How to calculate vapor pressure from a D86 distillation curve

Q: How does the CalebBell thermo library handle D86 conversion?

CalebBell's thermo library includes functions for petroleum fraction characterization. The `thermo.petroleum_fractions` module accepts TBP cut temperatures and specific gravities as inputs, characterizes each cut into a pseudocomponent with critical properties (using the Riazi-Daubert or other correlations), and returns objects you can pass directly into flash calculations. You must first convert D86 to TBP using a separate step—thermo does not do the D86-to-TBP conversion natively, but the API or Edmister correlation is straightforward to implement.

Calculating vapor pressure from an ASTM D86 distillation curve is a three-step process: convert the D86 curve to a true boiling point (TBP) curve, cut pseudocomponents from the TBP curve, then run a vapor pressure calculation using an equation of state on the characterized pseudocomponents. There is no single-step formula that goes directly from D86 temperatures to vapor pressure—the D86 test conditions introduce systematic distortions that must be corrected first.

The short answer

The D86-to-vapor-pressure workflow in brief:

D86 → TBP conversion using the Edmister-Okamoto or API Chapter 3 correlation. This corrects for the inefficient separation conditions of the D86 test.
TBP → pseudocomponents by slicing the TBP curve into cuts (typically 10–20°F boiling-point intervals) and characterizing each cut with a normal boiling point, specific gravity, and molecular weight.
Pseudocomponent characterization — compute Tc, Pc, and acentric factor for each cut using a correlation such as Riazi-Daubert or Lee-Kesler.
Vapor pressure calculation — run a Peng-Robinson flash or compute the vapor pressure of each cut independently using an Antoine-type correlation, then combine using Raoult's law or a full EOS flash.
Validate against a measured RVP if available—Reid Vapor Pressure from ASTM D323 is the most common field measurement you can use to spot-check the calculation.

Understanding what D86 actually measures

ASTM D86 is an atmospheric distillation test standardized for reproducibility in a QC lab, not for thermodynamic accuracy. The apparatus is a round-bottom flask over a burner, with a condenser and a graduated cylinder to catch distillate. The test specifies the heating rate, the condenser temperature, and the ambient conditions, but it does not impose a high reflux ratio. This means vapor and liquid coexist throughout the flask, and the distillation is not a true equilibrium stage process.

The practical consequence: D86 temperatures at a given volume-percent recovery are systematically lower than TBP temperatures for the same fraction—by roughly 10°C at the 10% point and up to 30°C at the 90% point, depending on the fraction's volatility distribution. If you take D86 temperatures at face value and treat them as boiling points, your pseudocomponents will have characteristically low boiling points and your vapor pressure estimate will be too high.

The Edmister-Okamoto correlation corrects D86 temperatures to TBP temperatures as a function of the slope of the D86 curve (d°F/vol%) and the volume fraction. The API Technical Data Book, Chapter 3, documents this conversion in detail. It is the industry standard; most simulators implement it internally when you enter a D86 assay.

A simpler but less accurate alternative is the API 50-year correlation, which fits a smooth polynomial to the D86-TBP offset as a function of percent distilled. It works reasonably well for straight-run naphthas and distillates but should not be used for cracked stocks or oxygenated blends.

Pseudocomponent generation

Once you have the TBP curve, you divide it into cuts. The standard approach in process simulation is to cut at fixed boiling-point intervals—10°F or 20°F cuts are common for refinery streams. Each cut is assigned:

A normal boiling point (Tb): the mid-volume-percent boiling temperature of the cut.
A specific gravity (SG): interpolated from the gravity versus boiling point curve for the crude or product.
A molecular weight: estimated from Tb and SG using a correlation such as Riazi (2005) or the Winn nomograph.

From Tb, SG, and MW, you estimate the critical properties using a characterization correlation. The Riazi-Daubert correlation (documented in the API Technical Data Book and in Riazi's Characterization and Properties of Petroleum Fractions, ASTM International, 2005) gives Tc, Pc, and Vc as functions of Tb and SG. The Lee-Kesler generalized correlation gives the acentric factor from Tc, Pc, and Tb.

These correlations have meaningful uncertainty for fractions with Tb above roughly 600°F—heavy gas oil and vacuum distillate cuts where the Watson characterization factor (Kw) and paraffinicity of the stock matter. Aromatic stocks will have different critical property relationships than paraffinic stocks at the same boiling point, and the generalized correlations do not capture this well. If your stream has unusual paraffinicity (e.g., a highly aromatic FCC LCO versus a straight-run paraffinic naphtha), expect larger errors in the characterization step.

Vapor pressure calculation from pseudocomponents

With Tc, Pc, and ω (acentric factor) for each pseudocomponent, you have two main paths:

Path 1: Per-cut vapor pressure + Raoult's law. Compute the vapor pressure of each pseudocomponent cut at the target temperature using the modified Antoine equation or the Riedel correlation. Combine using:

P_total = Σ (xᵢ · Pᵢˢᵃᵗ)

where xᵢ is the mole fraction of cut i and Pᵢˢᵃᵗ is its saturation pressure. This is Raoult's law and assumes ideal liquid behavior—a reasonable approximation for petroleum fractions where components are chemically similar. For light ends (C₄/C₅ + gasoline blend), the assumption is less accurate because of the significant non-ideality between light alkanes and aromatic components.

Path 2: Full EOS flash. Assemble all pseudocomponents into a mixture, assign binary interaction parameters (typically zero for pseudo-pseudo pairs, or estimated from a group-contribution method), and run a bubble-point flash at the target temperature. The flash gives you the bubble-point pressure, which equals the vapor pressure at that temperature for a system in two-phase equilibrium. Peng-Robinson with the Peneloux volume-translation correction is the standard choice for this calculation. Twu-modified PR improves accuracy for heavy fractions.

Path 2 is more rigorous and is what process simulators do internally. Path 1 is useful for quick estimates and for understanding sensitivity—you can see which cuts drive the overall vapor pressure and which are irrelevant.

Converting to Reid Vapor Pressure

If your end goal is Reid Vapor Pressure (RVP) for regulatory or blending purposes, the simulation result gives you a true vapor pressure (TVP) at 100°F. The relationship between TVP and RVP depends on the dissolved gas content (primarily butanes and lighter) and the air-saturation conditions of the ASTM D323 test.

For typical gasoline blends, TVP and RVP differ by 1–5 psi—RVP is generally higher than TVP because the dissolved air in the ASTM test contributes to the measured pressure. The API Technical Data Book provides conversion charts (API Chapter 5, vapor pressure conversion). For regulatory compliance purposes, the regulatory authority specifies which measurement method applies; do not substitute TVP for RVP without checking the applicable regulation or permit.

Light-ends correction: where D86 misses the story

A critical limitation of D86 is that very light components—anything lighter than pentane—do not fully condense in the standard apparatus and are largely lost from the distillate collection. This means a D86 curve on a full-range naphtha or blended gasoline systematically underrepresents the C₄ and lighter content, which are the dominant contributors to vapor pressure.

If you have a GC analysis of the full composition (including C₁–C₄), you should use it. The GC data gives you the light-end mole fractions directly; you use D86 only for the C₅+ fraction that the GC lumps as "heavy naphtha" or similar. Most refinery labs run both, and process engineers should reconcile the two before running a vapor pressure calculation.

If you have only D86 with no GC complement, your vapor pressure estimate for the light fraction will be systematically low. Adjust expectations accordingly, or request supplementary ASTM D5191 (dry vapor pressure equivalent) data from the lab.

When this gets harder

Blended or cracked stocks: D86 correlations were developed on straight-run petroleum fractions. FCC gasoline, reformate, and blended stocks can have non-monotonic gravity-versus-boiling-point relationships that violate the assumptions behind the Riazi-Daubert characterization. The correlations still produce numbers, but the uncertainty is larger—potentially 10–20% on vapor pressure for a complex blended stream.

Oxygenated blends (ethanol, MTBE): Oxygenates in gasoline form azeotropes with hydrocarbons, making the vapor pressure distinctly non-ideal. Raoult's law fails here. You need activity-coefficient models (NRTL, UNIFAC) for the liquid phase. The D86 test itself is affected: oxygenate-hydrocarbon azeotropes distill at temperatures that do not correspond to the TBP of either component alone.

Cutter stocks and variable gravity curves: Some process streams have two distinct density populations—for example, a lube distillate cut with a paraffinic base and a contaminating aromatic fraction. A single gravity-versus-TBP curve cannot represent both, and pseudocomponent characterization will be systematically biased. A GC-simulated distillation (ASTM D2887) gives better compositional resolution than D86 for these cases.

Temperature and pressure sensitivity: Vapor pressure is highly temperature-sensitive. A ±5°C error in pseudocomponent boiling point translates to roughly ±10–15% error in vapor pressure at 100°F for light fractions. The D86-to-TBP conversion itself carries ±3–5°C uncertainty for well-behaved stocks. These errors compound.

How Rankine handles this

Rankine accepts ASTM D86 assay data as a direct input to its petroleum fraction characterization workflow. You enter the D86 percent-distilled versus temperature table plus the bulk specific gravity, and Rankine performs the D86-to-TBP conversion, generates pseudocomponents at configurable cut intervals, runs the Riazi-Daubert characterization, and populates a stream with the resulting pseudocomponent mixture. The stream then flows into any unit operation in the flowsheet—a flash drum, a fractionation column stub, or a vapor pressure calculator block.

The EOS selection for petroleum fraction work is Peng-Robinson with Peneloux correction by default. The Twu alpha function is available as an override for heavy fraction accuracy. Every characterization step logs its inputs and outputs to the audit trail, so you can trace exactly what gravity and TBP curve produced the pseudocomponent set.

For engineers who need to understand how EOS choice affects the vapor pressure result—particularly when switching from a HYSYS model that used a different characterization method—the GERG-2008 vs Peng-Robinson article explains which EOS is appropriate for which stream type. If you are evaluating Rankine as a HYSYS replacement for refinery work, the HYSYS alternatives comparison covers the full simulator landscape. Learn more about Rankine at the homepage.

FAQ

Can you directly convert a D86 curve to vapor pressure?

Not directly—the D86 curve gives you atmospheric boiling-point distribution across the mixture, not a single-component vapor pressure. The standard workflow converts D86 to a TBP curve, then cuts pseudocomponents from the TBP curve, characterizes each pseudocomponent (Tc, Pc, acentric factor), and runs a flash or vapor pressure calculation using an equation of state or a correlation. Each step introduces uncertainty that accumulates.

What is the difference between D86 and TBP distillation?

ASTM D86 is a standardized atmospheric distillation test run at a fixed heating rate and geometry, designed for reproducibility in a QC lab. True Boiling Point (TBP) distillation—run at high reflux ratios under vacuum, or simulated by GC—produces a thermodynamically rigorous separation that maps directly to component boiling points. D86 temperatures run 10–30°C lower than TBP at the same volume fraction recovered because D86 conditions do not achieve true equilibrium separation. The D86-to-TBP conversion (Edmister-Okamoto or API Chapter 3 correlation) corrects for this.

What is Reid Vapor Pressure and how does it relate to D86?

Reid Vapor Pressure (RVP) is the vapor pressure of a petroleum product measured at 100°F (37.8°C) by ASTM D323 or D5191. D86 does not directly give you RVP, but RVP correlates empirically to the front end of the D86 curve—specifically the 5% and 10% boiling points—because light ends (butanes, pentanes) dominate the low-end distillate and drive vapor pressure. The API Technical Data Book documents correlations relating D86 front-end temperatures to RVP.

How does the CalebBell thermo library handle D86 conversion?

CalebBell's thermo library includes functions for petroleum fraction characterization. The thermo.petroleum_fractions module accepts TBP cut temperatures and specific gravities, characterizes each cut into a pseudocomponent with critical properties using the Riazi-Daubert or other correlations, and returns objects you can pass into flash calculations. You must first convert D86 to TBP using a separate step—thermo does not include the D86-to-TBP conversion natively, but the Edmister or API correlation is straightforward to implement.

Which equation of state is best for vapor pressure of petroleum fractions?

Peng-Robinson with volume translation (Peneloux shift) is the industry standard for petroleum fraction vapor pressure calculations. GERG-2008 is not applicable—petroleum pseudocomponents are not in GERG's 21-component scope. For heavier fractions where pseudocomponent critical properties are uncertain, the Twu or Mathias-Copeman alpha functions improve PR accuracy over the standard Boston-Mathias form.