Most wellbore stability problems are not surprises. They are the predictable consequence of drilling with a mud weight that was chosen before anyone built a stress model, or from a model that was never calibrated against the pressures and leak-off tests the well itself produced.
A one-dimensional mechanical earth model (1D MEM) is the standard answer. It is not exotic: it takes the logs you already have (compressional slowness, shear slowness, bulk density, resistivity) and turns them into a continuous profile of rock properties, pore pressure, and in-situ stress along the wellbore. From that profile falls out the number that actually matters at the rig: the range of mud weights that will neither collapse the hole nor fracture it.
The chain of calculations
A 1D MEM is a sequence, and each step depends on the one before it. Getting the order right matters more than any individual correlation.
1. Elastic properties. Dynamic Poisson's ratio and Young's modulus come from compressional and shear velocities with bulk density. Where a shear log is missing, which is often, shear slowness is synthesized from compressional slowness. The Castagna mudrock relation is the common default; the Greenberg–Castagna relations do better when you can specify lithology as sandstone, limestone, dolomite, or shale. Dynamic moduli are then scaled to static values with a factor typically between 0.5 and 0.8, because rock responds differently to a sonic pulse than to the slow loading of a wellbore.
2. Overburden stress. Vertical stress is the integral of bulk density from surface to depth. This sounds trivial and is where a surprising number of models go wrong: density logs rarely start at surface, so the shallow section must be extrapolated, and offshore wells need the water column handled at the seawater gradient with a proper mudline reference. An overburden curve that is wrong by a few hundred psi propagates into every horizontal stress below it.
3. Pore pressure. Two families dominate. Eaton compares the observed log, sonic or resistivity, against a normal compaction trend and raises the difference to an exponent, typically around 3.0. Bowers works through effective stress directly and, critically, can represent unloading behaviour where a section has been uplifted or where pressure was generated by a mechanism other than undercompaction.
The normal compaction trend is the part people rush. A trend taken from a textbook default rather than fitted to your own clean shale intervals will give you a pore pressure curve that is precisely calculated and wrong. Fit the trend to your data, then check the answer against measured points.
4. Rock strength. Unconfined compressive strength is estimated from log correlations chosen by lithology: McNally for sandstone, Horsrud for shale, Golubev and Rabinovich for carbonate. Internal friction angle is either set as a constant or derived from compressional velocity using Lal's relation. Tensile strength is conventionally taken as a small fraction of UCS, often around one twelfth.
These correlations carry real scatter. They are screening tools, and the honest way to use them is to know which one you selected and why, not to treat the output as a measurement.
5. Horizontal stresses. With overburden, pore pressure, and elastic properties in hand, minimum and maximum horizontal stress follow from a poroelastic formulation that includes tectonic strain terms. Those strain terms are the calibration handles; they are what you adjust so that the modelled Shmin honours your leak-off and formation integrity tests.
6. The mud weight window. Finally the useful part. The lower bound is the pressure below which shear failure, breakout, begins at the borehole wall. The upper bound is the pressure at which the wall goes into tension and the formation breaks down. Expressed in pounds per gallon equivalent mud weight, that interval is the drilling window.
Calibration is not optional
An uncalibrated MEM is a hypothesis. Three pieces of field data turn it into a model:
- RFT/MDT formation pressures test the pore pressure curve directly. Plot them over the prediction and the mismatch tells you whether your compaction trend or Eaton exponent needs work.
- Leak-off and formation integrity tests constrain minimum horizontal stress. If modelled Shmin sits well away from a good LOT, the tectonic strain terms are wrong.
- Drilling events (losses, tight hole, cavings, stuck pipe) are free calibration. A breakout interval that the model says should have been stable is telling you something about strength or stress that no correlation will.
The vertical well assumption breaks early
Textbook mud window derivations assume a vertical hole, where the borehole axis aligns with vertical stress and the arithmetic is simple. That assumption fails as soon as the well builds angle.
In a deviated or horizontal well the far-field stress tensor must be rotated into borehole coordinates using well inclination, well azimuth, and the azimuth of maximum horizontal stress. The stress concentration around the hole then depends on trajectory, and so does the safe window. Two wells through identical rock at identical depth can have materially different mud weight requirements purely because one is vertical and the other is drilling along the direction of minimum horizontal stress.
For any well beyond modest deviation, running the vertical approximation is not a conservative simplification. It can be optimistic in exactly the direction that gets you into trouble.
Sanity-check the stress state before anyone drills
The stress polygon, the Zoback construction, is the fastest check available. For a given vertical stress, pore pressure, and coefficient of friction, it bounds the horizontal stress pairs that are frictionally admissible: outside that boundary the crust would already have failed on optimally oriented faults. Plot your modelled Shmin and SHmax inside the polygon. If they land outside it, the model is not merely uncertain, it is physically impossible, and no amount of downstream analysis will fix that.
Beyond drilling: sanding and drawdown
The same stress model answers a completion question. As reservoir pressure depletes and drawdown increases, effective stress around the perforation rises until the rock fails and produces solids. The critical bottomhole flowing pressure marks that onset. Having it as a curve alongside the drilling window means the same mechanical model informs both how you get the hole down and how hard you can produce it afterwards.
Keeping it in one workspace
The practical obstacle to doing this well is rarely the physics. It is that geomechanics traditionally lives in a separate application from petrophysics, so curves get exported, depth references drift, units get converted twice, and the model is rebuilt from scratch every time the interpretation changes.
Running the mechanical earth model in the same workspace as the log interpretation removes that friction. The same depth reference, the same alias table, the same unit system, the same zones. When the shale volume curve is revised, the geomechanics is recomputed from it rather than from a stale export.
In Reservoir Risk Solutions, the 1D MEM is a compute mode inside the Formation Evaluation workspace. Shear synthesis, overburden integration, Eaton and Bowers pore pressure with fitted compaction trends, UCS correlations by lithology, poroelastic horizontal stresses, the deviated-well mud weight window, the stress polygon, and sanding onset pressure all run in the browser, on well data that stays in a project folder on your own machine. Calibration points import as CSV and overlay directly on the computed curves.
The goal is not a prettier plot. It is a wellbore stability answer that someone can interrogate, every correlation named, every parameter visible, and every prediction checked against what the well actually measured.