Eni revealed its new super computer, HPC5, during a ceremony on 6 February. This is an important time in the path toward the energy transition. It’s another step forward to the global goal that we share with our research and technology partners: making tomorrow’s energy an even closer reality.” The new supercomputer supports the previous system (HPC4), tripling its computing power from 18 to 52 petaflops, equivalent to 52 million billion mathematical operations per second, allowing Eni’s supercomputing ecosystem to reach a total peak power of 70 petaflops. The company said HPC5 is the world’s most powerful supercomputer infrastructure in the industrial sector and allows the company to achieve another milestone in its digitalization process.

The scope of emerging technology for the oil and gas industry could be summed up in just three areas: advanced computer programming, new physical toolsets, and alternative applications of upstream systems. An executive panel session held during the recent International Petroleum Technology Conference 2020 in Saudi Arabia featured examples of each category from several of the industry’s most technologically capable firms along with candid thoughts on how they might impact the industry’s people problem. Nasir Al-Naimi, Vice President of Petroleum Engineering and Development for Saudi Aramco, highlighted a number of ambitious technology initiatives that are poised to “revolutionize” his company’s core exploration operations. Among them is a recently launched digital program called GeoDrive, which is an integrated geophysical imaging and modeling program designed for intensive exascale computing—considered to be one of the next major milestones for high-performance computing. In 2016, Aramco debuted another in-house digital feat called TerraPowers that represents the industry’s first trillion-cell reservoir simulation and hydrocarbon migration algorithm.

Multilateral wells may vary in the number of laterals from two to even more than ten laterals. To control production on a lateral level, advance completion equipment, such as Inflow Control Valves (ICVs) and dual ported Permanent Down-hole Monitoring Systems (PDHMS) for real-time pressure measurements, are installed in the motherbore against each lateral to regulate reservoir fluids into the production tubing of the well from individual or collective set of laterals. Multilateral wells equipped with advance completion tools become complex wells (hereinafter called the "complex wells") in reservoir engineering. Recent complex wells involve multiple designs and architecture for which the reservoir exposure for laterals sums can be thousands of meters.

A new generation of smart multilateral well completions are Manara type wells where the laterals are divided into a number of segments or compartments using oil swell packers and Manara stations are placed against each segment for quantifying liquid rate, water cut and real-time reservoir pressure measurements. Each station is equipped with electrical control valve for controlling unwanted fluid production at a segment level. In this work, a new workflow is established to model and history match Manara well with complex modeling features using the GigaPOWERS (GP) simulator and a compositional full field model. The established segment level history matching workflow includes four important milestones to achieve. The first is advance well completion design at which the physical well completion details are translated and converted into a grid based completion details with the help of pre-processing tools. The second step covers the generation of three complex well-related input files for the GP simulator. The third step is the process of validating high frequency performance data including flow rate, choke size and pressure for the complex well stations. The last step involves conducting the history matching exercise on a segment level for every individual station to achieve the final history match model.

The conventional history match procedure includes generally three levels to match; field, group and well. In this workflow, station level match is done for the first time with a full field compositional model with a size of more than 61 million grid cells. The high definition history match is achieved at segment level for six stations that constitute a trilateral well with high frequency performance data. Complex well modeling in GP includes pressure drop calculations for several components. The pressure drops are related to friction, gravity, acceleration and advanced tools (i.e. ICV).

Modeling a complex well involves several pre- and post-simulation environment features at the segment level that should work in complete consistency. The achieved outcome enables a business impact evaluation for an accurate value proposition for complex well incremental rates, cash flow streams and sensitivity prediction cases.

Multilateral wells may vary in the number of laterals from two to even more than ten laterals. To control production on a lateral level, advance completion equipment, such as Inflow Control Valves (ICVs) and dual ported Permanent Down-hole Monitoring Systems (PDHMS) for real-time pressure measurements, are installed in the motherbore against each lateral to regulate reservoir fluids into the production tubing of the well from individual or collective set of laterals. Multilateral wells equipped with advance completion tools become complex wells (hereinafter called the "complex wells") in reservoir engineering. Recent complex wells involve multiple designs and architecture for which the reservoir exposure for laterals sums can be thousands of meters.

A new generation of smart multilateral well completions are Manara type wells where the laterals are divided into a number of segments or compartments using oil swell packers and Manara stations are placed against each segment for quantifying liquid rate, water cut and real-time reservoir pressure measurements. Each station is equipped with electrical control valve for controlling unwanted fluid production at a segment level. In this work, a new workflow is established to model and history match Manara well with complex modeling features using the GigaPOWERS (GP) simulator and a compositional full field model. The established segment level history matching workflow includes four important milestones to achieve. The first is advance well completion design at which the physical well completion details are translated and converted into a grid based completion details with the help of pre-processing tools. The second step covers the generation of three complex well-related input files for the GP simulator. The third step is the process of validating high frequency performance data including flow rate, choke size and pressure for the complex well stations. The last step involves conducting the history matching exercise on a segment level for every individual station to achieve the final history match model.

The conventional history match procedure includes generally three levels to match; field, group and well. In this workflow, station level match is done for the first time with a full field compositional model with a size of more than 61 million grid cells. The high definition history match is achieved at segment level for six stations that constitute a trilateral well with high frequency performance data. Complex well modeling in GP includes pressure drop calculations for several components. The pressure drops are related to friction, gravity, acceleration and advanced tools (i.e. ICV).

Modeling a complex well involves several pre- and post-simulation environment features at the segment level that should work in complete consistency. The achieved outcome enables a business impact evaluation for an accurate value proposition for complex well incremental rates, cash flow streams and sensitivity prediction cases.

Summary In this paper, we present a new well model for reservoir simulation. The proposed well model relates the volumetric flow rate and the bottomhole pressure (BHP) of the well to the reservoir pressure through a spatially distributed source term that is independent of the numerical method and the discrete mesh used to solve the flow problem. This is in contrast to the widely used Peaceman-type well models, which are inherently tied to a particular numerical discretization by the definition of an equivalent well radius. The proposed distributed well model does not require the calculation of an equivalent well radius. Hence, it can be readily applied to finite-difference, finite-volume (FV), or finite-element discretizations on arbitrarily unstructured meshes, which also makes it an attractive option for mesh-adaptation schemes. The new well model is demonstrated on a steady-state single-phase flow problem and an unsteady two-phase flow problem, using a conventional FV method and a high-order discontinuous Galerkin (DG) method. The distributed well model produces error-convergence behaviors that are very similar to the Peaceman well model on uniform structured meshes, but its applicability to high-order discretizations and mesh-adaptation schemes allows for higher convergence rates and more costefficient solutions, especially on adapted unstructured meshes. Introduction Representing the behavior of wells is an important part of the numerical simulation of fluid flows in the subsurface. The large disparity in length scales between a typical wellbore and a reservoir makes it computationally infeasible to explicitly model the near-well pressure behavior by increasing mesh resolution. Therefore, in most practical applications, a mathematical well model is used to capture the interaction between the wellbore and the reservoir, while still allowing the use of grid cells that are several orders of magnitude larger than the wellbore. One of the first theoretical studies of wells was performed by Peaceman (1978), in which a well model is presented for a cellcentered finite-difference method on square grids. The analysis provides an interpretation of the wellblock pressure and relates it to the flowing BHP of the well, under assumptions of single-phase flow in a homogeneous, isotropic reservoir.

A novel approach is introduced for simulation of multiphase flow, geomechanics, and fracture propagation on very general semi-structured grids. Complex networks consisting of both natural and hydraulically stimulated fractures are able to be represented using a diffusive zone model in large scale reservoirs. A mass conservative method called the enhanced velocity mixed finite element method is used to model multiphase flow with a fully-compositional equation-of-state model. Its recent reformulation on semi-structured, spatially non-conforming grids allows very general local refinement and dynamic mesh adaptivity.

Iteratively coupled geomechanics is simulated, which can predict fracture opening on fixed networks based upon induced stresses and poromechanical effects. In the most complex case, it is coupled with the phase field method to model nucleation and branching of non-planar fractures in highly heterogeneous media. Several examples are demonstrated to model fracture networks. The general semi-structured discretization can simulate flow and geomechanics on networks of fractures in large reservoirs with local resolution where desired. Dynamic adaptive mesh refinement can be used for both tracking transient flow features such as sharp the propagation of new fractures via hydraulic stimulation. This framework allows the seamless ability to switch from production to propagation scenarios, by varying the degrees of physics.

This work demonstrates a capability to perform high-fidelity simulations on complex fracture networks in large reservoirs at a reasonable computational cost. The gridding algorithms are straightforward extensions to traditional finite difference reservoir simulators. It can also be coupled with state-of-the-art complex phase field fracture propagation. This extends the capabilities of many legacy reservoir simulators to handle more physics.

A dynamic multilevel method for fully-coupled simulation of flow and heat transfer in heterogeneous and fractured geothermal reservoirs is presented (FG-ADM). The FG-ADM develops an advanced simulation method which maintains its efficiency when scaled up to field-scale applications, at the same time, it remains accurate in presence of complex fluid physics and heterogeneous rock properties. The embedded discrete fracture model is employed to accurately represent fractures without the necessity of unstructured complex grids. On the fine-scale system, FG-ADM introduces a multi-resolution nested dynamic grid, based on the dynamic time-dependent solution of the heat and mass transport equations. The fully-coupled implicit simulation strategy, in addition to the multilevel multiscale framework, makes FG-ADM to be stable and efficient in presence of strong flow-heat coupling terms. Furthermore, its finite-volume formulation preserves local conservation for both mass and heat fluxes. Multilevel local basis functions for pressure and temperature are introduced, in order to accurately represent the heterogeneous fractured rocks. These basis functions are constructed at the beginning of the simulation, and are reused during the entire dynamic timedependent simulation. For several heterogeneous test cases with complex fracture networks we show that, by employing only a fraction of the fine-scale grid cells, FG-ADM can accurately represent the complex flow-heat solutions in the fractured subsurface formations.

We present a software framework tailored to seismic applications embedding high performance features. Especially, we discuss the spatial cache-blocking algorithm for the Finite-Difference Time-Domain (FDTD) method and the protocol to find optimal parameters. We apply our modeling engine to perform 3D elastic seismic modeling on a large-scale 877 km² land survey. The velocity and density models are built with geological stratigraphic forward modeling. In total, 162300 shot gathers are computed with a maximum frequency of 20Hz and a maximum offset of 5 km. The main trait of the computed data is the complexity related to the surface-waves, especially their long and dispersive wave-train obscuring other wave arrivals. Computations were performed on the Kaust supercomputer Shaheen II. The modeling campaign took one month to complete using in average 2434 compute nodes in parallel. This achievement represents a workload of 59.6 ExaFLOP on a current PetaFLOP/s machine. This indicates that the next generation of supercomputers targeting the ExaFLOP/s sustained performance, would allow reducing the run-time of our application to one hour or less. With such performance, it is reasonable to predict that 3D elastic imaging will be a routinely used algorithm by seismic exploration in the years to come, while nowadays it still requires leading-edge hardware.

Is the Cloud Mature Enough for High-Performance Computing? The majority of Shell’s HPC work helps support its seismic imaging operations. The company supports 45 HPC applications, with the bulk of its processing and production workload taking place in-house. Oil and gas is in the midst of a pervasive digital transformation in which the industry is changing the way it manages assets, the way it interacts with customers, and the way it develops internal workflows. Perhaps one of the most significant impacts of this transformation, however, is the way in which companies characterize their subsurface data.

Is the Cloud Mature Enough for High-Performance Computing? The majority of Shell’s HPC work helps support its seismic imaging operations. The company supports 45 HPC applications, with the bulk of its processing and production workload taking place in-house. Oil and gas is in the midst of a pervasive digital transformation in which the industry is changing the way it manages assets, the way it interacts with customers, and the way it develops internal workflows. Perhaps one of the most significant impacts of this transformation, however, is the way in which companies characterize their subsurface data.