Model Comparisons Page 43

PWRI June 26, 2001

Model Comparisons Page 43

Model Comparisons

Hydraulic Fracturing in Produced Water Reinjection

SUMMARY

Reinjection of produced water has become a viable method for disposal, for support and for drive. Characteristic elements of these injection operations include long-term injection with consequent stress changes due to poro- and thermoelastic effects. Dilute concentrations of entrained particles in the produced water add another level of complexity. These micron-sized particles can plug the formation during matrix injection. During injection above fracturing pressure, these fines and carried-over oil will alter the near-fracture permeability, will afford development of external filter cake on the fracture faces and can plug the fracture tips or reduce the fracture conductivity itself. Successful produced water injection operations usually entail the intentional or unintentional development of hydraulic fractures. Success is measured on an economic basis and, as such, economic planning and performance evaluation require reliable predictions of fracture geometries and the capacity of fractures to accommodate fluid. The basic mechanisms for fracture growth during produced water injection, available in the public domain, are summarized. Hydraulic fracturing for, or as a result of, produced water reinjection is compared with hydraulic fracturing for stimulation. Finally, various public domain models for designing and evaluating produced water hydraulic fracturing are briefly summarized.

INTRODUCTION

Hydraulic fracturing simulators for stimulation have evolved substantially. Recently, some effort has been devoted to modeling fracturing processes that occur during flooding and disposal. For example, in maturing, water-drive oil fields, progressively increasing volumes of oily water are produced and must be disposed of. Reinjection is one disposal protocol that can be cost effective and environmentally attractive.[1] Declining well injectivity, often due to particles in the injected water, is one of the major factors in increasing costs of reinjection operations. In order to maintain injectivity, it is commonly necessary to inject above fracturing pressure. Economic forecasting is contingent on the fracture geometries that are created. The intent of this paper is to indicate some of the key differences between hydraulic fracturing for stimulation and hydraulic fracturing as a means for and a consequence of injecting produced water, as are currently available in the public domain. Also, the public domain methodologies for assessing fracture geometry and pressure during produced water injection will be summarized.

It can be surprising to realize the potential reduction in injectivity that can result from pumping dilute concentrations of small solids and oil. Wennberg, 1998,[2] described injection into unfractured, gravel packed injectors in an unconsolidated sand in the Gulf of Mexico (Figure 1). Despite high native permeability, initial injectivity was low and repeated stimulation treatments were performed. After each stimulation, injectivity increased dramatically but then declined progressively more rapidly. The half-life of some of these wells was approximately 50 days. This means that within 50 days, the injectivity had decreased by 50% - economically unsatisfactory. It was concluded that fines were the culprits. The injected seawater was deoxygenated, filtered to at least five microns and treated for bacteria as well as inhibited for scale. The solids content in the seawater at the wellhead ranged from less than 1 to 7 ppm. None of the particles was larger than 4 microns and the average diameter was 2 to 3 microns. Available models for understanding injectivity, even for such radial flow scenarios, are inadequate. Modeling of hydraulic fractures resulting from injection is also difficult. van der Zwaag and Øyno, 1996,[3] provided a field case that highlights the currently increasing perception that almost all successful injectors are knowingly or unknowingly hydraulically fractured. They described injection trials in the Ula field where the purpose of the injection was to supplement weak reservoir support. Seawater and seawater-produced water mixtures have been pumped. At the time of their publication they indicated rates of 200,000 BLPD into seven injectors. Additional information has been provided by Svendsen et al., 1991.[4] Typical injection water, reservoir and completion properties are provided in Table 1.

Figure 1. Injection decline for Well A09 (matrix injection, unconsolidated, Gulf of Mexico). (from Wennberg, 19982).

Table 1. Typical Injection Water Properties3

Property / Seawater / 50%SW:
50% PW
Total Suspended Solids, TSS (mg/l), including oil droplets / 0.6 - 13.0
Suspended Solids (mg/l) / 0.1-4.6 / 2.6-4.6
Mean Particle Diameter (microns) / 3.0 / N/A
Density (kg/m3) / 1023 / 1035
Viscosity (cP) / 1.011 / 1.116
Reservoir Height (feet) / 293
Hole radius (inches) / 4.25
re/rw / 1800
Perforations (spf) / 4
Perforation diameter (inches) / 0.5
Reservoir Permeability (md) / 173

Various injection scenarios were evaluated and it was eventually discerned that the only reason that injectivity had been maintained was because the reservoir had been thermally fractured. After 45 days of injection "fractures with 2.2 m full height and 20 m to 34 m half length were measured."3 There is some controversy over these dimensions. This is addressed by van den Hoek et al, 2000.[5]

DIFFERENCES

Table 1 suggests some of the differences between hydraulic fracturing for stimulation and hydraulic fracturing during water injection. Settari and Warren, 1994,[6] described modeling of waterflood-induced fractures and the features that distinguish this process from conventional hydraulic fracturing. First, there are basic philosophical differences. In produced water reinjection or waterflooding, injectivity can be maintained if fracturing occurs. However, the engineer must consider more than the immediate impact of stimulation. Production economics are an essential consideration. A fracture alters the displacement pattern and can potentially decrease (or increase) recovery. There are significant differences in the time scale of the operations and the injected fluid viscosity. In water injection, the efficiency can be close to zero. "As a result, waterflood fracturing is leak-off dominated as opposed to stimulation fracturing which is leak-off controlled."6 Settari and Warren suggested that the following factors might be important for produced water reinjection or waterflooding situations.

Ø  Significant pressure and saturation gradients may exist around the well from previous field injection or production. Reservoir properties may not be constant.

Ø  There can be large-scale reservoir heterogeneity and consequently leakoff variation.

Ø  Other offset producers or injectors will affect fracture propagation.

Ø  There can be thermally altered stresses, and changes in the fluid properties.

Ø  The average reservoir pressure can change during the time-scale of the injection operations.

Some of the relevant differences are described below. It is essential to recognize that an important difference between hydraulic fracturing for stimulation and hydraulic fracturing during waterflooding is that, during stimulation activities, the fracture will propagate (much) faster than the leakoff fluid front.

Mechanical Properties

As in stimulation scenarios, fracture length is strongly dependent on the leakoff. "However, net pressure in the fracture is affected by KC, fracture friction and the factors controlling fracture containment (height) such as the confining stress profile with depth and modulus contrast in the same manner as in conventional fracturing. Therefore, once the pfoc is fixed, the mechanical properties do not change the fracture length match. However, they determine the net pressure which is evident during pressure fall-off test (PFOT) analysis."6 The preceding is not necessarily true. Modulus strongly impacts thermal stresses. Also, Gheissary et al., 1998,54 and other researchers have established that, particularly for layered formations, different mechanical properties can yield entirely different fracture geometries for the same pressure.

Rate

Many stimulation hydraulic fracturing treatments are performed at injection rates ranging from 10 to 50 bpm. A paradigm shift is necessary when thinking of produced water reinjection - consider rates/volumes in terms of barrels per day rather than barrels per minute. Rates may be up to multiple tens of thousands of barrels per day (10,000 BLPD ~6.9 bpm). This implies that while the rates may be similar, the volumes injected and the time scale of the operations can be quite different. In terms of rates, van den Hoek et al, 2000,5 cite rates for injection in Oman of 15,000 to 20,000 m3/day (~65 to 87 bpm). Also consider the potential for periodic shutdowns that are inevitable in any long-term operation and the fact that the injection rate may vary in accordance with meeting voidage requirements. Potentially, injection rates may also be lower than for some typical stimulation operations, target formations may have very high permeability, and the injected fluid viscosity will be low, leading to low efficiency fracturing operations.

Time Scale

Most hydraulic fracturing stimulation operations are completed in a matter of hours. A few, large, specialty stimulations have injected large volumes at high rates. Produced water reinjection and waterflooding are ongoing operations that can last for years. The consequences include substantial possibilities for poroelastic and thermoelastic stress field alterations and interaction with remote producers and injectors. Figure 2 shows a typical field situation (from Detienne et al., 1995).47

Fluids

Stimulation treatments with no polymer in the base fluid are rare. Even slickwater treatments will have small polymer loading to minimize tubular friction. Produced water typically has dilute concentrations of solid particulate matter, droplets of oil (since this water has come from the production stream), and carried-over production chemicals. Many operators no longer do extensive filtering on injection water as it is anticipated that hydraulic fracturing will occur and that fractures will be able to accommodate the particulate material. The particles can be organic (bacteria, plankton, etc.) or inorganic (e.g., clay minerals, quartz, amorphous silica, feldspar, mica, carbonates, etc.). Additives to produced water injection streams will characteristically include biocides, scale inhibitors and sometimes drag reducers although the price of the latter can sometimes be prohibitive. The viscosity of heated water represents the viscosity, since the re-injected water will likely be hot. The reverse will be true if seawater is injected.

Figure 2. Events during a cold water injection program showing thermal fracturing (after Detienne et al., 199547).

Thermo- and Poroelastic Effects

Water injection is a long-term, low viscosity operation. There can be significant changes in the total stresses due to reservoir cooling (seawater), reservoir heating (possibly produced water) and pore pressure changes with the substantial injection volumes. Perkins and Gonzalez, 1984,[7] 1985,[8] provided a view of stress alteration due to cold water injection. "During ordinary hydraulic fracturing operations … leakoff is controlled so that injected fluid volumes will be minimized. As a result, pressure and temperature changes in the rock surrounding the fracture do not ordinarily have a very significant effect on the fracturing operation. Therefore, the primary concern has been the effect that temperature has on fracturing fluid rheology and leakoff behavior."8

"... in some cases injection of cold fluid can significantly reduce tangential earth stresses around an injection well. It follows that vertical hydraulic fractures can be initiated and propagated at lower pressures than would be expected for hydraulic fracturing of a nearby producing well. The injection well fracture, however, would tend to be confined to the low stress region that lies within the flooded zone surrounding the injection well. If the injection rate is sufficiently high, or if injected solids plug the face of the fracture, then the pressure within the fracture could rise, thus permitting the fracture to extend beyond the confines of the cooled region. After breakout, the fracture extension pressure should approach (and probably exceed because of the increased pressure field surrounding an injection well) the fracture extension pressure of nearby producing wells. The thermoelastic effect could have significant impact on fracture confinement at bounding zones. For injection wells, impermeable layers could confine fractures in vertical extent partly because the impermeable layers have not been cooled as much as the pay zone."8

Similar considerations apply to competing poroelastic effects. Detailed considerations of poroelastic calculations are available in the literature (for example, Detournay et al., 1989[9]). Their real significance may be in produced water reinjection. Stevens et al., 2000,[10] gave examples specifically relevant to produced water reinjection. "Cooling is principally due to convection, and since the rock heat capacity per unit reservoir volume is approximately twice that of the water, the thermal front advances at about one-third the rate of the water saturation front." These are two competing phenomena. Thermal changes in viscosity are also a factor.

Plugging

It is known that the solids in produced water can be injected. The literature indicates field operations where several fracture volume equivalents of solids contained in the injected water have been successfully pumped.[11],[12] In these situations, the fracture volumes were inferred from ancillary testing procedures (hydraulic impedance testing, falloff surveys ...). This brings to mind the dominant question: "Where do the solids go?" van den Hoek et al., 1996,[13] summarized the issue:

"An essential difference with simulation of conventional waterflood fracturing is that owing to fracture fill-up with injected solids the fracture conductivity cannot be assumed infinite any more. This relates to the important PWRI issue of where the injected solids go. Using our model, we show that the pressure drop over a finite conductivity fracture can lead to a significant increase in fracture volume without necessarily leading to a significantly higher pressure. Thus, a picture emerges in which the fracture conductivity 'adjusts' itself in order to accommodate injected solids. This picture allows the computation of well injectivity as a function of total injected water volume, solids loading, etc. This concept can also be used to qualitatively explain the PWRI field observation that injectivity appears to be partially or fully reversible as a function of water quality."13

Wennberg, 1998,2 and Wennberg et al., 1995,[14] presented the most comprehensive evaluation of water injection damage mechanics to date. The formation adjacent to the hydraulic fracture will be damaged due to particulate injection. Various empirical measurements have been made to facilitate representing injectivity decline as a function of injected volumes; particularly for matrix injection. Some of the highlights of these efforts are summarized below.

Donaldson et al, 1977,[15] showed that particles initially pass through the larger openings in a core and are gradually stopped by a combination of sedimentation, direct interception and surface deposition. They found that the larger particles initiate cake formation. Davidson, 1979,[16] found that the velocity required to prevent particle deposition is inversely related to the particle size (for the systems evaluated at least). Core measurements by Todd et al., 1984,[17] showed that the overall damage is related to the mean pore throat size. Cores damaged with aluminum oxide particles (with diameters up to 3 microns) exhibited damage along their entire length and as the particle size increased the damage gradually shifted to the injection end and external cake. Vetter et al, 1987,[18] found that particles with sizes from 0.05 to 7 microns caused damage and that the larger particles caused a rapid permeability decline with a short damaged zone. Permeability reduction with smaller particles was more gradual.