Numerical modeling of injection, stress and permeability enhancement during shear stimulation at the Desert Peak Enhanced Geothermal System

The problem

Heat can be extracted from hot rock only if water can flow through it. When drilled, well 27-15 at the Desert Peak geothermal field, Nevada, accepted water poorly. In 2010, it was stimulated to create an enhanced geothermal system (EGS), a network of fractures in hot rock through which water can circulate (Chabora et al., 2012). In shear stimulation, cold water is injected below the minimum principal stress so that existing fractures slip and their mismatched rough walls prop them open.

The wellhead pressure was raised in four steps, to at most 1.7, 2.4, 3.1 and 3.7 MPa. All stayed below the 5.2 MPa limit set to avoid hydraulic fracturing. Injectivity, the injection rate divided by the wellhead pressure, did not change at the first two steps. It first rose during the 3.1 MPa step and improved by an order of magnitude. Our goal was to model the stress and permeability changes it caused.

What we did

We built two models with FEHM, a simulator that couples fluid flow, heat transfer and rock deformation (Zyvoloski, 2007). The first models heat exchange between the injected water and the hot rock along 1030 m of well. It predicts the temperature and pressure of the water where it leaves the well.

These values are the inputs to the second model. Each point in its grid contains 100 fractures, with orientations sampled from the 72 fractures imaged in the well (Davatzes and Hickman, 2009). A fracture fails under the Mohr–Coulomb criterion when its shear stress exceeds its frictional strength. Its permeability then rises with slip, following laboratory sliding experiments on granite and marble (Lee and Cho, 2002). The injection rate is computed from the bottom-hole pressure and the evolving permeability.

What we found

From stresses measured in the well (Hickman and Davatzes, 2010) and friction coefficients of 0.65 to 0.96 from tests on core (Lutz et al., 2010), failure of the best-oriented fractures should begin once fluid pressure at depth exceeds its initial value by 2.1 to 4.8 MPa. The overpressure at depth was the wellhead pressure plus about 1.1 MPa from the weight of water filling the well. Even the first step therefore exceeded the lower end of this range, yet injectivity did not rise until the third step. The range applies only to the best-oriented fractures. For the friction coefficient of 0.75 in our model, it is about 2.8 to 3.9 MPa (our arithmetic). In our model, the water at the first two steps was also only about 20 °C cooler than the rock. Overpressure and cooling together were too small for fractures to fail.

The model reproduced the stable early injectivity and approximately captured the onset of injectivity gain near day 35. Matching the late injectivity required each fracture's permeability to rise by up to a factor of 50. This is probably an overestimate, because the model leaves out elastic opening of fractures.

The water at the first two steps warmed on its way down because injection was slow. At 3.1 MPa, the higher injection rate delivered colder water. Cooling makes rock contract and reduces the normal stress on fractures. Together with the fluid pressure, this brought fractures to failure. Failure raised permeability and the injection rate. Faster injection then delivered still colder water.

Overpressure fell off quickly with distance from the well. Several metres out it was too low to cause failure, yet failure continued. Because cold water crossed the stimulated rock with little warming, the temperature drop at the front remained sharp. The cooled rock is also mechanically coupled to the rock around it. As it contracts, it exerts stress beyond the cold zone. We call this non-local stress. At day 63, the drop in normal stress from cooling contributed most to failure next to the well. From 18 m out to the edge of the failure front at 30 m, non-local shear stress contributed most. This suggests that the failure front can sometimes be advanced by stress from the cooled rock behind it. However, in our model cooling put the rock near the well in tension. The model cannot represent tensile failure. Non-local stresses are therefore probably smaller and less extensive than modelled, although we expect they would still contribute to failure.

The modelled stimulated volume began to grow around day 37. By the end of the 3.1 MPa step it reached 46,000 m³, enough to fill a cube about 36 m on each side (our conversion). After about day 55, injectivity gains were small, but the volume kept growing. Its growth correlated closely with the volume of injected water colder than the rock.

Limitations

Injectivity fell by about a factor of two during each shut-in between steps, and our model cannot lose permeability. We therefore modelled only the first 67 days. Only one microearthquake was detected. Both observations would fit a stimulation driven entirely by elastic opening of fractures, a mechanism we cannot rule out. However, the abrupt rise after two stable steps suggests a pressure threshold, as shear failure requires. We also estimate that slip on fractures 4.4 to 20.5 m across, close to the size of the modelled stimulated zone, would have gone undetected.

The modelled volume probably underestimates the rock affected by injection, because the model assumes rock and water share one temperature.

Late in a stimulation, when injectivity gains become too small to measure, microseismic monitoring or tracer tests may be needed to follow the growth of the stimulated volume.

Glossary

An enhanced geothermal system (EGS) produces heat from hot rock whose permeability has been increased by stimulation.

Permeability measures how easily fluid flows through rock.

Shear stimulation is injection of water below the minimum principal stress so that existing fractures slip and become more permeable.

Injectivity is the injection rate divided by the wellhead pressure.

A megapascal (MPa) is a unit of pressure, about ten times atmospheric pressure.

The friction coefficient sets the shear stress a fracture can resist for a given normal stress. Higher values mean a stronger fracture.

The best-oriented fractures are those whose angle to the stresses brings them closest to slipping.

Overpressure is fluid pressure above its initial value in the rock.

The minimum principal stress is the smallest compressive stress in the rock. Fluid above this pressure can open a hydraulic fracture.

Elastic opening is reversible widening of a fracture as fluid pressure rises or the rock cools. The fracture closes again when the pressure falls or the rock reheats.

A hydraulic fracture is a new tensile crack opened by fluid pressure.

The Mohr–Coulomb criterion states that a fracture slips when its shear stress exceeds the friction coefficient times the normal stress, reduced by fluid pressure.

The normal stress on a fracture presses its walls together. The shear stress pushes them to slide past each other.

The failure front is the outer edge of the rock in which fractures have failed.

Non-local stress is stress induced in rock beyond the cooled region by the contraction of the cooled rock.

The stimulated volume is the volume of rock whose permeability has increased.

A shut-in is a period when injection stops and the well is closed.