Skip to content

Petroleum assays and pseudocomponents

process/assays.yaml holds sourced assays of petroleum fractions (naphtha and heavier). cheme process assay characterises each one into pseudocomponents with named correlations, refuses an assay that is insufficient or would need extrapolation, and flashes the pseudocomponents with Peng-Robinson through thermo inside a benchmarked claim. The code is in domains/process/thermodynamics/assay.py; the decision is ADR 0026. The pure-component routes (cheme process phase, cheme process simulate) do not take pseudocomponents.

The file

schema: cheme/assay/v1
kind: assays
assays:
  - id: HN-1
    title: heavy naphtha, unit 2 rundown
    source: "lab report LR-2026-114, ASTM D2892 TBP and D4052 density"
    temperature_unit: °C            # K, °C or °F, of the curve or the cuts
    distillation:
      method: TBP                   # or D86: exactly 0, 10, 30, 50, 70, 90, 100 %
      points: [[0, 70], [10, 85], [30, 100], [50, 115], [70, 130], [90, 150], [100, 175]]
    specific_gravity: 0.745         # or api_gravity, density_15C (kg/m3), or watson_k
    molar_mass: 108                 # measured, optional: a check
    pseudocomponents: 8             # cuts of equal liquid volume (default 8)
  - id: SUR-1
    source: "assay table 3 of ..."
    temperature_unit: K
    cuts:                           # instead of a curve: each cut with its own gravity
      - {name: C7 cut, liquid_volume_percent: 50, boiling_point: 371.5, specific_gravity: 0.688}
      - {name: C8 cut, liquid_volume_percent: 50, boiling_point: 398.8, specific_gravity: 0.707}

cheme schema process-assays prints the JSON schema.

Characterisation

Every step is deterministic and named in assay.CORRELATIONS, and every pseudocomponent property records which one gave it (provenance):

Step Method
D86 to TBP API Technical Data Book procedure 3A1.1 (Daubert 1994), as in Riazi (2005) ASTM MNL50
cuts equal liquid volumes of the TBP curve, linear between the points; a cut's normal boiling point is the curve's mean over its volume
cut gravities one Watson K per curve, from the fraction's specific gravity and the volume-weighted cube roots of the cut boiling points (the cut gravities reproduce the fraction's exactly), or the assay's watson_k
molar mass, Tc, Pc, C/H weight ratio Riazi and Daubert (1987), Ind. Eng. Chem. Res. 26, 755 (Tb-SG coefficients)
acentric factor Lee-Kesler (chemicals.acentric.LK_omega)
ideal-gas heat capacity Lastovka and Shaw (2013), through thermo, from the similarity variable of the C/H composition
mass and mole fractions from the cut volumes, gravities and molar masses

The correlations were checked against published worked examples: Riazi (2005) Example 3.3 and the API 3A1.1 example for the D86 conversion, and Ahmed's Example 2.2 for Riazi-Daubert. On the n-alkanes C6 to C10, Riazi-Daubert gives Tc within 1 %, Pc within 4 %, molar mass within 4 % and the C/H ratio within 2 % of the measured values (tests/test_assay.py).

Refusals

Rule When
cheme.assay-insufficient neither or both of curve and cuts; a curve not from 0 to 100 %, with fewer than 5 points (D86: not the 7 standard points), not increasing; no density (and no Watson K) or more than one; cuts that do not add up to 100 %, or with a fraction density
cheme.assay-extrapolation a pseudocomponent with a normal boiling point outside 80-650 °F, a specific gravity outside 0.63-0.97, a molar mass outside 70-300 g/mol, or a reduced boiling point above 0.8 (assay.RANGES)
cheme.assay-inconsistent warning: the pseudocomponents' molar mass misses a measured one by more than 5 %; info: a watson_k the density overrides
cheme.assay-schema, cheme.assay-duplicate-id the file does not follow the schema; two assays with one id

A refused assay produces no pseudocomponent. Light ends (C5 and lighter) and heavy fractions are outside the correlations' range and therefore refused.

Identity, fingerprint, record

A characterisation carries METHOD_VERSION and a 16-hex fingerprint of the method version and the assay's inputs (not its id, title or source). The pseudocomponent identity is pseudo:<fingerprint>:<n>: the same identity means the same method and the same data. assay.comparability(a, b) lists what differs between two characterisations; a flash result carries the fingerprint it was made from.

cheme process assay --fill writes process/assay-characterization.json (schema cheme/assay-characterization/v1): per assay the characterisation, the correlations with their citations, the tool version, the time and the revision of the assay file. cheme validate and cheme process assay --check report cheme.assay-record-stale for an entry made from other inputs or by another method version; --check also reports an assay without an entry (cheme.assay-record-missing, info). --fill --prune removes entries of removed assays.

Flash and its claim

cheme process assay --pressure P adds the bubble and dew temperatures at P kPa(a); --temperature T --pressure P adds a flash (phase, vapour fraction, enthalpy relative to the ideal gas at 298.15 K). The flash is Peng-Robinson with kij = 0 between pseudocomponents. It is claimed inside assay.FLASH_CLAIM only:

  • states at 300-550 K and 50-500 kPa(a), pseudocomponents with normal boiling points of 340-448 K and specific gravities of 0.66-0.875;
  • bubble and dew temperatures within 2.0 K; enthalpy differences of liquid, saturated and vapour states from the liquid at 300 K within 20 kJ/kg; mixture molar mass within 3 %; liquid density at 15 °C from the cut gravities within 1 %;
  • benchmark: two defined naphtha surrogates (n-hexane to n-decane; n-hexane to n-nonane with cyclohexane and toluene), each characterised from its cuts, against CoolProp 8.0.0's reference multi-fluid models (GERG-2008 parameters of Kunz and Wagner for the alkane pairs, Bell et al. 2016 for the pairs with cyclohexane or toluene). Largest deviations: 1.8 K, 18.2 kJ/kg, 1.5 % and 0.15 %.

Not claimed: the Peng-Robinson liquid density (about 3 % low), states inside the two-phase region, the liquid heat capacity on its own (enthalpy differences within the liquid are up to 6 % low, inside the absolute tolerance), mixtures with water, hydrogen or light gases. Results outside the claim are computed and listed with the reason (outside_claim).

A published naphtha

The characterisation is checked against a measured assay: TotalEnergies' Oman crude assay (assay date 10 October 2020, OMAN.pdf), whose crude TBP curve gives wt % and vol % distilled and whose naphtha cuts have measured densities. The 80-175 °C heavy naphtha, characterised from the curve and its 742 kg/m3:

  • mass basis: the cumulative mass fraction of the pseudocomponents at each cut boundary matches the assay's own wt % within 0.015 (both columns are published to 0.1 % of the crude);
  • density: its Watson K predicts the measured density of the 80-150 °C cut (734 kg/m3) within 2 kg/m3 (0.2 kg/m3 found), and the 80-150 °C cut's K that of the 80-175 °C cut;
  • boiling: at 101.325 kPa its bubble point lies between the curve's initial and 50 % points and its dew point between the 50 % and final points; its flash states at 300 and 500 K lie inside the claim.

The same assay's 375-550 °C vacuum distillate is refused (cheme.assay-extrapolation), and its 100-150 °C cut, with four curve points, as insufficient.

Gaps

  • No openly accessible measured naphtha assay with molar mass or phase-behaviour data (flash, enthalpy) was found: the published naphtha checks mass basis, density and the ordering of its boiling behaviour; the flash and enthalpy tolerances come from defined surrogates, not a real naphtha.
  • Pseudocomponents are not accepted by simulation.yaml: a flowsheet with a petroleum fraction still cannot be simulated in DWSIM.
  • Light ends and heavy fractions are refused rather than characterised.