Reservoir simulator RN-KIM is intended to create, calculate and analyze three-dimensional digital oil and gas field models.
Reservoir simulator RN-KIM is intended to create, calculate and analyze three-dimensional digital oil and gas field models.
In December 2019, it will be 15 years since the first state registration of the RN-KIM computer program. The RN-KIM hydrodynamic simulator is designed to create and analyze three-dimensional digital field models. A digital field model is used to calculate reserves and forecast hydrocarbon production. The model takes into account the geological and field information about the field, reproduces the operation of the wells and is a digital clone of the field for the analysis of “what if”.
Today, hydrodynamic modeling is characterized by a complication and increasing dimensions of models; multivariate parallel calculations on supercomputers are used all over the world. RN-KIM offers advanced computing technologies and daily use tools to support the development of large and giant reservoirs.
More than 1000 full-scale and sector models are prepared and updated annually at RN-KIM to solve wide range tasks.
Monthly issue updates of RN-KIM
In 2007, after passing independent tests, RN-KIM received the Gosstandart of Russia certificate of conformity - this was the beginning of the using own hydrodynamic modeling software in the preparation and protection of design documentation for field development by Rosneft.
With the increase in dimension and the complication of hydrodynamic models, the requirements for RAM and computer system performance increase. Speeding up model calculations is a priority for simulator development. The parallel multi-threaded version for workstations allows reducing the calculation time due to the use of multi-core processors. The simulator is adapted to work on cluster and supercomputer systems. So, when calculating at 32 cluster nodes for models with hundreds of millions of cells and tens of thousands of wells, the calculation was accelerated up to 24 times.
For automatic adaptation tasks, optimization algorithms are used. Using multivariate parallel model calculations, the effectiveness of development systems is evaluated. The optimization criterion is any complex technical and economic parameter, for example, oil recovery ratio or NPV. Other objective functions are set using the Python programming language. As a result, with the help of a simulator and routines in Python it is possible to solve a wide range of problems in the optimization of development history (History Matching).
To simulate filtration in fractured porous reservoirs, the simulator implements a model of double porosity and double permeability. Cracks and blocks of the matrix are considered as two media by embedding one into the other. The flow along cracks, the flow along blocks of matrices and flows between cracks and blocks of matrices are calculated separately. The mechanism of gravity impregnation and drainage was implemented according to the Gilman and Kazemi and Quandalle and Sabathier models. The option is especially relevant for the development of shales, Bazhenovskaya and Domanikov suite.
The simulator contains functionality for predictive calculations on the use of group control of wells and economic constraints. The user can determine the automatic actions performed according to a given condition, for example, for editing events of property cubes, using the Python interpreter built into the simulator. The built-in interpreter allows you to access the model parameters and simplifies the addition of new options with minimal programming knowledge.
The prepost processor supports the Eclipse™ (Schlumberger), Tempest™ (Emerson) data formats and allows to convert the model to the RN-KIM format. Also, the interface allows you to load calculations from the formats “RN-KIM”, “Eclipse”, “Tempest” and make comparisons between calculations.
The basis for solving the problem of filtering a viscous compressible multiphase mixture in a porous medium is the mass conservation equations and variations of the Darcy equation, which take into account the transition of oil and gas components into the liquid and gas phases. A completely implicit time sampling scheme and the Newton method of solving a nonlinear system of equations are used.
The implemented model allows taking into account flows along the wellbore for correct modeling of the simultaneous development of several reservoir layers.
Faults:
Aquifers:
The flow between the matrix and the crack is due to three mechanisms:
To model hydraulic fractures, a method is used based on the conjugation of the finite-difference approximation of the flow in the reservoir and the analytical solution in the vicinity of the fracture. A crack is modeled as a collection of drains (sources) located one at a time in each calculation block through which it passes.
Local grid refinement improves the accuracy of the solution near the wells. Grid enlargement can be used to combine those areas of the reservoir whose modeling accuracy is not significant.
For models of more than 2-4 million active cells, it is important to use several cluster nodes to reduce the calculation time. Also, the use of the cluster version allows you to remove restrictions on RAM for calculating giant models. In this case, the simulator automatically separates the model grid and provides data exchange along the boundaries.
Today, all personal computers has muticore processors. Our simulator parallelize calculations using threads and achieve good speedup.
At RN-KIM, the inflow to the fracture is modeled by the source method. The results are compared with the mesh grinding method, in which the mesh cells are ground to the size of the crack, and high permeabilities are set in the crack cells.
Conclusion: the accuracy of calculations was verified taking into account the effects of gas dissolution, changes in saturation pressure, gravity.
The difference between the calculation of RN-KIM and Eclipse for cumulative production in phases of less than 1%.
Conclusion: the accuracy of the calculations was verified taking into account the effects of the deposition of a heavier liquid phase in the formation and the subsequent evaporation of condensate due to the injection of dry gas.
In comparing RN-KIM and Eclipse, the maximum discrepancy in the cumulative integrated characteristics is less than 1%.
Conclusion: accuracy was verified, less than 1%, on a series of calculations of models with a horizontal well.
Conclusion: when comparing RN-KIM and Eclipse, the maximum discrepancy in the cumulative integrated characteristics is 1%.
Two-phase filtration in a reservoir with a highly inhomogeneous distribution of porosity and permeability is simulated.
Conclusion: when comparing RN-KIM and Eclipse, the maximum discrepancy in the cumulative integrated characteristics is less than 1%.
New functionality until 2020