Known limitations

Verification status

The calculations are covered by automated regression tests. The fluid-property calculator has been compared with independent reference data for water and steam, dry air, six light process fluids and C5 to C7 liquids; the findings are listed below. The hydraulic calculator has not been independently validated against a complete set of reference cases. Nothing here has been reviewed or certified by any standards body or licensed professional.

Hydraulic calculator

• Models a single sequential path. Branching or looped networks and transient behaviour are not solved.
• Two-phase pressure drop is not implemented. Liquid calculations assume single-phase liquid.
• Fluid properties are taken at the inlet and are not recalculated along the path.
• The connected isothermal gas model uses a fixed temperature and fixed property and compressibility assumptions.
• The adiabatic (Fanno) gas model covers horizontal, constant-bore, ideal-gas straight pipe only. Fittings, area changes, elevation, equipment and control valves are not supported in that model.
• The comparison between the Crane TP-410 Example 7-22 method and the Fanno model is unresolved; see the gas model audit. The site does not claim complete Crane compliance.
• Pipe catalog dimensions are nominal hydraulic geometry. They are not delivered-product measurements and do not establish a pressure rating. Some entries are still under review.

Fluid-property calculator

• Gas viscosity includes a density correction for pressure. Against reference data it is within 5% up to 5 MPa (725 psia). At 10 to 20 MPa (1450 to 2900 psia) some dense states are up to 11% off with Peng–Robinson and up to 24% off with SRK. The correction was developed for nonpolar gases.
• Dense fluid above its critical temperature has no sharp gas or liquid identity. The label shown for such a state is a convention and should not be relied on.
• Thermal conductivity has not been compared with reference data.
• Gas density from Peng–Robinson is within 1% of reference data for dry air up to 1 MPa (145 psia) and within 2% for six light process fluids up to 2 MPa (290 psia). Deviations grow with pressure and reach 3 to 5% at 20 MPa (2900 psia), and more above that.
• Liquid density is not taken from the cubic equation, which is 5 to 30% off for liquids. The method depends on the liquid and is shown with each result. This rule is provisional and will be retested as more models are added.
  • Hydrocarbon liquids, including those with dissolved nitrogen: the COSTALD correlation (Hankinson–Thomson, with a pressure correction). Within about 0.5% for six light process fluids and measured pentane, hexane and heptane liquids, and within about 1.5–1.8% with dissolved nitrogen. It applies up to 0.95 of the liquid's critical temperature; above that the uncorrected cubic value is shown with a warning and can be off by 10% or more. Hydrocarbons without fitted COSTALD parameters use estimated ones, which are less accurate and are flagged for review.
  • Water, alcohols, acids, ketones and other non-hydrocarbon liquids: each component's own data-fitted liquid-density correlation. Pure polar liquids are within 1% of Perry's handbook values. Mixtures add the component volumes (ideal mixing), which ignores the shrinkage on mixing: alcohol, acid and ketone solutions in water read up to about 4% low.
  • Any liquid containing carbon dioxide with other components: the uncorrected cubic value, with a warning. Against measured hydrocarbon + carbon dioxide liquids Peng–Robinson averages about 2% low (up to about 10%) and SRK about 10% low (up to about 17%), so use Peng–Robinson for these liquids.
• Water and steam reproduce the IAPWS-IF97 and IAPWS-95 verification tables, and vapor pressures of the six light fluids are within 2% of reference data.
• Every liquid value from Peng–Robinson and Soave–Redlich–Kwong is shown with its basis and a validation status: validated, approximate (with the error stated), not validated, or withheld. Liquid sound speed and the actual Cp/Cv are withheld: from the cubic equation the sound speed was 8 to 24% low at every light-hydrocarbon liquid point tested. Liquid heat capacity is about 3% off on average for light hydrocarbons (up to 6%, and more near the critical point) and has not been compared for other liquids. Liquid enthalpy, entropy, thermal conductivity and surface tension have not been compared with reference data; liquid viscosity has been compared only for six pure light fluids.
• GERG-2008 is available for mixtures made only of its 21 natural-gas components. It is CoolProp's implementation: GERG-2008 parameters for most component pairs, the later EOS-CG parameters for pairs among nitrogen, carbon dioxide, oxygen, argon, carbon monoxide and water, and reference equations for pure fluids, so it is close to but not identical with published GERG-2008. Its phase split is the app's own calculation.
  • Natural-gas and air densities are within 0.2% of measured and reference data up to 35 MPa (Peng–Robinson is off by up to 8%). Dry-air heat capacity and sound speed are within 0.3%.
  • Dense mixtures that are mostly carbon dioxide can be badly off: up to 24% for liquids near the carbon dioxide critical region and up to 8% for dense gas. The app warns for any mixture with at least 50% carbon dioxide at 7 MPa or more.
  • Nitrogen-rich methane/ethane liquids at 92–96 K came out 7 to 14% too dense. The app warns below 100 K.
  • Hydrocarbon liquids with carbon dioxide are mixed: GERG-2008 is better than Peng–Robinson for propane, decane and methane with carbon dioxide, and worse for heptane with carbon dioxide.
  • Viscosity and thermal conductivity are the same correlations as for the other models. One vapor and one liquid are modeled; a predicted liquid–liquid split is refused, and solids and hydrates are not modeled.
• Binary interaction parameters that are not available are set to zero, with a warning.
• Mixtures are solved for one vapor and one liquid phase. Liquid–liquid, solid and electrolyte equilibria are not modelled, and only the density of aqueous or strongly polar mixtures has been compared with reference data. Peng–Robinson and SRK can wrongly split such liquids into two phases; acetone + water does so at about half of the compositions tested.
• Two-phase transport properties are not averaged or estimated.
• The first calculation downloads a large scientific runtime and needs a network connection.

Methods and third-party software

Methods and review notes are recorded in the formula review and the gas model audit. Property calculations use the open-source CoolProp, Thermo, Chemicals and Fluids libraries under their own licenses.