The Small Effect of Poroelastic Pressure Transients on Triggering of Production‐Induced Earthquakes in the Groningen Natural Gas Field

Abstract

Over the past decade, a steep increase in the number of seismic events has been observed in the Groningen gas field, the Netherlands. It is generally accepted that these are induced by compaction of the reservoir rock due to extensive depletion, causing a buildup of strain energy in faults that may be released seismically. We address the possible triggering of fault slip by the transient pressure field surrounding a well that has undergone a sudden rate change. Assuming a unilateral decoupling between displacement and pressure, numerical experiments are conducted using a sequential finite volume‐finite element solution strategy that fully incorporates second‐order terms in the radial flow equation. We investigate an idealized Groningen‐like geometry to discover whether the hypothesized possibility of triggering‐induced seismicity exists, explore how some controllable and uncontrollable variables influence its severity, and possibly provide clues on how production strategy might be able to avoid its occurrence. The results demonstrate that sudden production changes can indeed trigger near‐well seismic events, but that the effect is very small compared to other potential causes. Changing from sudden to gradual changes in well rates is therefore not expected to lead to a significant reduction in the number or magnitude of production‐induced earthquakes in the Groningen field.

Seismogenic Index of Underground Fluid Injections and Productions

“…The density of the hydrocarbon in situ is approximately ρ sg =650 kg/m 3 . Here we follow the estimate used by Postma and Jansen () . Therefore, the produced in situ volume is 2.3 × 10 9 m 3 .…”

Inelastic Deformation of the Slochteren Sandstone: Stress‐Strain Relations and Implications for Induced Seismicity in the Groningen Gas Field

“…During Stage 2, inelastic porosity reductions were generally small (0.4% to 0.6%) but still constitute 30% to 50% of the total porosity reduction measured in this stage (Figure ). In many geomechanical modelling studies (Bourne et al, ; Dempsey & Suckale, ; Lele et al, ; Postma & Jansen, ; Wassing et al, ; Zbinden et al, ), these inelastic strains are ignored, while the stress versus total strain behavior is quantified using be assumed elastic constants. We emphasize that while this assumption of poroelastic behavior in the near‐linear Stage 2 may originate from the small absolute inelastic strains developing here, the relative inelastic contribution to the total deformation behavior (30%–50%) is significant and should therefore be considered alongside the elastic behavior.…”

“…On the above basis, notably the latest experiments on Slochteren sandstone (Hol et al, , ; Pijnenburg et al, ) and earlier tests (Bernabe et al, ), it is increasingly clear that inelastic deformation contributes significantly to the compressive deformation of sandstone at the small strains (<1%) relevant for producing reservoirs. However, most geomechanical models addressing induced subsidence and seismicity ignore any inelastic contribution to the deformation of the reservoir and describe compaction using a simple compaction coefficient, in effect a poroelastic stiffness or compliance constant (Bourne et al, ; Dempsey & Suckale, ; Lele et al, ; Mulders, ; Postma & Jansen, ; van Eijs et al, ; Wassing et al, ; Zbinden et al, ). When an inelastic contribution is included, it is typically described using plasticity theory, originally developed to represent the inelastic deformation behavior of incohesive soils (Chan et al, ; Crawford et al, ; Crawford & Yale, ; Fredrich & Fossum, ; Han et al, ).…”

Using machine learning for model benchmarking and forecasting of depletion-induced seismicity in the Groningen gas field

“…Forecasting of induced seismicity requires a detailed understanding of both the physical mechanisms governing depletion-induced seismicity, as well as reservoir properties in time and space. On physical mechanisms, research from lab to field experiments and from experimental to complex numerical models has provided insights into the mechanisms behind induced seismicity in general [8][9][10][11][12], in particular depletion-induced seismicity [13][14][15][16][17][18][19] and injection-induced seismicity [20,21]. On reservoir properties, data acquisition efforts [22][23][24] can improve data quantity and quality, but as properties can only be measured directly in wells (which gives very limited spatial coverage) or using seismic reflection data (which needs to be interpreted, and has limits in applicability in regions with induced seismicity), these datasets carry large intrinsic uncertainties.…”