Code-to-Code Benchmarks of OpenFLUME
Achieved agreement with published GFSSP and SINDA/FLUINT results, and with the analytic references used alongside them
Generated by scripts/benchmarks-report.ts. All model numbers come from live solves of the current solver. The CI gates are in src/ui/tests/gfssp-benchmarks.test.ts (A–C), leeMartin.test.ts (D), lh2StorageTank.test.ts (E) and nureth.test.ts (F); those tests log the same achieved values.
Abstract
User manual §8.3 lists six benchmarks against other codes. A gate is a pass band that a CI test enforces, not a measured agreement, so this report states for every gated quantity the reference it is compared with, the type of that reference, the error the current solver achieves, and the gate. The types matter: Benchmark A is gated against an analytic solution and is analytic verification, not code-to-code. Benchmark B's duty is gated against ε-NTU; only its outlet temperatures are published GFSSP output. Benchmark C compares with GFSSP's published output, Benchmark D with a GFSSP figure and an independent Runge–Kutta integration of the same lumped model, and Benchmark E with SINDA/FLUINT's published sample-problem output. Benchmark E is reproduced only through an emulation built from registers, logic rules, and boundary-node controllers (user manual §3.8), not by the core formulation. Benchmark F reproduces published trends only and carries no error figure.
Reference types
| Type | Meaning |
|---|---|
| GFSSP output | A value GFSSP published for the same problem; agreement inherits GFSSP's assumptions |
| SINDA/FLUINT output | A value from the SINDA/FLUINT sample-problem deck |
| Analytic | A closed-form solution of the modeled equations (verification, not code-to-code) |
| ε-NTU | The effectiveness–NTU closed form for the modeled exchanger (analytic) |
| Independent RK4 | A separate Runge–Kutta integration of the same lumped equations |
| Conservation | An internal balance; checks the solve, not the agreement |
Results
| Benchmark | Quantity | Reference type | Reference value | OpenFLUME (achieved) | Gate |
|---|---|---|---|---|---|
| A. GFSSP Example 13, conduction rod with convection | Solid temperatures, 5 nodes | Analytic (fin solution) | T(x) of the Example 13 fin equation | largest deviation 2.59 K (+0.67 K, +0.15 K, −0.23 K, −0.91 K, −2.59 K) | 3.0 K |
| A. GFSSP Example 13, conduction rod with convection | End conduction heat flows | Analytic (fin solution) | −kA dT/dx at the ends | cold end −4.44 %, hot end −5.64 % | 7 % |
| A. GFSSP Example 13, conduction rod with convection | Global energy balance | Conservation | Q_hot = Q_cold + Q_conv | 0.31 % | 0.5 % |
| B. N₂–N₂ counterflow heat exchanger | Duty | Analytic (ε-NTU) | 5.34 kW | 5.30 kW (−0.84 %) | 5 % |
| B. N₂–N₂ counterflow heat exchanger | Hot / cold outlet temperature | GFSSP output | 390.3 K / 298.2 K | 389.92 K (−0.38 K) / 298.53 K (+0.33 K) | 2.5 K |
| B. N₂–N₂ counterflow heat exchanger | Hot/cold duty balance | Conservation | Q_hot = Q_cold | 0.000 % | 1 % |
| C. GFSSP Example 5, water–water heat exchanger | Hot outlet temperature | GFSSP output | 295.67 K | 296.11 K (+0.44 K) | 1.0 K |
| C. GFSSP Example 5, water–water heat exchanger | Cold outlet temperature | GFSSP output | 290.94 K | 291.13 K (+0.19 K) | 0.5 K |
| C. GFSSP Example 5, water–water heat exchanger | Duty | GFSSP output | 25.6 kW | 24.88 kW (−2.80 %) | 3 % |
| D. Lee & Martin entrapped air | First pressure peak (Δt = 0.01 s) | GFSSP output (figure) | ≈ 1.90 MPa | 1.690 MPa (−11.1 %) at t = 0.70 s | ±15 %; peak at 0.6–0.9 s |
| D. Lee & Martin entrapped air | Oscillation period (Δt = 0.01 s) | GFSSP output (figure) | ≈ 0.55–0.60 s | 0.50–0.51 s | 0.5–0.65 s |
| D. Lee & Martin entrapped air | First peak, its time, first period (Δt = 0.003 s) | Independent RK4 (same lumped ODE) | RK4, h = 1 ms | −6.1 %, +0.1 %, +0.0 % | 8 % each |
| E. SINDA/FLUINT Sample F, LH₂ no-vent fill (emulated with logic rules) | Fill fraction at 15 min | SINDA/FLUINT output | 0.739 | 0.733 (−0.8 %) | 5 % |
| E. SINDA/FLUINT Sample F, LH₂ no-vent fill (emulated with logic rules) | Tank pressure at 15 min | SINDA/FLUINT output | 59.72 psia | 59.52 psia (−0.3 %) | 0.5 % |
| E. SINDA/FLUINT Sample F, LH₂ no-vent fill (emulated with logic rules) | Vapor temperature at 15 min | SINDA/FLUINT output | 47.39 R | 47.79 R (+0.8 %) | 5 % |
| E. SINDA/FLUINT Sample F, LH₂ no-vent fill (emulated with logic rules) | Liquid temperature at 15 min | SINDA/FLUINT output | 38.10 R | 38.88 R (+2.1 %) | 5 % |
| E. SINDA/FLUINT Sample F, LH₂ no-vent fill (emulated with logic rules) | TVS vent flow at 15 min | SINDA/FLUINT output | 0.0522 lbm/hr | 0.0512 lbm/hr (−1.9 %) | 5 % |
| F. NURETH-16 gas-cooled reactor loop | Natural-circulation trends | Published trends | — | Trends reproduced (see nureth.test.ts); geometry is representative | Qualitative |
Notes:
- A is analytic verification: GFSSP Example 13's own documentation compares with the same fin solution. It is listed here because the problem is a GFSSP example.
- B: the loop pressure is raised to 2.0 MPa from the published schematic (whose flow rates at 1 atm give about 500 m/s); see the example's source comment.
- D: the default polytropic index of the gas cushion, 1.03, was chosen to recover the published first-peak amplitude and period at the default step (
buildLeeMartininsrc/ui/examples.ts), so the GFSSP-figure rows are partly calibrated. The RK4 rows are not. - E is emulated: FLUINT's twinned tanks and moveable ties are reproduced with registers, logic rules, and boundary-node controllers (
src/ui/lh2StorageTank.tslists what is emulated and what is not). All five deck quantities are shown. - F has no quantitative reference: the paper's geometry tables are not available, so the model is representative and only the published trends are checked.
References
- Majumdar, A. K., LeClair, A. C., and Moore, R., Generalized Fluid System Simulation Program, Version 6.0, NASA/TM-2013-217492, §6.13 (Example 13, which is Example 11, §5.11 of the Version 5 Educational manual, NASA/TM-2011-216470); and Majumdar, TFAWS 2004.
- Majumdar, A. K., and LeClair, A. C., JANNAF 2024 conjugate heat-exchanger paper, as cited in user manual §8.3 (Benchmark B).
- GFSSP Version 5 user manual, Example 5 (water–water heat exchanger), via Patel 2011, as cited in
src/ui/tests/gfssp-benchmarks.test.ts. - Lee & Martin entrapped-air problem: GFSSP Figure 10 as cited in user manual §8.3; the example source cites Majumdar et al., AIAA 2015-3850, Fig. 11.
- C&R Technologies, SINDA/FLUINT Sample Problems, Sample Problem F ("LH2 storage tank with internal HX and vapor cooled shield"), modeling Anderson, J. E., et al., "Evaluation of Long-term Cryogenic Storage System," Cryogenic Engineering Conference, 1989.
- Williams, W. C., and McLean, J., "Using NASA's GFSSP Code for Steady State and Transient Modeling of Gas Cooled Reactor Passive Safety Systems," NURETH-16, Chicago, 2015, paper 13066.