The paper focuses on the sub-class of issues which can affect/dominate the performance of hydraulic fracturing in (gas or oil) reservoirs with higher-permeability, of order lmD (10mD) or more in gas (oil) reservoirs. These issues are placed in the broader context of five major aspects which have been newly extracted from data-analysis over the past five years and which have greatly polarized/reversed industry thinking on hydraulic fracturing: heretofore, primary implications have been for lower-permeability reservoirs, where convection and tortuosity constraints impose a narrow (or non-existent) window for successful design of fracture treatments, especially in light of properly-interpreted higher net fracturing pressures (than conventional industry models) induced by nonlinear rock response to (multiple) fractures. A much wider design window has made successful stimulation of higher-permeability reservoirs relatively easy and economically more rewarding, a result surprising to some. Perhaps more surprising are conclusions that appropriate proppant concentration may vary somewhat inversely with reservoir permeability and that fractures may preferentially grow (vs. initiate) in lower- permeability strata/sections, the main theme of this paper. Fluid rheology also plays a much more important role in (leak-off behavior during) fracturing of high-permeability reservoirs vs. a lower-permeability role limited to near-wellbore (tortuosity) response. Some issues are illustrated with simulations performed by a unique computerized system, which allows credible matching of (treatments and production) data over a wide range of reservoir, fluid and job parameters - thereby allowing establishment of conclusions with case-histories, supported by some careful laboratory experiments. With the aid of this computerized capability, dramatic results have been achieved in reducing job costs and improving production, especially for low- to medium-permeability reservoirs, but many of the implications can be carried over to high-permeability reservoirs.

Conventional thinking has associated hydraulic fracturing with marginally (un)economical or problem wells: examples vary from low-permeability (U.S.) gas reservoirs (Refs. 1-4) to more recent implementation of "frac-packs" for simultaneous solution of sand-control and production problems (Refs. 5-7). However, substantial success has also been reported in the use of fracturing (using proppant or acid) for improving production in wells which are already economical (e.g. Refs. 8, 9): indeed, some (e.g. Ref. 10) have been early in recognizing the retrospectively obvious fact that economical wells may actually allow for (much) greater return on investment (RoI) from fracturing - when RoI is calculated properly to account for accelerated cash flow, flexibility of production schedules (e.g. in variable gas/oil markets) and operating costs; of course, such calculations should also (actuarially) account for operational risks, logistical considerations and potential damage.

The latter consideration of risk has been the main force retarding broader application of hydraulic fracturing, especially in (even marginally) economical wells. The reasons for this have varied from management conservatism to poor understanding of the hydraulic fracturing process itself. Many misunderstandings have been identified and resolved over the recent past few years (e.g. Refs. 11-17). The major purpose of this paper is to delineate the many major issues involved and to clarify the various realms of appropriate application, using case-studies and unique simulation capabilities (Refs. 18-27) developed over the past ten years of extensive laboratory and field applications. The emphasis will be on higher-permeability applications (e.g. Refs. 5-10, 28-30) but space restrictions require that most points be made by extensive references to more general work and explained in the context of overall dominant fracturing issues. As well, unfortunately, presentation will be somewhat inhibited by delayed approval for use of field data and somewhat incomplete laboratory experiments.

Our careful analysis of data over the past few years has isolated at least five dominant issues which had hitherto been essentially unrecognized, although critical for successful design, execution and evaluation of fracturing treatments:

  1. Convective re-arrangement of fluid stages with variable density (e.g. proppant-laden), as described in Ref. 12.

  2. Near-wellbore tortuosity, which limits proper placement of adequate proppant, as delineated in Ref. 16.

  3. High net pressures (e.g. due to nonlinear rock dilatancy), reducing the role of stress barriers (e.g. Ref. 11).

  4. Greatly reduced effects of frac-fluid rheology and flowrate (except for #1, 2, and fluid leak-off in 5.), Ref. 13.

  5. Major permeability/variation effects on fracture growth.

As indicated, all of these mechanisms have been described previously, except #5 - the primary focus of this paper, Recognition of those issues as dominant has come from data- analysis, as against purely speculative theories. Adequate elaboration on each issue would require long and detailed development (including many supporting field data-sets), which must await more comprehensively substantiated work on the subject. This paper will merely summarize the first four issues, as background for #5, attempting only to place some perspective on their various realms of importance.

The first issue (convection, Ref. 12) is dominant in low-permeability reservoirs; the associated increasingly-documented lack of proppant placement opposite pay/perforations explains many disappointing results of fracturing in such environments (where the most "experience" has been gained and the greatest relative reward should be expected): the proppant (or acid) drops out of the pay zone and/or even fails to form a connection back to perforations from which the fracture has emanated. Indeed, the phenomenon has been observed also (e.g. flowback of early non-resin-coated and/or tagged proppant) even in higher-permeability environments, even though convection should be most readily prevented there, by development of a proppant pack in the fracture - which is easy to achieve in high-permeability environments, even without the high concentrations of proppant often assumed necessary.

Ironically, Issue No. 2 (tortuosity) came to attention primarily because of the high proppant concentrations assumed necessary for high-permeability environments, but also because of the deviated wells associated with many such environments (e.g. offshore and/or arctic conditions, Refs. 8, 9). However, careful collection and evaluation of adequate data-sets from many different environments (e.g. Ref. 16) have shown that tortuosity can be present in very many situations and is especially harmful in lower-permeability environments, where convection forbids the use of excess fluid volumes - common usage of which has hitherto obscured the common presence of tortuosity. Taken together, Issues 1 and 2 delineate required "design windows" - outside of which job success is unlikely.

When jobs have been successful, it is most likely a result of modified "envelope boundaries", which create/widen design windows: thus, for low-permeability environments, either tortuosity must be removed (e.g. Refs. 16,17) or the fracture must be contained in the pay-zone. However, the presumption of containment is rendered questionable by Issue No. 3 - which requires substantially higher stress barriers for containment than previously thought: for instance, containment should not be expected at depths substantially less than 10,000 ft., unless other mechanisms intercede (e.g. Ref. 19, 25) - such as slippage at very shallow interfaces.

No reasonable manipulation of job execution parameters (rheology and/or flow-rate, Issue No. 4) can greatly affect the resulting large-scale fracturing geometry in low-permeability reservoirs: only the placement of proppant can be affected and that can be achieved only by opening a window between obstacles imposed by convection and tortuosity; a high enough total overall average proppant concentration must be allowed (by reducing tortuosity, if necessary, Ref. 16) that substantial convection is not mechanically possible on the time frame to fracture closure (of order hours for lower-permeability vs. minutes for higher-permeability). The practical result is that high proppant concentrations must be used to properly place proppant in low-permeability reservoirs: - this is in complete apposition to typical theoretical "design" concentrations required only to create adequate conductivity needed to carry lower production rates from such reservoirs.

A similar apparent contradiction (reality vs. theory) applies to the high-permeabilitv application: high proppant concentrations are suggested by the high required conductivity but, in fact, low concentrations in the fluid pumped may actually lead to better/easier placement of successful fractures, certainly in thick pay-zones accessed by deviated wellbores. The reason for this is that the slurry dehydrates much faster in higher- permeability environments, thus becoming immobile and even preventing other proppant from convecting. Thus, even if tortuosity is not removed (e.g. using techniques in Ref. 16), use of a lower concentration may still allow achievement of a successful pack back to the perforations, with adequate resulting width and proppant conductivity. Only if specific problems exist (e.g. Ref. 29) must reasonable care be taken to achieve more precise proppant placement. It is then necessary to actually understand the physics of fracture growth in such environments of (extreme) variation in permeability/leak-off.

This latter realization (with emphasis on the role of Issues No. 4, 5) will be the main theme of the current paper. The development will proceed from fundamental physical arguments (Section 3) to field case-studies (Sections 4,5), ending with conclusive support from laboratory data (Section 6).

It is generally understood that increasing permeability implies increasing leak-off from the fracture, hence reducing fraction of pumped volume remaining (i.e. reduced efficiency). Indeed, the earliest Carter-type design models (e.g. Ref. 1) were based solely on a consideration of leak-off, with mass balance (for an assumed fracture height and width) determining the created fracture length. Despite the evolution/proliferation of many models (e.g. see review and criticisms in Ref. 31), purporting to calculate height and width, most current industry models do not greatly improve on such Carter models, even for calculating the fracture dimensions: this is especially true in the context of our present topic, i.e. when the permeability (variation) plays a dominant role in the growth of the fracture.

The problem is exacerbated by "practical" approaches which assign a "leak-off coefficient" to the whole (or part of) a fracture treatment, implicitly assuming that this is relatively constant for the reservoir conditions and treatment fluids in use. In fact this leak-off (which can be directly related to permeability and treatment/reservoir fluid properties, for any particular unit area with constant permeability, e.g. as in Refs. 20-21) varies as the fracture grows across typical strata/lense geometries with varying permeabilities. Although common interpretation techniques (e.g. Nolte/Smith) allow, in principle, for some variations (e.g. resulting fracture efficiency), they cannot handle most realistic/representative reservoir geometries.

Some models do, implicitly, allow for variation of leak-off as fractures cross stratification/variations: however, they still miss essential features which have taken us many years to characterize: not alone does the varying permeability affect leak-off, it also affects the growth/shape of the fracture. For instance, even the most "advanced" of the current industry models might calculate a circular fracture shape (e.g., when there are no stress or modulus variations); they may even correctly compute the variation of leak-off with permeability over any fracture face; but they all fail to capture the actual (e.g non-circular) shape which the permeability variation induces. Such model departure from reality becomes increasingly important with increase of the largest permeability (i.e. .associated leak-off coefficient) that the fracture encounters.

Extreme examples may best illustrate the issue: Figure 1(a) indicates a fracture initiating in a lower-permeability stratum, encountering high stress and higher-permeability strata, respectively, as it grows upward and downward. If the contrast in leak-off coefficient of the adjacent strata were increased, the fracture would grow increasingly elliptical until eventually it could be perfectly contained within the low-permeability stratum. (See Section 5, 6 later). Of course, for any given permeability and reservoir/injection fluid properties, continuation of pumping, for a long enough period, would eventually cause it to grow far enough into the permeable strata that most/all of the injected fluid would be lost before reaching the outward-growing perimeter section in the lower permeability zone: the fracture acts as a time/space adjustable/regulating valve which may be used to control the distribution of fluid (e.g. by ceasing to pump until the "valve" closes, and then re-cycling the injection as desired). Such a possibility may be attractive to those who wish to preferentially flood the lower-permeability sections of their reservoir without breaking/watering out into high-permeability sections (e.g see water-injection case-studies in Section 5).

Figure 1(a)

Illustraiioo of (mini-) fracture growing in multi-layer zone, with (partial) permeability barrier below 15620 ft. (see also Fig.2(a))

Figure 1(a)

Illustraiioo of (mini-) fracture growing in multi-layer zone, with (partial) permeability barrier below 15620 ft. (see also Fig.2(a))

Close modal

This prior observation may also be appealing to those who have long sought reasons for their claims/convictions that fractures remain contained in the target pay-zone: the presence of high-permeability streaks offer the potential for such containment, if present (and if properly exploited). Unfortunately, the foregoing mechanism is rarely allowed the opportunity (to operate) by typical job designs, which involve large pad fluid volume injections before proppant could arrive and possibly dehydrate at such streaks. Indeed, a strongly negative potential result of such large fluid/proppant volume ratios (e.g. imposed by fear of screen-out) is shown in Figure 1(b), where the converse of a lower "permeability barrier" exists: preferred downward growth of the fracture then allows the proppant to drop away from the pay-zone and, worse still, it may even lose connection with the perforations.

Figure 1(b)

Preferential downward fracture growth and optimistic proppant placement below the higher permeability zone (around 16430 ft).

Figure 1(b)

Preferential downward fracture growth and optimistic proppant placement below the higher permeability zone (around 16430 ft).

Close modal

This tendency of fractures to grow in lower-permeabilitv strata has important implications, even for fracturing of high- permeability (limited) strata: adjacent lower-permeability sections may be preferentially fractured and accept most of the proppant, unless an aggressive design schedule (i.e. minimum fluid/proppant volume ratio) is adopted. This is one case where most/all of the problems associated with fracturing of low-permeability pay zones are present - as illustrated also in Figure 1(b), adapted from an actual job executed recently, having failed to acquire (one major partner’s) agreement for procedures needed to achieve pack-off (Refs. 16, 17).

However, in general, the successful fracturing of higher- permeability (thick! reservoirs pay-zones is not as difficult as for lower-permeabilitv reservoirs: this may explain the success of most (including poorly rationalized) designs. Indeed, the higher-permeability environment generally even helps to (favorably) control fracture growth profiles: an example of this is the success achieved in staying out of "bottom water" (e.g. Ref. 29), even when designs are developed on the basis of simplistic models, with some associated simulations which are demonstrated to be seriously in error (by the discrepancy between pressure data and simulation predictions, honestly presented as such in the some of these papers). We will use an accurate consistent simulator (commercially referred to as FRACPRO) to illustrate some of these issues for just a few case-studies, in succeeding sections: although this simulator is being continually improved, it does match most (treatment and production) data very well, with realistic assumptions.

To illustrate the last paragraph of the last section, along with a number of the other issues, we begin with an example of a job conducted in a deep German gas reservoir (already conveniently described in more detail in Ref. 15). The minifrac (Figures 1(a), 2(a)) was conducted with a downhole pressure gauge present, allowing detailed interpretation of the pressure response (which can only be estimated during the main fracture, as shown in Figure 2(b)). In particular, as well, some flow-rate changes were performed, to allow for the measurement of tortuosity and consequent better evaluation of the true net fracturing pressure: even when downhole gauges (and/or dead-strings) have been available, investigators rarely perform such flow-rate variations - certainly never as required to accurately characterize tortuosity (Ref. 16). Indeed, even this job had inadequate data for such characterization, but it nevertheless serves to illustrate the intense variation of tortuosity in the early stages of injection: this may be seen by comparing the measured data to on-site (early versions of) rpodel calculations, as in Figure 2(a), showing the kind of discrepancy - due to tortuosity variation but not removal - that we have seen on many other jobs.

Figure 2(a)

Match of mini-frac data: early disparity is due to model deviation from highly intense near-wellbore torrtuosity variation.

Figure 2(a)

Match of mini-frac data: early disparity is due to model deviation from highly intense near-wellbore torrtuosity variation.

Close modal
Figure 2(b)

Sample match with (treating) pressure data for main fracture, following mini-frac of Figs. (1a. 2a); pack-off is a mirage.

Figure 2(b)

Sample match with (treating) pressure data for main fracture, following mini-frac of Figs. (1a. 2a); pack-off is a mirage.

Close modal

However, the more important point illustrated by this job may be observed by comparing the pressure fall-off at the end of the mini-frac with that found at the end of the main-frac (which is all too often obscured by flowback on other jobs): the efficiency of the main treatment (16%) is, (untypically), substantially lower than that for the mini-frac (40%). The most reasonable explanation (with later resulting excellent match of all the data) may be achieved by modelling the reservoir layers with increasing downward permeability, consistent with log/core data, as shown schematically in Figure 2(c) - where an associated (on-site) very optimistic estimate for the created propped fracture geometry is also shown.

Figure 2(c)

Optimistic proppant distribution for pressure match in Fig. 2(b): more realistic profile would be density-stratified (e.g. Ref. 12).

Figure 2(c)

Optimistic proppant distribution for pressure match in Fig. 2(b): more realistic profile would be density-stratified (e.g. Ref. 12).

Close modal

In fact, the downward - increasing permeability is considered to have been (mainly) responsible for the success of this job, by preventing excessive downward growth, and associated resulting (excessive) detrimental convection - which was previously illustrated in Figure 1(b), for instance. Although the size of the pad for this job was reduced by 50% (from the original service-company design), after on-site analysis of the mini-frac, there was still too much fluid present to achieve the desirable pack-off, which is required for most successful fractures: this was so, ironically, even though the pad which proper mini-frac interpretation would have indicated for packoff would be less than the pad which would have actually achieved pack-off. For example, a design with 50% further pad reduction is shown in Figure 2(d); this produces a very similar hydraulic geometry but (more realistic proppant distribution with) better pack than in Figure 2(c). Such major discrepancies between reality and most current software calculations (e.g. Ref. 31) still forces too many compromises in the field.

Figure 2(d)

Slightly more aggressive design to achieve something like Fig 2(c): more aggression has required proppant slugs (Ref. 16).

Figure 2(d)

Slightly more aggressive design to achieve something like Fig 2(c): more aggression has required proppant slugs (Ref. 16).

Close modal

Even more informative, however, is the "apparent pack-off" which seems to occur at end of the main job (Fig. 2b): this seems to be at odds with the statements above and with the model calculations, which indicate little pack-off, even when the actual pressure fall-off (i.e. permeability profile) has been matched. In fact, this rapid pressure increase is not a pack-off but rather comes from near-wellbore (tortuosity) behavior; the discontinuous changes of the slope correspond closely with the arrival of the various proppant concentration steps at the perforations. Although large volumes of fluid have been pumped, and despite the relatively small deviation (15°) of the wellbore (Ref. 15), tortuosity remains and exhibits itself at relatively low concentrations - compared to what we have pumped with much less preceding fluid volumes, using techniques in Ref. 16. Indeed, tortuosity can be found even in vertical wells (Ref. 16), certainly when small enough fluid volumes precede proppant (i.e. to prevent convection, Ref. 12); but tortuosity has mainly been noticed on deviated wells.

The latter issue (tortuosity vs. pack-off) is one of the most common sources of misunderstanding, especially in fracturing of deviated wellbores - even in high-permeability reservoirs. It is quite easy to identify such near-wellbore responses (see also Ref. 30), even among jobs which are specifically designed to be "frac-packs" (e.g. Refs. 6, 7). In any case, the so-called design of "tip screen-outs" is certainly a misnomer: the simultaneous blocking of growth at all parts of the perimeter is, in fact, not achievable. However, even though it may represent the difference between success and failure of such jobs, sometimes it may also not matter greatly whether or not such observed responses are really dominantly near-wellbore; such an example is illustrated in Figures 3(a, b), which shows one FRACPRO simulation of a job already described in Ref. 5: although the upward growth of the fracture is likely to be even less (and although Ref. 5 shows a simple circular calculation of the fracture, as reproduced approximately by FRACPRO in Figures 3 (c, d)), it is easy to show that the production difference is negligible: this is due to the "choke" effect created by the limited fracture conductivity available to carry the high available flow rates (as the simplified flow/reservoir model in Ref. 5 may even be used to show). Ultimately, the true test of job performance (and relevance) may be found in such production data. Here, again, higher- permeability reservoirs are typically much more tolerant of sub-optimal jobs and/or lend themselves (i.e. are less sensitive) to the numerous interpretations of created fracture geometry. Indeed, even lower-permeability reservoirs are often argued to have successful fractures (e.g. based on initial pre-frac/damaged production and/or downward revisions of kH). However, eventually, the true results reveal themselves, e.g. in Fetkovich-type cumulative production plots, which often display the errors in simplified type curve analysis - e.g. based on limited (2D) models of kH, or even on complicated 3-D (even multi-phase) models which make extremely non-unique matches to a limited (early) sample of production/test data.

Figure 3(a)

Sample match of data from job in Ref. 5: minor issue of friction removal (offset gage) affects interpretation, not overall result.

Figure 3(a)

Sample match of data from job in Ref. 5: minor issue of friction removal (offset gage) affects interpretation, not overall result.

Close modal
Figure 3(b)

Fracture geometry for match in Fig. 3(a); consistent interpretation (with less height growth) is possible depending on rheology.

Figure 3(b)

Fracture geometry for match in Fig. 3(a); consistent interpretation (with less height growth) is possible depending on rheology.

Close modal
Figure 3(c)

Match for Fig. 3(a) with "conventional" model, (e.g Ref. 31). ignoring permeability barriers: note (semi-) circular upward growth.

Figure 3(c)

Match for Fig. 3(a) with "conventional" model, (e.g Ref. 31). ignoring permeability barriers: note (semi-) circular upward growth.

Close modal
Figure 3(d)

Fracture geometry for Fig3(c); proppant model is "unconventional" (Ref. 12). distinct from rheology-dependent growth model.

Figure 3(d)

Fracture geometry for Fig3(c); proppant model is "unconventional" (Ref. 12). distinct from rheology-dependent growth model.

Close modal

One of the major problems with the many field data-sets, including those described (or referred to) in the previous section, is that the complexity of fluids (e.g. Ref. 28) and other job conditions (e.g. limited downhole pressure data, when available, and poor use of data/models) prevent a conclusive determination of the actual physical processes involved.

The clearest and best data-sets available to us for evaluation of the role of permeability variation in relatively high-permeability reservoirs have been those for sea-water injection in North Sea wells (e.g. Ref. 32). Indeed, the observations of North Sea operators on such wells were primarily responsible for developing many unique modelling features, such as "permeability barriers" in the FRACPRO simulator, which was developed originally for lower-permeability applications.

Figure 4(a) shows a data-set collected on one of the Statoil-operated wells referred to in Ref. 32; a great advantage was availability of a permanent downhole gauge - so that all predictions can be checked continuously, throughout the life of the well. Figure 4(b) shows a match of the data, achieved with the model parameters set to values which have also been used to match numerous other data-sets and laboratory results.

Figure 4(a)

Sample injection and downhole pressure gage data (BHP) on North Sea sea-water injection well (e.g. as described in Ref. 32).

Figure 4(a)

Sample injection and downhole pressure gage data (BHP) on North Sea sea-water injection well (e.g. as described in Ref. 32).

Close modal
Figure 4(b)

Matching of "observed" net pressure data (calculated from BHP in Fig. 4(a)): near-wellbore model leaves extreme tortuosity.

Figure 4(b)

Matching of "observed" net pressure data (calculated from BHP in Fig. 4(a)): near-wellbore model leaves extreme tortuosity.

Close modal

The resulting fracture dimensions are shown in Figure 4(c, d): the former shows the fracture acting as a "valve", i.e. fracture dimensions varying with time during injection and shut-in cycles - while the latter shows the (deduced) reservoir/fracture geometry in which this process/measurement is occurring. Many other informative aspects of the data may be pointed out; primary among these is the very clear variation of nearwellbore tortuosity as the injection continues: this can be deduced by observing the dramatic reduction of the nearwellbore friction (Δp/ΔQ) on the downhole gauge, at each successive shut-in, with increasing volumes injected. This has been observed by us on very many wells, but these volumes are the greatest that we have found to be required in order to reduce the tortuosity to "steady-state" levels; however this is not surprising (Ref. 16) in light of the inviscid fluid employed and the deviation of the wellbore (65°).

Figure 4(c)

Fracture dimensions computed for pressure match in Fig. 4(b): also analyzed longer injections (and fractures) for other wells.

Figure 4(c)

Fracture dimensions computed for pressure match in Fig. 4(b): also analyzed longer injections (and fractures) for other wells.

Close modal
Figure 4(d)

Schematic of reservoir and fracture growth for dimensions in Fig. 4(c): note upper higher permeability zone acting as barrier.

Figure 4(d)

Schematic of reservoir and fracture growth for dimensions in Fig. 4(c): note upper higher permeability zone acting as barrier.

Close modal

Even more informative were the consistent "PLT" plots available during injection on this and other (e.g. Saga-operated) wells: these showed that the fluid injection profiles go from relatively uniform dispersion over the perforated interval to dominant injection into the upper perforations, thereby reducing the number of fractures (and tortuosity). Although initially interpreted as upward fracture migration, our analysis showed that the fracture had not yet grown up into the high- permeability zone. Of course, if water injection is continued for long enough, the fracture finally will/does grow upward (e.g. as confirmed later); therefore, our major task (next section) has been to carefully quantify the associated "time-constant" of this "valve" process, in order to provide operations with quantified options for such procedures.

The realization that definitive field measurements and deductions about fracture geometry would be long delayed, by practical considerations and technology limitations, has led us to conduct detailed laboratory experiments over the past fifteen years (Refs. 14, 23-27), with special recent emphasis on the role of permeability variation on the growth of hydraulic fractures. Overall descriptions and final details of the latter work may be found in Ref. 27; only some representative examples will be included here (Figures 5(a-c)), to illustrate the major point of the paper.

Figure 5(a)

Particular laboratory test result for "transition" region of (higher) permeability contrast with lower-permeability injection zone.

Figure 5(a)

Particular laboratory test result for "transition" region of (higher) permeability contrast with lower-permeability injection zone.

Close modal
Figure 5(b)

Sample scaling to field (approx lmD gas. 10mD oil at 1 cp.) of Fig 5(a), showing comparison to simulator used in Figures 1-4.

Figure 5(b)

Sample scaling to field (approx lmD gas. 10mD oil at 1 cp.) of Fig 5(a), showing comparison to simulator used in Figures 1-4.

Close modal
Figure 5(c)

Lab recall for higher permeability barrier effects (scaled approx, to 10mD gas. 100mD oil at 1 cp. in higher permeability zone).

Figure 5(c)

Lab recall for higher permeability barrier effects (scaled approx, to 10mD gas. 100mD oil at 1 cp. in higher permeability zone).

Close modal

The particular apparatus used (Figure 5(d)) has been dubbed DISLASH (for Desktop Interface Separation Laboratory Apparatus for the Simulation of HydraFrac). It has been shown to be an accurate physical model of the dominant field processes, except for nonlinear rock dilatancy, which we have otherwise simulated: (literally) tens of thousands of experiments, each quick/repeatable and very low cost, have been performed to test numerical models (e.g. Refs. 14, 27).

Figure 5(d)

Schematic of (DISLASH) test apparatus used to acquire data in Figs. 5 (a. c): more details are in Refs. (14. 27).

Figure 5(d)

Schematic of (DISLASH) test apparatus used to acquire data in Figs. 5 (a. c): more details are in Refs. (14. 27).

Close modal

Because of limited space (and ongoing experimentation), we have selected one particular simulation to very simply illustrate the transition from "conventional" model behavior to the actual physics associated with permeability variation. The results are shown in Figures 5(a, b): Figure 5(a) shows the behavior observed when growing fractures in a low-permeability environment (bold curves, see also Ref. 14) vs. particular observations when growing fractures in a layered medium (dashed lines, connecting specific but representative data points); these particular results are then "scaled" appropriately (Ref. 27), in Figure 5(b), to a corresponding field environment and compared with the FRACPRO simulator, which uniquely represents the physics of the process with good accuracy.

The particular choice of parameters in Figures 5(a, b) shows the length growth (in the low-permeability middle layer) closely following radius for the uniform low-permeability environment, while the height growth goes asymptotically to the (uniform) radius growth in the higher-permeability environment, for similar pumping conditions; there is only a slight delay as the fracture encounters the boundary of the "permeability barrier". However, even this apparently simple behavior cannot be captured by other ("conventional") industry models; further, as the permeability (barrier) increases, the delay at the barrier is increasingly stretched out in time (as shown by Figure 5(c) and many other laboratory simulations in Ref. 27, also by many FRACPRO simulations, e.g. Figure 4(c,d) earlier); eventually, a perfect barrier is formed for the whole periods of the injection, for strong enough (high) permeability barriers and/or short enough periods of injection. Hence there is a complex "time-space" valve behavior.

Quantifying the latter valve process has been the focus of much experimental/ modelling effort, now concluding (Ref.27).

A number of primary conclusions may be drawn from the foregoing analysis and review of many other data-sets:

  1. Permeability variation affects the leak-off behavior from a fracture in a complex way; suitably chosen shut-ins can determine this variation of efficiency (and hence deduce the fracture geometry, at least with simple fluids).

  2. The leak-off behavior of fracturing fluids in higher permeability formations may be quite complex; credible evaluations require use of simpler fluids. (Such simpler/inexpensive fluids may also be a better choice for many applications, such as "frac-packs").

  3. The pressure response associated with a pack-off (often misrepresented as physically - unrealistic "tip screen-outs") may also be confused with that associated with nearwellbore tortuosity: resulting failures to correct job design/ execution may lead to (much) less than potential performance. The simplest (often adequate) means of avoiding such near-wellbore screen-out (without proppant slugs, Ref. 16), may be to use low proppant concentration for high permeability applications - where conditions should still yield adequate packs in most cases, provided that potential risks (e.g. water-zones, gas caps) are avoided (e.g. with realistic simulation/design and execution).

  4. Permeability (i.e. leak-off) variation also affects the actual growth of the fracture: fractures prefer to grow in impermeable (i.e. low leak-off) rock. Such "permeability barriers" (besides many other issues, e.g. Ref. 31) must be included in any realistic hydraulic fracturing simulator.

  5. Permeability barriers may play favorable or unfavorable roles, depending on the application: fractures may tend to stay out of bottom-water (or gas-caps), but it may be difficult to drive into (or keep proppant/fracture in) high- permeability pay-zones, e.g. in multi-zone fracturing.

In general, the simulation of fracturing in higher-permeability reservoirs is (even) more complex than corresponding efforts in lower-permeability reservoirs (e.g. Ref. 31, although many errors are demonstrated in conventional industry simulators even for such simpler environments). However, ironically, the higher-permeability environment is much more forgiving of poor job design and execution, i.e. the consequences (e.g. for production) are less severe; this may help to explain the more substantial success of fracturing, compared to realistically assessed lower-permeability environments, where conditions are much less forgiving (e.g. as delineated in Refs. 11-17).

However, it is hoped that resolution of the issues, as delineated in this paper, will contribute to improved understanding (and even increased success) for increasingly-popular application of hydraulic fracturing to higher permeability environments. Indeed, despite (actually because of ) highly unreliable oil and gas prices, it may be logically argued - in a much longer document - that intelligent use of hydraulic fracturing is the single most powerful weapon in the technological and economic environment, which we expect to characterize an efficiency-driven petroleum-production business of the future.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the authors). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A.

The author is grateful to RWE-DEA (Germany) and Statoil (Norway) for permission to publish the data. The development of many concepts and the simulator (FRACPRO) used in the paper was made possible by support from the US Gas Research Institute (GRI) and the paper preparation was greatly assisted by personnel at RES and especially Tim Quinn at MIT. Personnel at these and numerous other companies (e.g. Saga, Norway) are also acknowledged for co-operative interaction and data on these topics: the author hopes that they will (co-)author many individual case histories, including production data to fully verify the reality of the various issues.

Howard
,
C.G.
and
C.R.
Fast
, "
Hydraulic Fracturing
",
SPE Monograph
,
1970
.
Gidley
,
J.L.
,
S.A.
Holditch
,
D.E.
Nierode
and
R.W.
Veatch
, (Editors)
"
Recent Advances in Hydraulic Fracturing
",
SPE Monograph
,
1989
.
Saunders
,
B.F.
,
B.M.
Robinson
,
S.A
Holditch
and
R.E.
Peterson
, "
Hydraulic Fracturing Research in the Frontier Formation Through the Gas Research Institute’s Fourth Staged Field Experiment
",
SPE Paper No. 24854
,
1992
.
Cleary
,
M.P.
, "
The Engineering of Hydraulic Fractures- State of the Art and Technology
’,
Journal of Petroleum Technology,
13
21
,
January
1988
.
Bale
,
A.
,
K.
Owren
and
M.B.
Smith
, "
Propped Fracturing as a Tool for Sand Control and Reservoir Management
",
SPE Paper No. 24992, Europec
,
1992
.
Hannah
,
R.R.
,
E.I.
Park
,
R.E.
Walsh
,
D.A.
Porter
,
J.W.
Black
and
Frank
Waters
, "
A Field Study of a Combination Fracturing/Gravel Packing Completion Technique on the Amberjack, Mississippi Canyon 109 Field
",
SPE Paper No. 26562, Annual Technical Conference
,
1993
.
Jennings
,
A.
, "
Good Wells Make Better Fracturing Candidates
",
SPE Production Engineering
,
371
376
,
November
1991
.
Cleary
,
M.P.
,
C.A.
Wright
, and
T.B.
Wright
, "
Experimental and Modelling Evidence for Major Changes in Hydraulic Fracturing Design and Field Procedures
",
SPE Paper No. 21494
,
1991
.
Cleary
,
M.P.
and
A.
Fonseca
, Jr
, "
Proppant Convection and Encapsulation in Hydraulic Fracturing: Practical Implications of Computer and Laboratory Simulations
",
SPE Paper No. 24825
,
1992
.
Wright
,
T.B.
,
W.W.
Aud
,
C.
Cipolla
,
D.N.
Meehan
,
K.F.
Perry
and
M.P.
Cleary
, "
Identification and Comparison of True Net Fracturing Pressures Generated by Pumping Fluids with Different Rheology into the Same Formations
",
SPE Paper No. 26153
,
1993
.
Johnson
,
D.E.
and
M.P.
Cleary
, "
Implications of Recent Laboratory Experimental Results for Hydraulic Fractures
",
SPE Paper No. 21846
,
1991
.
Roodhart
,
L.P.
,
P.A
,
Fokker
,
D.R.
Davies
,
J.
Shlyapobersky
and
G.K.
Wong
, "
Frac and Pack Stimulation: Application, Design, and Field Experience From Gulf of Mexico to Borneo
",
SPE 26564
,
1993
.
Martins
,
J.P.
J.C.
Abel
,
C.G.
Dyke
,
C.M.
Michel
and
G.
Stewart
, "
Deviated Well Fracturing and Proppant Production Control in the Prudhoe Bay Field
",
SPE Paper No. 24858, SPE Annual Conference
1992
.
Pospisil
,
G.
,
K.W.
Lynch
,
C.M.
Pearson
and
J.A.
Rugen
, "
Results of a Large-Scale Refracture Stimulation Program, Kuparuk River Unit, Alaska
,"
SPE 24857
,
1992
.
Johnson
,
D.E.
,
C.A.
Wright
,
A.
Stachel
,
H.
Schmidt
and
M.P.
Cleary
, "
On-Site Real-Time Analysis Allows Optimal Propped Fracture Stimulation of a Complex Gas Reservoir
",
SPE Paper No.25414
,
1993
.
Cleary
,
M.P.
,
D.E.
Johnson
,
H-H.
Kogsbpll
,
K.A.
Owens
,
K.F.
Perry
,
C.J.
de Pater
,
A.
Stachel
,
H.
Schmidt
and
M.
Tambini
, "
Field Implementation of Proppant Slugs to Avoid Premature Screen-Out of Hydraulic Fractures with Adequate Proppant Concentration
",
SPE 25892
,
1993
.
Johnson
,
D.E.
,
T.B.
Wright
,
M.
Tambini
,
R.
Maroli
and
M.P.
Cleary
, "
Real-Data On-Site Analysis of Hydraulic Fracturing Generates Optimum Procedures for Job Design and Execution
",
SPE Paper No. 25920
,
1993
.
Cleary
,
M.P.
"
Comprehensive Design Formulae for Hydraulic Fracturing
",
SPE Paper No. 9259
,
1980
.
Cleary
,
M.P.
, "
Analysis of Mechanisms and Procedures for Producing Favorable Shapes of Hydraulic Fractures
",
SPE Paper No. 9260
,
1980
.
Crockett
,
A.R.
,
R.M.
Willis
, and
M.P.
Cleary
, "
Improving Hydraulic Fracture Predictions by Real-Time HistoryMatching on Observed Pressures
",
SPE 15264
,
1986
.
Crockett
,
A.R.
,
N.M.
Okusu
, and
M.P.
Cleary
, "
A Complete Integrated Model for Design and Real-Time Analysis of Hydraulic Fracturing Operation
",
SPE Paper No. 15069
,
1986
.
Cleary
,
M.P.
,
D.T.
Barr
, and
R.M.
Willis
, "
Enhancement of Real-Time Hydraulic Fracturing Models with Full 3-D Simulation
",
SPE Paper No. 17713
,
1988
.
Narendran
,
V.M.
and
M.P.
Cleary
, "
Analysis of Growth and Interaction of Multiple Hydraulic Fractures
",
SPE Paper No. 12272
,
1983
.
Papadopoulos
,
J.M.
,
V.M.
Narendran
and
M.P.
Cleary
, "
Laboratory Studies of (Multiple) Hydraulic Fracture Growth and Interaction
",
SPE Paper No. 11618
,
1983
.
Cleary
,
M.P.
, "
Effects of Depth on Rock Fracture
",
Proceedings International Symposisum on Rock at Great Depth
(
Pau
,
1989
) published by
A.A.
Balkema
(Editors
V.
Maury
&
D.
Fourmaintraux
), 1990.
De
Pater
,
C.J.
,
L.
Weijers
,
M.P.
Cleary
,
T.S.
Quinn
,
D.E.
Johnson
,
D.T.
Barr
and
M.
Savic
, "
Experimental Verification of Dimensional Analysis for Hydraulic Fracturing
",
SPE Paper No. 24994, Europec
,
1992
.
Quinn
,
T.
, "
The Role of Permeability Variation in the Growth of Hydraulic Fractures
",
Ph.D. Thesis
,
MIT
,
1994
.
McGowen
,
J.M.
,
S.
Vitthal
,
M.A.
Parker
,
A.
Rahimi
and
W.E.
Martch
Jr.
, "
Fluid Selection for Fracturing High- Permeability Formations
",
SPE Paper No. 26559
,
1993
.
Martins
,
P.J.
,
P.J.
Collins
,
M.
Rylance
,
O.E.
Ibe
,
R.T.
Kelly
and
P.A.
Bartel
, "
Small, Highly, conductive Fractures Near Reservoir Fluid Contacts: Application to Prudhoe Bay
",
SPE Paper No. 24856
,
1992
.
De Pater
,
C.J.
,
J.J.
Hagoort
,
I.S.
Abou-Sayed
and
R.S.
Donders
, "
Propped Fracture Stimulation in Deviated North Sea Gas Wells
",
SPE Paper No. 26794
,
1993
.
Warpinski
N.R.
,
Z.A.
Moschovidis
,
C.D.
Parker
and
I.S.
Abou-Sayed
, "
Comparison Study of Hydraulic Fracturing Models: Test Case - GRI-Staged Field Experiment No.3
",
SPE 25890, (see also response to SPE 25890 by Cleary, M.P.) for publication in Production and Facilities Quarterly Journal, February
1994
.
Unneland
,
T.
and
T.
Haugland
, "
Permanent Downhole Gauges Used in Reservoir Management of Complex North Sea Oil Fields
",
SPE Paper No.26781
,
1993
.