传统油藏模拟技术难以准确预测油藏信息，新方法或许可以解决这一难题！ 编译 | 惊蛰 在油田开发策略的制定阶段，油藏模拟模型至关重要。这些模型需要先进行历史数据拟合，才能用于可靠的预测。油藏历史数据拟合的传统方法是，在可接受的范围内改变油藏模型中的不确定参数，以匹配该井位处观察到的产量与压力数据。参数可以大致分为两类：静态与动态。静态参数包括渗透率、孔隙度、产层有效厚度等。动态参数则包括油/水界面、断层传导性、相对渗透率曲线以及流动路径。 确定性建模只侧重于油藏模型的单一情景，忽略了所考虑参数中不确定性的影响。概率历史拟合是通过改变不确定的静态与动态参数，来选择非唯一的、多历史拟合的储层模型，从而获得一系列可能的预测结果。通常，在概率拟合工作流程中，目标函数是由井底压力模拟值与测量值之间的差异，以及各个单相流的流量组成，用于识别重要因素或不确定性。这些函数通常是基于井位处获得的监测数据，因此可能对远离油井的关键储层的不确定性不敏感。对于给定的油藏模型，与生产数据的良好拟合并不能确保井动态预测的稳健性。 储层非均质性可显著影响排水型式、波及形态、井产量的预测。特别是在油田生产的早期阶段，与压力、流量等生产数据相匹配的历史数据，往往缺乏必要的信息，来充分解决这些关键的非均质性，进而可能会对产量造成重大影响。即使再优秀的概率历史拟合方法也无法包含非均质性的多样性与复杂性。由于缺乏远离油井的数据，而且存在可能会影响产量的静态与动态不确定性，因此非均质性对后期产量会造成重大影响。 监视地震数据或*D地震数据可提供井间动态流体与压力分布的空间解释，便于准确选择出井间区域的流体分布模型。为了准确获取水、气前缘的相对位置，可将*D地震数据集成到油藏模拟模型中，以提高模拟的可靠性。通常，用于解释饱和度与压力变化的*D地震数据，已被用于辅助油藏模型的定性与确定性更新。孔隙度与渗透率一般是更新的主要变量，以拟合出井位之间驱替前缘的变化。 随机建模技术，包括基于聚类与基于流线的方法。其利用*D地震数据来更新油藏模型，在业内得到了广泛的应用。然而，这些方法需要大量模拟才能有效工作并避免崩溃，从而增加了所需的计算资源与时间。目前，业内已经提出了一种基于相关性的自适应分布方案。该方法使用相关性的空间分布来更新储层模型。近年来，一般都采用基于流线的数据集成方法，通过在模拟网格单元级别上校准渗透率场，将*D地震数据吸纳到油藏模拟模型中。 定量方法减少不确定性 有人提出了一种基于历史数据拟合的油藏模拟模型与过滤工作流程（*DHAM），利用*D地震数据生成空间约束，以减少不确定性，改善生产动态。该方法综合了驱替前缘的空间位置，在全范围内处理静态与动态参数的不确定性，并对历史数据拟合进行了定量与概率分析。与其他方法相比，该方法的主要优点之一是能够以相对较少的工作量或时间，来减少不确定性。 若是知晓了地震预测中流动特性的不确定性，那么，将地震数据集成到概率历史数据拟合工作流程中就相对简单了。地震数据可以通过以下三种不同的方法集成到储层模型中：*. 通过正演地震模型拟合地震响应；*. 弹性性质变化与地震反演的拟合；*. 拟合解释压力与解释饱和度的变化。 第一种方法计算量大，需要地震正演模拟，通常会增加地震数据相关的不确定性。第二种方法也是计算密集型，需要将模型与给定的石油弹性模型耦合，利用地震数据的反演，来预估弹性性质的变化。在这两种方法中，储层模型都必须转换成与地震数据相对比的空间，这限制了历史数据拟合的速度与效率。多数情况下，都使用正演模拟技术来计算阻抗变化或地震属性变化，这些变化可表示为储层性质（如孔隙度、孔隙压力、饱和度）的函数。上述两种方法都会产生非唯一解，这使得约束油藏模拟模型成为了具有计算量的迭代过程。地震数据与综合模型数据带来的分辨率差异，也增加了实现历史数据拟合的难度。 第三种方法假设地震数据可以解释或反演为压力与饱和度的变化，避免了必须修改油藏模型以拟合地震数据的额外步骤。该方法可以快速分析数据间的不匹配，并有助于推进模型的更改或选择，以改善历史数据的拟合。其使用*D地震数据来解释压力与饱和度变化，并将其与模拟结果进行比较，从而无需为历史数据拟合建立完整的耦合模型。 首先利用实验设计方法，通过改变不确定参数的范围，生成概率历史数据拟合模型。从生产历史数据拟合出的模型中提取饱和度图，再将之与*D预测的波及异常进行比较。然后，建立一种自动提取方法，用于协调*D地震数据与模拟结果之间的空间采样差异。将解释的*D地震数据与模拟结果进行比较，并将产生的不匹配用作为*D滤波器，对油藏模拟模型进行优化。所选择的模型用于识别油藏模拟参数，这些参数对形成良好的拟合非常敏感。 案例分析 在两个非洲海上深水油藏采用*DAHM过滤工作流程，不仅降低了不确定性，还提供了关键性能指标的信息，有助于获得可靠的历史数据拟合，为加密井的钻井决策提供更详细的信息。 历史数据拟合是通过改变静态与动态特性来实现的。利用*DAHM工作流程，建立了侵蚀接触面与砂面之间的联系。同时，测试了孔隙体积、流动系数以及倍增范围。采用拉丁超立方体采样，生成了***个概率模拟，并且在水饱和度发生变化之后，提取了压力变化。产量与压力的拟合，将***个模拟减少至**个子集，应用*D地震滤波器后，进一步减少到**个模拟模型。 利用*D过滤之前和之后的模拟，分别分析原始地质储量）、预估最终采收率、初始参数分布，并进行量化对比。分析表明，*D过滤使P**模型的原始地质储量增加了**%，并增加了P**模型的预估最终采收率，有助于可靠分析出靶点位置处的波及形态。除了降低原始地质储量的不确定性之外，该工作流程还通过提供校正后的饱和度变化S曲线，来帮助量化加密井的最佳靶点位置。它还节省了大量周期时间。? Reservoir-simulation models play an essential role in generating optimal field-development strategies, but they need to be history-matched before they can be used for reliable forecasting. Traditional history matching of a reservoir involves matching observed production and pressure data at well locations by changing the uncertain parameters in the reservoir model within the acceptable range. The parameters can be classified broadly as static and dynamic. Static parameters include permeability, porosity, and net to gross, among many others. Dynamic parameters may include oil/water contacts, fault transmissibilities, relative permeability curves, and flow pathways.? Deterministic modeling focuses on a single scenario of the reservoir model, ignoring the effect of uncertainty in the parameters considered. Probabilistic history matching is the process of selecting non-unique and multiple history-matched reservoir models by altering the uncertain static and dynamic parameters to obtain a range of possible forecasts. Typically, in probabilistic work flows, objective functions composed of a combination of differences in measured and simulated bottomhole pressure data and individual fluid-phase flow rates are used to identify significant factors or uncertainties. The functions are typically based on surveillance data obtained at well locations and therefore may not be sensitive to key reservoir uncertainty away from wells. A good match with production data for a given reservoir model does not ensure robustness in making performance forecasts. Reservoir heterogeneities—high--permeability pathways, barriers and baffles, or vertical connections forged from geologic erosion—can significantly affect drainage and swept patterns and well-production forecasts. Especially during early stages of field production, history matching to pressure and fluid-rate production data often lacks information necessary to resolve these critical heterogeneities fully, which may significantly affect production. Even a well-posed probabilistic history-matching approach cannot incorporate the variety and complexities of heterogeneities that are yet to have a strong imprint on future production because of insufficient data away from wells and the sheer number of static and dynamic uncertainties that can affect production. Surveillance seismic data, or *D seismic, provides a spatial interpretation of the dynamic fluid and pressure distributions between wells and facilitates choosing models that mimic the interwell areal fluid distributions accurately. Integration of *D seismic data into reservoir-simulation models has been used typically to improve the reliability of simulation by aiming to capture relative locations of water and gas fronts accurately. Traditionally, *D seismic data in the form of interpreted saturation and pressure changes have been used to aid in qualitative and deterministic updating of reservoir-simulation models. Porosity and permeability typically are the main variables updated to match the evolution of flood fronts between well locations. Stochastic modeling techniques, including ensemble- and streamline-based methods, have gained popularity by using *D seismic data to update the reservoir models. These methods require a large set of simulations to work efficiently and avoid collapsing, which increases the computational resources and time needed. A correlation-based adaptive localization scheme that uses the spatial distributions of correlations to update the reservoir models has been proposed. More recently, a streamline-based data-integration method was used to assimilate *D data into reservoir-simulation models by calibrating the permeability field at the simulation-grid-cell level. Quantitative Method Reduces Uncertainty The methodology presented in this paper uses history-matched reservoir--simulation models and a filtering work flow (called *DHAM by the authors) to use spatial constraints generated from *D seismic data to reduce uncertainty and improve production performance. The method endeavors to incorporate spatial locations of flood fronts, address uncertainty in the full range of static and dynamic parameters, and analyze the history matching quantitatively and probabilistically. In comparison with other methods, one of the main advantages of this method is the ability to reduce uncertainty with a relatively small amount of work or time? If the uncertainty in the seismic prediction of flow properties is understood, incorporating seismic data in a probabilistic history-matching work flow is relatively straightforward. Seismic data can be integrated into reservoir models in three different approaches:Matching the seismic response through forward seismic modeling Matching the changes in elastic properties with seismic inversion Matching interpreted pressure and saturation changesThe first approach is computationally extensive, requiring seismic forward modeling, which usually adds uncertainties associated with the seismic data. The second approach also is computationally intensive because it requires the simulation model to be coupled with a given petroelastic model, as well as an inversion of the seismic data to estimate changes in elastic properties. In both approaches, the reservoir models must be transformed into a space to be compared with the seismic data, which inhibits speed and efficiency in history matching. A variety of cases have used forward modeling techniques to calculate impedance changes, or seismic attribute changes, as a function of reservoir properties such as porosity, pore pressure, and saturations. These techniques produce nonunique solutions, which makes constraining reservoir-simulation models an iterative procedure with computational overheads. Resolution differences that arise from seismic and synthetically modeled data from simulation models increase the challenges of achieving a history match. The third approach—proposed in this paper—assumes the seismic data can be interpreted or inverted for pressure and saturation changes, avoiding additional steps whereby the reservoir-simulation model must be modified to determine a match to seismic. Mismatches can be analyzed quickly and used to drive changes or selections to improve the history match. The proposed work flow uses pressure and saturation changes interpreted from *D surveillance data and compares them with simulation outcomes, eliminating the need to have a fully coupled model for history matching. Design of experiments is used initially to generate the probabilistic history-match simulations by varying the range of uncertain parameters. Saturation maps are extracted from the production history-matched simulations and then compared with *D-predicted swept anomalies. An automated extraction method was created and is used to reconcile spatial sampling differences between *D data and simulation output. Interpreted *D data are compared with simulation output, and the mismatch generated is used as a *D filter to refine the suite of reservoir-simulation models. The selected models are used to identify reservoir-simulation parameters that are sensitive for generating a good match. The methodology and the steps involved in the work flow are discussed in detail in the complete paper. Deepwater Africa Case Study A detailed case study and results from adopting the *DAHM filtering work flow in two different deepwater reservoirs offshore Africa are presented in the complete paper. Using the proposed methodology not only reduced uncertainty but also provided information on key performance indicators critical in obtaining a robust history match to informing decisions on infill drilling opportunities.? History matching was achieved by changing both dynamic and static properties. Erosional contacts and sand-on-sand connections were established through *DAHM efforts. Pore volume, transmissibility, and multiplier ranges were tested. Latin hypercube sampling was used to generate *** probabilistic simulations and, subsequent to changes in water saturation, changes in pressure were extracted. Production rate and pressure matching reduced the *** simulations to a subset of **, which were further reduced to ** simulation models after applying a *D filter. Original oil in place (OOIP), estimated ultimate recovery (EUR), and initial parameter distributions were analyzed for the simulations before and after *D filtering to quantify the effect (Fig. *). The analysis indicated that *D filtering resulted in an increase of **% P** OOIP, an increase in P** EUR, and assistance in reliably estimating the sweep patterns at the target locations. In addition to a reduction of OOIP uncertainty, the work flow helped quantify the optimal infill well target location by providing updated S-curves of saturation change. It also produced significant cycle-time savings.