The Research And Application About Reservoir Modeling Algorithm Class Library

Through studying with the general process of reservoir modeling, the subject has combined with geostatistical analysis system of reservoir GASOR software modeling algorithm, and this paper improves and expands some of the existing reservoir modeling algorithms in laboratory such as Kriging and Multi-point Statistic algorithm, and it uses common design patterns to design and encapsulate the reservoir modeling algorithms, thus systematically build the reservoir algorithm class library. The modeling system which constructed from the library helps users to quickly build the reservoir model and helps them to understand and describe the heterogeneity of reservoir. At the same time, it has a great significance to developing reservoir modeling software with own intellectual. The main innovation and contents of this paper are as follows:Firstly, it designs the whole framework of the reservoir algorithm class library. The framework introduces the MVC structure, and incorporates the algorithm based on plug-in technology design. In accordance with the principle of Interface-oriented programming, the whole algorithm is abstracted. The reused code is separated from the algorithm through the way of reconstruction and packed using strategies module to make the algorithm easier to expand and update. In addition, through defining the interface of the algorithm plug-in, it laid the foundation for framework to achieve plug-in, making the class library has a good maintainability and scalability.Secondly, because the algorithm of the reservoir modeling algorithm class library needs to operate a large number of reservoir model objects, so this paper designs and implements the object management system of reservoir modeling based on smart pointer to simplify the programming model for reservoir modeling system, and it effectively solves the problem of memory leaks; The creation, destruction, inquiry and calling of every module object in algorithm class library have been unified management through improving the factory model and apply it to object management. It could reduce duplication of costs arising from building objects, the coupling between the various modules block, and facilitate the algorithm plug-in management. This pattern which combines the plug-in design with the MVC design is more conducive to an easy reuse, easy to maintain and expand the algorithm class library.Thirdly, it designs and implements a mechanism which separates the algorithm parameter settings and user interface in algorithm class library in order to solve the dependence of algorithm on the user interface. And the user interface is separated from reservoir algorithm through the parameter file of XML, which greatly reduces the dependence of algorithm on the user interface. At the same time, the method that dynamic generate algorithm interface is used to support the extensions of the algorithm plug-in. When user expands the algorithm, they only need to load their own algorithm plug-in and algorithm parameters into the system. Furthermore, they can save the algorithm parameters of some algorithm which could help users to comparison and usage in the future.Fourthly, it reconstructs the implementation of the class library in the algorithm by the application of common design patterns, which enhances the logic of the library and makes it easier to maintain and expand. It applies the improved factory pattern in the object management module that make the object creation avoids the weakness of code redundancy in client; the display of data model and interface layer are unified through the observe-model effectively; the modules, such as strategic model in algorithm and single mode in the object tree management make the library more reusability.