Yechun Zhang

Virtual milk for modelling and simulation of dairy processes

The modeling of dairy processing using a generic process simulator suffers from shortcomings, given that many simulators do not contain milk components in their component libraries. Recently, pseudo-milk components for a commercial process simulator were proposed for simulation and the current work extends this pseudo-milk concept by studying the effect of both total milk solids and temperature on key physical properties such as thermal conductivity, density, viscosity, and heat capacity. This paper also uses expanded fluid and power law models to predict milk viscosity over the temperature range from 4 to 75°C and develops a succinct regressed model for heat capacity as a function of temperature and fat composition. The pseudo-milk was validated by comparing the simulated and actual values of the physical properties of milk. The milk thermal conductivity, density, viscosity, and heat capacity showed differences of less than 2, 4, 3, and 1.5%, respectively, between the simulated results and actual values. This work extends the capabilities of the previously proposed pseudo-milk and of a process simulator to model dairy processes, processing different types of milk (e.g., whole milk, skim milk, and concentrated milk) with different intrinsic compositions, and to predict correct material and energy balances for dairy processes.

A Numerical Modelling Approach for Dual Reflux PSA Separation of N2 and CH4 in LNG Production

Abstract In this paper, a novel, full Dual Reflux PSA (PSA) model is proposed for separating methane and nitrogen mixtures. The model is a full, integrated ODE model, in contrast to our former work where stripping PSA and enriching PSA were simulated separately and were combined using the total material balance (referred to as a transfer function model). The full model was built rigorously in a commercial simulator in terms of mass balances, energy balances and pressure-flow relationships, and it was numerically solved by the ODE solver which was integrated in the simulator. The transfer function model which was proven to have a close match with experimental data was compared to the results from the full DR-PSA model using the same key parameters, such as feed composition, total throughput, reflux ratios and cycle time, and resulted in a close match. The impact of feed position, feed temperature, feed composition and cycle time on the system was also studied with the new model.

Modelling batch bioreactions with continuous process simulators

Process models are increasingly becoming necessary for process understanding and optimizing. However, in the field of bioprocessing, modelling using a generic process simulator suffers from shortcomings given that many simulators cannot model batch mode bioreactors with varying reaction rates based on microbial metabolism. Bioprocesses, mostly batch, are difficult to model because most of the available unit operations in process simulators operate continuously. Our aim is to provide a general simulation platform to model these batch bioprocesses in commercial process simulators and explain how to transform batch processes with varying reaction rate into continuous unit operations for simulation purposes. Two typical fermentation processes, lactose fermentation and glucose/xylose co-fermentation, were simulated in steady state as case studies. The results are discussed as examples using the proposed approach. The potential of how to extend the simulation platform is also explained in detail.