Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Emulation”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 793 records · Page 44

A generic method for interfacing IEDs using low voltage interfaces to real-time simulators with hardware in the loop

This study presents an interface method (IM) to interconnect real-time simulators (RTSs) without amplifiers for different intelligent electronic devices (IEDs), such as protective relays, smart meters, and other devices. It is a significant topic for the power systems protection community because there are many protection devices with different requirements. These protection devices have pinouts that measure bus voltages/currents and breaker pole states and generate trip/close pulse signals. Most of the emulation test beds with RTSs and HIL need amplifiers to connect IEDs from different product manufacturers because the low-voltage interface levels are not available in some instruction manuals. Additionally, commercial amplifiers have cutoff frequency limitations and have an excessive cost when several IEDs are wired onto test beds. Then, the validation of protection and control systems with RTSs and IEDs in the loop from different vendors without amplifiers results in an unfeasible alternative. In this paper, the IM was based on using an interface box, power source, and RTS to identify IED inputs/outputs. Here, the sequence to recognize IED pinouts and calculate the current/voltage scaling factors is described. The IM was validated satisfactorily with RTS and a protection device (with unknown pinouts) in the loop.

42 ENGINEERING↗

Accelerating Laboratory-Based, All-Solid-State Battery Research and Development: Industrially Relevant Small-Batch Dry-Processing, Small and Low-Cost Test Fixtures, and Short-Tolerant Separators

This work presents improvements to mixing (for dry-processed electrodes), separator robustness, and cell testing jigs for all-solid-state batteries (ASSBs). These developments combine synergistically to enable rapid research and development of ASSB catholytes. A mechanical "kneader" is designed and built which emulates the mixing and fibrillation of poly-tetrafluoroethylene binder that occurs in industrially relevant twin-screw extruders. A modest improvement in the mixing and dispersion catholyte materials is observed using micro-resolution X-ray computed tomography (X-ray CT). Catholytes are paired with an In-metal anode and separated by a novel, dual-layer, polyaramid-fiber-supported, sulfide-solid electrolyte (Li6 PS5 Cl) separator. This separator achieves high short tolerance which enabled a 95% success rate across 20 attempted cells and to date over 100 successful cells have been built. ASSBs were tested in a 2032 coin-cell format. Simultaneous cycling of a large number of cells is further enabled by a compact and low cost (~ 38 USD per jig) pressure application jig of which 100 have been constructed to date. The utility of these advances is demonstrated through a brief study on the effect cathode-side conductive carbon and current collector type has on capacity and rate performance.

25 ENERGY STORAGE↗

Synergistic foam stabilization and transport improvement in simulated fractures with polyelectrolyte complex nanoparticles: Microscale observation using laser etched glass micromodels

Inaccessibility to direct pore scale observation in hydrocarbon recovery of tight shale formations poses a great challenge to water-energy nexus initiatives and necessitates the use of high throughput technologies to emulate environmentally friendly processes. Herein, we employ a precise glass micromodel fabrication and visualization method to isolate the supercritical CO 2 bubbles surrounded by CO 2 -water lamella prepared in saline produced water stabilized with molecular complexation of zwitterionic surfactants (ZS) and polyelectrolyte complex nanoparticles (PECNP). The Selective Laser Enhanced Etching (SLE) technique was selected for micromodel simulation of high-pressure flow. Two representative designs, (1) fracture/micro-crack network 28 and (2) fracture/matrix were etched on fused silica glass with a laser printing machine and scCO 2 foam was injected to study the foamability, propagation, stability, and fluid loss properties. The highly monodispersed and uniformly distributed array of scCO 2 bubbles were detected in flow of scCO 2 foam in highly saline brine containing ionic complexes of positively charged PECNPs and ZS, whereas foam flow with the lamella containing ZS in fractures offered a noticeably large and polydisperse array of scCO 2 bubbles. scCO 2 bubble motion and deformation were traced, and local description of foam flow was visually examined. The confined array of scCO 2 bubbles stabilized by ZS in microcracks was affected by bubble growth and coalescence, whereas the super-populated array of monodispersed scC O2 bubbles with lamella containing complexes of PECNP and ZS were able to fill the channels with stable configurations within the timeframe of comparative stability measurements. The ability of complex fluid to prevent the formation damage was evaluated through fluid loss visualization in micromodels. Probing scCO 2 foam transport in homogenous porous media revealed smaller volume leak-off for scCO 2 foam containing PECNP-ZS ionic complexes.

04 OIL SHALES AND TAR SANDS↗

An investigation of reaction parameters on geochemical storage of non-pure CO2 streams in iron oxides-bearing formations

Hematite deposit that is the main FeIII-bearing mineral in sedimentary red beds was proposed as a potential host repository for converting CO2 into carbonate minerals such as siderite (FeCO3), when CO2–SO2 gas mixtures are co-injected. This work investigated CO2 mineral trapping using hematite and sensitivity of the reactive systems to different parameters, including particle size, gas composition, temperature, pressure, and solid-to-liquid ratio. Experimental and modelling studies of hydrothermal experiments were conducted, which emulated a CO2 sequestration scenario by injecting CO2-SO2 gas streams into a NaCl-NaOH brine hosted in iron oxide-containing aquifer. This study provides novel information on the mineralogical changes and fluid chemistry derived from the co-injection of CO2-SO2 gas mixtures in hematite deposit. It can be concluded that the amount of siderite precipitate depends primarily on the SO2 content of the gas stream. Increasing SO2 content in the system could promote the reduction of Fe3+ from the hematite sample to Fe2+, which will be further available for its precipitation as siderite. Moreover, siderite precipitation is enhanced at low temperatures and high pressures. The influence of the solid to liquid ratio on the overall carbonation reaction suggests that the conversion increases if the system becomes more diluted.

CO2 storage↗

Architecture for the Next Generation System Management Tools

To get more results or greater accuracy, computational scientists execute their applications on distributed computing platforms such as Clusters, Grids and Clouds. These platforms are different in terms of hardware and software resources as well as locality: some span across multiple sites and multiple administrative domains whereas others are limited to a single site/domain. As a consequence, in order to scale their applica- tions up the scientists have to manage technical details for each target platform. From our point of view, this complexity should be hidden from the scientists who, in most cases, would prefer to focus on their research rather than spending time dealing with platform configuration concerns. In this article, we advocate for a system management framework that aims to automatically setup the whole run-time environment according to the applications needs. The main difference with regards to usual approaches is that they generally only focus on the software layer whereas we address both the hardware and the software expecta- tions through a unique system. For each application, scientists describe their requirements through the definition of a Virtual Platform (VP) and a Virtual System Environment (VSE). Relying on the VP/VSE definitions, the framework is in charge of: (i) the configuration of the physical infrastructure to satisfy the VP requirements, (ii) the setup of the VP, and (iii) the customization of the execution environment (VSE) upon the former VP. We propose a new formalism that the system can rely upon to successfully perform each of these three steps without burdening the user with the specifics of the configuration for the physical resources, and system management tools. This formalism leverages Goldberg s theory for recursive virtual machines by introducing new concepts based on system virtualization (identity, partitioning, aggregation) and emulation (simple, abstraction). This enables the definition of complex VP/VSE configurations without making assumptions about the hardware and the software re- sources. For each requirement, the system executes the corresponding operation with the appropriate management tool. As a proof of concept, we implemented a first prototype that currently interacts with several system management tools (e.g., OSCAR, the Grid 5000 toolkit, and XtreemOS) and that can be easily extended to integrate new resource brokers or cloud systems such as Nimbus, OpenNebula or Eucalyptus for instance.

99 GENERAL AND MISCELLANEOUS↗

Algorithms and data structures to accelerate network analysis

As the sheer amount of computer generated data continues to grow exponentially, new bottlenecks are unveiled that require rethinking our traditional software and hardware architectures. Here in this paper we present five algorithms and data structures (long queue emulation, lockless bimodal queues, tail early dropping, LFN tables, and multiresolution priority queues) designed to optimize the process of analyzing network traffic. We integrated these optimizations on R-Scope, a high performance network appliance that runs the Bro network analyzer, and present benchmarks showcasing performance speed ups of 5X at traffic rates of 10 Gbps.

97 MATHEMATICS AND COMPUTING↗

Software-Defined Network for End-to-end Networked Science at the Exascale

Domain science applications and workflow processes are currently forced to view the network as an opaque infrastructure into which they inject data and hope that it emerges at the destination with an acceptable Quality of Experience. There is little ability for applications to interact with the network to exchange information, negotiate performance parameters, discover expected performance metrics, or receive status/troubleshooting information in real time. The work we presen here is motivated by a vision for a new smart network and smart application ecosystem that will provide a more deterministic and interactive environment for domain science workflows. The Software-Defined Network for End-to-end Networked Science at Exascale (SENSE) system includes a model-based architecture, implementation, and deployment which enables automated end-to-end network service instantiation across administrative domains. An intent based interface allows applications to express their high-level service requirements, an intelligent orchestrator and resource control systems allow for custom tailoring of scalability and real-time responsiveness based on individual application and infrastructure operator requirements. This allows the science applications to manage the network as a first-class schedulable resource as is the current practice for instruments, compute, and storage systems. Deployment and experiments on production networks and testbeds have validated SENSE functions and performance. Emulation based testing verified the scalability needed to support research and education infrastructures. Key contributions of this work include an architecture definition, reference implementation, and deployment. This provides the basis for further innovation of smart network services to accelerate scientific discovery in the era of big data, cloud computing, machine learning and artificial intelligence.

97 MATHEMATICS AND COMPUTING↗

Deep Learning Explicit Differentiable Predictive Control Laws for Buildings

We present a differentiable predictive control (DPC) methodology for learning constrained control laws for unknown nonlinear systems. DPC poses an approximate solution to multiparametric programming problems emerging from explicit nonlinear model predictive control (MPC). Contrary to approximate MPC, DPC does not require supervision by an expert controller. Instead, a system dynamics model is learned from a small dataset of recorded observations of the perturbed system's dynamics and the control law is optimized offline by interaction with the learned system model. The DPC method is based on two sequential steps, i) system identification using a constrained neural state-space model, and ii) optimization of an explicit control law parametrized by another neural network in closed-loop simulation with the identified neural state-space model. The combination of a differentiable closed-loop system and penalty methods for constraint handling of system outputs and inputs allows us to optimize the control law's parameters directly by backpropagating economic MPC loss through the learned system model. By incorporating domain knowledge and leveraging established techniques from optimal control, our method leverages deep neural networks as nonlinear function approximators for system identification and control while avoiding concomitant costs of intractably large datasets, and computationally expensive over-parametrized models. The scalability, data efficiency, and constrained optimal control capability of the proposed DPC method are demonstrated in simulation using a multi-zone building emulator.

Drgona, Jan↗

Coordinated frequency regulation among asynchronous AC grids with an MTDC system

With the increasing penetration level of renewable power, the power systems’ inertias are decreasing. To ensure that the frequency is above the threshold of the under frequency load shedding, one effective way is to share the spinning reserves among asynchronous AC grids through HVDC systems. In this paper, a coordinated frequency regulation scheme that enables spinning reserves sharing is proposed for such hybrid AC/DC grids. This scheme is a communication-free method that consists of the droop control, the inertia emulation control and the frequency safety control. The combined response of the three controls can increase not only the frequency nadir but also the settled frequency of the disturbed grid. In addition, the hysteresis comparators and the power injection limits are adopted in the frequency regulation scheme to reduce the coupling effect among AC grids. Thus, under small and harmless disturbance, the frequency regulation scheme will be inactive; the disturbance in strong AC grids would not affect the weak AC systems. The frequency regulation scheme coordinates well with the automatic generation control of the grids. Furthermore, the feasibility and effectiveness of the proposed frequency regulation scheme are verified through simulation studies.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Adaptive critic design-based reinforcement learning approach in controlling virtual inertia-based grid-connected inverters

In this report, an adaptive critic design (ACD) approach is proposed to control the phase and voltage of a grid-connected virtual synchronous generator (VSG). The penetration of fast responding inertia-less power converters significantly affect the stability of the power system, especially weak systems such as micro grids. The concept of virtual inertia addresses this concern by virtually emulating the behavior of a synchronous generator. However, the conventional VSG is designed based on two conditions: (i) fixed operating point and (ii) inductive grid connections. The performance of VSGs in low-voltage semi-resistive microgrids is far from optimal. To overcome the aforementioned concerns, a heuristic dynamic programing (HDP) approach is proposed to optimally control grid-connected VSGs. The neural-network-based inherence of the HDP enables the proposed technique to adapt to any impedance angle. The HDP controller includes two subnetworks: (i) the action network that controls the system optimally and (ii) the critic network, which evaluates the effectiveness of the action network. The simulation and experimental results are provided to evaluate the effectiveness of the proposed technique. As shown, the HDP-based approach illustrates a better performance in comparison with the conventional PI-based VSG in various operating conditions.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Reduced order models for assessing CO 2 impacts in shallow unconfined aquifers

Risk assessment studies of potential CO 2 sequestration projects consider many factors, including the possibility of brine and/or CO 2 leakage from the storage reservoir. Detailed multiphase reactive transport simulations have been developed to predict the impact of such leaks on shallow groundwater quality; however, these simulations are computationally expensive and thus difficult to directly embed in a probabilistic risk assessment analysis. Here we present a process for developing computationally fast reduced-order models which emulate key features of the more detailed reactive transport simulations. A large ensemble of simulations that take into account uncertainty in aquifer characteristics and CO 2 /brine leakage scenarios were performed. Twelve simulation outputs of interest were used to develop response surfaces (RSs) using a MARS (multivariate adaptive regression splines) algorithm (Milborrow, 2015). A key part of this study is to compare different measures of ROM accuracy. We then show that for some computed outputs, MARS performs very well in matching the simulation data. The capability of the RS to predict simulation outputs for parameter combinations not used in RS development was tested using cross-validation. Again, for some outputs, these results were quite good. For other outputs, however, the method performs relatively poorly. Performance was best for predicting the volume of depressed-pH-plumes, and was relatively poor for predicting organic and trace metal plume volumes. We believe several factors, including the non-linearity of the problem, complexity of the geochemistry, and granularity in the simulation results, contribute to this varied performance. The reduced order models were developed principally to be used in probabilistic performance analysis where a large range of scenarios are considered and ensemble performance is calculated. We demonstrate that they effectively predict the ensemble behavior. But, the performance of the RSs is much less accurate when used to predict time-varying outputs from a single simulation. If an analysis requires only a small number of scenarios to be investigated, computationally expensive physics-based simulations would likely provide more reliable results. Finally, if the aggregate behavior of a large number of realizations is the focus, as will be the case in probabilistic quantitative risk assessment, the methodology presented here is relatively robust.

54 ENVIRONMENTAL SCIENCES↗

A numerical study of an atomizing jet in a resonant acoustic field

By applying state-of-the-art, high-fidelity methods for compressible multiphase flows, this work features a parametric study of a turbulent liquid jet atomizing under the influence of standing acoustic waves with five different amplitude values. We perform these simulations in a semi-periodic domain to isolate interactions between acoustic and atomization processes and to emulate a pure liquid fuel spray placed at the pressure node of a resonant acoustic field. Characterization of the flow includes measurements of the mean dimensions, droplet number, and surface area of the liquid distribution, alongside qualitative depictions of the interface evolution, breakup events, and hydrodynamic fields. As acoustic forcing is provided to the flow, the added energy provokes atomization when shear would otherwise be insufficient for breakup. The magnitude of the forcing determines the rate of droplet production, which accelerates as the evolving liquid surface modifies the induced motion of the gas phase. At the nodal plane, the acoustic field produces an organizing effect on the liquid structures.

42 ENGINEERING↗

Peridynamics and surrogate modeling of pressure-driven well stimulation

In this work we use the peridynamics theory of solid mechanics to simulate fracture in an annular rock domain subject to an in-situ stress and create surrogate models that predict the area of the resulting cracks. Peridynamics is a non-local formulation of continuum mechanics that naturally accommodates material discontinuities. Furthermore, unlike other fracture modeling techniques there is no need to provide information about the crack path. We utilize the peridynamics code Peridigm and take a two-stage approach to fracture modeling. First an implicit solve is performed to compute the in-situ stress state. We then execute an explicit solve where a pressure loading designed to emulate fluid-driven hydraulic fracture is applied at the borehole and transmitted to the pre-stressed rock. We present results from polynomial and single and multi-level Gaussian process surrogate models constructed from a sampling study of the peridynamics model. The surrogates predict crack area given a measure of the in-situ stress anisotropy and rise time and amplitude of the pressure loading. These surrogates take a minuscule fraction of peridynamics model's running time to evaluate and are a step towards enabling advanced optimization and uncertainty quantification workflows that require many model evaluations.

42 ENGINEERING↗

What Is the Best Way to Contour Lung Tumors on PET Scans? Multiobserver Validation of a Gradient-Based Method Using a NSCLC Digital PET Phantom

Purpose: To evaluate the accuracy and consistency of a gradient-based positron emission tomography (PET) segmentation method, GRADIENT, compared with manual (MANUAL) and constant threshold (THRESHOLD) methods. Methods and Materials: Contouring accuracy was evaluated with sphere phantoms and clinically realistic Monte Carlo PET phantoms of the thorax. The sphere phantoms were 10-37 mm in diameter and were acquired at five institutions emulating clinical conditions. One institution also acquired a sphere phantom with multiple source-to-background ratios of 2:1, 5:1, 10:1, 20:1, and 70:1. One observer segmented (contoured) each sphere with GRADIENT and THRESHOLD from 25% to 50% at 5% increments. Subsequently, seven physicians segmented 31 lesions (7-264 mL) from 25 digital thorax phantoms using GRADIENT, THRESHOLD, and MANUAL. Results: For spheres <20 mm in diameter, GRADIENT was the most accurate with a mean absolute % error in diameter of 8.15% (10.2% SD) compared with 49.2% (51.1% SD) for 45% THRESHOLD (p < 0.005). For larger spheres, the methods were statistically equivalent. For varying source-to-background ratios, GRADIENT was the most accurate for spheres >20 mm (p < 0.065) and <20 mm (p < 0.015). For digital thorax phantoms, GRADIENT was the most accurate (p < 0.01), with a mean absolute % error in volume of 10.99% (11.9% SD), followed by 25% THRESHOLD at 17.5% (29.4% SD), and MANUAL at 19.5% (17.2% SD). GRADIENT had the least systematic bias, with a mean % error in volume of -0.05% (16.2% SD) compared with 25% THRESHOLD at -2.1% (34.2% SD) and MANUAL at -16.3% (20.2% SD; p value <0.01). Interobserver variability was reduced using GRADIENT compared with both 25% THRESHOLD and MANUAL (p value <0.01, Levene's test). Conclusion: GRADIENT was the most accurate and consistent technique for target volume contouring. GRADIENT was also the most robust for varying imaging conditions. GRADIENT has the potential to play an important role for tumor delineation in radiation therapy planning and response assessment.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Automation and Intensity Modulated Radiation Therapy for Individualized High-Quality Tangent Breast Treatment Plans

Purpose: To demonstrate the large-scale clinical implementation and performance of an automated treatment planning methodology for tangential breast intensity modulated radiation therapy (IMRT). Methods and Materials: Automated planning was used to prospectively plan tangential breast IMRT treatment for 1661 patients between June 2009 and November 2012. The automated planning method emulates the manual steps performed by the user during treatment planning, including anatomical segmentation, beam placement, optimization, dose calculation, and plan documentation. The user specifies clinical requirements of the plan to be generated through a user interface embedded in the planning system. The automated method uses heuristic algorithms to define and simplify the technical aspects of the treatment planning process. Results: Automated planning was used in 1661 of 1708 patients receiving tangential breast IMRT during the time interval studied. Therefore, automated planning was applicable in greater than 97% of cases. The time for treatment planning using the automated process is routinely 5 to 6 minutes on standard commercially available planning hardware. We have shown a consistent reduction in plan rejections from plan reviews through the standard quality control process or weekly quality review multidisciplinary breast rounds as we have automated the planning process for tangential breast IMRT. Clinical plan acceptance increased from 97.3% using our previous semiautomated inverse method to 98.9% using the fully automated method. Conclusions: Automation has become the routine standard method for treatment planning of tangential breast IMRT at our institution and is clinically feasible on a large scale. The method has wide clinical applicability and can add tremendous efficiency, standardization, and quality to the current treatment planning process. The use of automated methods can allow centers to more rapidly adopt IMRT and enhance access to the documented improvements in care for breast cancer patients, using technologies that are widely available and already in clinical use.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

On the Bayesian calibration of computer model mixtures through experimental data, and the design of predictive models

For many real systems, several computer models may exist with different physics and predictive abilities. To achieve more accurate simulations/predictions, it is desirable for these models to be properly combined and calibrated. Here, we propose the Bayesian calibration of computer model mixture method which relies on the idea of representing the real system output as a mixture of the available computer model outputs with unknown input dependent weight functions. The method builds a fully Bayesian predictive model as an emulator for the real system output by combining, weighting, and calibrating the available models in the Bayesian framework. Moreover, it fits a mixture of calibrated computer models that can be used by the domain scientist as a mean to combine the available computer models, in a flexible and principled manner, and perform reliable simulations. It can address realistic cases where one model may be more accurate than the others at different input values because the mixture weights, indicating the contribution of each model, are functions of the input. Inference on the calibration parameters can consider multiple computer models associated with different physics. The method does not require knowledge of the fidelity order of the models. We provide a technique able to mitigate the computational overhead due to the consideration of multiple computer models that is suitable to the mixture model framework. We implement the proposed method in a real-world application involving the Weather Research and Forecasting large-scale climate model.

97 MATHEMATICS AND COMPUTING↗

A physics-informed and hierarchically regularized data-driven model for predicting fluid flow through porous media

This paper presents a new deep learning data-driven model for predicting structure dependent pore-fluid velocity fields in rock. The model is based on a Convolutional Auto-Encoder (CAE) artificial neural network capable of learning from image data generated by direct numerical simulations of fluid flow through pore-structures, such as by Lattice Boltzmann or molecular dynamics methods. The main novelty of the model in comparison to previous CAE-based data-driven approaches consists of three parts. The first is a methodology for decomposing the full-domain of the porous media into sub-regions, or “sub-domains”, in order to reduce the overall size of the CAE, batch process the sub-domains in parallel, and enable the CAE to learn local and generalizable nonlinear mappings of pore-fluid velocities. The second consists of embedding the finite difference solutions of the incompressible Navier-Stokes and continuity equations into convolutional layers prior to the CAE in order to provide the CAE with knowledge of fluid dynamics physics (PhyFlow). The third main novelty is that the training of the CAE is regularized with a hierarchical loss function that encourages the learning of fluid flow patterns (in a way similar to ranked modes in principal component analysis), ranking from most to least important. This is shown to increase the stability in learning, reduce over-fitting, and promote interpretability of the CAE neural network layers (HierCAE). The comprehensive new data-driven model, which we call the PhyFlow-HierCAE model, is shown to exhibit improved accuracy and generalizability of flow field predictions over conventional CAE models, attributable to the embedded physical knowledge and the hierarchical regularization, as well as realize orders of magnitude speed-ups in computation times as a surrogate for the direct numerical simulations. Examples of training and forward predictions on unseen pore-structures are provided and evaluated for data from Lattice Boltzmann and molecular dynamics simulations of pore-fluid flow. The model is shown to be a fast and accurate emulator (or “surrogate”) for predicting effective permeability of unseen pore-structures based on learning from relatively small direct numerical simulation datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A gradient-based deep neural network model for simulating multiphase flow in porous media

We report simulation of multiphase flow in porous media is crucial for the effective management of subsurface energy and environment-related activities. The numerical simulators used for modeling such processes rely on spatial and temporal discretization of the governing mass and energy balance partial-differential equations (PDEs) into algebraic systems via finite-difference/volume/element methods. These simulators usually require dedicated software development and maintenance, and suffer low efficiency from a runtime and memory standpoint for problems with multi-scale heterogeneity, coupled-physics processes or fluids with complex phase behavior. Therefore, developing cost-effective, data-driven models can become a practical choice, and in this work, we choose deep learning approaches as they can handle high dimensional data and accurately predict state variables with strong nonlinearity. In this paper, we describe a gradient-based deep neural network (GDNN) constrained by the physics related to multiphase flow in porous media. We tackle the nonlinearity of flow in porous media induced by rock heterogeneity, fluid properties, and fluid-rock interactions by decomposing the nonlinear PDEs into a dictionary of elementary differential operators. We use a combination of operators to handle rock spatial heterogeneity and fluid flow by advection. Since the augmented differential operators are inherently related to the physics of fluid flow, we treat them as first principles prior knowledge to regularize the GDNN training. We use the example of pressure management at geologic CO 2 storage sites, where CO 2 is injected in saline aquifers and brine is produced, and apply GDNN to construct a predictive model that is trained with physics-based simulation data and emulates the physics process. We demonstrate that GDNN can effectively predict the nonlinear patterns of subsurface responses, including the temporal and spatial evolution of the pressure and saturation plumes. We also successfully extend the GDNN to convolutional neural network (CNN), namely gradient-based CNN (GCNN), and validate its capability to improve the prediction accuracy. GDNN has great potential to tackle challenging problems that are governed by highly nonlinear physics and enable the development of data-driven models with higher fidelity.

42 ENGINEERING↗