Spring 2004 |
Most senior design courses include a component of steady-state process simulation in the synthesis and analysis of a process flowsheet. Many of the processes developed have recycle streams. One of the most challenging aspects of developing and using a steady-state simulation of a process with recycle is the convergence of the recycle loops, which requires the solution of a very large number of simultaneous nonlinear algebraic equations. Despite the fact that all commercial process simulation software products have sophisticated convergence algorithms, convergence is all too frequently not achieved. In many flowsheets, the user spends an inordinate amount of time and energy trying to find a way to attain convergence: changing the maximum number of iterations, trying different algorithms (Wegstein, Broyden, Newton, etc.), changing error tolerances and the like.
The purpose of this paper is to suggest an alternative approach that uses a
dynamic simulation of the process to converge to a steady state. The basic
idea is not new. Relaxation algorithms have been around for decades. However,
there are two recent developments that make this approach more practical and
effective. The first is the availability of commercial dynamic simulators to
accomplish the job. The second is an easily applied plant-wide control
procedure, which is one of the prerequisites of any dynamic simulation.
The suggested procedure is to develop the steady-state flowsheet with the
recycle loops "open", adjusting operating conditions and design
parameters until the assumed values of the "tear streams" are
reasonably close to the calculated values. Then the simulation is converted into
dynamic form with the necessary dynamic sizing parameters supplied (column
diameters, surge-drum sizes, etc.). Finally the recycle loops are
connected, the plant-wide control structure is applied and the dynamic
simulation is run out to steady-state conditions. The proposed method is
illustrated in an example (the dimethyl ether process).