|Title||A Versatile Simulation Method for Complex Single Mixed Refrigerant Natural Gas Liquefaction Processes|
|Publication Type||Journal Article|
|Year of Publication||In Press|
|Authors||Vikse M, Watson HAJ, Gundersen T, Barton PI|
|Journal||Industrial & Engineering Chemistry Research|
Natural gas liquefaction is an energy intensive process with very small driving forces particularly in the low temperature region. Small temperature differences in the heat exchangers and high operating and capital costs require the use of an accurate and robust simulation tool for analysis. Unfortunately, state-of-the-art process simulators such as Aspen Plus and Aspen HYSYS have significant limitations in their ability to model multistream heat exchangers, which are critical unit operations in liquefaction processes. In particular, there exist no rigorous checks to prevent temperature crossovers from occurring in the heat exchangers, and the parameters must therefore be determined through a manual iterative approach in order to establish feasible operating conditions for the process. A multistream heat exchanger model that performs these checks, as well as area calculations for economic analysis, has previously been developed using a nonsmooth modeling approach. In addition, the model was used to successfully simulate the PRICO process with the Peng-Robinson equation of state. However, the PRICO process is one of the most basic single mixed refrigerant processes, and it is therefore necessary to investigate whether the nonsmooth framework is capable of also simulating larger and more complex single mixed refrigerant processes. In this article, the nonsmooth multistream heat exchanger model is used to simulate three different single mixed refrigerant processes of varying complexity. Different case studies are performed, each solving for a different set of unknown variables. Several different variables were considered in the analysis to investigate whether the models obtained feasible solutions even for ostensibly challenging cases such as varying the mixed refrigerant composition. The solutions are then validated using results from Aspen Plus and Aspen HYSYS. The simulations in Aspen Plus gave nearly identical solutions to the nonsmooth models. Results in HYSYS, on the other hand, correlated well at high temperatures but deviated from the nonsmooth solution at cold temperatures. The disparity was caused by different ideal gas enthalpy correlations used by the two simulation tools.