Use of Numerical Models To Develop and Operate Gas Storage Reservoirs

Abstract

Performance of gas storage reservoirs is affected by a combination of well placement and operational strategy. This paper illustrates the use of a two-dimensional, single phase dry gas reservoir simulator in the study of such a phase dry gas reservoir simulator in the study of such a reservoir. Results are presented that illustrate the effects of well spacing and operational planning on the ability of a reservoir to meet certain requirements. Introduction In the design and operation of gas storage reservoirs, the placement of wells with respect to the most advantageous operational stratagem is of prime importance. The economic consequences of failing to give adequate attention to the effects of interference among wells can be great. In addition, the impact of operational stratagems on the desired performance of the reservoir must be considered. The level of season-end performance is subject to the effects of the method of operation during the season. In this study, the placement of new wells in an existing dry gas storage reservoir is treated concurrently with the search for an operational stratagem that would permit the reservoir to meet certain withdrawal requirements. These investigations were carried out using a two-dimensional, single-phase reservoir simulator based on Eq. 1. .......(1) (The reservoir model is described in detail in the Appendix.) This study also illustrates one approach to the problem of extracting a workable reservoir description from problem of extracting a workable reservoir description from meager data. Definition of Problem The object of the study was a nearly depleted dry gas Oriskany sand reservoir that had been converted to gas storage. In operation as a storage field, the main portion of the reservoir contained 41 wells and, under a given seasonal operational plan, was capable of delivering 140 MMcf/D on the last day of withdrawal. It was required that turnover of gas be increased and that the last-day capacity be upgraded to 300 MMcf/D. This was to be accomplished by drilling additional wells and adding compressor horsepower so that field gathering-line pressure could be lowered. Plans were made for extra compression and 38 new wells, so the problem reduced to one of correct well placement with the possibility that perhaps one or more wells could be eliminated from the plan and yet the required deliverability could be met. Description of Reservoir The computing grid was superimposed on the reservoir map as shown in Fig. 1. The block size was chosen as a compromise between computing speed and definition. Also shown in Fig. 1 are deliverability areas defined from observed performance of the wells. These areas are labeled in order of quality; i.e., Area 1 contains the wells with the highest deliverability and Area 6 contains those with the lowest. These area definitions show that the reservoir decreases in quality outward from a central zone. These data, along with geologic interpretations, led to a contour map of gas-filled porosity that generally followed the same pattern. Values of porosity were entered on the grid at pattern. Values of porosity were entered on the grid at selected control points and a statistical regression technique was used to obtain values over the entire grid. Log picks indicated that a constant value of 7 ft for net pay thickness over the entire grid was reasonable. Obtaining a representation of permeability posed some problems. Core data were available for one well and problems. Core data were available for one well and pressure vs injection/production data were recorded only on a pressure vs injection/production data were recorded only on a cumulative reservoir basis. Thus, in the absence of individual well drawdown or buildup data, the only available means of obtaining a detailed reservoir description was through the use of deliverability curves for each well. Initial values of permeability were selected on the basis of flow capacity, and the regression routine was used to obtain values over the full grid, with the wells serving as control points. Using constant pressure boundary conditions, we then set up the model to compute deliverability curves for each well. We computed three points on these curves during each pass, each time restoring the grid to initial pressure and using a different limiting pressure. To shift the computed curve so as to match the observed curve, we made hand adjustments to grid block values of permeability at wells. To smooth the entire grid, we used an arithmetic moving average procedure that held the values at these control points as constant. JPT P. 1239

Effective Permeabilities of Abandoned Oil and Gas Wells: Analysis of Data from Pennsylvania

“…We use these data, along with additional recent measurements in Venango, Lawrence, and Allegheny Counties, to estimate effective permeabilities of AOG wells. These measurements provide additional quantitative information on the ability of fluids to flow through these wells.The new effective permeability data can be used in modeling studies to evaluate the leakage potential of AOG wells in applications such as geologic storage of carbon dioxide, − natural gas storage, − and oil and gas development. − …”

Gas Storage Facility Design Under Uncertainty

“…McVay and Spivey (2001) provide concise descriptions of several investigations that proposed and demonstrated methods for gas storage facility design optimization. These works include Duane (1967), Henderson et al (1968), Coats (1969) (gas-producing field), Van Horn and Wienecke (1970), and Wattenbarger (1970). After this initial work, there appears to be a gap in the research literature.…”

Effect of Reservoir Properties on the Performance and Design of Gas Storage Wells

“…al. 11 illustrated the use of a two-dimensional, single-phase dry gas reservoir simulator to optimize the gas storage field. They varied the location of wells to increase the production for the study area for peak demands.…”