Solving highly-looped systems in Flarenet
Refinery flare systems can be very complex, potentially allowing for the relief of hundreds of PSVs into a system composed of many laterals and headers routing flow to flare(s) or for routine flow possibly to a flare gas recovery unit.
A flare modeling application such as Aspen Flare System Analyzer (also known as "Flarenet") allows the user to model such systems, providing significant flexibility in terms of the size and scope of the model, as well as the scenarios which can be analyzed for that system.
However, these systems can potentially include many jumpovers, looped lines and multiple flares which be quite a challenging problem for a solver.
First, one must be careful when using Flarenet that the system has actually converged to reasonable tolerances. The software is dangerously subtle with respect to the convergence tolerances as it appears on the surface that the case has successfully converged. However, it is important to ALWAYS check the trace window which would indicate whether or not the case has successfully converged, as well the Results (Messages option, Solver tab). Scrolling down to the bottom of the Solution page, the user can determine whether or not Flarenet has properly converged. If the convergence message close to the bottom of the list indicates "Solved" then one can be certain that the errors are lower than the specified tolerances. Typically this means that the pressure errors should be smaller than 1e-2.
However, if the trace window indicates "Scenario Failed to Converge", and a convergence message on the Solver tab indicates something to the effect of "Loop Failure" then the case has not properly converged.
When solving difficult systems, we recommend the following procedure:
Try solving the scenario based on the previous solution.
If this does not work, clear the estimates for the current Scenario (this can be done by editing the Scenario from the Build menu, and clicking the "Clear" button on the Estimates tab), then re-run the scenario.
If the case does not converge after that, typically flow or pressure tolerances will not have been met at junctions (tees). Find one of these junctions (usually the one with the largest error) and ignore one of the pipes in the vicinity of the tee.
Re-run the case and repeat the last step as needed until the case converges. This may require several pipes to be ignored.
Once converged, un-ignore one pipe at a time; typically the case will re-solve successfully as this is done.
Finally once the last pipe is un-ignored you will usually have a successfully converged case.
In our experience, this is a procedure that consistently works. If it doesn't work for you, then you may need to provide initial values on the Estimates tab of the scenario (Aspen recommends using the area ratio to split the flow).
Finally, if that doesn't work, contact Aspentech Technical Support!
Flarenet provides a great deal of flexibility and can be an extremely powerful application for analyzing flare systems. For very complex systems you may run into difficulties but with patience these can usually be resolved.
Do you have questions or comments regarding this article? Click here to contact us.