NPSS cycle solver: residual convergence without a documented positive mass-flow guard

Question

NPSS closes an off-design thermodynamic cycle by driving residual and target equations to zero with an iterative solver. The documented stopping rule is a residual tolerance combined with an iteration limit, not a test for physical feasibility. This means the solver can report convergence at a mathematically valid but non-physical root—for example, non-positive mass flow or pressure—depending on the initial guess and solver settings.

The configuration surface that matters is the solver tolerance and iteration controls together with user-supplied initial guesses. No single documented switch guarantees a physically valid operating point for every cycle. A desired behavior would be an optional per-variable bound or residual guard that rejects iterations producing non-positive mass flow or pressure, so that convergence implies feasibility. Whether such a guard exists as a documented default or configurable option is release-sensitive and must be confirmed against the manual for the installed NPSS version.

Are per-variable bounds or feasibility guards documented in the current NPSS release for off-design cycle solvers?
If not, is the only mitigation to supply physically plausible initial guesses and post-check converged mass flows and pressures?
Which manual section or configuration block defines the exact stopping rule and any available physical-bound options for a specific NPSS version?

Tasadduq Burney · Answer

NPSS off-design cycle solvers converge by driving residual and target equations to zero. The documented stopping rule is a residual tolerance combined with an iteration limit, not a test for physical feasibility. There is no documented per-variable bound or global positive mass-flow guard in the core solver that rejects iterations producing non-positive mass flow or pressure. The solver can therefore report convergence at a mathematically valid but non-physical root.

Confirmed Facts vs. Likely Behavior
The residual-tolerance and iteration-limit stopping criteria are the documented convergence rules. The absence of a built-in feasibility guard is consistent with the solver architecture, but the exact availability of bound or limit parameters is release-sensitive and must be confirmed against the manual for your installed NPSS version.

Required Mitigation Steps

Supply physically plausible initial guesses close to the expected operating point to keep the iteration path in a feasible region.
Post-check converged mass flows, pressures, and temperatures after the solver call; flag any non-positive values as invalid.
If you maintain custom component models, you can add penalty terms to residual functions to discourage zero-crossing, but this is user logic rather than a global solver setting.

Where to Verify the Stopping Rule
Inspect the solver-specific configuration blocks that define the solve call for maximum iterations and absolute or relative tolerance thresholds. Consult the Solver Reference or Iterative Methods chapter of the NPSS User Manual for your installed version to confirm whether any variable constraint or boundary limit parameters exist.

Because option names and defaults vary by release, confirm your installed NPSS version if you need the exact manual section and parameter names.

NPSS cycle solver: residual convergence without a documented positive mass-flow guard

1 answer

Confirmed Facts vs. Likely Behavior

Required Mitigation Steps

Where to Verify the Stopping Rule

Your perspective belongs here.

0 question comments