Recent developments in well completion technologies have transformed the unconventional reservoir systems into economically feasible reservoirs. However, the uncertainty associated with production forecasts and non-uniqueness related with well/reservoir parameter estimation, are the main issues in future development of these reservoirs. In addition, recent operational methods such as restricting rates by decreasing the choke size add up to the uncertainty in production forecasts. This work attempts to investigate the effect of production practices on ultimate recovery. It is observed that wells producing in the Haynesville shale gas play exhibit severe productivity loss throughout their producing life. Production practices such as controlling the drawdown or restricting rates by decreasing the choke size are employed by several operators to deal with the severe productivity loss. In this work our main objective is to investigate the issues (such as stress dependent permeability, proppant embedment, operational problems, etc.) contributing to decreasing well productivity over time. In particular, from modeling standpoint, we focus on stress-dependent permeability as a mechanism, which affects well performance over time. Using a horizontal well with multiple fractures numerical simulation model coupled with geomechanics, we generate synthetic simulation cases including several drawdown scenarios. It is shown that high drawdown cases result in higher effective stress fields around the well and fracture system. We therefore infer that higher effective stress fields result in lower well productivity over time. Based on this hypothesis and diagnostics of field data, we model two different scenarios (i.e., high drawdown and low drawdown cases) for a horizontal well with multiple fractures using two different permeability decay functions and same well/formation model parameters. Our modelling results indicate that low drawdown case yields higher recovery suggesting that rate restriction could be a mitigating factor in decreasing well productivity over time.

This paper presents a new insight into rate transient analysis using the beta-derivative function (β-derivative). Production rates and flowing pressures from tight gas and shale gas wells were analyzed using various implementations of the betaderivative to emphasize different features of the data and, as a result, reveal characteristic information about flow regimes and the extent to which the reservoir has been drained. The beta-derivative was applied to rate, pressure and normalized rate, and the effect of skin on the β-derivative was also investigated. The intent was to determine which format is the most useful for diagnosing the dominant flow regimes or the sequence of flow regimes that have occurred while producing from an unconventional hydrocarbon reservoirs (tight gas, shale gas and light tight oil). It was found that the classic signature of the β-derivative is altered by the presence of skin. Also, the derivative based on constant rate is different from that based on constant pressure. The beta-derivative's diagnostic value was compared to that of the Bourdet Derivative and the Primary Derivative The β-derivative has significant diagnostic value for identifying power-law type of flow regimes (such as wellbore storage, linear flow, bilinear flow, boundary-dominated flow, etc) because it possesses a recognizable unique character for each of these flow regimes. For instance, the β-derivative is 0.5 for linear flow, 0.25 for bilinear flow and 1.0 for boundary dominated flow. In addition, since the β-derivative is dimensionless, it can be used to differentiate the performance of wells producing from the same field or from different resource plays. The new plotting functions presented in this paper are not intended to replace existing diagnostic functions but can be used in conjunction with them to enhance production data analysis.

Arps Decline Curve Analysis (DCA) has been the standard for evaluating expected ultimate recovery (EUR) in oil and gas wells since the 1950's. Although this empirical method has served the conventional petroleum industry well, its misapplication to wells in ultra-low permeability plays, most notably shale, often yields ambiguous results due to invalid assumptions. Depending on the application of DCA, these forecasts, with constant hyperbolic decline exponent (‘b-value’) assumptions, have proven to be either overly optimistic or overly pessimistic. The applicability of the Arps method is limited to wells exhibiting boundary dominated flow. However shale wells exhibit transient flow for years, making them unsuitable candidates for the conventional approach of DCA. With the industry now focused on shale, new tools are required to help reduce uncertainty in well forecasts. This paper presents a two-part study on the forecasting of shale gas well production. Part one utilizes a recently developed trilinear flow model (Ozkan et al. 2009) to improve forecast reliability while preserving simplicity for practical application. It is designed to account for 1) considerations of multi-fractured horizontal wells (MFHW), 2) improved localized effective permeability within the stimulated reservoir volume (SRV), 3) regional permeability contributions to the SRV and, 4) potential contributions of adsorbed gas to expected ultimate recovery. A workflow using this model is proposed and exemplified using production data from two gas wells from the Marcellus Shale (North-East Pennsylvania). Although analytical models provide a deeper understanding of shale gas reservoirs as well as more reliable forecasting capability, Arps decline curves are, and will continue to be, the language of the reserves evaluation. That is why it is useful to understand the dynamic behavior of the decline exponent resulting from flow regime transitions during the transient period. Part two of this paper highlights an important consideration in the application of DCA to shale: the effect of desorption on the b-value. It will be shown that when desorption is invoked, the b-value will increase. This difference has been quantified for a MFHW in the Marcellus using a Langmuir adsorption isotherm representative of the region.

The application of "Decline Curve Analysis" (DCA) in unconventional reservoirs is almost always problematic. The Arps relations (hyperbolic and exponential relations) have been the standard for evaluating estimated ultimate recovery (EUR) in petroleum engineering applications for more than 80 years. However; with the pursuit of low and ultra-low permeability plays, these relations often yield ambiguous results due to invalid assumptions (e.g., existence of the boundary-dominated flow regime, presumption of a constant bottomhole pressure, etc.). Misapplications of the Arps' relations to production data exhibiting long-term, transient flow generally results in significant overestimates of reserves-specifically when the hyperbolic relation is extrapolated unconstrained, using an Arps b-value greater than 1. We note that the "modified hyperbolic" relation-one with an initial (unconstrained) hyperbolic trend used during early times, coupled with an exponential decline trend using a standard terminal decline can be used effectively (with proper care) for predicting EUR and production extrapolations. However; we note that this approach is "non-unique" in the hands of most users, and often yields widely varying estimates of reserves with time, and/or "consistent" estimates of reserves, which are highly biased. In short, the modified hyperbolic relation can be effectively applied to production data from low/ultra-low permeability reservoirs systems, these analyses must be based on diagnostic interpretations of the data (as we have proposed earlier [Ilk et al. (2008)]), where multiple data functions are used to define the analyses. The use of diagnostics is a necessary, not a sufficient condition-the underlying models must be able to characterize the selected flow regimes, and there must also be constraints applied to production extrapolations and EUR predictions. The issues related to the use of Arps' rate decline relations have led various authors [Ilk et al. (Power Law Exponential, 2008), Valkó (Stretched Exponential, 2009), Clark et al. (Logistic Growth Model, 2011), and Duong (2011)] to propose various rate decline relations which attempt to properly model the time-rate behavior-specifically early transient and transitional flow behavior. However, none of these equations can be considered sufficient to forecast production for all unconventional plays, due to the characteristics and operational conditions of each play and the behavior of the time-rate equation. In other words, one equation could work very well in a specific play, but could possibly perform poorly in another play. Under these circumstances, it is critical to understand the behavior of each equation, and to apply these relations appropriately for production forecasts. This work presents guidelines for the application of the various time-rate relations currently being deployed in the petroleum industry. The results of time-rate analyses of wells from three different plays are presented, and the advantages/ disadvantages of each time-rate relation are disc...

The Montney gas play in NE British Columbia and NW Alberta is undergoing intense development utilizing multi-stage fracturing of horizontal wells. Problems persist in determining an optimal development strategy for each area within the Montney. For the Groundbirch area, operators are in the early stages of development. As a consequence they have invested significantly in well monitoring including microseismic, pressure build-ups, real time rate and pressure measurements and offsetting pressure observation wells. Conventional core analysis indicates that reservoir rocks have permeability in the micro-Darcy range. Lab experiments indicate permeability is also a function of the changing effective stress. Effective stress increases and permeability decreases as production occurs. A coupled geomechanical/reservoir study was performed after 400 days of production history in order to match behavior and determine the sensitivity of conventional production forecasts to the changing stresses within the reservoir. This paper focuses on the performance analysis of a single Upper Montney well in the Groundbirch area. The approach is staged, with the analysis proceeding from the simple to the more complex. The first step involved an improved version of production analysis. After the identification of linear flow, the utility of a long neglected plot, dm/q versus superposition linear time is illustrated. The ability of this plot to accurately split production well performance into reservoir properties and skin effects introduced by the fracture treatments is shown. This has wide ranging implications for the industry. Finally a simple practical method has been found to compare performance of various types of horizontal multi-stage fracture treatments where the impact of reservoir properties has been removed.In the second step of the study various descriptions of the study well were developed for all the processes the well encountered. These reservoir simulation models successfully matched treatment placement, an early build-up test distorted by clean-up effects and long term production behavior. Models types ranged from constant reservoir permeability through stress dependent permeability variation. Techniques were developed that eliminated the need to perform the time consuming geomechanical calculations. This allows the use of conventional reservoir models with pressure dependent permeability functions of a special form, to be used in day to day analysis. Sensitivity of the predicted production response with these models was then quantified.