Large Scale Steam Flood Performance Forecast using Type Curve Analysis and Multi Temperature Fractional Flow

Abstract

Abstract In the current move to a low-carbon and low-crude-oil-price world, numerous existing steam flood projects need to be economically and environmentally re-evaluated. Because of the high cost of steam generation and high associated carbon emissions, these projects will require increased attention and assessment. Traditionally, reservoir simulation is a crucial part of field evaluation and project economic forecasting. However, large-scale steam flood projects can cover thousands of acres, making simulation challenging. Forecasting of steam flood performance based on reservoir simulation is effective when addressing the complicated geological structure and heterogeneous reservoir properties in real reservoirs. However, large scale simulation of steam floods is still not computationally efficient. A fine grid is required to capture the steam/condensing waterfront, thus the simulation efficiency is difficult to improve by enlarging the size of the grid. At field-scale, running a simulation in the millions or even billions of grids is not a practical approach. Thus, this paper uses a semi-conceptual method named "Pattern-Based Modeling (PBM)" to generate the fast and accurate forecast for large steam projects. Instead of using grids, PBM uses patterns as the foundation of modeling. A pattern is a flow regime with the special arrangement of producers and injectors. For steam flood projects, the size of a pattern is measured in acres which is thousands of times greater than the size of grid. The upscaling from grid to pattern is valid because (a) the heavy crude field usually has high permeability and porosity sandstone; (b) heavy crude oil is almost immobile unless heated by the steam while steam has a huge mobility ratio. Those physics phenomena substantially reduces the effect of heterogeneity. In the field and lab observations, steam always goes to the impermeable overburden immediately due to the vast density differences. The steam chest is formed after steam reaches the bottom of producers. The steam chamber, which contains saturated steam, pushes the oil below using the latent heat. In a single pattern, the oil rate type curve is constructed by two mechanisms: (a) the frontal displacement (Marx-Langenheim) before steam breakthrough; (b) the gravity drainage (Vogel) after steam breakthrough. Then the oil rate type curve is created based on these two mechanisms. The paper also provided a multi-temperature fractional flow method to forecast the water cut. The oil and water rate forecast in the single pattern is validated by reservoir simulation. The paper runs numerous pattern simulations in multiple scenarios. The results indicate that the oil and water rate is independent of well placement after assuming the homogeneity of reservoir properties in the single pattern. The number of injector and producer will also not affect the oil and water rate, which means that the well arrangement (invert five-spot, nine-spot, etc.) would likely to generate similar results. Although the reservoir properties (porosity, permeability, initial saturation, etc.) in a single pattern are homogeneous, they can vary between patterns. The PBM has the flexibility that for single pattern, the type curve can be generated based on the individual reservoir properties of that pattern. Each pattern is assigned a starting date. Then the field rate is the accumulation of each pattern. The single pattern reservoir properties can be adjusted during the rate history match. For the steam performance forecast, the paper introduced a statistical correlation, which indicated that besides the reservoir properties, the surface steam operation would also affect the ultimate recovery (EUR). High steam injection rate results in high steam chest pressure (or temperature), and thus reduces the thermal latent heat, and thus decrease EUR. The paper has conducted a case studies for the demonstration: the Kern River Field (USA) with a long history of steam flood. The paper showed that PBM generated a more than 90% correlation coefficient during the history matching. Based on the correlation, the Kern River thermal EUR within current operation is about 87 % of in place resources.