\n
Finding the solution to the isothermal flash problem is a time consuming and complex element of compositional transient reservoir and process simulation. The conventional method (Michelsen, 1982b) formulates the problem as an unconstrained minimisation of the reduced Gibbs energy.
\nThe molar amount of one phase is made dependent through the material balance to eliminate the constraints. For multiphase problems the dependent phase must be selected individually for each component, complicating the implementation. Furthermore chemical reactions cannot be easily incorporated into the conventional flash algorithms.
\n
\nTwo algorithms are proposed for isothermal flash calculations, modified RAND and Vol-RAND. The former uses the equilibrium elemental potentials and phase fraction as independent variables and is similar to non-ideal RAND (Greiner, 1991), while the latter co-solves a pressure explicit equation of state with the equilibrium equations using the equilibrium elemental potentials and phase volumes.
\n
\nThe advantages of these methods are: the algorithms are quadratically convergent; the elemental balance constraints are met after each iteration therefore the Gibbs energy can be monitored and a line search performed if necessary; all components in all phases are treated in the same manner leading to a simple implementation; and chemical reactions can be included or the methods simplified for purely phase equilibria.
\n
Finding the solution to the isothermal flash problem is a time consuming and complex element of compositional transient reservoir and process simulation. The conventional method (Michelsen, 1982b) formulates the problem as an unconstrained minimisation of the reduced Gibbs energy.
\nThe molar amount of one phase is made dependent through the material balance to eliminate the constraints. For multiphase problems the dependent phase must be selected individually for each component, complicating the implementation. Furthermore chemical reactions cannot be easily incorporated into the conventional flash algorithms.
\n
\nTwo algorithms are proposed for isothermal flash calculations, modified RAND and Vol-RAND. The former uses the equilibrium elemental potentials and phase fraction as independent variables and is similar to non-ideal RAND (Greiner, 1991), while the latter co-solves a pressure explicit equation of state with the equilibrium equations using the equilibrium elemental potentials and phase volumes.
\n
\nThe advantages of these methods are: the algorithms are quadratically convergent; the elemental balance constraints are met after each iteration therefore the Gibbs energy can be monitored and a line search performed if necessary; all components in all phases are treated in the same manner leading to a simple implementation; and chemical reactions can be included or the methods simplified for purely phase equilibria.