Accurate Forecasting of Liquid Rich Gas Condensate Reservoirs with Multiphase Flow

Khanal, Aaditya (Chemical Engineering, University of Houston) | Khoshghadam, Mohammad (Chemical Engineering, University of Houston) | Lee, W. John (Texas A&M)

OnePetro 

Summary

The conventional Arps hyperbolic decline model was developed to estimate ultimate recovery for conventional reservoirs which quickly enter the boundary dominated flow regime. Arps' model is usually optimistic when applied to low permeability reservoirs, depending on the empirical exponent (b-value) selected to match the long-duration transient flow in these reservoirs and on the time at which the analyst switches from the original b value to a lower value (often zero) to implement the modified Arps model. In addition, this standard procedure in complicated by the poorly understood effects of multi-phase flow in gas condensate reservoirs. Given the uncertainty in results from traditional models, new tools to supplement the ones in use today are required to improve the accuracy of our production forecasts.

In this paper, we used compositional reservoir simulation for retrograde condensate fluids to generate synthetic production histories. These production data, based on robust numerical simulation, were then used to evaluate forecasts using the initial history from simulation for matching and forecasting with each of several models used in the industry including Arps, Duong, and Stretched Exponential (Power Law) decline method. In addition to this we investigated the use of rate transient analysis (RTA) and principal component analysis (PCA), both of which use readily available field production data to history-match and forecast the future production data. RTA matching and forecasting algorithms are often based on the Ozkan's analytical trilinear flow model, which cannot be used to nonlinear multiphase flow case such as the gas condensate reservoirs without modifications. Unlike rate transient analysis, PCA is a non-parametric method which can be used to forecast and predict the production from gas condensate reservoirs. The simulated production histories were analyzed by this method to history match and forecast the production of simulated wells. In addition to this, the method was verified by using field data from the Eagle Ford Shale. We also investigated the similarities and differences between conventional and unconventional reservoirs and how multiphase flow is manifested in production diagnostic plots. For multiphase flow in gas condensate reservoirs, liquid dropout and relative permeability effects below the dew-point pressure have to be incorporated to accurately model the flow of both condensate and gas

As expected, we found that no approximate method reproduces the forecasts from compositional simulation. Nevertheless, we recognize that more rapid approximate methods will be required for routine analysis. Understanding the limitations of different approximate methods in given circumstances, as identified in this paper, should lead to optimal use of these methods.