Production from shale gas reservoirs in the USA has become an important component in the increase of natural gas supply. The Haynesville shale, in particular, is a major contributor in gas supply due mainly to its relatively higher initial deliverability compared to other gas shale plays. One of the critical questions in developing a play efficiently and economically is the well spacing. There are several approaches to addressing this question. The paper looks at one approach, namely the process behind building a calibrated, history matched multi well reservoir model. The model is run in prediction mode with different sensitivities to answer the well spacing issue. The model honors the initial static and dynamic conditions, is capable of running in a reasonable time and, most importantly, has been useful to management in the decision making process. In this field case, a half section in the state of Louisiana has been drilled and completed with 4 horizontal multistage producing wells and 2 vertical microseismic monitoring wells, 1 of which was subsequently converted to a downhole pressure monitoring well. During the entire hydraulic fracturing operation, downhole microseismic data were simultaneously recorded in both observation wells. The pressure data from the monitor well acquired during production was entered into the reservoir model as another history matching variable. The microseismic data were used to calculate the fracture parameters and as a limiting constraint in the process. This dual porosity model is a practical example application of the methodology previously described in SPE paper 132180 by Du et al. (2010). The sections in this paper describe a method for building a reservoir model that honors the static boundary conditions. The model was built in two parts according to the Dual Porosity nature of it: a conventional geological model representing the initial porosity and permeability of the rock matrix, and a second part that models the fracture network generated by the stimulation operations and the pre-existing natural fractures. The paper then explains how this model was tuned to enable a production and pressure history match. There is also a section devoted to the generation and utilization of a critical correlation between pressure depletion and a combined fracture half-length times the square root of permeability (Xf.√k) parameter which greatly reduced the uncertainty caused by the non-uniqueness of the match. Finally, some general conclusions regarding the results are presented.

While well interference is a known phenomenon in shale gas plays, it is often overlooked or only considered when its effects are readily apparent. Spacing tests are often performed when starting to appraise or develop a new area where a qualitative evaluation of interference between drilled wells will determine the appropriate well spacing. While numerical reservoir modelling and build-up analysis can be applied in the Marcellus shale play they are not appropriate for determining optimal spacing due to very low, nano-darcy reservoir permeabilities and uncertainty over hydraulic fracture geometry. The challenge was to develop a new approach to measure and estimate the impact of interference on gas recovery and optimize well spacing. The impact of interference was initially evaluated by assessing the change in the productivity index of wells due to offset wells being added, put on production or shut/in. In order to have a more meaningful way of estimating its impacts on economics, interference was also quantified based on projected future five year cumulative production using Arps decline curve, Rate Transient Analysis and Pressure Normalized Rate methods. Results were compared and a statistical workflow was used to estimate optimal spacing. By relating the degree of interference with overlap between wells, the nature of interference was also investigated. Interference can be due to a Stimulated Rock Volume (SRV) overlap. This is characterized by a proportionality increasing interference with SRV overlap. The impact of natural features such as faults or high permeability streaks that can act as conductivity highways across many wells is usually not proportional to overlap. This study in Marcellus shale play demonstrated that measurable interference occurred at wide spacing and that 1,000 ft spacing should result in an increased FYFCP (Five Year Forecasted Cumulative Production) of 10% over existing spacing assumptions, key at low gas prices. Therefore, 1,000 ft spacing has been recommended for future well placement. The workflow outlined in this paper is currently being used to evaluate well spacing for other assets and can be used by Reservoir Engineers to evaluate spacing in tight or shale gas/oil plays.