Published on: 04/19/2017
This a new post of our series of publications in Kraken’s blog related to Reservoir Engineering. This article is to briefly introduce the reader to the concept of reservoir management and its role in developing a production plan.
As discussed in this previous post, numerical simulation has become an indispensable practice to ensure good performance at all stages of a reservoir life. A reservoir simulation model provides insights into the reservoir’s behavior and allows us to forecast its future performance.
Many scholars and specialists define the reservoir’s life as an integrated process of exploration, followed by delineation, development, production and abandonment, as illustrated in the figure below.
In that light, “reservoir management” has emerged as a key concept. It entails a wide range of challenges to optimize reservoir production while juggling various technicals, economic, political and environmental constraints. According to Satter et. al. , a successful operation throughout the reservoir’s life requires integration of the engineering and geoscience disciplines and their multi-disciplinary teams.
The management process is defined by setting a realistic and achievable goal which varies within the company’s strategies. However, those strategies can change throughout the field lifetime (which makes the goals definition a dynamic process). In all cases, the aim is to optimize available resources by maximizing profits while minimizing investments and expenses. So, it is said to be a complex decision-making process, where planning plays a key role.
As stated by Satter and Thakur , setting a development strategy strongly depends upon the nature of the reservoir. Therefore, a solid knowledge of geology, rock/fluid properties, recovery mechanisms, drilling and well completion, and historical production are required. Additionally, a thorough understanding of the environment as a whole – corporate, economic and social – is crucial for developing suitable management strategies. Another critical factor is the evolution and adoption of available technology, as the success of reservoir management largely depends on the proper utilization of technology as it fits the project needs.
Developing a reservoir optimization plan should also consider environmental and ecological aspects. In particular, regulatory agency constraints need to be satisfied in order to obtain proper licensing which can be a complex and lengthy process. Moreover, the economic viability of a project is greatly influenced by reservoir performance and forecasts: material balance, decline curve analysis, and EOR simulators are important tools to analyze performance and estimate reserves in order to assess the project’s feasibility.
Before beginning production, geological (static) data – provided by cores, plugs evaluation, well logging and seismic data – are used to construct a suitable reservoir model. These data provide information about reservoir structure, petrophysical properties and help determine the hydrocarbon potential. As production begins, dynamic data from the asset and time-lapse seismic data are incorporated (Oldenziel ). The reservoir model is then used to predict production scenarios based upon rigorous knowledge of fluids composition and geochemistry. This explains why a reservoir model is an integrated model, given that the data cannot be analyzed separately.
Building a model requires a technique called “upscaling“. The reservoir grid is divided into grid blocks that cannot simulate fluid flow initially. They are usually at a fine-scale, and then converted into a coarse-scale model. This is desirable in terms of simulation goals, provided that a fine-scale model requires large memory usage and high computational run times.
The upscaling process converts a heterogeneous region into an equivalent homogeneous region, with a coarser grid and property values. Designing a coarse-grid cell considers geological features, such as faults and fractures. The process also requires selecting a coordinate system, orientation, block geometry and local grid refinement (LGR) in key regions of interest.
In order to construct the model, the concept of integration plays a vital role: a wide range of data with different accuracies, resolutions and conditioning volumes need to be integrated. Efficient data management is a time-consuming task and requires careful acquisition, analysis, validation, and storage. Data gathered from various sources need to be stored in a common database, accessible to the multidisciplinary team to carry out management activities (Satter et. al. ).
Additionally, a successful reservoir management requires constant monitoring in order to check if the ongoing performance matches with the development plan. To achieve this, production needs to be compared to simulation forecasts, and divergences are reported and investigated by the integrated team. That process is called “history matching” and consists of comparing data through a misfit function, commonly known as “objective function“.
Another important assignment used in history matching is the “sensitivity analysis”. It consists of determining the variation of simulations outputs when different inputs are introduced. Thus, it makes possible evaluating which parameters have greater influences on the objective function. The most sensible parameters will impact considerably the results, hence they will be modified during history matching.
To sum up, integrated reservoir management has received significant attention in recent years, especially on the synergism between a range of disciplines. The advanced computational improvements are providing more efficient techniques to enhance oil recovery which has been used to increment reserves as well as upsurge competition between companies and economies worldwide. In view of that, Kraken comes up as a powerful data integration software that can support engineers to perform tasks like reservoir simulation analysis, comparison between field data and the dynamic reservoir model, create forecasts and, by making use of Kraken’s API and Python Language, extend its use to perform Assisted History Matching and Production Optimization. Those more advanced uses will be better explored in future posts.