SPE 54630 Non-Darcy and Multiphase Flow in Propped Fractures: Case Studies Illustrate the Dramatic Effect on Well

Productivity
Michael C. Vincent, SPE, Insight Consulting; C. Mark Pearson, SPE; and John Kullman, SPE, CARBO Ceramics, Inc.

Copyright 1999, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the 1999 SPE Western Regional Meeting held in Anchorage, Alaska, 2628 May 1999. 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 author(s). 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. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

substantial volumes of free gas in the fracture. This paper shows that non-Darcy and multiphase flow effects frequently result in an effective fracture conductivity 50% to 98% lower than the reference value obtained from the API testing procedure. Incremental reductions for gel damage and proppant embedment often result in fracture permeabilities or conductivities (the product of fracture permeability and effective fracture width) of 1 to 10% of the published values. This paper examines the mistakes that are likely to be made when ignoring these effects, and estimates the additional cashflow which can be obtained by optimizing fracture design with consideration of multiphase and non-Darcy flow. Introduction The American Petroleum Institute has published standard testing procedures 3 which involve flowing a single phase liquid (water with 2% KCl) through a 7 x 1 linear proppant pack cell at extremely low flowrates of 1 to 10 ml/min. These rates correspond to a range of 4 to 40 bpd of oil, or 13 to 130 mscfd of gas at 1000 psi p bh , produced from two fracture wings of a 30 ft. high fracture. The superficial velocity of the water during the test is on the order of 0.2 to 2.0 inches per minute; while in actual fractures, the true fluid velocity can be several feet per second. The API recognized that these tests are not representative of actual conditions and placed the following disclaimer with the procedures: CAUTION: The testing procedures in this publication are not designed to provide absolute values of proppant conductivity under downhole reservoir conditions. 3 The conductivity tests were designed for operational simplicity and selected to be in the laminar flow regime to improve repeatability. Unfortunately, proppant suppliers in the industry have solely used these procedures to publish permeability values for their products and many engineers have used these reference values in fracture design without making appropriate adjustments for the non-Darcy and mutiphase flow effects. Erroneously using the published proppant permeability values without compensation for non-Darcy and multiphase flow effects results in: Invalid fracture conductivity predictions which may be overestimated by more than an order of magnitude,

Abstract The dramatic effects of non-Darcy and multiphase flow in propped hydraulic fractures have been documented by several authors.1,2,7,9-19 However, many engineers disregard these effects when designing fracture treatments under the assumption that they only apply to high rate wells. This paper shows these effects are significant even in wells considered to be low rate by current industry standards. Ignoring these effects will often lead to inaccurate production forecasts, suboptimal fracture design, and selection of an inappropriate proppant type. These mistakes result in lost revenues which can exceed $2 million per fracture treatment for typical gas and oil well fracture treatments conducted in North America. Proppant permeability values reported by the industry and used in most fracture design models are measured with a single-phase fluid at extremely low velocities. The laboratory rates stipulated in the American Petroleum Institute3 (API) Recommended Practice number 61 typically correspond to superficial fluid velocities ranging from 0.2 to 2.0 inches per minute. However, in real fractures, the actual fluid velocity may exceed two feet per second, approximately 1000 times greater than in the laboratory measurements. Although the API warned that the lab-measured values would not be realistic under actual fracture conditions, the industry has largely failed to incorporate correction factors into production models to compensate for non-Darcy flow effects. In addition to non-Darcy effects, the measured proppant permeability values fail to incorporate the effects of multiphase flow. Most gas wells will produce some free liquid (water or condensate), and almost all oil wells will be produced below the bubblepoint of the oil, resulting in

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

Design of a sub-optimal stimulation treatment including a shorter effective producing fracture half-length, Significant overestimation of the post-stimulation production rates, Potential for selection of the incorrect proppant for the fracture treatment design, and Significant loss in production from the fractured well compared to what could have been achieved.

Non-Darcy Flow Effects Henry Darcy initially developed his familiar correlation by observing water draining through a sand column.4 His studies showed that at low velocities, the pressure drop through porous media is proportional to the fluid velocity: p/L = v / k ... (1) where: p/L = pressure drop per length of proppant pack = fluid viscosity v = superficial fluid velocity (as if porosity were 100%) k = permeability of porous media However, at the higher velocities typical in a propped fracture, the pressure gradients become proportional to the square of the velocity, as represented by Forchheimers equation: 5 p/L = v / k + v2 .. (2) where: = coefficient of inertial resistance = fluid density Forchheimers equation states that the pressure gradient is the sum of the viscous forces ( v/k) and the inertial forces (v 2 ). At low velocities, where inertial forces are small, Forchheimers equation reduces to Darcys Law. Conceptually, Darcys Law states that the pressure drop in porous media is controlled by friction. At low velocities, the viscous drag of the molecule of fluid upon the proppant controls the pressure drop. Forchheimers equation recognizes that additional energy loss is incurred by the repeated acceleration and deceleration of the fluids. A molecule of oil or gas will travel ~50,000 proppant grain diameters within 100 ft. of fracture length. At each encounter with a proppant grain, the molecule must change direction to travel around the grain. The molecule will undergo substantial acceleration to travel through the narrow pore throat and decelerate into the larger pore body. Forchheimers equation recognizes that the energy losses due to this repeated acceleration and deceleration of the molecule are highly significant at realistic fluid velocities. Using the above equations, the pressure drop per foot of fracture length can be calculated as a function of velocity in the fracture or production rate through a given cross-sectional area of fracture. Figure 1 shows the pressure drop through a bi-wing fracture with a 50 ft. frac height when propped with 20/40 Jordan sand at 2 lb/ft 2 . It is apparent that Darcys Law is not valid for calculation of the pressure drop down the fracture at even moderate rates below 500 mscfd. Figure 1

was developed for a bottom hole pressure (p bh ) of 2000 psi. At lower p bh , the gas will expand and subject the fluid to increasingly higher velocities, resulting in even greater inertial pressure loss. Examination of Figure 1 reveals that the inertial pressure drops are much more significant than the viscous losses, suggesting that beta is more important to ultimate productivity than the laminar conductivities reported by proppant manufacturers using the API test procedure. Interestingly, Forchheimers equation for pressure drop in porous media is similar in form to the equations used to calculate the pressure drop in pipes. At low (laminar) flow rates, viscous forces dictate the pressure drop, with p/L being proportional to the fluid velocity. But at higher (turbulent) flow rates, the pressure drop becomes proportional to the square of the velocity. Because of this similarity, the nonDarcy flow effects are often referred to as turbulent effects. This is not entirely accurate. A turbulent flow regime typically begins to occur in pipe flow at Reynolds numbers above 2300.6 However, in porous media, the non-Darcy effects become significant even if the overall velocity is within the laminar flow regime and has been observed with Reynolds numbers as low as one.7 The nature of flow in porous media causes a very complex flow regime even at low velocities. The non-Darcy inertial forces are primarily due to the acceleration and deceleration of the fluid as it travels through the tortuous flow path of the porous media. The oil, water or gas in the fracture must continually change direction, accelerating through pore throats and decelerating in the larger pore spaces. The force necessary to effect those accelerations can also be derived from Newtons Laws of Motion, in which the kinetic energy is equal to the mass multiplied by the square of the velocity.8 Therefore, the form of Forchheimers experimentally derived relationship is intuitively reasonable. It is important to recognize that both Darcys and Forchheimers equations are based on the superficial velocity of the fluid, which effectively assumes 100% porosity in the fracture. The same fluid velocity is assumed for both a damaged fracture containing poorly sorted, severely crushed, highly angular proppant and a theoretically ideal propped fracture containing perfectly sphericial proppant of uniform size. In reality, the pressure loss in the fracture is clearly related to changes in the true fluid velocity, which is a function of the proppant characteristics. Forchheimers equation incorporates the differences in proppant characteristics via the beta factor (). Beta Factor. Alternatively referred to as the inertial flow coefficient, the beta factor is a proportionality coefficient that is determined by laboratory measurements. Beta is essentially a measure of the tortuosity of the flow path.9 Beta is determined by flowing more realistic velocities through the API conductivity cell and solving Equation 2 for beta to match the observed pressure drops. Many factors influence the inertial flow effects for single phase flow including: 9-12

SPE 54630

NON-DARCY AND MULTIPHASE FLOW IN PROPPED FRACTURES: CASE STUDIES ILLUSTRATE THE DRAMATIC EFFECT ON WELL PRODUCTIVITY

Initial proppant permeability, Porosity of the proppant, Curvature of streamlines (proppant angularity), Relative aperture of pore throat to pore space, Proppant size distribution, Heterogeneities, and Surface roughness.

Figure 1 indicates that the pressure drop due to friction is minimal compared to inertial losses for most gas wells with moderate production rates. A well producing 500 mscfd of dry gas through two 50 ft. high fracture wings at a p bh of 2000 psi will have a total pressure drop approximately five times greater than Darcys Law would predict. To minimize pressure drop through the fracture, it is necessary to minimize the v2 term from Equation 2. Figure 2 shows that the beta factors for common proppants can vary by a factor of six. Since the pressure drop in the fracture is typically dominated by non-Darcy effects, the beta factor can be more important than reference permeability when selecting a proppant. Unfortunately, beta factors are not typically measured and reported by the industry. Selecting a proppant by reference permeability alone can be misleading. It is possible for proppants with similar reference permeabilities to have drastically different effective permeabilities when subjected to the realistic fluid velocities in typical fractures. Multiphase Flow Effects Several researchers have attempted to quantify the effect of multiphase flow in fractures with laboratory work or by analyzing production data.13-18 While these authors report differing results regarding the absolute value of the conductivity loss, all conclude that the effects are substantial and should not be ignored. Although he recommends measuring multiphase effects for specific cases, Penny estimates that the effective conductivity is typically reduced by a factor of three for each 4 barrels of liquid produced per mmscfd.13 Evans reported that beta factors increase by tenfold with very small mobile liquid saturations.14 Milton-Tayler reported order of magnitude reductions in effective conductivity due to multiphase flow.15 Schubarth reviewed actual production results from 550 gas wells and found a strong correlation between condensate yield (multiphase flow) and reduced productivity.16 In the literature, as in this paper, the effect of multiphase and non-Darcy flow can be reported in several ways. If the pressure drop across the proppant pack increases by ten-fold, this is synonymous with reporting a ten-fold reduction in effective conductivity or a ten-fold increase in beta. Converting the pressure drop to an effective conductivity is one technique to account for multiphase and non-Darcy effects in a production model utilizing Darcys law.19 Incremental pressure drop due to multiphase flow is believed to be the result of three primary causes: saturation changes, relative permeability effects and phase interactions

within complex flow regimes. For oil wells produced below bubble point, the evolution of solution gas to free gas will also drastically increase the fluid velocities in the fracture. It is relatively simple to calculate the saturation changes and relative permeability effects due to multiphase flow in the fracture. However, the complex interactions resulting from multiphase flow are exceedingly difficult to model. Due to marked viscosity differences, the gas phase travels at substantially higher velocity than the liquid phase in the fracture. However, due to the tortuous flow path inside the proppant pack, the two phases often remain in intimate contact. Liquid droplets may be picked up (entrained) by the gas and substantial energy is consumed as the liquid is accelerated to many times its superficial velocity. When the droplet collides with a proppant grain, it undergoes substantial deceleration only to be accelerated again by the gas phase. The extreme inefficiencies in flow path and inertial losses are responsible for the surprisingly large pressure drops observed in the laboratory with addition of a few volume percent of liquid to a gas system. A common mistake with oil wells is failure to recognize that the pressure in the fracture is typically far below bubble point. Although the average reservoir pressure can often be maintained at or above bubble point, the bottom hole pressure (p bh ) of the producing well is typically reduced to maximize drawdown and production rates from the well. It is important to realize that an oil well with a gas-oil ratio (R) of 560 scf/bbl produces 99% gas by volume and only 1% liquid when evaluated at atmospheric pressure conditions. The bottom hole pressure may be 150 psi for wells with submersible pumps and may yield 90% gas and 10% liquid by volume. At some distance away from the wellbore, the pressure in the fracture may be 1500 psi; in which case, the fracture may be filled with 50% liquid and 50% gas. Unless the production model is linked to a compositional simulator, it is unlikely that it rigorously handles the evolution of solution gas in the body of the fracture. Failing to consider that 50% to 90% of the fracture can be filled with free gas has obvious implications regarding the effective capacity of the fracture to produce oil. Experimental measurements of multiphase and non-Darcy flow effects have been documented by Penny.13 As shown in Figure 3, the proppant conductivity measured at extremely low flowrates according to API specifications can be exceedingly optimistic. Even with a fairly modest well (1 mmscfd from a bi-wing 50 ft. fracture height), the non-Darcy flow effects reduce the effective conductivity of the proppant pack by approximately 70%. Addition of a mere 10 bwpd to the 1 mmscfd causes an incremental reduction of approximately 25% of the reference conductivity. Although 10 bbls of water in 1 mmscfd appears to be a negligible volumetric percentage (0.006%) when evaluated at atmospheric pressure, the volumetric percentage is ~1% when evaluated at 3000 psi. Note, however, that adding a mere 1% liquid to the system increased the pressure drop by 550%! Saturation changes and relative permeability changes alone cannot account for this dramatic effect. It is believed that the

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

primary factor is the increased complexity of the flow regime due to phase interactions. Additional damage to the proppant pack due to gel damage, embedment, long-term degradation, fines migration and other factors can result in effective conductivities that are 99% lower than suggested by the API testing procedures. Production Models Few production models accurately handle non-Darcy flow. Although sophisticated 3-phase models can predict the effects of saturation changes and relative permeability, it appears that no mechanistic model currently available can accurately predict the extreme pressure drop due to phase interactions within the intimately mixed flow regimes typical in fractures. Production models which assume infinite fracture conductivity are often unsuitable for these analyses. The assumption of infinite fracture conductivity provides reasonable results when the dimensionless fracture conductivity (CfD ) is greater than 500.20 In concept, infinite conductivity models assume that the pressure drop in the fracture is negligible compared to the pressure drop in the formation. For a constant fracture length, Holditch has shown that to optimize economic return, the C should be in the fD range of 10 to 30.21 Elbel has shown that fracture designs with a CfD of three or less cannot be improved significantly by increasing fracture length with the same conductivity. If the CfD is greater than 30, increasing length will be more beneficial than increasing conductivity.22 The case studies in this paper indicate that the pressure drop in the fracture is frequently significant and fracture conductivities obtained in practice will typically be far lower than optimal. Production and economic forecasting in this paper was prepared with Stim-Labs SLFrac production model. This tool was chosen because it has the highest quality permeability and beta factor data in the industry and the production model is available and used by the 40+ companies who are members of the Stim-Lab proppant consortium. All the major service companies are members of the consortium and may use the program to evaluate potential fracture designs. The production model is based on Agarwal type curves, so it includes implicit assumptions regarding an infinite reservoir with no pressure support.20 However, if incremental benefits of multiple cases are compared, the inaccuracies of the simplified type curve solution will essentially cancel out. At this time, the SLFrac program has the ability to estimate multiphase effects in gas wells by correlating trends in measured laboratory data, but does not accurately predict the conductivity reduction for oil wells with p bh below bubble point. If an engineer does not have access to a program such as SLFrac, it may be possible to compensate manually for these effects. Some finite-conductivity models will allow the user to modify the reference proppant conductivity tables or input a damage factor to compensate for multiphase and non-Darcy effects.

Results - Case Studies: Following are four specific case studies which have been highly simplified for this paper. For actual analyses, the authors recommend comparing a wider range of proppant types, sizes and performing sensitivities on uncertain values such as fracture width, oil/gas prices and formation permeability to develop a better understanding of the effect of each uncertainty upon fracture design. These wells were not selected because of particularly innovative frac designs, but rather to depict four very different wells with modest flowrates which illustrate the effects of non-Darcy and multiphase flow in a variety of flowing conditions. Assumptions for Case Studies. The Appendix contains a complete listing of reservoir characteristics and product prices used for all analyses. To simplify the paper, the results for only three proppants are shown in Cases 1 through 4: natural frac sand, a premium resin-coated sand (RCS) and light weight ceramic (LWC). The three-dimensional plots for Case 5 include both standard and premium RCS, and also include heavy weight ceramics. These analyses do not consider the problem of proppant flowback. In unconsolidated formations, it is often necessary to use curable resins to maintain proppant pack integrity. The analyses presented in this paper assume that consolidation of the pack is not required. Case 1: Deep Gas well. Case 1 is a mid-continent gas well in the Anadarko Basin, Oklahoma, at 12,500 total depth, with 25 of pay and 0.1 md perm. The service companys predicted fracture geometry is 555 half-length, 50 frac height, with an average proppant concentration of 1.5 lb/ft 2 . This geometry equates to 83,250 lbs. proppant. Production history-matching on adjacent wells indicated that a gel damage factor of 50% provided reasonable forecasts. As shown in Table 1, the initial stress on the proppant for this deep well was calculated to be 9279 psi. At this stress, the standard industry practice is to place ceramic proppant. The third column of Table 1 shows the predicted producing characteristics of the well when non-Darcy and multiphase effects are ignored. Disregarding multiphase and non-Darcy flow, simplistic production models would indicate that 1.5 lb/ft 2 LWC would provide a C of 12.9, yielding a fracture fD approaching infinite conductivity relative to the formation. Expected well rates with LWC are 3.8 mmscfd, which generates superior economic returns compared to natural frac sand or resin coated sand (RCS). Columns four and five show the effects of including nonDarcy and multiphase flow. Although LWC is still predicted to be the most economic proppant, the effective C is 0.16 fD and the predicted well rate is reduced to 1.6 mmscfd. Although disregarding the non-Darcy effects allowed selection of the correct proppant, it is clear that 1.5 lb/ft 2 of LWC does not provide an infinite conductivity fracture, and that increasing fracture width or placing higher permeability proppants should be considered. Disregarding these flow

SPE 54630

NON-DARCY AND MULTIPHASE FLOW IN PROPPED FRACTURES: CASE STUDIES ILLUSTRATE THE DRAMATIC EFFECT ON WELL PRODUCTIVITY

effects results in the loss of 2 mmscfd of potential production rate and lost revenue of $2,000,000 within the first three years of production. Case 2: Shallow Gas well. The preceding example showed that an optimized fracture design will be flawed unless multiphase and non-Darcy effects are considered. However, the proppant choice for Case 1 did not change. Based on experience and rules of thumb, the traditional wisdom is to use premium proppants with deep gas wells. But what happens to this analysis with a shallow well? As detailed in the Appendix, Case 2 models the same 25 of pay at 3000 depth. Reservoir pressure has been reduced to 1800 psi, and fracture geometry has been maintained with 1.5 lb/ft 2 average concentration. For this well, a p bh of 800 psi is assumed, which generates an effective stress on the proppant of merely 1780 psi. Traditional wisdom specifies that natural sand would suffice in this shallow, low stress well. Table 2 shows the predictions for the well with each assumption. As expected, with laminar flow assumptions, the 20/40 sand provides ample conductivity to handle the predicted 700 mscfd, with a C of 32. Any investment to fD purchase premium proppant is wasted, as it results in less than 5 mscfd incremental rate. However, with multiphase and nonDarcy flow effects, the sand fracture is shown to be conductivity limited, with a C of 2.4 and total production fD rate would be constrained to under 500 mscfd. Here the ceramic proppant is shown to produce an incremental 59 mscfd (a 13% increase in initial productivity), which pays out the incremental investment within one year and produces a 88% return on the investment within 3 years. While the economics are not outstanding, many companies would consider an investment with expected annual returns exceeding 20%. These results are counter to current industry design guidelines. Ceramic proppant is the most economic proppant for a shallow well with 1780 psi stress and producing only 500 mscfd? These results indicate that the pressure drop down the fracture is significant when propped with 1.5 lb/ft 2 natural sand and subjected to multiphase flow. It is desirable to increase the conductivity in some manner. In these analyses, selecting a higher permeability proppant was evaluated. An alternative remedy may be to increase fracture width if operational constraints allow. It is interesting to note that the selection of proppant had little to do with proppant strength or crush resistance, but was due to the lower beta factor with the less angular, high porosity ceramic proppant. While it may be obvious that deep wells merit higher strength proppant, it is more challenging to convince that premium proppants should be considered at very low stresses. Case Study 2 suggests that proppant selection should not be made by strength, but rather by effective conductivity required to handle the expected flowrates.

Case 3: Moderate Depth Oil Well. The third case study is a moderate stress oil well from the Kuparuk River Field, Alaska. Total well depth is 6000 feet and expected p bh is 800 psi. With a normal frac gradient of 0.7 psi/ft, the effective closure stress is calculated to be approximately 3400 psi. As shown in Table 3, the formation productivity is good, resulting in oil rates often exceeding 200 bopd for any of the 20/40 proppants studied. Several SPE papers have been written showing the advantage of increasing fracture width and proppant permeability in moderate permeability reservoirs.23-25 A typical fracture geometry assumed for the Kuparuk case study was 180 half-length, 55 frac height and 5 lb/ft 2 proppant concentration equating to approximately 100,000 lbs. of proppant per stimulation. Table 3 shows that using a laminar production model, an engineer may recognize that the well was conductivity limited with natural sand and could justify the incremental expense of ceramic proppant. At these low closure pressures, ceramic proppant is selected not for strength but for improved permeability. Although use of a laminar flow model allowed the correct proppant to be selected, it seriously over-estimates production rates that can be produced through the fracture. While a Darcy model will predict 700 bopd to be produced through the fracture with modest pressure drops, inclusion of non-Darcy effects indicates that 300 bopd is more representative of actual flow capacity even when propped with 5 lb/ft 2 20/40 sand or RCS. As has been proven from production data, increased use of larger, higher conductivity ceramic proppants and tip screen out techniques were very beneficial in development of this field.23,24 The SLFrac model does not have a compositional simulator and, therefore, cannot accurately predict how much free gas will be present in the fracture. However, considering only non-Darcy flow effects of single-phase oil, it can be shown that the ceramic proppant can generate a 8500% return on incremental investment over three years. If necessary, PVT data could be examined to estimate the volumetric percent of solution gas that would be present in the fracture as free gas. However, the analyses of non-Darcy flow effects alone have indicated that the fracture is conductivity-limited and the presence of gas would only exaggerate the decision to use a LWC proppant and increase fracture width where possible. Case 4: Lower Rate, Low Stress Oil Well. The fourth case study is a low stress oil well in Lake Maracaibo, Venezuela. Total well depth is 6500 feet, expected p bh is 350 psi and, with a relatively low frac gradient of 0.54 psi/ft, the effective closure stress is calculated to be less than 3200 psi. As shown in Table 4, the productivity is modest, resulting in oil rates of ~200 bopd per frac zone. The wells are typically propped with 4 lb/ft 2 20/40 proppants. Which proppant type will be most economic? Table 4 shows that using a laminar flow production model, an engineer would recognize that the well was conductivity-limited with natural sand, but could not justify

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

the incremental expense of premium ceramic proppant. A Darcy model suggests that ceramic proppant will yield an additional 13 bopd, which takes more than a year to payout the incremental investment. Although the ceramic will ultimately generate a higher cash flow, the incremental oil rate would be hard to document and justify. However, when non-Darcy flow effects are considered, the expected production rates drop to below 100 bopd for the natural sand frac compared to 266 bopd for 20/40 lightweight ceramic. Placing natural sand as recommended by the Darcy production model will result in selection of the incorrect proppant type, sacrificing over 160 bopd and over $1,000,000 of additional cash flow during the first three years of production. The SLFrac model cannot be used to predict the effects of free gas present in the fracture. However, considering only non-Darcy flow effects, it can be shown that the 20/40 ceramic proppant will generate a 2600% return on incremental investment over three years. Although the data are not included here, evaluation of larger mesh ceramic proppants show additional production benefit for this field. Again, these results suggest that traditional rules of thumb for proppant selection do not agree with more rigorous engineering analyses and economic optimization. Case 5: Three Dimensional Plots To further illustrate how proppant selection is affected by non-Darcy and multiphase assumptions, Figure 4 was developed. The preceding analyses demonstrate that proppant selection should not be made solely on strength. It is hypothesized that the fluid velocity in the fracture might best correlate to proppant selection. However, to generate a more user-friendly correlation, three-dimensional graphs were made that relate proppant selection to closure stress and formation permeability for a specific fracture geometry. Again, these plots are highly simplified and include the following assumptions: Gas well with 50 pay; 200,000 lbs of 20/40 proppant (500 half-length, 100 frac height, 2 lb/ft 2 ) 250F, Proppant prices = 40% discount from book price, Gas price = $2/mscf. The vertical axes display the incremental cash flow (after proppant cost) relative to sand after three years of production. Therefore, sand is shown as a horizontal plane at an incremental cash flow of zero, and RCS and ceramic are displayed as intersecting surfaces relative to this plane. Figure 4a shows the projected cash flow benefits for each proppant when multiphase and non-Darcy effects are ignored. This plot illustrates that at low closure pressure and low formation perm, natural sand is the most economic proppant. Curves for RCS and ceramic result in a negative cash flow relative to sand at these conditions. However, with the scale of the plot to $35 Million, the incremental cost of the premium proppants (up to ~$100,000) is hard to discern. As fracture closure pressure and/or permeability increase, the benefit of premium proppants is seen to increase. It is also noted that the $35mm benefit to ceramic at 1 md and 9000 psi stress is overstated by the Darcy model. Ignoring multiphase

and non-Darcy effects allows prediction of unrealistically high gas rates exceeding 30 MMSCFD through a fracture containing 2 lb/ft 2 20/40 proppants. Figure 4b shows the forecast for an identical fracture when non-Darcy effects are included in the calculation. Notice that the scale of the vertical axis is reduced to a maximum benefit ~$10mm, which reflects more reasonable incremental gas rates through the fracture. Figure 4c shows the impact of including both non-Darcy and multiphase flow effects in the fracture. Notice that the relative area where natural sand is the preferred proppant is reduced due to the additional pressure drops of these flow effects. These plots are not meant to be used as precise tables to determine the most economic proppant for a given well as they are rigorous only for the fracture geometry and product prices shown. Instead, they are presented to illustrate the fallacy of existing rules of thumb and to illustrate the dramatic impact of non-Darcy and multiphase flow on proppant selection and/or fracture design. Figure 5 shows 3-d plots in the same format for an oil well. Input values are shown in the Appendix. The range of permeabilities have been increased by one order of magnitude to more accurately depict the characteristics of economic oil reservoirs. Interestingly, the shape of the gas and oil cases are similar when non-Darcy effects are ignored. However, with non-Darcy flow, the benefit to increased permeability is more pronounced in oil wells at moderate stresses. A plot with multiphase flow has been omitted, as the SLFrac program does not accurately handle the evolution of solution gas from the liquid phase below bubble point. Interpretation of Results Ignoring multiphase and non-Darcy effects can lead to incorrect decisions regarding the required fracture width and proppant selection. It is common that the CfD is below 10. Infinite conductivity fracture models are frequently invalid. Economic potential of the fracture design is not optimized. Proppant selection should not merely be a function of stress, well depth, or other simplistic rules of thumb. Economic optimization will lead to selection of a preferred fracture geometry and proppant based on stress, formation deliverability and other factors. Although highly simplified, these analyses illustrate that fracture design will frequently be flawed if multiphase and non-Darcy effects are ignored. Alternative Strategy for Frac Design/Prop Selection In lieu of applying rules of thumb to design fracture treatments, the following alternative is proposed: Estimate reservoir and fracture parameters, bracket uncertain values. Use a production model which includes calculation of non-Darcy and multiphase effects.

SPE 54630

NON-DARCY AND MULTIPHASE FLOW IN PROPPED FRACTURES: CASE STUDIES ILLUSTRATE THE DRAMATIC EFFECT ON WELL PRODUCTIVITY

With available production data from offset wells, perform sensitivities to history match effective fracture length, width and gel damage. Determine what conductivity is needed and the most economic way to achieve this. Optimize economics (cash flow, Net Present Value, payback period, ROI, etc.) by adjusting fracture geometry and proppant selection within ranges dictated by operational constraints.

due to imperfect proppant distribution concentrating the stress on fewer proppant grains. Conclusions 1. Non-Darcy and multiphase flow effects are substantial and should be considered in fracture design. 2. The actual pressure drop down the fracture may be 100x greater than predicted using Darcys Law with reference proppant permeabilities measured under API test conditons. 3. If a production model is incapable of modeling the complex phase interactions involved with multiphase and non-Darcy flow, these effects need to be approximated by reducing the effective proppant permeability. 4. Multiphase flow effects can be substantial. Even a few percent (by volume) of a second phase can increase the pressure drop by up to an order of magnitude. These changes are much more dramatic than intuition would predict. It is believed that the pressure loss is due to repeated acceleration of the denser phase. 5. Failure to consider these factors can lead to inadequate fracture conductivity, which may account for short effective fracture lengths. 6. Applying rules of thumb for frac design appears to frequently give the wrong answer. Proppant selection should be made by optimizing economic returns, not as a function of depth or stress alone. The most economic proppant is highly dependent upon formation permeability. Even if the most economic proppant is determined for a reservoir with low formation permeability (e.g. 0.1 md), it will frequently not be the most economic proppant for a similar depth formation with a higher permeability. Rules of thumb that relate proppant selection to a single variable, such as well depth or proppant stress, are inherently flawed. 7. A simple production model is available which can handle multiphase and non-Darcy flow in gas wells. The model is easy to use and can give a 10-minute analysis that displays the dramatic result of these flow effects. It is not necessary to rely on inaccurate rules of thumb. Nomenclature p/L = pressure drop per length of proppant pack = fluid viscosity v = superficial fluid velocity (if porosity were 100%) k = permeability of porous media = coefficient of inertial resistance = fluid density (v / k) = viscous term (v 2 ) = inertial term p bh = bottom hole pressure R = gas-oil ratio (producing) KCl = potassium chloride g = gram

How to handle uncertainties: Try to bracket unknown values. If the expected proppant concentration is 2 lb/ft 2 ., it will be instructional to perform sensitivities at 1 and 3 lb/ft 2 . concentrations to determine if that estimate changes your decision. If the directional answer remains the same (i.e., if increased frac length is beneficial in both cases), then the precise value for concentration appears not to change the decision. Potential of multiple fractures. If multiple fractures are expected, then fracture conductivity predicted by traditional planar models will be optimistic. Cycling of pressure/stress on proppant. Permeability data gathered according to API specifications are for single-cycle crush tests and conductivity tests performed for ~50 hours per stress level in a conductivity cell. It may be appropriate to reduce these values further if multiple pressure cycles or long-term degradation are expected. Additional stress due to induced fracture width or net pressure generated during a screen out. Substantial stress can be applied to the formation face during the fracturing treatment. After the fluids leak off, the proppant must be competent to support this additional stress. The fracture gradient typically used in the industry is measured at a fracture width of zero. If the fracture is not allowed to close, that additional stress must be supported by the proppant. These additional stresses may be calculated from rock properties or can be estimated from observing the net pressure during the fracture.26 For a fully-packed fracture, many engineers consider that the fracture is unable to relax and the full net pressure applied during the treatment is added to the effective fracture closure stress. Varying pressure conditions. Proppant crushing should be evaluated at the lowest pbh that may occur during the productive life of the well. In a gas reservoir, this will typically occur late in the life of the field as the reservoir pressure has dropped resulting in an increase in the effective stress on the proppant. If a cleanout is performed following the frac job, that p bh may be lower than typical production pressures and expose the proppant to substantially greater effective stress. Non-uniform proppant distribution. Conductivity testing is performed with uniform proppant packs. In actual fractures, crush should be expected to be greater

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

References
1. 2. Cooke, C.E., Conductivity of Fracture Proppants in Multiple Layers, paper SPE 4117, JPT, Sep 1973. Martins, J.P., et al.: The Effects of Non-Darcy Flow in Propped Fractures, paper SPE 20709 presented at the 1990 Annual Technical Converence and Exhibition, Sep 23-26. American Petroleum Institute.: Recommended Practices for Evaluating Short Term Proppant Pack Conductivity, API RP 61, Oct. 1989. Frick, Thomas C., et al.: Petroleum Production Handbook, Millet the Printer, Dallas (1962) Vol 2, p 23-11. Forchheimer, P.: Wasserbewegung durch Boden, ZVDI (1901) 45, 1781. Welty, J.R., et al.: Fundamentals of Momentum, Heat, and Mass Transfer, John Wiley & Sons, New York (1984) p 164. Stim-Lab Proppant Consortia Notes, 1990, p 12-v. Tipler, P.A.: Physics, Worth Publishers, New York (1982) 165. Geertsma, J.:Estimating the Coefficient of Inertial Resistance in Fluid Flow through Porous Media, paper SPE 4706, SPEJ, Oct 1974. Pursell, et al.: Lab Investigation of Inertial Flow in High Strength Proppants, paper SPE 18319 presented at the 1988 Annual Technical Conference and Exhibition, Oct. 2-5. Milton-Tayler, David: Non-Darcy Gas Flow: From Laboratory Data to Field Prediction, paper SPE 26146 presented at the 1993 Gas Technology Symposium, June 28-30. Stim-Lab Proppant Consortium 1994-1998 Reports. Penny, G.S., and L Jin: The Development of Laboratory Correlations Showing the Impact of Multiphase Flow, Fluid, and Proppant Selection Upon Gas Well Productivity, paper SPE 30494 presented at the 1995 Technical Conference and Exhibition, Oct. 22-25. Evans, E.V and R.D. Evans: Influence of an Immobile or Mobile Saturation on Non-Darcy Compressible Flow of Real Gases in Propped Fractures, paper SPE 15066, JPT, Oct. 1988. Milton-Tayler, David: Realistic Fracture Conductivities of Propped Hydraulic Fractures, paper SPE 26602 presented at 1993 Annual Technical Conference and Exhibition, Oct. 3-6.

3.

4. 5. 6. 7. 8. 9.

10.

11.

12. 13.

14.

16. Schubarth, S.K., et al: Moxa Arch Frontier Development Success Through Increased Fracture Conductivity Part 2, paper SPE 30717 presented at the 1995 Annual Technical Conference, Oct 22-25. 17. Jin, L, and G.S. Penny: A Study on Two Phase, Non-Darcy Gas Flow through Proppant Packs, paper SPE 49248 selected for presentation at 1998 Annual Technical Conference and Exhibition, Sep. 27-30. 18. Holditch, S.A. and R.A. Morse: The Effects of Non-Darcy Flow on the Behavior of Hydraulically Fractured Gas Wells, paper SPE 5586, JPT, Oct 1976. 19. Gidley, J.L.: A Method for Correcting Dimensionless Fracture Conductivity for Non-Darcy Flow Effects, paper SPE 20710, SPEPE, Nov 1991. 20. Agarwal, R.G. et al,: Evaluation and Performance Prediction of Low-Permeability Gas Wells Stimulated by Massive Hydraulic Fracturing, paper SPE 6838, JPT, Mar. 1979. 21. Holditch, S.A.: Criteria of Proppant Agent selection, prepared for the Norton Company, 1979. 22. Elbel, J.L.: Considerations in Fracture Design, in Economides and Noltes Reservoir Stimulation, Prentice Hall, New Jersey (1989) p 9-1. 23. Pearson, C.M. et al,:Optimal Fracture Stimulation of a Moderate Permeability Reservoir, Kuparuk River Unit, paper SPE 20707 presented at the 1990 Technical Conference, Sept 23-26. 24. Pospisil, G. et al, Results of a Large-Scale Refracture Stimulation Program, Kuparuk River Unit, Alaska, paper SPE 24857. 25. Martins, J.P. et al, Tip Screenout Fracturing Applied to the Ravenspurn South Gas Field Development, paper SPE 19766. 26. Schubarth, S.K. et al, Understanding Proppant Closure Stress, paper SPE 37489 presented at the 1997 Production Operations Symposium, March 9-11.

15.

Fig. 1 - Comparison of Viscous and Inertial Flow Effects

50 Inertial Forces (Forchheimer Correction) Pressure Drop per Foot of Frac Length (psi/ft) 40 Viscous Forces (Darcy's Law)
Conditions: 50' Frac Height

30

2#/sq ft ->0.15" after embedment 2000 psi BHFP, 200 F 20/40 Jordan Sand 2000 psi closure stress Reference Perm = 243 Darcy 55% gel damage

20

10

0 0
Reference Conductivity measured at this rate

3 Well Flowrate (MMSCFD)

SPE 54630

NON-DARCY AND MULTIPHASE FLOW IN PROPPED FRACTURES: CASE STUDIES ILLUSTRATE THE DRAMATIC EFFECT ON WELL PRODUCTIVITY

Fig. 2 - Beta Factor Comparison, Common Proppants

0.006
(atm sec^2/gram)

0.005 0.004 0.003 0.002 0.001 0.000

LWC 20/40 LWC 16/20 RCS 20/40 RCS 16/30 Sand 20/40 Sand 16/30

Beta

Data: SLFrac, July 98, Gas Well, 6000 psi stress, 200F, 3e6 Modulus, 50% gel damage, 2#/sq ft

Fig. 3 - Impact of Multiphase Non-Darcy Flow

20/40 Proppants at 2 lb/sq ft, 6500 psi and 225 F Adapted from SPE 30494
4002

4000 Jordan Sand Resin Coated Sand Light Weight Ceramic 3000

Conductivity (md-ft)

2032

2000
1354

1000
646 467 152 27 246 112 14 56 123

0
Reference Conductivity, Laminar Flow With Non-Darcy Flow With Multiphase Non-Darcy Flow With 50% Gel Damage (assumed)

???
With Long Term degradation, multiple fractures, pressure cycling

Reference Conductivity Conditions: Velocity equivalent to 65 MSCFD dry gas from bi-wing, 50 fracture height at 3000 psi bottom hole pres. Non-Darcy Flow Conditions: Velocity equivalent to 1 MMSCFD dry gas from bi-wing 50 fracture height at 3000 psi bottom hole pressure. Multiphase Flow Conditions: Equivalent of 10 bwpd added to 1 MMSCFD of dry gas at same conditions.

10

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

Table 1. Deep Gas Well, Proppant Stress = 9279 psi Disregarding Includes non-Darcy Multiphase and Effects, Non-Darcy Effects Disregards multiphase Flow 2781 423 3614 820 3763 1756 78 1.5 387 4.2 717 15.4 1.4 0.03 7.0 0.08 12.9 0.28 3773 717 4457 1352 4662 2718 20/40 LWC 20/40 RCS 20/40 LWC 20/40 RCS 20/40 LWC < 2 months < 2 months 3600% 3419% 20/40 LWC < 2 months < 2 months 3342% 7696% Including both non-Darcy and Multiphase flow* 354 470 1578 0.4 2.1 8.8 0.01 0.04 0.16 602 776 2507 20/40 LWC < 2 months < 2 months 916% 7327%

Analytical Predictions: Initial Production Rate (x1000 scfd) Effective Conductivity (md-ft) Dimensionless Fracture Conductivity (Fcd) Cash Flow after 3 years of production ($000, net of proppant cost) Most Economic Proppant for assumed frac geometry Time required to payout incremental investment for premium proppant % return on incremental investment after three years

Proppant 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC

Table 2. Shallow Gas Well, Proppant Stress = 1780 psi Disregarding Includes non-Darcy Multiphase and Effects, Non-Darcy Effects Disregards multiphase Flow 711 614 711 502 716 635 1769 282 1886 180 2633 387 32 5.1 34 3.2 47 7.0 815 722 797 615 794 711 20/40 Sand 20/40 RCS 20/40 LWC 20/40 RCS 20/40 LWC never never -95% -81% 20/40 Sand never ~ 4 years -563% -42% Including both non-Darcy and Multiphase flow* 461 428 520 132 86 198 2.4 1.6 3.6 599 550 622 20/40 LWC never 1 year -258% 88%

Analytical Predictions: Initial Production Rate (x1000 scfd) Effective Conductivity (md-ft) Dimensionless Fracture Conductivity (Fcd) Cash Flow after 3 years of production ($000, net of proppant cost) Most Economic Proppant for assumed frac geometry Time required to payout incremental investment for premium proppant % return on incremental investment after three years 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC

SPE 54630

NON-DARCY AND MULTIPHASE FLOW IN PROPPED FRACTURES: CASE STUDIES ILLUSTRATE THE DRAMATIC EFFECT ON WELL PRODUCTIVITY

11

Table 3. Moderate Depth Oil Well, 5 lb/sq ft 20/40 Proppants, Proppant Stress = 3400 psi Disregarding Includes non-Darcy Multiphase and Effects, Non-Darcy Effects Disregards multiphase Flow 724 280 734 293 747 654 5903 484 7863 515 16336 1677 0.66 0.05 0.87 0.06 1.82 0.19 5607 2482 5635 2556 5704 5124 20/40 LWC 20/40 RCS 20/40 LWC 20/40 RCS 20/40 LWC 1.25 years 5 months 122% 313% 20/40 LWC 6 months < 1 month 322% 8523%

Analytical Predictions: Initial Production Rate (bopd) Effective Conductivity (md-ft) Dimensionless Fracture Conductivity (Fcd) Cash Flow after 3 years of production ($000, net of proppant cost) Most Economic Proppant for assumed frac geometry Time required to payout incremental investment for premium proppant % return on incremental investment after three years 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC

Table 4. Lower Rate Oil Well, 4 lb/sq ft 20/40 Proppants, Proppant Stress = 3160 psi Disregarding Includes non-Darcy Multiphase and Effects, Non-Darcy Effects Disregards multiphase Flow 280 97 283 104 293 266 3193 261 4089 300 8392 931 0.53 0.04 0.68 0.05 1.40 0.16 2043 728 2031 753 2067 1813 20/40 LWC 20/40 RCS 20/40 LWC 20/40 RCS 20/40 LWC never 1.25 years -40% 59% 20/40 LWC 1.6 years < 1 month 83% 2646%

Analytical Predictions: Initial Production Rate (bopd) Effective Conductivity (md-ft) Dimensionless Fracture Conductivity (Fcd) Cash Flow after 3 years of production ($000, net of proppant cost) Most Economic Proppant for assumed frac geometry Time required to payout incremental investment for premium proppant % return on incremental investment after three years 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC 20/40 Sand 20/40 RCS 20/40 LWC

12

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

Fig. 4 Incremental Cash Flow due to Proppant Selection Gas Well

4a) Disregards Multiphase and Non-Darcy Effects 4b) Disregards Multiphase Flow, Includes Non-Darcy Effects 4c) Includes Multiphase and Non-Darcy Effects

SPE 54630

NON-DARCY AND MULTIPHASE FLOW IN PROPPED FRACTURES: CASE STUDIES ILLUSTRATE THE DRAMATIC EFFECT ON WELL PRODUCTIVITY

13

Fig. 5 Incremental Cash Flow due to Proppant Selection Oil Well

5a) Disregards Multiphase and Non-Darcy Effects 5b) Includes Multiphase and Non-Darcy Effects

14

M. C. VINCENT, C. M. PEARSON, J. KULLMAN

SPE 54630

Appendix: Input Data for All Analyses in Report

Shallow Gas Well, Same perm and frac Table 2 Shallow Gas with and w/o with and w/o Gas 3000 25 10% 0.1 1800 150 0.6 40% 0/10 0/8 n/a 800 30 API 800 0.7 180 55 2.5 3400 0 9 40% off book Sand 20/40 "Premium" RCS 20/40 Light Weight Ceramic 20/40 5 20% 0 8 40% off book Sand 20/40 "Premium" RCS 20/40 Light Weight Ceramic 20/40 4 50%
very low stress, low rate Decision changes w/ mp, nD

Results Shown Case Include Non-Darcy Effects? Include Multiphase Flow? Well Depth (ft TVD) Net pay (ft) Form Porosity Form Perm (md) Initial Res pressure (psi) Res Temp (F) Gas gravity Form Water Saturation Water Rate (bpd) Cond/Oil Rate (bpd) Oil API Gravity Bottom hole pressure (psi) 25 API 350 0.54 200 80 3 3160 0.86 555 50 5 9279 1.8 0 40% off book Sand 20/40 "Premium" RCS 20/40 Light Weight Ceramic 20/40 1.5 50%
Even at low stress, ceramic makes if formation perm high enough Decision changes w/ mp, nD Pick Ceramic regardless Overestimate rate if disregard mp and D

Mid-Continent Deep Gas Well Table 1 Deep Gas with and w/o with and w/o Gas 12534 25 10% 0.1 7000 220 0.6 40% 0/10 0/8 n/a 1500 0.86 555 50 5 1780 varies with depth/stress 500 100 4.5 2000 to 9000 2 0 40% off book Sand 20/40 Standard and Premium RCS Light Weight Ceramic 20/40 2 50%
Shows where inaccurate models lead to incorrect proppant l ti

Oil; moderate perm (Kuparuk) Table 3 A Sand, Oil with and w/o partial Oil 6000 30 22% 50 3000 155 0.6 20% 0

Low Rate Oil (Venezuela) Table 4 Low Rate with and w/o partial Oil 6500 60 16% 30 1500 150 0.7 25% 10

3-dimensional plots-GAS Figure 4 Various depths with and w/o with and w/o Gas Various depths 50 20% 0.001 - 1.0 varies with depth/stress 250 0.65 30% 0/10 0/10 n/a varies with depth/stress

3-dimensional plots-OIL Figure 5 Various depths with and w/o partial Oil Various depths 25 20% .01 - 10 varies with depth/stress 200 0.65 20% 0/10 30 API varies with depth/stress varies with depth/stress 250 50 3.5 2000 to 9000 0 12 40% off book Sand 20/40 Standard and Premium RCS Light Weight Ceramic 20/40 2 50%
Shows where inaccurate models lead to incorrect proppant l ti

Frac gradient (psi/ft) Frac half-length (ft) Frac height (ft) Young's Modulus (million psi) Stress on Proppant

Gas Price ($/mscf) Oil Price ($/bbl) Prop Pricing Scenario Proppant Type 1 Proppant Type 2 Proppant Type 3 Proppant Conc (#/sq ft) Prop damage (gel, %)

1.8 0 40% off book Sand 20/40 "Premium" RCS 20/40 Light Weight Ceramic 20/40 1.5 50%

conclusions: (data/results in Figures)

Pick Ceramic regardless Overestimate rate if disregard mp and D