-
Deep Learning-Accelerated 3D Carbon Storage Reservoir Pressure Forecasting Based on Data Assimilation Using Surface Displacement from InSAR
Authors:
Hewei Tang,
Pengcheng Fu,
Honggeun Jo,
Su Jiang,
Christopher S. Sherman,
François Hamon,
Nicholas A. Azzolina,
Joseph P. Morris
Abstract:
Fast forecasting of reservoir pressure distribution in geologic carbon storage (GCS) by assimilating monitoring data is a challenging problem. Due to high drilling cost, GCS projects usually have spatially sparse measurements from wells, leading to high uncertainties in reservoir pressure prediction. To address this challenge, we propose to use low-cost Interferometric Synthetic-Aperture Radar (In…
▽ More
Fast forecasting of reservoir pressure distribution in geologic carbon storage (GCS) by assimilating monitoring data is a challenging problem. Due to high drilling cost, GCS projects usually have spatially sparse measurements from wells, leading to high uncertainties in reservoir pressure prediction. To address this challenge, we propose to use low-cost Interferometric Synthetic-Aperture Radar (InSAR) data as monitoring data to infer reservoir pressure build up. We develop a deep learning-accelerated workflow to assimilate surface displacement maps interpreted from InSAR and to forecast dynamic reservoir pressure. Employing an Ensemble Smoother Multiple Data Assimilation (ES-MDA) framework, the workflow updates three-dimensional (3D) geologic properties and predicts reservoir pressure with quantified uncertainties. We use a synthetic commercial-scale GCS model with bimodally distributed permeability and porosity to demonstrate the efficacy of the workflow. A two-step CNN-PCA approach is employed to parameterize the bimodal fields. The computational efficiency of the workflow is boosted by two residual U-Net based surrogate models for surface displacement and reservoir pressure predictions, respectively. The workflow can complete data assimilation and reservoir pressure forecasting in half an hour on a personal computer.
△ Less
Submitted 26 January, 2022; v1 submitted 21 January, 2022;
originally announced January 2022.
-
A Deep Learning-Accelerated Data Assimilation and Forecasting Workflow for Commercial-Scale Geologic Carbon Storage
Authors:
Hewei Tang,
Pengcheng Fu,
Christopher S. Sherman,
Jize Zhang,
Xin Ju,
François Hamon,
Nicholas A. Azzolina,
Matthew Burton-Kelly,
Joseph P. Morris
Abstract:
Fast assimilation of monitoring data to update forecasts of pressure buildup and carbon dioxide (CO2) plume migration under geologic uncertainties is a challenging problem in geologic carbon storage. The high computational cost of data assimilation with a high-dimensional parameter space impedes fast decision-making for commercial-scale reservoir management. We propose to leverage physical underst…
▽ More
Fast assimilation of monitoring data to update forecasts of pressure buildup and carbon dioxide (CO2) plume migration under geologic uncertainties is a challenging problem in geologic carbon storage. The high computational cost of data assimilation with a high-dimensional parameter space impedes fast decision-making for commercial-scale reservoir management. We propose to leverage physical understandings of porous medium flow behavior with deep learning techniques to develop a fast history matching-reservoir response forecasting workflow. Applying an Ensemble Smoother Multiple Data Assimilation framework, the workflow updates geologic properties and predicts reservoir performance with quantified uncertainty from pressure history and CO2 plumes interpreted through seismic inversion. As the most computationally expensive component in such a workflow is reservoir simulation, we developed surrogate models to predict dynamic pressure and CO2 plume extents under multi-well injection. The surrogate models employ deep convolutional neural networks, specifically, a wide residual network and a residual U-Net. The workflow is validated against a flat three-dimensional reservoir model representative of a clastic shelf depositional environment. Intelligent treatments are applied to bridge between quantities in a true-3D reservoir model and those in a single-layer reservoir model. The workflow can complete history matching and reservoir forecasting with uncertainty quantification in less than one hour on a mainstream personal workstation.
△ Less
Submitted 10 January, 2022; v1 submitted 9 May, 2021;
originally announced May 2021.
-
Using monodromy to statistically estimate the number of solutions
Authors:
Jonathan D. Hauenstein,
Samantha N. Sherman
Abstract:
Synthesis problems for linkages in kinematics often yield large structured parameterized polynomial systems which generically have far fewer solutions than traditional upper bounds would suggest. This paper describes statistical models for estimating the generic number of solutions of such parameterized polynomial systems. The new approach extends previous work on success ratios of parameter homot…
▽ More
Synthesis problems for linkages in kinematics often yield large structured parameterized polynomial systems which generically have far fewer solutions than traditional upper bounds would suggest. This paper describes statistical models for estimating the generic number of solutions of such parameterized polynomial systems. The new approach extends previous work on success ratios of parameter homotopies to using monodromy loops as well as the addition of a trace test that provides a stop** criterion for validating that all solutions have been found. Several examples are presented demonstrating the method including Watt I six-bar motion generation problems.
△ Less
Submitted 2 October, 2020; v1 submitted 24 April, 2020;
originally announced May 2020.
-
Data-driven Super-resolution on a Tactile Dome
Authors:
Pedro Piacenza,
Sydney Sherman,
Matei Ciocarlie
Abstract:
While tactile sensor technology has made great strides over the past decades, applications in robotic manipulation are limited by aspects such as blind spots, difficult integration into hands, and low spatial resolution. We present a method for localizing contact with high accuracy over curved, three dimensional surfaces, with a low wire count and reduced integration complexity. To achieve this, w…
▽ More
While tactile sensor technology has made great strides over the past decades, applications in robotic manipulation are limited by aspects such as blind spots, difficult integration into hands, and low spatial resolution. We present a method for localizing contact with high accuracy over curved, three dimensional surfaces, with a low wire count and reduced integration complexity. To achieve this, we build a volume of soft material embedded with individual off-the-shelf pressure sensors. Using data driven techniques, we map the raw signals from these pressure sensors to known surface locations and indentation depths. Additionally, we show that a finite element model can be used to improve the placement of the pressure sensors inside the volume and to explore the design space in simulation. We validate our approach on physically implemented tactile domes which achieve high contact localization accuracy ($1.1mm$ in the best case) over a large, curved sensing area ($1,300mm^2$ hemisphere). We believe this approach can be used to deploy tactile sensing capabilities over three dimensional surfaces such as a robotic finger or palm.
△ Less
Submitted 26 February, 2018;
originally announced February 2018.
