... performance. A less computationally expensive alternative to the brute-force search is to use cudaOccupancyMaxPotentialBlockSize in CUDA or hipOccupancyMaxPotentialBlockSize in HIP, which suggest thread block sizes that maximize occupancy. However, in some cases, multiple configurations may achieve...

... parallel simulator and uses two libraries to improve its computational performance: the Library of Iterative Solvers (LIS) for linear system ( Nishida, 2010 ) to solve the set of simultaneous equations described by Eq. 54 , and METIS ( Karypis, 2013 ) for a computationally efficient domain...

... The main operations involved for Black Oil property calculation is local cell computations and tables look-up operations. These operations are computationally cheap and, thus, are largely memory-bound. However, the fragmentation of cells between various rock and fluid regions makes it hard to have a good...

..., focusing on the details of how to deliver the best performance possible with CUDA and GPU. For the year 2019, our focus is mainly on learning CUDA and Proof of Concept (PoC) exercise for computationally intensive component/module (e.g. reservoir linearization and linear solver), limited to Black-Oil...

... Abstract Despite the rapid rise of computing power and advances in computational techniques in past decades, it is still challenging in reservoir simulation to model complex and detailed features that are represented by small cells with large permeability values, for example, fractures, multi...

..., and one new object is created to store the operation's result. The result is used as the next corresponding operator's input argument and is freed once the next operator has been processed. This strategy is very redundant and computationally costly. To overcome this drawback, ET builds several temporary...

..., modern parallel computing relies on complex hardware architectures. Those are composed of several multi-core processors and massively parallel computation units such as GPU accelerators. In the context of solving linear systems, the legacy algorithms are not capable of fully exploiting these highly...

.... Although high-resolution models for well and fracture are required to achieve adequate resolution, they are computationally too demanding for practical field models with many stages of hydraulic fracture. The paper aims to innovate localization strategies that take advantage of locality on timestep...

... dynamics computation algorithm upstream oil & gas gpu accelerator chrono petroleum engineer preconditioner reservoir simulation scalability application matrix flow in porous media gpu-accelerated linear solver equation linear system implementation fsai gpus Introduction...

... formulation of the Darcy problem then reads: Find ( q , p ) ϵ V (Ω) × Q (Ω) such that Abstract In many applications the accurate representation of the computational domain is a key factor to obtain reliable and effective numerical solutions. Curved interfaces, which might be internal, related...

.... On the other hand, if the VLP table is too dense, the construction of this table can become computationally expensive. Since in our framework the VLP tables are constructed dynamically during the simulation and only for the range of operating conditions, we can construct the table efficiently even if a high...

... for these grids (e.g. Geiger et al., 2009 ). Other, more computationally efficient approaches are therefore needed to in order to compare and rank the different models and their underlying concepts, both in terms of their static and (quasi-)dynamic properties. Streamlines can be readily computed for fast flow...

... an efficient parallel implementation procedure as to how to enrich a given multiscale formulation with additional local basis functions. These additional basis functions, constructed in parallel computational platform, aim to resolve large error components for a generic fine-scale system with no right-hand...

... Abstract This paper discusses the parallel implementation of complex wells modeling in a large-scale simulation. Due to intensive computational requirements of giant reservoirs with hundreds of these complex wells, parallel implementation is essential for practical simulation studies...

... these challenges, the computational complexity and the dissimilarity in the scales between the simulation model and the production data stand out. The construction of lower fidelity, less computationally intensive models is a common practice to alleviate the computational complexity. Upscaling ( Durlofsky, 2005...

... al., 2015 ). While we could also apply a formal optimization algorithm for the outer loop (determination of T *), we have found that a simple interpolation approach performs well and is computationally efficient. Our nested optimization procedure is outlined in Algorithm 1. Step 8 of Algorithm 1...

... of the computational loop in modern simulators, defining which physical mechanisms and assumptions are employed. This significantly complicates the implementation of simulation codes for heterogeneous computing hardware, which promises significant improvements in simulation time. In this work, we use the recently...

... is the solution method of choice. The advances in hardware for high performance computing have in recent years enabled an enormous increase in the complexity of reservoir simulation. This increase in complexity is due to the addition and interconnection of physical processes that are encountered in reservoir...

... Abstract Compositional flow simulation, which is required for modeling enhanced oil recovery (EOR) operations, can be very expensive computationally, particularly when the geological model is highly resolved. It is therefore difficult to apply computational procedures that require large numbers...

... Jacobian as always. If it is to be treated explicitly, it is instead subtracted from the explicit matrix, G E . This leaves two computationally intensive steps – inversion of G I and computation of the eigenvalues of G . The code used to generate the results presented here is written in Octave...