Breaking News

# Seismic Velocity Changes in the Groningen Reservoir Associated with Remote Drilling

As mentioned earlier, the temporary decrease in the travel time of the P waves to geophone 10 and the increase in the delay time of the PS waves could be explained by a temporary upward movement of the gas-water contact (GWC). Here we explore this interpretation in more detail.

### GWC elevation from seismic observations

The variation of seismic velocities due to the substitution of gas by water in a porous sandstone can be calculated with Gassmann’s model of fluid substitution25, 26. The bulk modulus of a fluid-saturated rock is related to the porosity and bulk moduli of the mineral matrix, interstitial fluid, and dry rock framework. The mass modulus of the fluid will increase in case of gas-water substitution, which will increase the effective mass modulus of the rock, and therefore the velocity P. The shear modulus, on the other hand, will not change because it mainly depends on the solid rock structure. However, due to the small increase in density, a slight decrease in S velocity is expected.

Quantitative estimation of the change in GWC level using the Gassmann model would require accurate values ​​of the mass and shear moduli of the matrix, fluids (gas and brine), and rock framework for the local rock. . Since these estimates are unknown and approximate would have large uncertainties, we took a more practical approach. We estimated the average P velocity above and below the GWC from sonic logging data and found P velocities of 3321 m/s and 3688 m/s, respectively (Supplementary Material, Section 5) . Assuming these values, a decrease in P-wave travel time of 0.7 ms would correspond to a rise in GWC of 23 m. We further investigated whether this rise in GWC could also explain the increase in PS lag time ((Delta (t_{PS}-t_P)) (simeq) 1.0ms). Assuming a 23 m GWC offset, an increase in S-wave travel time of 0.3 ms, vertical propagation with an S-wave velocity of 2000 m/s for sandstone with gas12, we find a decrease in the speed of the S wave of only 52 m/s (2.6%). This decrease would be solely the effect of the increase in density on the shear rate caused by the replacement of gas by water. Although the values ​​seem realistic, it should be noted that the uncertainties are large and that 23 m should only be interpreted as an indication of the elevation of the GWC inferred from our measurements.

The other observation is the rapid decrease in the noise level, as well as its rapid return to the normal level observed for geophone 10 compared to the other geophones (Fig. 4). These rapid changes are easily accomplished by changes in the level of the GWC, although it is unclear how this would reduce the noise level.

### Relationship to drilling operations at Borehole ZRP-3

If a temporary elevation of the GWC can explain the seismic observations, the question arises as to what caused the elevation of the GWC. Gas production data in the area has been verified, but does not show any correlation with our data. Since the timing of the anomaly appeared to be correlated with the drilling of the ZRP-3 well at a distance of 4.5 km, we reviewed the detailed drilling report provided by NAM.

Drilling started on May 23 (2015) and the reservoir was reached by drilling in the Ten Boer clay on July 13 (Fig. 3a). Downhole drilling mud losses occurred on July 18 and the early hours of July 19. Deeper drilling took place for limited periods on single days between July 23 and August 21 when the maximum depth of 3284 m was reached. GWC depth was reached on July 31 and Carboniferous Shale was drilled on August 11. The cementing of the borehole took place on August 28 and 29 and the borehole was left on August 30 after the cement hardened.

The first conclusion is that there is no correlation between our observations and actual drill intervals. First, the drilling and coring periods were scattered over time, while our observations show a trend over a month of generally decreasing travel times (Fig. 3a). Second, borehole noise would affect travel times between all geophone pairs, but this is not observed (Supplementary Material Fig. S3). Thus, drilling noise cannot explain the observations. We also considered downhole losses that occurred while drilling in Ten Boer clay. However, these downhole losses began 30 hours after the anomalous observations began.

A more likely cause is pore pressure variations caused by drilling. NAM provided us with downhole static pressure data ((BHP_s)), calculated from the depth of the borehole (h), the density of the drilling mud ((rho _m)) and gravitational acceleration (g): (BHP_s=rho _m gh). These data are shown in Figure 3a. Note that (BHP_s) represents only a portion of the total downhole pressure at the wellhead (BHP) because the effects of ram pressure are not included. Rapid decrease of (BHP_s), for example between July 19 and 23 related to mud losses, will have been dynamically compensated by the circulation of the drilling fluid to stabilize BHP. From July 23 to August 19, when drilling depths increased from 2919 to 3267 m, there is a gradual increase in (BHP_s) from 36 to 39 MPa, as indicated by the dotted blue box in Figure 3a. The progressive trend of increasing BHP is anti-correlated with the decrease in travel time of the P wave from geophone 8 to 10 and correlated with the increase in the PS delay time at geophone 10 (Fig. 3a,b ).

Assuming that our anomalous observations at SDM-1 are related to downhole pressure (BHP) at ZRP-3, it is likely that they are related by changes in pore pressure. An elevation of the GWC of (sim) 23 m would correspond to an increase in the pore pressure in the aquifer part of the sandstone of (sim) 0.23MPa ((Delta P = rho _w g Delta h)). By relating the start and end times of the drilling operations in the reservoir to the start and end times of our anomalous observations, we calculated the time it took for the pressure front to propagate from ZRP-3 to SDM- 1. Drilling in the Ten Boer Clay took place on July 13, between 7:45 a.m. and 5:00 p.m., while anomalous seismic observations began on July 17 at (sim) 00:00 (Fig. 4a). This gives a delay of 3 days and 7 to 16 hours. A similar calculation can be made for the end of the abnormal period. The hardening of ZRP-3 cement took place on August 30 (00:00-07:30). After the cement hardened, the well was sealed and there was no further influence from drilling operations. Combining this with the end of the anomaly at SDM-1 on September 2 at 7:00 p.m. (Fig. 4b) gives a delay of 3 days and 11.5 at 7:00 p.m.

### Diffusion of pore pressure

From our seismic observations and their correlation with downhole pressure at ZRP-3, it is inferred that variations in pore pressure may have caused changes in the GWC level in SDM-1. Next, it should be verified that the pore pressure diffusion process can explain the delay between the drilling of the reservoir at ZRP-3 and the GWC response at SDM-1 at 4.5 km distance.

In case of isotropic and spherical diffusion, the hydraulic diffusivity (D) associated with pore pressure diffusion in a fluid-carrying porous medium can be estimated from time (you) it takes the pressure front to reach a certain distance (r)27

begin{aligned} r = sqrt{4 pi D t}. end{aligned}

(1)

The pore pressure diffusivity is estimated from the propagation time of the pressure front, given the time frames of 3 days and 7–16 h (beginning) and 3 days and 11.5–19 h (ending). The greatest times (3 days and 7 p.m.) and the smallest (3 days and 7 a.m.) give diffusivities of 4.9 m(^2)/s and 5.7m(^2)/s, respectively.

An independent estimate of the hydraulic diffusivity (D) can be calculated from material properties, including average porosity (0.1528) and permeability (120 mD29) with more details provided in the “Method” section. We find a diffusivity of the pore pressure of 3.9 m(^2)/s, which is similar to, although somewhat smaller than, the previously estimated diffusivity range of 4.9–5.7 m(^2)/s. For this diffusivity range, permeabilities of 151 to 176 mD are required, somewhat higher than our adopted value of 120 mD, but within the wide range of 1 to 1000 mD measured for the Groningen gas reservoir30. Thus, it is concluded that the diffusion of the pore pressure in the aquifer part of the reservoir can explain the delay between the overpressure at ZRP-3 caused by the drilling and the change of the GWC at SDM-1.

Groningen Reservoir is heavily faulted, and faults can either act as barriers or effective conduits of pore pressure depending on direction: permeability is generally high in the damaged zone parallel to the fault and low across the fault31. The NAM fault map for the top of the reservoir (Fig. 5a) shows a fault with an offset of approximately 150 m midway between SDM-1 and ZRP-3 separating two compartments of the reservoir with the pits on either side and d other (Fig. 5b). This defect is likely to hinder the direct diffusion of the pore pressure through the gaseous parts between the two compartments. On the other hand, the (sim) Change in level of 20 m from the level of the GWC at 4.5 km distance from the drilling site and the high diffusivity ((sim) 5 meters(^2)/s) suggest a high permeability conduit between the two locations. The NAM fault map does not show a connecting fault, although speculatively there are two ENE-WSW trending fault segments that could be related to the bottom of the reservoir (Fig. 5a).

It is important to know that SDM-1 is an open top and bottom and perforated well at reservoir depths between 2965 and 2995 m. Loading a column of high-density brine inside the well prevents reservoir gas from flowing inward through the perforations. Since the well is an open system, it is sensitive to variations in hydrostatic pressure in the reservoir. Our speculative hypothesis is that the pressure front, caused by overpressure at the far wellhead and propagated through the aquifer portion of the reservoir, reached SDM-1 and moved the brine column upwards, raising the water table. in the well. Following the level change in the perforated well, the GWC in its immediate vicinity was also elevated, which was detected by seismic data.

While we realize that parts of our interpretation are highly speculative, we have been unable to find another plausible explanation. Nevertheless, it seems obvious that the observations are related to distant drilling, an unexpected effect and which may be important for other drilling activities.

## How much will Halloween candy cost you this year?

The National Retail Federation expects Americans to spend \$3 billion this year on Halloween candy. …