Most convergence issues can be traced back to the usual culprits: stream inputs, UnitOp specifications, or thermodynamic selections. Sometimes, though, convergence issues are rooted in how the simulation is attempting to solve the flowsheet.
CHEMCAD offers control over the calculation sequence and convergence methods, which define how the simulation engine works through your flowsheet. Tweaking these settings can often provide the missing piece that solves what changing inputs and specs alone can’t fix.
Calculation Sequence: Why the Order Matters
CHEMCAD offers a choice in the calculation sequence, which defines the order in which the software will solve the flowsheet.
Sequential modular is the default in CHEMCAD, and it’s often a good choice for flowsheet calculation sequence. With this setting, each unit operation is solved in order, one at a time, from upstream to downstream. In this way, each downstream unit operation depends on the results of the one before it. This gives you more control over the flow of information to solve the flowsheet, and makes it easier to troubleshoot the simulation.
Sometimes, flowsheets contain complex non-sequential dependencies that make sequential, step-by-step solving impractical. In some cases, downstream conditions influence upstream behavior, so a sequential solving path creates “undefined” inputs or specifications. Using a different sequence makes proper calculation possible. Some common situations where this comes into play include:
- Recycle loops, where streams loop back into the upstream sections of the flowsheet
- Heat integration, where the energy balance forms dependencies between multiple interconnected UnitOps
- Controllers, which can adjust upstream variables based on downstream targets
- Pressure nodes, where flow and pressure balance can create linked dependencies among multiple UnitOps
In these cases, a simultaneous modular calculation sequence would be a better fit. Instead of solving each UnitOp in sequence, the solver considers each UnitOp to be an element in a matrix. When the simulation is run, CHEMCAD attempts to solve all UnitOps simultaneously, using matrix methods. This allows upstream and downstream parts of the flowsheet to update together during each iteration, resulting in faster, more stable solutions.
Convergence Methods: The Next Step in Solving Complex Loops
When a flowsheet contains recycles, controllers, or other complex dependencies, changing the calculation sequence can certainly help. In complex simulations such as these, you might also consider changing the convergence method. Convergence methods tell CHEMCAD how to iteratively adjust calculations and compare results until the model settles on a solution. There are three convergence methods to choose from:
- Direct substitution, where each calculation iteration result feeds directly into the next. This is the default method and it works well for most calculations, but can be too simple and slow for complex loops.
- Wegstein, which attempts to predict where the answer is headed to speed up convergence. This method requires that you set appropriate bounds.
- Dominant Eigenvalue (DEM), which accounts for the interdependence of variables and uses damping to stabilize to a solution.
Some methods are better suited to predict the next values more intelligently than others; the right choice can make the difference between quick convergence and endless iterations. If the recycle calculation is slow, try switching to Wegstein for speed and accuracy. If there are nested recycle or controller loops, try switching to DEM.
The Convergence Parameters Screen
Both the Calculation sequence field and the Recycle Convergence Methods settings are found on the Convergence Parameters dialog box. To bring up this dialog, find the Setup group on the Home tab and click Convergence.
Understanding how a process simulator converges to a solution can help you run and troubleshoot simulations more efficiently. When you strategically adjust the calculation sequence and convergence method, CHEMCAD can better navigate complex problems and solve even the most challenging process flowsheets.