Discussion of the Paper SPE-187038-MS: Fracture Closure Stress: Reexamining Field and Laboratory Experiments of Fracture Closure Using Modern Interpretation Methodologies

In recent years, there has been discussion in the literature regarding methods of estimating the magnitude of the minimum principal stress from subsurface fracture injection tests, commonly called Diagnostic Fracture Injection Tests (DFITs). McClure et al. (2016) used modeling and mathematical analysis to propose that changes should be made to common interpretation techniques. Subsequently, Craig et al. (2017) attempted to refute those findings using field and laboratory observations. This discussion paper reviews the interpretations from Craig et al. (2017) and concludes that their conclusions are unsupported by the data presented.

💡 Research Summary

This discussion paper critically evaluates the claims made by Craig et al. (2017) that the traditional “tangent method” for interpreting diagnostic fracture injection tests (DFITs) provides reliable estimates of the minimum principal stress (Sh min). The authors, led by Mark McClure, argue that Craig et al.’s conclusions are unsupported by the data and suffer from several methodological flaws.

The background section explains that DFITs are used to infer Sh min from pressure‑time data. Historically, the “tangent method” (Barree et al., 2009) constructs a G·dP/dG plot and selects the closure pressure at the point where a line from the origin is tangent to the curve. This method assumes that the closure pressure equals Sh min, but it lacks a solid theoretical basis and performs poorly in low‑permeability formations where the pressure transient deviates from the ideal square‑root‑time behavior.

McClure et al. (2016) introduced the “compliance method,” derived from the chain‑rule decomposition dP/dG = (dP/dV)(dV/dG). When fracture walls contact, system compliance drops sharply, causing dP/dG to rise. Because real fractures retain some aperture due to surface roughness, the pressure at contact is typically a few tens of psi higher than Sh min. The compliance method therefore predicts a closure pressure that is Sh min + ~75–150 psi, and it is supported by detailed physics‑based simulations that include fracture propagation, wall contact, and leak‑off.

The paper then dissects Craig et al.’s attempt to refute the compliance method using field tilt‑meter data and laboratory shut‑in experiments. Eight major criticisms are presented:

1. Improper Plotting – Craig et al. plotted tilt versus time rather than tilt versus pressure, obscuring the true relationship between fracture stiffness and pressure.

2. Misidentification of Re‑opening Points – In their reproduced tilt‑pressure plot (Figure 2), the only clear deflections occur at points A (≈2500 psi) and C (≈3200 psi), which correspond to the compliance‑method predictions. The tangent‑method pressure (≈2900 psi, point B) shows no change in slope, contradicting Craig et al.’s claim that it marks re‑opening.

3. Selective Literature Citation – Earlier studies (Gulrajani & Nolte, 2000; Barree et al., 2009) that analyzed the same data and supported the compliance method are omitted, while Craig et al. cite only sources that favor the tangent method.

4. Neglect of Rough‑Surface Fluid Storage – Real fractures can store fluid even when walls are in contact due to asperities. Ignoring this leads to under‑estimation of the closure pressure.

5. Inappropriate G‑Function Application to Lab Data – The laboratory shut‑in tests exhibit strong deviation from Carter leak‑off because pressure drops dramatically (≈50 % of ISIP). The G‑function assumes constant fracture stiffness and Carter leak‑off; applying it under these conditions invalidates the analysis.

6. Misinterpretation of “Pressure‑Dependent Leak‑off” – Craig et al. label the observed early‑time high permeability as pressure‑dependent leak‑off, whereas it is simply the breakdown of the Carter model when pressure falls.

7. Biased Closure Pick – They select a closure point in the middle of a flat G·dP/dG segment, which conveniently matches the known Sh min, suggesting a post‑hoc fitting rather than an objective interpretation.

8. Incomplete Reporting of Blind Tests – Only one blind‑test result (which happened to be close to the true Sh min) is presented, while other blind tests with poorer matches are omitted, indicating selective reporting.

The authors reinforce that the compliance method accounts for fracture roughness, residual aperture, and the physics of wall contact, providing a mathematically consistent framework. Recent in‑situ core‑across studies (Gale et al., 2018) and joint‑stiffness experiments (Barton et al., 1985) further validate the assumptions underlying the compliance approach.

In conclusion, the discussion paper asserts that Craig et al.’s (2017) arguments against the compliance method are based on flawed data presentation, inappropriate analytical techniques, and a fundamental misunderstanding of fracture mechanics. The compliance method remains the only rigorously justified and empirically validated technique for estimating Sh min from DFIT data, especially in low‑permeability reservoirs where traditional tangent‑method interpretations are unreliable.