The Underground Storage Department at Union Gas Limited, a Spectra Energy Company, has long recognized that well interference effects during extended flow periods are difficult to simulate using simplified tank models. Although more complex and rigourous reservoir models can be used to predict well interference, these models are cumbersome when simulating multiple reservoirs flowing to common delivery points at varying operating pressures. Using seasonal field data from SCADA including flow rates and station pressures for each reservoir, partial regression coefficients were developed to predict well interference using simplified tank type model methodology. The results of this study show that the simplified methodology incorporated into a hydraulic modeling software program using linear regression can provide reasonable well interference results. The results lead to a direct improvement in forecasting reservoir performance which will better predict: Storage withdrawal capability Storage pool inventory depletion Compressor horsepower requirements
Pinnacle reef natural gas storage reservoirs contain gas in porous, heterogeneous rock. The ability of a pinnacle reef reservoir to cycle gas is a function of pressure and flow rate governed by reservoir properties such as porosity, permeability, water saturation, etc. A reservoir's cycling capability is also a function of the number of injection/withdrawal wells, well diameters and well spacing. The cycling capability of a pinnacle reef reservoir is variable ranging from high to low performance depending on the reservoir properties and facilities. The size and pressure operating range of a pinnacle reef storage reservoir is easily defined on a pressure/z vs. inventory graph where z represents the compressibility of gas. Using a tank model as an ideal example of a storage reservoir, the pressure/z vs. inventory curve follows a straight line defined by the reservoir index "K" as the reservoir is filled and emptied. A pinnacle reef's reservoir pressure falls along the reservoir index after a period of stabilization as measured at a ‘key’ or observation well ("observation well"). Shut-in stabilized reservoir pressures for a typical pinnacle reef reservoir are plotted in Figure 1. Under normal operating conditions (unstablized conditions), measured field pressures at static observation wells do not follow along the path of the reservoir index. The deviation of the measured static field pressures from the reservoir index during injections and withdrawals is referred to as the hysteresis effect as shown in Figure 2. Tutt and Dereniewski (1978) proposed a practical regression model of hysteresis to forecast deliverability for Michigan Stray sand reservoirs. This study evaluates the accuracy of using observation well pressures (i.e. hysteresis) as a predictive tool in estimating deliverability in pinnacle reef reservoirs.
Stabilized steady-state deliverability of individual wells in a reservoir can be determined using a back-pressure equation as defined by Equation 2 (Tek, 1987). The performance coefficient (c) can be calculated using wellhead pressures when pipe friction is negligible (Katz et al., 1959). For shallow pinnacle reef reservoirs (less than 2,500 ft) Equation 2 can be modified to reasonably calculate the flow performance based on wellhead pressures as defined in Equation 3 (Katz & Coats, 1968).
