tasks associated with chemical process development such as batch crystallization and chemical reaction kinetics modeling. The standard approach uses a reduced model which assumes the system can be adequately represented by a single overall heat transfer coefficient which is independent of time; however, the performance of reduced models for predicting the evolution of process temperature is rarely discussed. Laboratory scale (0.5 and 5 L) experiments were conducted using a Huber thermoregulator to deliver a thermal fluid at constant flow to a heat transfer jacket. It is demonstrated that the relative specific heat contribution of the reactor and inserts represent an increasing obstacle for these transient models with decreasing scale. However, a series of experiments implied that thermal losses were the limiting factor in the performance of a single coefficient reduced model at laboratory-scale. A diabatic model is presented which accounts for both thermal losses and the thermal inertia of the reactor vessel and inserts by incorporating a second coefficient and a modified heat capacity term. The mean absolute error in predicted process temperature was thereby reduced by a factor of 8, from 2.4 to 0.3 K, over a 150 min experiment.