Water-fracs, consisting of proppant pumped with un-gelled fluid are the type of stimulation used in many low-permeability reservoirs throughout the United States. The use of low viscosity, Newtonian, fluids allows the creation of long narrow fractures in the reservoir without the excessive height growth often seen with cross-linked fluids. Proppant transport is a central issue in all these treatments because of the low viscosity of the fracturing fluid. New models for proppant transport and settling in hydraulic fractures were developed and implemented in a 3-d hydraulic fracturing code. It is shown that a simple Stokes settling model is grossly inadequate. The proppant settling models developed in this paper account for the effects of fracture walls, changes in settling velocities and rheology caused by changes in proppant concentration, turbulence effects due to high fluid velocities and inertial effects associated with large relative velocities between the proppant and the fluid. Narrower fractures, higher proppant concentration and smaller proppant size reduce settling whereas turbulence leads to an increase in settling. Results are presented to show how the settling velocities are impacted by fluid velocity, proppant size, fluid rheology and fracture width. In most instances settling velocities differ significantly from the Stokes settling velocity. The new proppant settling model was incorporated into a 3-D hydraulic fracture simulator (UTFRAC-3D). Simulation results show that when settling is accounted for, significantly shorter propped lengths are obtained. The narrow fractures associated with water-fracs alter settling and thereby alter the proppant placement significantly. Although increasing fluid viscosity can reduce settling rates, increased height growth reduces the distance to which proppant can be placed. This clearly suggests a need to optimize fluid rheology. The improved fracture simulator can be used to better design fracture treatments (fluid rheology, injection rates, proppant concentration and size) for better proppant placement under a given set of in-situ stress conditions. Introduction Water-fracs are commonly applied in low permeability gas reservoirs. These treatments involve pumping low viscosity ungelled fracture fluids. The low viscosity of the slick water leads to long created fracture lengths. However, due to high settling velocities of the proppant in the low viscosity fluid, the propped lengths achieved can be very small. Modifications to the water-frac stimulation design are needed to transport proppant further out into the fracture. This requires suspending the proppant until the fracture closes without generating excessive fracture height. Proppant transport clearly is a central issue in all these treatments. An improved proppant transport model is presented that can accurately model proppant transport when either un-gelled or cross-linked fluids are used to place the proppant. The use of this proppant transport model will allow engineers to customize treatment designs for individual wells. The complete model for proppant transport in hydraulic fractures was incorporated into UTFRAC-3D, a fully three-dimensional hydraulic fracture simulator. The proppant transport equations were solved on an adaptive finite element mesh. The settling of the proppant was modeled taking into account the change in settling velocities and rheology due to changes in proppant concentration, turbulence effects due to high fluid velocities, and inertial effects associated with large relative velocities between the proppant and the fluid. Inertial effects become significant at high settling velocities (Rep>2) and are discussed in the next section. The effects of particle concentration, fracture width and turbulence are discussed in the following sections. An example calculation is shown to demonstrate the importance of each of the correlation factors applied to the Stoke's settling velocity. Finally, the settling correlations are incorporated into a proppant transport model in a fully 3-D fracture simulator (UTFRAC-3D). Results from the model are discussed in the last section. The project is co-funded by the US Department of Energy - National Energy Technology Laboratory.

In order to better understand the process of fracturing tight gas sands, a comprehensive data set was gathered and analyzed for the Bossier formation in the Dowdy Ranch field. The dataset collected on the APC Anderson #2 well represents one of the most comprehensive datasets ever collected for a commercial gas well. The entire interval was cored, and a complete set of core analysis was performed across the sands. Stress profiles derived from dipole sonic logs across the pay zone and in the shales below were calibrated with stress tests. The frac jobs were micro-seismically monitored with downhole geophones and included breakdown and mini-frac stages. Post fracture data collection included pressure buildup testing, production logs with multiple passes, and tracer logs with multiple isotopes. Additional data was collected on five offset wells in the field. Results from these wells will be presented in a companion paper. The bottomhole treating pressures were found to be higher than expected based on the measured stress profiles. However, the higher treating pressures encountered did not result in excessive fracture height growth. This may be partially attributed to unexpected faulting providing a conduit for fluid leak-off, resulting in low efficiency and narrow fractures. Propped or effective fracture lengths derived from pressure buildup analysis and history matching production data were significantly shorter than designed frac lengths (or those predicted from uncalibrated frac models). The net pressure plots showed some evidence of proppant bridging even at low proppant concentrations, again indicating only limited fracture widths were being achieved. The data collected and analyzed provide valuable insight into the performance of water and hybrid fracs in tight gas formations. Recommendations for the design of future fracture treatments are made based on the findings. Introduction Water-fracs and hybrid-fracs are the type of stimulations used in many low-permeability reservoirs in East Texas and throughout the United States. Improving the performance of gas wells in marginal gas plays typically requires improvements to fracturing technologies. To ensure that these gas wells remain economically attractive even at modest gas prices, completion and fracturing practices must be optimized. Research aimed at improving the performance and reducing the cost of fracture treatments by acquiring and analyzing fracture treatment data from the Bossier play in East Texas has been undertaken by The University of Texas at Austin and Anadarko Petroleum Corporation (APC). The project is co-funded by the US Department of Energy - National Energy Technology Laboratory. Six gas wells (data wells) were identified in the Dowdy Ranch field for a detailed analysis of the well and fracture treatment data prior to, during, and after a fracture treatment. This paper focuses on the data acquired in the APC Anderson #2 well, summarizes our findings thus far and presents recommendations for better fracture treatments based on the lessons learnt. Some relevant results from the offset wells are also discussed here. A detailed analysis of data from the other wells will be presented in a later publication. Background In many low-permeability gas reservoirs, water-fracs and hybrid fracs are common stimulation methods. These treatments involve creating fractures using slick water rather than cross-linked gels. The proppant is carried into the fracture either with a gelled fluid (hybrid frac) or slick water (water frac). In theory, low viscosity pad fluids allow creation of more confined fractures. However, the low viscosity promotes proppant settling and, therefore, can lead to poor proppant placement and also leads to narrower fractures, which in some cases can cause proppant bridging.

Most water injection wells in waterflooded reservoirs have fractures that grow with time. These fractures can have a significant impact on the reservoir performance (oil production rates, o/w ratio and ultimate recovery). A single well model that predicts the length of injection well fractures by modeling fracture growth due to fracture face plugging and thermal stresses is coupled with a reservoir simulator to simulate injection wells that have fractures that dynamically grow with time. The relative importance of the injected water quality, formation permeability, injection rate and the temperature of injected water on the rate of fracture growth are demonstrated. The single well model accurately accounts for all the physics of fracture growth at the injector while the reservoir simulator accounts for the large-scale reservoir structure. The presence of high conductivity fractures in the injectors affects the waterflood sweep efficiency. Results indicate that fracture orientation, rate of fracture growth, injection water quality and reservoir heterogeneity play an important role in determining the oil production rates and ultimate recovery. The results of the simulations can be used to set injection well pressures and rates, specify water quality and to select injection well patterns to maximize oil recovery. 1. Introduction Waterflooding is the most widely used improved oil recovery method. Injection of sea and surface water is common in mature fields. Water injection is also used for pressure maintenance. Most water injection wells have fractures that grow with time. Fractures can be initiated in injection wells due to thermal stresses, changes in pore pressure and an increase in injection pressure due to particulate plugging. These fractures may have a significant impact on the reservoir performance. Unlimited fracture growth can have a number of undesirable consequences. The presence of growing high permeability fractures in the injectors distorts the water flood fronts. Depending on the position of the injection wells, this may result in poor sweep efficiency and consequently in premature water breakthrough. Unconstrained vertical growth of fractures can connect water and hydrocarbon bearing zones. An important factor to be considered here is that most injection wells are not fractured at the commencement of injection. Rather, fracturing is induced during the course of injection and these fractures grow with time. Hence, in determining the effect of injection well fractures on the reservoir, these injection well fractures cannot be modeled as static fractures of fixed length and conductivity. The growth of these fractures with time needs to be taken into account. Traditional reservoir simulators allow only for fixed fracture lengths in injectors and producers. Hence, our approach to modeling the effect of injection well fracturing involvesDeveloping a single well model to predict fracture growth in injection wells due to particle plugging, thermal and pore pressure effects.Coupling the single well model to a reservoir simulator to study the effect of injection well fracturing on oil recovery.

Summary In conventional gel-fracturing treatments, the damage induced by the gel can have a significant impact on well performance, particularly in low-permeability gas formations. Slick-water fracs have been shown to be more successful in some tight gas formations because of reduced gel damage and limited height growth. Proppant placement is a major concern in water fracs. Hybrid water fracs (i.e., using slick water as the pad fluid and gel to place the proppant) provide improvements to the performance of water fracs. This paper proposes a new method, reverse-hybrid fracs (RHF), for the efficient placement of proppant deep into created fractures while minimizing gel-induced damage. Experiments were conducted in a simulated fracture (i.e., slot cell) to study the transport of proppant. Slick water was injected first into the slot, followed by gel. Finally, slick water containing proppant was injected to displace the gel. The water that carried the proppant quickly formed viscous fingers through the gel. The gel was observed to form long, thin layers that effectively hindered proppant settling and helped transport the proppant further into the slot. This resulted in the formation of proppant packs above the gel layers. In this paper, experimental results are presented to show how the gel layers distribute in the slot and how proppant distribution is affected by the gel layers. A transparent cell composed of rough walls was set up to investigate the effect of fracture wall roughness. The effects of fluid viscosity ratio, fracture wall roughness, and gel pad volume were investigated. Based on the experiments and scaling relations, recommendations are made for the pumping sequence and the size of the gel and slick-water stages in reverse-hybrid fracture treatments. This method of proppant placement requires less gel than conventional gel fracs. Other possibleadvantages include:less gel damage to the proppant pack,limited height growth, andless penetration of the pad fluid into the formation resulting in shorter cleanup time following fracture treatment. Introduction In a fracturing treatment, the effective fracture lengths achieved (e.g., measured by matching the production response or from pressure-transient tests) can quite often be significantly smaller than the created fracture lengths (e.g., measured by fracture-mapping techniques). This difference can be attributed to inadequate proppant transport and/or insufficient fracture cleanup. In hybrid fracs, low-viscosity and slick water are used to create the fracture, while a high-viscosity gelled fluid is used in the proppant stage. In some fields, the use of hybrid fracs has resulted in a significant increase in effective fracture lengths and well productivity (Sharma et al. 2004). The use of hybrid fracs can, under certain conditions, result in the increased possibility of tip screenout. Because the fracture is being created with a high leak-off, low-viscosity fluid, the smaller fracture widths can result in premature tip screenout. This is the primary motivation for using reverse-hybrid fracs. As the name suggests, the sequence of fluid injection is the reverse of what is used in hybrid fractures; a high-viscosity polymer (e.g., linear or cross-linked) fluid is used to create the fracture while the proppant is pumped, behind this high viscosity pad, into a low-viscosity fluid. This kind of treatment is referred to as reverse-hybrid fracs (RHF) in this paper.

A model for the velocity of proppant particles in slot flow is presented. The proppant is either retarded or accelerated relative to the fluid depending on the ratio of the proppant size to the fracture width. It has been found that when this ratio is small, the proppant travels faster than the average fluid velocity at that location because the proppant tends to be confined to the center of the flow channel where the fluid velocity is higher. As the proppant size increases, the effect of the fracture walls becomes more important and the proppant is retarded by the walls. The retardation of particle relative to the fluid is greater for larger particles and greater proximities to the fracture walls due to the hydrodynamic stress exerted on the sphere by the walls in the narrow gap. A higher proppant concentration restricts the area available to flow and increases the drag forces on the particles. A model is presented for the effect of fracture walls and proppant concentration on proppant transport. The effect of this increased drag force is accounted for by modifying the wall - particle interaction. The influence of the surrounding proppant spheres on the drag force on a particle is estimated from the effect of a wall on the drag force acting on a single particle. The equivalent hydraulic diameter is then used to determine the proppant retardation. The effects of wall roughness and fluid leakoff are discussed. Models are suggested that capture these first order effects. The new model for proppant retardation has been incorporated into a 3D fracture simulator. Results show that the proppant placement is substantially different when proppant retardation/acceleration is considered. Comparisons of propped fracture lengths obtained with the new model agree much better with propped and effective fracture lengths reported in the field. 1.Introduction Hydraulic fracturing is a commonly used stimulation technique. Proppant transport is a key factor in determining the productivity of these fractured wells. Water fracs are common stimulation treatments for low permeability gas reservoirs. These treatments use low viscosity Newtonian fluids to create long narrow fractures in the reservoir, without the excessive height growth that is often seen with cross-linked fluids. The low viscosity fluid and the narrow fractures introduce some significant challenges for proper proppant placement. The low viscosity of the carrying fluid leads to high settling velocities for the proppant. The narrow fractures created can have widths comparable to the diameter of the proppant and can alter proppant transport significantly due to the hydrodynamic forces acting on the proppant because of the fracture walls. Other proppant particles create additional hydrodynamic drag forces leading to retardation. Fracture diagnostic studies that have been reported in the literature have observed that the effective propped lengths for both water fracs and conventional gelled fracs are sometimes significantly different than those predicted by fracture models. Designed and created fracture lengths are usually much longer than the effective fracture lengths obtained from post production analysis[1–4]. They can sometimes be an order of magnitude lower. Proppant transport is a key factor determining the effective propped lengths and therefore the productivity of these fractured wells. In current hydraulic fracture models, the proppant is assumed to flow with the fluid in the direction of fracture propagation. It is shown in this paper that the proppant usually flows at a different velocity than the fluid, particularly in narrow fractures.It is important to develop reliable models to predict proppant transport. A detailed model for proppant settling in water fracs was presented earlier by the authors5. Several correlations for modeling proppant settling in water fracs were presented. These UTFRAC correlations allow fracture models to correct the settling velocity for inertial effects, proppant concentration, fracture width and turbulence. The models were implemented in a 3-D hydraulic fracture simulator and results showed that propped fracture lengths could vary significantly when settling was properly accounted for.