Aquifer test interpretation with special emphasis on the drawdown evaluation for wells within fracture networks smaller than the representative elementary volume (rev)

[摘要] Fractured aquifers are characterized by the fact that most of the water flows alongfractures, faults, open bedding planes, or other geological features. These features areembedded in a matrix that has either porous nature, like in sandstone, or is almostimpermeable (inert), as in the case of granite.It is often observed that in fractured aquifers the measured air lift yield is a strongoverestimation of the long-term sustainable yield of the well. The explanation for thiseffect is that the water extracted initially is provided by a geological feature that ishigh yielding but limited in its extension, while the long-term sustainable yield is theresponse of the matrix. Such a geological feature can be among others, a singlevertical fracture or a fracture network, which usually acts as a preferential flow path.Pumping tests in primary and secondary aquifers are widely used by the groundwater industry because they provide important information on the reservoir and thewell performance. Various researches in the oil and ground water industries havefound that the presence of single preferential flow paths results in characteristicdrawdown curves. However, a lack of research is encountered, when it comes to morecomplex fracture networks. This work investigates the behavior of drawdown curvesin fracture set ups below the representative elementary volume (REV), which isdefined as the smallest volume of aquifer that can be considered as a homogeneousfractured unit. Emphasis is given to the importance of a thorough diagnosis of the datato be able to adequately estimate the aquifer properties.Chapter 2 of the present work summarizes the basic knowledge on ground waterflow in fractured reservoirs, where the REV, fracture connectivity, and conductivitycontrast between fracture and matrix are defined and explained. Thereafter, the flowbehavior in fractured media (linear, radial, and spherical) are described. This chapterends with the review of various well and reservoir boundary effects, such as well borestorage, well bore skin, partial penetration skin, fracture skin, pseudo-skin, fracturedewatering, and reservoir boundaries.Chapter 3 gives practical advice for the planning and performance of pumping testsand stresses the necessity of time correction in the case of variable discharge rateduring the test. The importance of the pseudo-skin effect originated by the presence ofa single vertical fracture is highlighted. It is shown that pseudo-skin effects are thereason for the apparent dependence of the storage coefficient (S) on the distancebetween the observation borehole and the single vertical feature, when the commonevaluation methods are used for the estimation of S. Furthermore, the radial-actingflow phase and in relation to the REV is explained. This chapter ends with thedescription of various diagnosis tools, which allow, among others, the determinationof the flow phases from pumping test data influenced by preferential flow paths.These tools are included in the computer program Test Pumping Analysis (TPA),which was compiled under the umbrella of this thesis. It is explained that dataconsistency can be rapidly analysed with the comparison between drawdown andrecovery data and any discrepancy must be investigated additionally. The use ofstraight-lines, especial plots, and curves derivatives is described.Chapter 4 presents the most important analytical and semi-analytical availablesolutions for the analysis of pumping test data in fractured aquifers, which areincluded in TPA. For each case, the mathematical solution is first described. Theinfluence of well bore and reservoir effects are explained using TPA, based ontheoretical and field examples. Special emphasis is given to the various skin analysesand to the possible misinterpretation of drawdown curves. The solutions presentedare: double porosity model of Moeneh (1984)�?single vertical fracture with infinite conductivity and finite extent of Gringarten etal. (1974)�?single vertical fracture with finite conductivity and finite extent of Cinco-Ley etal. (1978)�?single vertical dike with finite conductivity and infinite extent of Boonstra &Boehmer (1986)�?bedding plane fracture with infinite conductivity and finite extent of Gringarten &Ramey (1974)�?generalized radial flow model for fractured reservoirs of Barker (1988)Chapter 5 investigates more complex fracture situations with help of numericalmodelling based on the Darcian law. Synthetic pumping tests are simulated and theirdrawdown behavior is analysed. The single vertical fracture case is first computed toensure that the model set up leads to the analytical and semi-analytical solutions ofGringarten et al. (1974) and Cinco-Ley et al. (1978), respectively. To investigate theinfluence of wider fault zones, which are assumed as a homogeneous fractured zone,faults with increasing width are modelled. It is found that:�?for large storage capacities and finite conductivity, the drawdown at early timeshows a radial-acting flow phase within the fault, which could be easilymisinterpreted as double porosity. However, this effect occurs most likely underunconfined conditionsThe model is then modified to include parallel vertical fractures. It is found that:�?parallel vertical structures with infinite conductivity have no influence on thedrawdown at the well�?parallel vertical structures with finite conductivity show minor influences at earlytime, if the dimensionless relative separation Sr (Sr = df/xf) is less than 0.125Thereafter, the model is modified to represent a crossed fracture case and a bendfracture case, both vertical and with infinite conductivity. The computed drawdowndiffers significantly from the drawdown measured in the single straight fracture. It isfound that:�?this drawdown is comparable to that obtained with the uniform flux solution ofGringarten et al. (1974), although the influx along the fracture is not uniform.However, the authors mentioned that some field data from hydraulic fracturing fitbetter to the uniform flux solution. The results of this work give reasons to believethat such field data are attributed to more complex fracture networks similar tothose studied here.The horizontal bedding plane case is also investigated. First, the model is run tocompute the infinite and finite flux solutions from Valkó & Economides (1997). Themodelled curves fit adequately the data for their solutions, although a labelling errorin the published data is identified. Further, the influence of the fracture geometry isanalysed. It is found that:�?horizontal penny-shape fractures and square features with equivalent influx areahave the same drawdown�?rectangular horizontal features have a significant influence on the drawdownbehavior The investigation of parallel bedding planes shows that:�?the shape of the drawdown curve in parallel horizontal fractures is equivalent tothat of the single horizontal bedding plane. Therefore, without additional on-siteinvestigations (e.g. fluid logging or flow meter measurements) it is impossible todetermine whether the drawdown belongs to a single fracture or to a series ofparallel features�?The analysis of drawdown curves produced by parallel horizontal fractures usingtype curves for single horizontal fractures leads to an over estimation of thefracture radius. This effect is important among others, for the design of protectionzonesFinally, intersections of a single vertical fracture and a single horizontal beddingplane are modelled. It is found that:�?the obtained drawdown curves could be misinterpreted with drawdown curves ofsingle cases. Therefore, it is concluded that additional information is necessary tocorrectly identify the geological set up. This issue is highly important for both thedesign of well protection zones and the estimation of the transport time

[发布机构] University of the Free State