Estimating oil and gas production from assets is essential to carry oil recovery processes. From the many techniques and tools used for such a task, the compositional simulation model is important for miscible displacement and problems with complex phase behavior. In this model, the fluid flow in porous media is described by a set of conservation equations. The solution of these problems involves spatial and time discretization schemes and approaches for handling the coupling of fluid flow and phase behavior. As a consequence, several solution algorithms arise from combining different selection of primary variables and equations, phase behavior decoupling techniques, and spatial/time discretization, which define the computational performance of these algorithms. In this work, a new Adaptive Implicit Method (AIM) is proposed by combining a global intensive variables Fully Implicit (FI) formulation and an IMPEC (IMplicit Pressure, Explicit Compositions) approach. In this approach, gridblocks are dynamically selected as FI or IMPEC based on a stability analysis algorithm. The fully implicit part considers pressure, water saturation, and overall compositions as primary variables and the IMPEC approach considers pressure and total number of moles for each component as primary variables. A new stability analysis algorithm is proposed and used for up to four-phases, but it is general enough to be used for any number of phases. The eigenvalues of the error matrix are used for the stability analysis and are computed using the Power's iteration method. The new approach is implemented in the in-house simulator called UTCOMPRS. The AIM formulation is compared to the fully implicit and IMPEC versions. We observed considerable improvement in the computational performance of UTCOMPRS with the new AIM when compared to the pure FI and IMPEC formulations.

从资产估算油气产量对于进行采油过程至关重要。从用于此任务的许多技术和工具来看，成分模拟模型对于可混溶位移和复杂相行为的问题非常重要。在该模型中，多孔介质中的流体流动通过一组守恒方程来描述。这些问题的解决方案涉及空间和时间离散方案以及用于处理流体流动和相行为耦合的方法。结果，通过组合不同的主要变量和方程式选择，相行为解耦技术以及空间/时间离散化，出现了几种求解算法，这些算法定义了这些算法的计算性能。在这项工作中，通过结合全局密集变量完全隐式（FI）公式和IMPEC（隐式压力，显式成分）方法，提出了一种新的自适应隐式方法（AIM）。在这种方法中，基于稳定性分析算法，将网格块动态选择为FI或IMPEC。完全隐含的部分将压力，水饱和度和总体组成视为主要变量，而IMPEC方法将每个组件的压力和总摩尔数视为主要变量。提出了一种新的稳定性分析算法，该算法可用于多达四个阶段，但是它足够通用，可以用于任意数量的阶段。误差矩阵的特征值用于稳定性分析，并使用幂的迭代方法进行计算。新方法在称为UTCOMPRS的内部模拟器中实现。将AIM公式与完全隐式和IMPEC版本进行比较。与纯FI和IMPEC公式相比，我们发现使用新AIM的UTCOMPRS的计算性能有了显着提高。