Gassmann fluid substitution: predicting the seismic response of fluid changes

The central insight

When a pore fluid is replaced by another fluid, the rock’s saturated bulk modulus $K_{\text{sat}}$ changes — stiff fluids (brine) make the rock harder to compress; soft fluids (gas) make it easier. But the saturated shear modulus $\mu_{\text{sat}}$ does NOT change. Fluids cannot resist shear (μ_fluid = 0), so substituting one zero-μ fluid for another leaves the rock’s shear behaviour identical.

This is Gassmann’s key result. It seems trivially obvious in retrospect but it has profound consequences: the entire AVO classification framework, and most of QI, depends on the fact that fluid substitution affects K and density but not μ. We will see in §5.4 how this lets pre-stack seismic separate fluid effects from lithology effects.

Gassmann’s equation

For a porous rock with porosity $\phi$, mineral matrix bulk modulus $K_{\text{min}}$, dry-frame bulk modulus $K_{\text{dry}}$, and pore-fluid bulk modulus $K_{\text{fluid}}$, the saturated bulk modulus is:

**

**

It looks intimidating but the structure is intuitive: the saturated K is the dry-frame K plus a positive “fluid stiffening” term. The fluid stiffening is large when $K_{\text{fluid}}$ is large (brine) and small when $K_{\text{fluid}}$ is small (gas). The factor $(1 - K_{\text{dry}}/K_{\text{min}})^{2}$ in the numerator measures how much room the dry frame leaves for the fluid to contribute — stiffer dry frames give the fluid less room to matter.

And critically:

$\mu_{\text{sat}} = \mu_{\text{dry}}$

One line. Often the most important line in the entire equation set.

How density changes

Density mixes linearly: a rock with porosity $\phi$ and mineral density $\rho_{\text{min}}$ saturated with fluid of density $\rho_{\text{fluid}}$ has bulk density:

**

**

For mixed-fluid pore spaces (some brine, some gas), the bulk-density mixing is the simple weighted sum: density mixes by volume. Bulk modulus, on the other hand, mixes by inverse-volume (Reuss average for uniform mixing): $1/K_{\text{fluid}} = S_w/K_{\text{brine}} + S_g/K_{\text{gas}}$.

The non-linear mixing of $K_{\text{fluid}}$ has a famous consequence: even a small gas fraction (5–10%) drops $K_{\text{fluid}}$ dramatically because the soft gas dominates the inverse-K average. This is what produces the fizz effect — a small amount of gas is almost as visible on Vp as a large amount.

What this means for Vp, Vs, and ρ

Combining Gassmann’s K equation with the velocity-from-modulus formulas:

• Vp drops with gas. $V_p = \sqrt{(K + 4\mu/3)/\rho}$. K drops sharply (Gassmann); μ unchanged; ρ drops linearly. The K drop dominates the ρ drop, so Vp falls.
• Vs INCREASES with gas. $V_s = \sqrt{\mu/\rho}$. μ unchanged; ρ drops. Vs goes UP. This is the counterintuitive result that surprises everyone the first time they see it.
• ρ drops linearly with gas saturation.
• Vp/Vs drops sharply. Both Vp drop and Vs rise contribute. (Vp/Vs)² = K/μ + 4/3, so as K falls, the ratio falls. Combined with the Vs rise effect, Vp/Vs typically drops 0.2–0.4 across full gas substitution — a huge shift in seismic terms.

Exercise — see the fluid effects

• The widget starts with clean sandstone, plotting Vp vs water saturation Sw. The brine baseline (dashed) is at Sw=100%; the curve drops as Sw decreases (more gas). Notice the curve’s shape: a sharp drop in the first 5–10% gas, then a long flat tail. This is the fizz effect.
• Slide Sw from 100% down to 95% (just 5% gas). Watch how much Vp drops in that small range — typically over half the total drop you would see at 100% gas. Five percent gas can produce nearly the full amplitude anomaly of a real reservoir, which is why “fizz water” (sub-economic gas saturations) is a notorious DHI false positive.
• Now switch the Property dropdown to Vs. Slide Sw down. Notice that Vs RISES as gas replaces brine — this is the Gassmann counterintuitive result. μ didn’t change but ρ dropped, so Vs went up.
• Switch to Density. Sliding Sw down gives a clean linear decrease — density mixes volumetrically, no fancy physics.
• Switch to Vp/Vs. The drop here is dramatic and almost entirely concentrated at low Sw (high gas). This is why Vp/Vs from inversion is one of the most powerful gas indicators in QI work.
• Try other rock frames. Shaley sandstone shows similar shapes but smaller magnitudes — lower porosity gives the fluid less room to matter. Limestone shows much smaller fluid effects because the dry-frame stiffness is so high; carbonates are notoriously hard to interpret with Gassmann.

Why the fizz effect happens

The fluid bulk modulus $K_{\text{fluid}}$ is what couples into Gassmann’s formula. The Reuss average for a brine-gas mixture is:

**

**

With $K_{\text{brine}} \approx 2.95$ GPa and $K_{\text{gas}} \approx 0.05$ GPa, the gas term is 60 times larger than the brine term per unit saturation. Even at 5% gas (Sw=0.95), the $1/K_{\text{fluid}}$ is dominated by the gas inverse: $1/K_{\text{fluid}} = 0.95/2.95 + 0.05/0.05 = 0.32 + 1.00 = 1.32$, giving $K_{\text{fluid}} = 0.76$ GPa — only 25% of pure brine, mostly because of that 5% gas.

The seismic consequence: 5% gas drops Vp almost as much as 100% gas. The interpreter cannot distinguish between an economic reservoir (50–80% gas) and a sub-economic fizz reservoir (5–10% gas) on Vp alone. Density and Vs help — ρ drops linearly, so 5% gas drops ρ by only 5% of the full drop, while Vp drops by 80% of the full drop. The combination of Vp + density + Vs gives a fighting chance at distinguishing fizz from pay; Vp alone does not.

Common Gassmann pitfalls

• The dry-frame K and μ must be known or estimated. They are usually back-computed from a brine-saturated state (well-log Vp, Vs, ρ + assumed brine K, density) using inverse Gassmann — the exact procedure the widget uses internally. Errors in the brine-saturated input propagate.
• K_min varies with mineralogy. Quartz K ≈ 37 GPa; calcite ≈ 70 GPa; dolomite ≈ 95 GPa. A clean sandstone is mostly quartz; a feldspathic sand is harder; a shale uses a different framework altogether (Gassmann is not strictly valid for shales). Check the K_min you assume.
• Frequency dispersion. Gassmann is a low-frequency theory; ultrasonic-lab measurements (MHz) over-stiffen the rock. For seismic frequencies (10–100 Hz), Gassmann is excellent. For sonic logs (10 kHz) it is good. Don’t back-substitute Gassmann’s rules into ultrasonic-frequency data without correction.
• Carbonates are hard. The pore-space geometry in carbonates (mouldic, vuggy, fractured) violates Gassmann’s implicit assumption of well-connected pore space. Production carbonate interpretation often uses a more general framework (Brown-Korringa, anisotropic) instead.
• Patchy vs uniform saturation. Gassmann’s standard form assumes uniform mixing (Reuss average for fluids). When fluids mix patchily at sub-wavelength scales, the effective K_fluid is the Voigt average instead, which is much higher. Patchy saturation gives smaller Vp drops than uniform saturation — this is one reason field measurements sometimes show less Vp drop than Gassmann predicts.

The QI workflow that uses Gassmann

A typical quantitative-interpretation pipeline:

• Acquire pre-stack seismic data; perform AVO analysis to extract intercept and gradient volumes (§5.4).
• From a well log within the survey: read Vp, Vs, ρ in the brine-saturated reference state. Compute K_sat and μ from these.
• Inverse Gassmann: back out the dry-frame K_dry. μ is already in hand.
• Forward Gassmann: substitute different fluid scenarios (full gas, full oil, mixed saturations) at the same dry frame. Predict Vp, Vs, ρ for each scenario.
• Forward AVO modelling: use the predicted Vp, Vs, ρ to compute the expected AVO response for each scenario.
• Compare with the observed AVO. Pick the scenario whose predicted response best matches the data.

The whole loop depends on Gassmann being able to predict the velocity changes accurately. When it can’t (carbonates, very dispersive rocks, anisotropic shales), the entire interpretation is uncertain, and that is when interpreters reach for full elastic inversion or seek calibration with more wells.