Benchmark Reference Data
Every Standard and Comprehensive NeqSim task must compare its results against
independent reference data before the numbers are reported. In practice that
step gets rewritten in every benchmark notebook, with the reference values pasted
in as bare literals, no citation, an invented tolerance, and no check that the
reference is actually independent of the model being tested.
This skill supplies the parts that should not be rewritten: where the reference
came from, whether it is allowed to be used as a benchmark here, how far the
model may deviate before the comparison fails, and what the result block must
look like so the report generator and the CI gate accept it.
It does not compute NeqSim results. It provides the other side of the comparison
and the verdict.
When to Use
- A task is writing its
XX_benchmark_validation.ipynb notebook.
- A
benchmark_validation block must be produced for results.json.
- A property, critical constant, or saturation condition from a NeqSim fluid
needs an independent, citable reference value.
- A number quoted in a report needs a traceable provenance chain from value to
citation.
- An existing benchmark claim must be audited: is the reference independent, is
the deviation larger than the reference's own uncertainty, are there enough
points?
When Not to Use
- As a property engine. CoolProp and the anchor table are references, not the
model — do not use them to produce the answer the task is asked for.
- For mixtures outside a reference formulation's validated range. The
applicability gate will refuse, and it is right to refuse.
- As a substitute for measured data on the actual fluid. A reference EOS for a
pure component does not validate a characterised reservoir fluid; it validates
the pure-component limit of the model.
- To manufacture a PASS. Loosening the tolerance until a comparison passes is
the failure mode this skill exists to make visible.
Inputs
| Input |
Meaning |
model_value |
the number the task computed (NeqSim, or any model under test) |
reference |
a ReferencePoint from the anchor table or the CoolProp backend |
tolerance_pct |
acceptance band in percent; defaults per property when omitted |
model_tier |
authority tier of the model basis, default "correlation" |
model_label |
label used in the output block, default "neqsim" |
informational |
mark a comparison as context only, never graded |
ReferencePoint carries fluid, property_name, value, unit, state,
source_key, and the citation reached through the source registry.
Outputs
| Output |
Meaning |
BenchmarkResult.status |
PASS, WARN, FAIL, or INFO |
deviation, deviation_pct |
signed model-minus-reference difference |
within_source_uncertainty |
whether the deviation is smaller than the reference's own stated uncertainty |
independent |
whether the reference outranks the model basis |
citation |
full reference citation for the report's reference list |
BenchmarkReport.overall_status |
roll-up: FAIL beats WARN beats PASS |
BenchmarkReport.blockers() |
why the benchmark may not be presented as validated |
to_results_json() |
the benchmark_validation block for results.json |
to_markdown() |
the table for the notebook cell and the report body |
Engineering Method
1. Authority tiers. Sources are ranked
primary_standard > reference_eos > measured_data > published_case >
correlation. A benchmark reference must rank strictly above the model basis
it tests. Comparing SRK against SRK, or GERG-2008 against a CoolProp HEOS
evaluation of the same formulation, is a consistency check, not a benchmark — it
is graded INFO, never PASS, and appears in blockers().
2. Applicability gate. Each source declares the fluids and the temperature
and pressure ranges over which it is validated (for example Span-Wagner CO2 from
the triple point 216.592 K to 1100 K and up to 800 MPa). sources_for(...)
returns only sources valid at the requested state, so a reference is never quoted
outside its published range.
3. Deviation and grading. Deviation is
$\delta = x_\text{model} - x_\text{ref}$ and
$\delta_% = 100,\delta / |x_\text{ref}|$. With tolerance $\tau$:
$$
\text{status} =
\begin{cases}
\text{PASS} & |\delta_%| \le \tau \
\text{WARN} & \tau < |\delta_%| \le 2\tau \
\text{FAIL} & |\delta_%| > 2\tau
\end{cases}
$$
A zero reference value cannot be graded relatively and returns INFO.
4. Resolution check. Agreement cannot be claimed finer than the reference's
own uncertainty. within_source_uncertainty records whether
$|\delta_%|$ is below the source's stated uncertainty for that property, so a
report does not claim 0.01 % agreement against a source good to 0.5 %.
5. Three-point minimum. The task rules require at least three independent
comparison points. meets_minimum_points() counts only graded results, so
padding a thin benchmark with INFO rows does not satisfy the gate.
6. Offline anchors. The anchor table holds published constants — critical
points, triple points, normal boiling points, the CO2 sublimation point, and
water density at 25 °C and 1 atm — taken directly from the cited reference
formulations. It exists so a benchmark is still possible with no optional
dependency and no network. It is a smoke-test set; a real study extends it with
CoolProp, lab data, or a published case for the fluid actually modelled.
Python Usage Pattern
Offline, no optional dependency:
from benchmark_reference_data import BenchmarkReport, compare, find_anchor
report = BenchmarkReport(description="SRK pure-component anchors")
report.add(compare(
"CO2 critical temperature",
fluid.getPhase(0).getComponent("CO2").getTC(), # NeqSim value
find_anchor("co2", "critical_temperature"),
))
report.add(compare(
"Methane critical temperature",
fluid.getPhase(0).getComponent("methane").getTC(),
find_anchor("methane", "critical_temperature"),
))
report.add(compare(
"Water density at 25 C, 1 atm",
water_density_kg_m3,
find_anchor("water", "density", temperature_K=298.15, pressure_Pa=101325.0),
))
print(report.to_markdown())
results["benchmark_validation"] = report.to_results_json()
With the optional CoolProp backend, at states the task actually operates at:
from benchmark_reference_data import BenchmarkReport, compare
from benchmark_reference_data.coolprop_backend import is_available, reference_grid
if is_available():
references = reference_grid("co2", "density", (280.0, 300.0, 320.0), (5.0e6, 10.0e6))
report = BenchmarkReport(description="CO2 density vs CoolProp HEOS")
for ref in references:
state = ref.state
report.add(compare(
"CO2 density at {:g} K, {:g} bar".format(
state["temperature_K"], state["pressure_Pa"] / 1.0e5),
neqsim_density(state["temperature_K"], state["pressure_Pa"]),
ref,
))
Auditing an existing benchmark before trusting it:
if not report.meets_minimum_points():
raise AssertionError(report.blockers())
for issue in report.blockers():
print("BLOCKER:", issue)
results["references"] = [{"id": "ref", "text": c} for c in report.citations()]
Validation Checklist
Common Mistakes
Benchmarking a model against itself. A CoolProp HEOS density is not an
independent check on a NeqSim GERG-2008 density — both evaluate the same
formulation. Set model_tier honestly and let the gate say so.
Pasting reference values without a citation. A number with no source
cannot be defended in review; use the registry so the citation travels with
the value.
Tuning the tolerance to the result. Set the tolerance from the acceptance
criteria or the source uncertainty before running the comparison.
Claiming agreement finer than the reference. Check
within_source_uncertainty.
Padding to three points with INFO rows. Only graded results count.
Quoting a reference outside its range. Span-Wagner CO2 does not extend
below the triple point at 216.592 K; the applicability gate refuses for a
reason.
Validating a characterised reservoir fluid with pure-component anchors.
That validates the pure-component limit only; say so in the report.
Checking a lab water PVT (PVTW) against a cubic/CPA water. NeqSim CPA
pure water at 139.8 °C and 542 bara gives 978 kg/m3 against 953 kg/m3 from
IAPWS-95 (CoolProp), so Bw comes out 1.020 instead of 1.048 — a 2.7 %
bias that would wrongly flag a correct lab Bw. Use IAPWS-95 as the
reference for water Bw and compressibility; CPA viscosity (0.220 cP against
0.210 cP) is closer. Expect a brine Bw slightly below the pure-water
value and a brine viscosity a few percent above it.
Grading a tubing/flowline outflow (VLP) against measured well pressures with a tight tolerance and unknown
geometry. Gauge depth, deviation, tubing ID and gas-lift injection depth are usually unknown, so a 15 % band
just reports the geometry guess as FAIL. Set the tolerance to the engineering scatter of an untuned empirical
correlation (about 30 %) before running, record the signed bias per well (all wells over-predicting dP
means the deliverability is conservative), and carry that bias into the report as a stated direction of error.
Limitations
- The anchor table is intentionally small and pure-component only. It is a smoke
test, not a validation dataset.
- CoolProp is optional and not vendored; without it, reference values are limited
to the anchor table.
- No mixture reference data is bundled. GERG-2008 is registered as a source, but
values must come from CoolProp, a publication, or a lab report.
- Tolerances are screening defaults derived from typical reference-EOS
uncertainties. They are not project acceptance criteria.
- The skill grades a comparison; it does not decide whether the property being
compared is the one that matters for the engineering decision.
Related NeqSim Functionality
neqsim.thermodynamicoperations.ThermodynamicOperations#TPflash() followed by
SystemInterface#initProperties() produces the model values compared here —
transport properties are zero without initProperties().
neqsim.thermo.system.SystemGERG2008Eos and
neqsim.thermo.util.gerg.NeqSimGERG2008 give NeqSim's own GERG-2008
evaluation. Because it is the same formulation as the gerg2008 source, use
model_tier="reference_eos" so the comparison is correctly graded INFO.
neqsim.thermo.util.steam.Iapws_if97 and neqsim.thermo.phase.PhaseWaterIAPWS
are NeqSim's IAPWS implementations, with the same independence caveat against
the iapws95 / iapws_if97 sources.
neqsim.util.agentic.TaskResultValidator validates the emitted
benchmark_validation block in Java; devtools/validate_task_results.py is
the equivalent CI gate.
neqsim.thermodynamicoperations.ThermodynamicOperations#calcPTphaseEnvelope
supplies the cricondenbar and cricondentherm that a lab CME/CVD or GERG-2008
reference is compared against.
Related Skills
neqsim-fluid-quality-check — run before benchmarking; a composition that
fails the quality check makes any benchmark meaningless.
neqsim-pvt-regression-characterization-factor,
neqsim-pseudocomponent-split-characterization — supply the characterised
fluid whose pure-component limits this skill checks.
neqsim-e300-fluid-io — supplies the fluid basis a benchmark is run on.
neqsim-cfd-coupling, neqsim-fem-coupling — same gate-before-quoting
discipline applied to CFD and FEM studies rather than property values.
References
- Wagner, W., & Pruss, A. (2002). The IAPWS formulation 1995 for the
thermodynamic properties of ordinary water substance for general and scientific
use. Journal of Physical and Chemical Reference Data, 31(2), 387-535.
- IAPWS R7-97(2012). Revised Release on the IAPWS Industrial Formulation 1997
for the Thermodynamic Properties of Water and Steam.
- Span, R., & Wagner, W. (1996). A new equation of state for carbon dioxide
covering the fluid region from the triple-point temperature to 1100 K at
pressures up to 800 MPa. Journal of Physical and Chemical Reference Data,
25(6), 1509-1596.
- Setzmann, U., & Wagner, W. (1991). A new equation of state and tables of
thermodynamic properties for methane covering the range from the melting line
to 625 K at pressures up to 100 MPa. Journal of Physical and Chemical
Reference Data, 20(6), 1061-1155.
- Span, R., Lemmon, E. W., Jacobsen, R. T., Wagner, W., & Yokozeki, A. (2000). A
reference equation of state for the thermodynamic properties of nitrogen.
Journal of Physical and Chemical Reference Data, 29(6), 1361-1433.
- Bucker, D., & Wagner, W. (2006). A reference equation of state for the
thermodynamic properties of ethane. Journal of Physical and Chemical Reference
Data, 35(1), 205-266.
- Lemmon, E. W., McLinden, M. O., & Wagner, W. (2009). Thermodynamic properties
of propane. III. A reference equation of state. Journal of Chemical and
Engineering Data, 54(12), 3141-3180.
- Kunz, O., & Wagner, W. (2012). The GERG-2008 wide-range equation of state for
natural gases and other mixtures. Journal of Chemical and Engineering Data,
57(11), 3032-3091. Adopted as ISO 20765-2 and ISO 20765-3.
- Bell, I. H., Wronski, J., Quoilin, S., & Lemort, V. (2014). Pure and pseudo-pure
fluid thermophysical property evaluation and the open-source thermophysical
property library CoolProp. Industrial & Engineering Chemistry Research,
53(6), 2498-2508.
- Linstrom, P. J., & Mallard, W. G. (Eds.). NIST Chemistry WebBook, NIST
Standard Reference Database Number 69. National Institute of Standards and
Technology, Gaithersburg MD.
- ASME V&V 20-2009, Standard for Verification and Validation in Computational
Fluid Dynamics and Heat Transfer — comparison error and validation
uncertainty, the basis for the resolution check.
1---2name: neqsim-benchmark-reference-data-23calculation_basis: "data-retrieval"4version: "0.1.1"5description: "Reproducible, source-traceable benchmark validation for NeqSim tasks: a registry of reference sources with authority tiers and uncertainties (IAPWS, Span-Wagner, GERG-2008, CoolProp, NIST), an offline anchor table, PASS/WARN/FAIL grading with the three-point minimum, and the benchmark_validation block the report and CI gate consume. USE WHEN: validating NeqSim against independent data, writing a benchmark notebook or block, tracing a deviation to a citable source, or checking a benchmark claim."6last_verified: "2026-09-25"7requires:8 python_packages: []9 java_packages: []10 env: []11 network: []12metadata:13 source: https://skillmd.com/skills/equinor/neqsim-benchmark-reference-data-214---1516# Benchmark Reference Data1718Every Standard and Comprehensive NeqSim task must compare its results against19independent reference data before the numbers are reported. In practice that20step gets rewritten in every benchmark notebook, with the reference values pasted21in as bare literals, no citation, an invented tolerance, and no check that the22reference is actually independent of the model being tested.2324This skill supplies the parts that should not be rewritten: *where the reference25came from*, *whether it is allowed to be used as a benchmark here*, *how far the26model may deviate before the comparison fails*, and *what the result block must27look like* so the report generator and the CI gate accept it.2829It does not compute NeqSim results. It provides the other side of the comparison30and the verdict.3132## When to Use3334- A task is writing its `XX_benchmark_validation.ipynb` notebook.35- A `benchmark_validation` block must be produced for `results.json`.36- A property, critical constant, or saturation condition from a NeqSim fluid37 needs an independent, citable reference value.38- A number quoted in a report needs a traceable provenance chain from value to39 citation.40- An existing benchmark claim must be audited: is the reference independent, is41 the deviation larger than the reference's own uncertainty, are there enough42 points?4344## When *Not* to Use4546- As a property engine. CoolProp and the anchor table are references, not the47 model — do not use them to produce the answer the task is asked for.48- For mixtures outside a reference formulation's validated range. The49 applicability gate will refuse, and it is right to refuse.50- As a substitute for measured data on the actual fluid. A reference EOS for a51 pure component does not validate a characterised reservoir fluid; it validates52 the pure-component limit of the model.53- To manufacture a PASS. Loosening the tolerance until a comparison passes is54 the failure mode this skill exists to make visible.5556## Inputs5758| Input | Meaning |59| --- | --- |60| `model_value` | the number the task computed (NeqSim, or any model under test) |61| `reference` | a `ReferencePoint` from the anchor table or the CoolProp backend |62| `tolerance_pct` | acceptance band in percent; defaults per property when omitted |63| `model_tier` | authority tier of the model basis, default `"correlation"` |64| `model_label` | label used in the output block, default `"neqsim"` |65| `informational` | mark a comparison as context only, never graded |6667`ReferencePoint` carries `fluid`, `property_name`, `value`, `unit`, `state`,68`source_key`, and the citation reached through the source registry.6970## Outputs7172| Output | Meaning |73| --- | --- |74| `BenchmarkResult.status` | `PASS`, `WARN`, `FAIL`, or `INFO` |75| `deviation`, `deviation_pct` | signed model-minus-reference difference |76| `within_source_uncertainty` | whether the deviation is smaller than the reference's own stated uncertainty |77| `independent` | whether the reference outranks the model basis |78| `citation` | full reference citation for the report's reference list |79| `BenchmarkReport.overall_status` | roll-up: FAIL beats WARN beats PASS |80| `BenchmarkReport.blockers()` | why the benchmark may not be presented as validated |81| `to_results_json()` | the `benchmark_validation` block for `results.json` |82| `to_markdown()` | the table for the notebook cell and the report body |8384## Engineering Method8586**1. Authority tiers.** Sources are ranked87`primary_standard` > `reference_eos` > `measured_data` > `published_case` >88`correlation`. A benchmark reference must rank *strictly above* the model basis89it tests. Comparing SRK against SRK, or GERG-2008 against a CoolProp HEOS90evaluation of the same formulation, is a consistency check, not a benchmark — it91is graded `INFO`, never `PASS`, and appears in `blockers()`.9293**2. Applicability gate.** Each source declares the fluids and the temperature94and pressure ranges over which it is validated (for example Span-Wagner CO2 from95the triple point 216.592 K to 1100 K and up to 800 MPa). `sources_for(...)`96returns only sources valid at the requested state, so a reference is never quoted97outside its published range.9899**3. Deviation and grading.** Deviation is100$\delta = x_\text{model} - x_\text{ref}$ and101$\delta_\% = 100\,\delta / |x_\text{ref}|$. With tolerance $\tau$:102103$$104\text{status} =105\begin{cases}106\text{PASS} & |\delta_\%| \le \tau \\107\text{WARN} & \tau < |\delta_\%| \le 2\tau \\108\text{FAIL} & |\delta_\%| > 2\tau109\end{cases}110$$111112A zero reference value cannot be graded relatively and returns `INFO`.113114**4. Resolution check.** Agreement cannot be claimed finer than the reference's115own uncertainty. `within_source_uncertainty` records whether116$|\delta_\%|$ is below the source's stated uncertainty for that property, so a117report does not claim 0.01 % agreement against a source good to 0.5 %.118119**5. Three-point minimum.** The task rules require at least three independent120comparison points. `meets_minimum_points()` counts only *graded* results, so121padding a thin benchmark with `INFO` rows does not satisfy the gate.122123**6. Offline anchors.** The anchor table holds published constants — critical124points, triple points, normal boiling points, the CO2 sublimation point, and125water density at 25 °C and 1 atm — taken directly from the cited reference126formulations. It exists so a benchmark is still possible with no optional127dependency and no network. It is a smoke-test set; a real study extends it with128CoolProp, lab data, or a published case for the fluid actually modelled.129130## Python Usage Pattern131132Offline, no optional dependency:133134```python135from benchmark_reference_data import BenchmarkReport, compare, find_anchor136137report = BenchmarkReport(description="SRK pure-component anchors")138139report.add(compare(140 "CO2 critical temperature",141 fluid.getPhase(0).getComponent("CO2").getTC(), # NeqSim value142 find_anchor("co2", "critical_temperature"),143))144report.add(compare(145 "Methane critical temperature",146 fluid.getPhase(0).getComponent("methane").getTC(),147 find_anchor("methane", "critical_temperature"),148))149report.add(compare(150 "Water density at 25 C, 1 atm",151 water_density_kg_m3,152 find_anchor("water", "density", temperature_K=298.15, pressure_Pa=101325.0),153))154155print(report.to_markdown())156results["benchmark_validation"] = report.to_results_json()157```158159With the optional CoolProp backend, at states the task actually operates at:160161```python162from benchmark_reference_data import BenchmarkReport, compare163from benchmark_reference_data.coolprop_backend import is_available, reference_grid164165if is_available():166 references = reference_grid("co2", "density", (280.0, 300.0, 320.0), (5.0e6, 10.0e6))167 report = BenchmarkReport(description="CO2 density vs CoolProp HEOS")168 for ref in references:169 state = ref.state170 report.add(compare(171 "CO2 density at {:g} K, {:g} bar".format(172 state["temperature_K"], state["pressure_Pa"] / 1.0e5),173 neqsim_density(state["temperature_K"], state["pressure_Pa"]),174 ref,175 ))176```177178Auditing an existing benchmark before trusting it:179180```python181if not report.meets_minimum_points():182 raise AssertionError(report.blockers())183for issue in report.blockers():184 print("BLOCKER:", issue)185results["references"] = [{"id": "ref", "text": c} for c in report.citations()]186```187188## Validation Checklist189190- [ ] At least three **graded** comparisons (`meets_minimum_points()` is true).191- [ ] `blockers()` is empty, or every entry is explained in the report.192- [ ] Every reference outranks the model basis (`independent` is true).193- [ ] Every reference state lies inside the source's validated range.194- [ ] Tolerances are the project's acceptance criteria, not the screening195 defaults, whenever acceptance criteria exist.196- [ ] `within_source_uncertainty` is checked before claiming tight agreement.197- [ ] `report.citations()` is copied into `results["references"]`.198- [ ] The emitted block passes `python devtools/validate_task_results.py <task>`.199- [ ] The benchmark notebook includes a parity or deviation plot alongside the200 table.201202## Common Mistakes203204- **Benchmarking a model against itself.** A CoolProp HEOS density is not an205 independent check on a NeqSim GERG-2008 density — both evaluate the same206 formulation. Set `model_tier` honestly and let the gate say so.207- **Pasting reference values without a citation.** A number with no source208 cannot be defended in review; use the registry so the citation travels with209 the value.210- **Tuning the tolerance to the result.** Set the tolerance from the acceptance211 criteria or the source uncertainty *before* running the comparison.212- **Claiming agreement finer than the reference.** Check213 `within_source_uncertainty`.214- **Padding to three points with `INFO` rows.** Only graded results count.215- **Quoting a reference outside its range.** Span-Wagner CO2 does not extend216 below the triple point at 216.592 K; the applicability gate refuses for a217 reason.218- **Validating a characterised reservoir fluid with pure-component anchors.**219 That validates the pure-component limit only; say so in the report.220- **Checking a lab water PVT (`PVTW`) against a cubic/CPA water.** NeqSim CPA221 pure water at 139.8 °C and 542 bara gives 978 kg/m3 against 953 kg/m3 from222 IAPWS-95 (CoolProp), so `Bw` comes out 1.020 instead of 1.048 — a 2.7 %223 bias that would wrongly flag a correct lab `Bw`. Use IAPWS-95 as the224 reference for water `Bw` and compressibility; CPA viscosity (0.220 cP against225 0.210 cP) is closer. Expect a brine `Bw` slightly **below** the pure-water226 value and a brine viscosity a few percent **above** it.227228- **Grading a tubing/flowline outflow (VLP) against measured well pressures with a tight tolerance and unknown229 geometry.** Gauge depth, deviation, tubing ID and gas-lift injection depth are usually unknown, so a 15 % band230 just reports the geometry guess as FAIL. Set the tolerance to the engineering scatter of an untuned empirical231 correlation (about 30 %) **before** running, record the signed bias per well (all wells over-predicting dP232 means the deliverability is conservative), and carry that bias into the report as a stated direction of error.233234## Limitations235236- The anchor table is intentionally small and pure-component only. It is a smoke237 test, not a validation dataset.238- CoolProp is optional and not vendored; without it, reference values are limited239 to the anchor table.240- No mixture reference data is bundled. GERG-2008 is registered as a source, but241 values must come from CoolProp, a publication, or a lab report.242- Tolerances are screening defaults derived from typical reference-EOS243 uncertainties. They are not project acceptance criteria.244- The skill grades a comparison; it does not decide whether the property being245 compared is the one that matters for the engineering decision.246247## Related NeqSim Functionality248249- `neqsim.thermodynamicoperations.ThermodynamicOperations#TPflash()` followed by250 `SystemInterface#initProperties()` produces the model values compared here —251 transport properties are zero without `initProperties()`.252- `neqsim.thermo.system.SystemGERG2008Eos` and253 `neqsim.thermo.util.gerg.NeqSimGERG2008` give NeqSim's own GERG-2008254 evaluation. Because it is the *same formulation* as the `gerg2008` source, use255 `model_tier="reference_eos"` so the comparison is correctly graded `INFO`.256- `neqsim.thermo.util.steam.Iapws_if97` and `neqsim.thermo.phase.PhaseWaterIAPWS`257 are NeqSim's IAPWS implementations, with the same independence caveat against258 the `iapws95` / `iapws_if97` sources.259- `neqsim.util.agentic.TaskResultValidator` validates the emitted260 `benchmark_validation` block in Java; `devtools/validate_task_results.py` is261 the equivalent CI gate.262- `neqsim.thermodynamicoperations.ThermodynamicOperations#calcPTphaseEnvelope`263 supplies the cricondenbar and cricondentherm that a lab CME/CVD or GERG-2008264 reference is compared against.265266## Related Skills267268- `neqsim-fluid-quality-check` — run before benchmarking; a composition that269 fails the quality check makes any benchmark meaningless.270- `neqsim-pvt-regression-characterization-factor`,271 `neqsim-pseudocomponent-split-characterization` — supply the characterised272 fluid whose pure-component limits this skill checks.273- `neqsim-e300-fluid-io` — supplies the fluid basis a benchmark is run on.274- `neqsim-cfd-coupling`, `neqsim-fem-coupling` — same gate-before-quoting275 discipline applied to CFD and FEM studies rather than property values.276277## References278279- Wagner, W., & Pruss, A. (2002). The IAPWS formulation 1995 for the280 thermodynamic properties of ordinary water substance for general and scientific281 use. *Journal of Physical and Chemical Reference Data*, 31(2), 387-535.282- IAPWS R7-97(2012). *Revised Release on the IAPWS Industrial Formulation 1997283 for the Thermodynamic Properties of Water and Steam.*284- Span, R., & Wagner, W. (1996). A new equation of state for carbon dioxide285 covering the fluid region from the triple-point temperature to 1100 K at286 pressures up to 800 MPa. *Journal of Physical and Chemical Reference Data*,287 25(6), 1509-1596.288- Setzmann, U., & Wagner, W. (1991). A new equation of state and tables of289 thermodynamic properties for methane covering the range from the melting line290 to 625 K at pressures up to 100 MPa. *Journal of Physical and Chemical291 Reference Data*, 20(6), 1061-1155.292- Span, R., Lemmon, E. W., Jacobsen, R. T., Wagner, W., & Yokozeki, A. (2000). A293 reference equation of state for the thermodynamic properties of nitrogen.294 *Journal of Physical and Chemical Reference Data*, 29(6), 1361-1433.295- Bucker, D., & Wagner, W. (2006). A reference equation of state for the296 thermodynamic properties of ethane. *Journal of Physical and Chemical Reference297 Data*, 35(1), 205-266.298- Lemmon, E. W., McLinden, M. O., & Wagner, W. (2009). Thermodynamic properties299 of propane. III. A reference equation of state. *Journal of Chemical and300 Engineering Data*, 54(12), 3141-3180.301- Kunz, O., & Wagner, W. (2012). The GERG-2008 wide-range equation of state for302 natural gases and other mixtures. *Journal of Chemical and Engineering Data*,303 57(11), 3032-3091. Adopted as ISO 20765-2 and ISO 20765-3.304- Bell, I. H., Wronski, J., Quoilin, S., & Lemort, V. (2014). Pure and pseudo-pure305 fluid thermophysical property evaluation and the open-source thermophysical306 property library CoolProp. *Industrial & Engineering Chemistry Research*,307 53(6), 2498-2508.308- Linstrom, P. J., & Mallard, W. G. (Eds.). *NIST Chemistry WebBook, NIST309 Standard Reference Database Number 69*. National Institute of Standards and310 Technology, Gaithersburg MD.311- ASME V&V 20-2009, *Standard for Verification and Validation in Computational312 Fluid Dynamics and Heat Transfer* — comparison error and validation313 uncertainty, the basis for the resolution check.