Multiphase isenthalpic flash: General approach and its adaptation to thermal recovery of heavy oil

Abstract

Isenthalpic flash is a basic equilibrium calculation in process simulation. The recent interest in isenthalpic multiphase flash is mainly caused by the need for simulating thermal recovery of heavy oil. We present here systematic solutions to multiphase isenthalpic flash with full thermodynamics (such as EoS models) or with correlations for K‐factors, and discuss how to tailor the general methods to systems encountered in thermal recovery of heavy oil. First, for the general situation with full thermodynamics we recommend a solution strategy which uses Newton's method for rapid convergence in the majority of cases and Q‐function maximization to safeguard convergence when Newton's method fails. The solution procedure is a generalization of Michelsen's state function based two‐phase flash to multiple phases. The general solution does not need special considerations for the components in the system and is not limited to the selected thermodynamic models and the number of phases. For thermal recovery processes where gas, oil, and aqueous phases are typically involved, the stability analysis and initialization steps are tailored to improve the efficiency. Second, as it is quite common in thermal reservoir simulators to describe phase equilibrium and heat properties with temperature‐dependent K‐factors and separate correlations for heat capacities, we propose a formulation as an extension of the ideal solution isothermal flash formulation to solve such problems. It uses a Newton–Raphson procedure to converge in the majority of cases and a nested loop procedure with the outer loop for a temperature search as a fallback approach for convergence. If the correlations for K‐factors and for heat capacities are thermodynamically consistent, the outer loop can be treated as a maximization. Finally, we present systematic tests of the proposed algorithms using examples with full thermodynamics or K‐factor based thermodynamics. The algorithms prove robust and efficient even in challenging cases including a narrow‐boiling system, a degenerate system, and a four‐phase system. The additional computational cost relative to the corresponding isothermal flash is modest and would be suitable for the purpose of thermal reservoir simulation. © 2018 American Institute of Chemical Engineers AIChE J, 65: 281–293, 2019