Article Preview
Top1. Introduction
Reservoir modeling is the sequence of data modeling activities aiming the simulation of a petroleum reservoir in order to get an estimation of oil production. The exploration and production activities of petroleum industry generates a huge amount of data every day that come through many sources supported by information systems of different software providers. The industry relies on the efficient use of these data to build computational models that helps to reduce the uncertainty and risk in taking decision (Werlang, Abel, Perrin, Carbonera, & Fiorini, 2014).
Figure 1 shows a set of models that are usually developed in order to build a reservoir model. It shows at least eight different models which are linked usually as being inputs for the next activity in the workflow. The models start from interpretation of seismic and well log data until a flow simulation model is produced, where reservoir engineers can estimate the production of petroleum in the studied reservoir.
Figure 1. Set of models needed to simulate a petroleum reservoir. Extracted from (Abel, Perrin, & Carbonera, 2015)
Due to this great amount of data existent in the domain, a problem that arises is the data interoperability. According to (Wache, et al., 2001), data interoperability is the ability of offering to the user an uniform access of geological data and information created and managed by different information systems, allowing the performance of unified analysis over these data. In other words, interoperability is the capability of distinct information systems to communicate each other, sharing data, information and knowledge in efficient and safe way.
There are two kinds of data in this domain: indirect rock data, and direct rock data. Indirect data is the data that comes from some tool analysis of the rock, like seismic data, well logs and petrophysics. Directly data is the data that comes after directly observation of the rock, like well cores, thin sections and rock samples. Direct rock data is important because it provides more reliable information, decreasing the uncertainty of the models, supporting a better prediction of the reservoirs.
Unfortunately, direct rock data is often ignored by reservoir modelers. The main reasons are that it has a high cost of acquisition, requires a specialist to interpret the data and, because of the scale of analysis and qualitative nature, it is very hard to input it into the reservoir models.
A common approach to achieve the data interoperability is the use of ontologies. An ontology is an explicit formal specification of a sharing conceptualization (Studer, Benjamins, & Fensel, 1998). In (Abel, Perrin, & Carbonera, 2015), the authors claim that conceptual tools provided by ontologies theories help to identify and model the invariant and consensual parts of the geological knowledge used to build models.
In order to facilitate the use of direct rock data in reservoir models, we propose here a framework to support the integration of direct rock data into reservoir models. The framework is based on a domain ontology, a well log segmentation algorithm, well correlation algorithms and the geological concept of reservoir petrofacies.
Our contribution in this work is to define a set of tasks which can guide the modelers involved in the reservoir modeling activity to insert direct rock data in the models, and trace this data along the various models existent in this activity.
We have presented partially the framework in (Garcia, Graciolli, Ros, & Abel, 2016). Here, we present the domain ontology of Petrography we developed and further details of the framework itself.
The paper is structured as the following. In section 2, we propose the use of reservoir petrofacies, a geological concept, in the computer science context. In section 3 we present a domain ontology in the Petrography domain, with the goal to support the description of thin sections and definition of reservoir petrofacies to be used in our framework. In section 4 we present the log segmentation algorithm that we use in various steps of our framework. In section 5 we present the proposed framework and finally, in section 6 we are making some conclusions of proposed framework and domain ontology, as well as defining future works.