Conference Papers
Recent Submissions
-
Deep Reinforcement Learning Based Beamforming Codebook Design for RIS-aided mmWave Systems(IEEE, 2023-03-17) [Conference Paper]Reconfigurable intelligent surfaces (RISs) are envisioned to play a pivotal role in future wireless systems with the capability of enhancing propagation environments by intelligently reflecting the signals toward the target receivers. However, the optimal tuning of the phase shifters at the RIS is a challenging task due to the passive nature of reflective elements and the high complexity of acquiring channel state information (CSI). Conventionally, wireless systems rely on pre-defined reflection beamforming codebooks for both initial access and data transmission. However, these existing pre-defined codebooks are commonly not adaptive to the environments. Moreover, identifying the best beam is typically performed using an exhaustive search that leads to high beam training overhead. To address these issues, this paper develops a multi-agent deep reinforcement learning framework that learns how to jointly optimize the active beamforming from the BS and the RIS-reflection beam codebook relying only on the received power measurements. To accelerate learning convergence and reduce the search space, the proposed model divides the RIS into multiple partitions and associates beam patterns to the surrounding environments with low computational complexity. Simulation results show that the proposed learning framework can learn optimized active BS beamforming and RIS reflection codebook. For instance, the proposed MA-DRL approach with only 6 beams outperforms a 256-beam discrete Fourier transform (DFT) codebook with a 97% beam training overhead reduction.
-
DDOS attacks detection based on attention-deep learning and local outlier factor(IEEE, 2023-03-14) [Conference Paper]One of the most significant security concerns confronting network technology is the detection of distributed denial of service (DDOS). This paper introduces a semi-supervised data-driven approach to the detection of DDOS attacks. The proposed method employs normal events data without labeling to train the detection model. Specifically, this approach introduces an improved autoencoder (AE) model by incorporating a Gated Recurrent Unit (GRU) based on the attention mechanism (AM) at the encoder and decoder sides of the AE model. GRU enhances the AE's ability to learn temporal dependencies, and the AM enables the selection of relevant features. For DDOS attacks detection, the local outlier factor (LOF) anomaly detection algorithm is applied to extracted features from the improved AE model. The performance of the proposed approach has been verified using DDOS publically available datasets.
-
On-chip photoacoustic transducer based on monolithic integration of piezoelectric micromachined ultrasonic transducers and metasurface lenses(SPIE, 2023-03-09) [Conference Paper, Poster]An on-chip photoacoustic transducer is proposed by monolithically integrating piezoelectric micromachined ultrasonic transducers (PMUTs) on metasurface lenses for applications such as single-cell metabolic photoacoustic microscopy (SCM-PAM)1 . As shown in Figure 1a, every PMUT cell has a ring-shaped top electrode, and the membrane center is transparent without piezoelectric and electrode materials. The laser beam, therefore, can travel through a PMUT cell after being focused by a metasurface lens bonded on the backside of the PMUT (see Figure.3). The on-chip photoacoustic transducer fully leverages current PMUT and metasurface technologies and does not rely on transparent piezoelectric and electrode materials like typical transparent ultrasonic transducers2 . Moreover, the on-chip photoacoustic transducer has a monolithic integrated achromatic metasurface lens (see Figure 3), which can easily and efficiently focus the visible light (wavelength range: 400-700 nm) at the same focus point. Design and process this and preliminarily test the performance of PMUT and metasurface.
-
Cloud Native Applications Profiling using a Graph Neural Networks Approach(IEEE, 2023-03-08) [Conference Paper]The convergence of Telecommunication and industry operational networks towards cloud native applications has enabled the idea to integrate protection layers to harden security posture and management of cloud native based deployments. In this paper, we propose a data-driven approach to support detection of anomalies in cloud native application based on a graph neural network. The essence of the profiling relies on capturing interactions between different perspectives in cloud native applications through a network dependency graph and transforming it to a computational graph neural network. The latter is used to profile different deployed assets like micro-service types, workloads' namespaces, worker machines, management and orchestration machines as well as clusters. As a first phase of the profiling, we consider a fine-grained profiling on microservice types with an emphasis on network traffic indicators. These indicators are collected on distributed Kubernetes (K8S) deployment premises. Experimental results shows good trade-off in terms of accuracy and recall with respect to micro-service types profiling (around 96%). In addition, we used predictions entropy scores to infer anomalies in testing data. These scores allow to segregate between benign and anomalous graphs, where we identified 19 out of 23 anomalies. Moreover, by using entropy scores, we can conduct a root cause analysis to infer problematic micro-services.
-
Data-Driven Machine Learning Modeling of Mineral/CO2/Brine Wettability Prediction: Implications for CO2 Geo-Storage(SPE, 2023-03-07) [Conference Paper]CO2 wettability and the reservoir rock-fluid interfacial interactions are crucial parameters for successful CO2 geological sequestration. This study implemented the feed-forward neural network to model the wettability behavior in a ternary system of rock minerals (quartz and mica), CO2, and brine under different operating conditions. To gain higher accuracy of the machine learning models, a sufficient dataset was utilized that was recorded by conducting a large number of laboratory experiments under a realistic pressure range, 0 – 25 MPa and the temperatures range, 298 – 343 K. The mica substrates were used as a proxy for the caprock, and quartz substrates were used a proxy for the reservoir rock. Different graphical exploratory data analysis techniques, such as heatmaps, violin plots, and pairplots were used to analyze the experimental dataset. To improve the generalization capabilities of the machine learning models k-fold cross-validation method, and grid search optimization approaches were implemented. The machine learning models were trained to predict the receding and advancing contact angles of mineral/CO2/brine systems. Both statistical evaluation and graphical analyses were performed to show the reliability and performance of the developed models. The results showed that the implemented ML model accurately predicted the wettability behavior under various operating conditions. The training and testing average absolute percent relative errors (AAPE) and R2 of the FFNN model for mica and quartz were 0.981 and 0.972, respectively. The results confirm the accuracy performance of the ML algorithms. Finally, the investigation of feature importance indicated that pressure had the utmost influence on the contact angles of the minerals/CO2/brine system. The geological conditions profoundly affect rock minerals wetting characteristics, thus CO2 geo-storage capacities. The literature severely lacks advanced information and new methods for characterizing the wettability of mineral/CO2/brine systems at geo-storage conditions. Thus, the ML model's outcomes can be beneficial for precisely predicting the CO2 geo-storage capacities and containment security for the feasibility of large-scale geo-sequestration projects.
-
An Intelligent Safe Well Bottom-Hole Pressure Monitoring of CO2 Injection Well into Deep Saline: A coupled Hydro-Mechanical Approach(SPE, 2023-03-07) [Conference Paper]Geological Carbon Sequestration (GCS) in deep geological formations, like saline aquifers and depleted oil and gas reservoirs, brings enormous potential for large-scale storage of carbon dioxide (CO2). The successful implementation of GCS requires a comprehensive risk assessment of the confinement of plumes and storage potential at each storage site. To better understand the integrity of the caprock after injecting CO2, it is necessary to develop robust and fast tools to evaluate the safe CO2 injection duration. This study applied deep learning (DL) techniques, such as fully connected neural networks, to predict the safe injection duration. A physics-based numerical reservoir simulator was used to simulate the movement of CO2 for 170 years following a 30-year CO2 injection period into a deep saline aquifer. The uncertainty variables were utilized, including petrophysical properties such as porosity and permeability, reservoir physical parameters such as temperature, salinity, thickness, and operational decision parameters such as injection rate and perforation depth. As mentioned earlier, the reservoir model was sampled using the Latin-Hypercube sampling approach to account for a wide range of parameters. Seven hundred twenty-two reservoir simulations were performed to create training, testing, and validation datasets. The DNN model was trained, and several executions were performed to arrive at the best model. After multiple realizations and function evaluations, the predicted results revealed that the three-layer FCNN model with thirty neurons in each layer could predict the safe injection duration of CO2 into deep saline formations. The DNN model showed an excellent prediction efficiency with the highest coefficient of determination factor of above 0.98 and AAPE of less than 1%. Also, the trained predictive models showed excellent agreement between the simulated ground truth and predicted trapping index, yet 300 times more computationally efficient than the latter. These findings indicate that the DNN-based model can support the numerical simulation as an alternative to a robust predictive tool for estimating the performance of CO2 in the subsurface and help monitor the storage potential at each part of the GCS project.
-
Momentum-based ICA for Self Interference Cancellation in In-Band Full-Duplex Systems(IEEE, 2023-03-07) [Conference Paper]Recently, Independent Component Analysis (ICA) has proven its effectiveness as a self-interference cancellation method for in-band full duplex systems. However, ICA could suffer slow convergence due to the iterative estimation of the independent components which limits its usage in real-time applications. In this paper, we introduce a momentum-based ICA to accelerate convergence via incorporating gradient history. The proposed momentum-based ICA is evaluated and tested on different ICA algorithms including real-valued and complexvalued FastICA and entropy bound minimization based ICA. The results show significant speedup improvement compared to native ICA based on gradient descent approach. The proposed algorithm shows consistent results under different transceiver non-linearity and for different frame lengths.
-
Application of Image Processing Techniques in Deep-Learning Workflow to Predict CO2 Storage in Highly Heterogeneous Naturally Fractured Reservoirs: A Discrete Fracture Network Approach(SPE, 2023-03-07) [Conference Paper]Naturally fractured reservoirs (NFRs), such as fractured carbonate reservoirs, are commonly located worldwide and have the potential to be good sources of long-term storage of carbon dioxide (CO2). The numerical reservoir simulation models are an excellent source for evaluating the likelihood and comprehending the physics underlying behind the interaction of CO2 and brine in subsurface formations. For various reasons, including the rock's highly fractured and heterogeneous nature, the rapid spread of the CO2 plume in the fractured network, and the high capillary contrast between matrix and fractures, simulating fluid flow behavior in NFR reservoirs during CO2 injection is computationally expensive and cumbersome. This paper presents a deep-learning approach to capture the spatial and temporal dynamics of CO2 saturation plumes during the injection and monitoring periods of Geological Carbon Sequestration (GCS) sequestration in NFRs. To achieve our purpose, we have first built a base case physics-based numerical simulation model to simulate the process of CO2 injection in naturally fractured deep saline aquifers. A standalone package was coded to couple the discrete fracture network in a fully compositional numerical simulation model. Then the base case reservoir model was sampled using the Latin-Hypercube approach to account for a wide range of petrophysical, geological, reservoir, and decision parameters. These samples generated a massive physics-informed database of around 900 cases that provides a sufficient training dataset for the DL model. The performance of the DL model was improved by applying multiple filters, including the Median, Sato, Hessian, Sobel, and Meijering filters. The average absolute percentage error (AAPE), root mean square error (RMSE), Structural similarity index metric (SSIM), peak signal-to-noise ratio (PSNR), and coefficient of determination (R2) were used as error metrics to examine the performance of the surrogate DL models. The developed workflow showed superior performance by giving AAPE less than 5% and R2 more than 0.94 between ground truth and predicted values. The proposed DL-based surrogate model can be used as a quick assessment tool to evaluate the long-term feasibility of CO2 movement in a fracture carbonate medium.
-
Improved Amott Tests Help Quantify Primary Driving Forces in Spontaneous Imbibition in Water-Wet and Oil-Wet Limestone Rock(SPE, 2023-03-07) [Conference Paper]The improved oil recovery techniques, such as customized ionic composition waterflood or "smart-water" flood, are being developed to increment crude oil production. Counter-current spontaneous imbibition of brine into oil-saturated rock is a critical mechanism of recovery of the crude oil bypassed in highly-heterogeneous carbonate rocks. In laboratory, spontaneous imbibition in the Amott cell experiment is the main instrument to explore oil recovery from oil-saturated core plugs at different wettability conditions. The classical Amott test, however, masks a number of flaws that hinder interpretation of the physical phenomena in recovery dynamics and precise modeling of the cumulative recovery profiles. In this work, we identify these flaws in the spontaneous imbibition experiments with mixed-wet limestone samples saturated with crude oil. We describe an improved Amott method and study crude oil recovery from mixed-wet carbonate core plugs. The introduced modifications of the Amott test ensure reliable and reproducible results for both non-wetting mineral and crude oils. Finally, we show that the resulted smooth recovery profiles of oil production can be described with a mathematical model with high accuracy. For the first time, we show that generalized extreme value (GEV) distribution can be applied to model cumulative oil production from mixed-wet carbonate core samples.
-
Minimizing Drill-String-Induced Wellbore Instability(SPE, 2023-03-07) [Conference Paper]Objective/Scope: Every year the petroleum industry spends more than an estimated $6 billion in mitigating wellbore instabilities that account for nearly half of the drilling-related NPT (non-productive time). Researchers believe wellbore instability problems occur primarily due to physical and chemical interactions between rocks and drilling fluid and mostly neglect the impact of drill string vibrations on wellbore stability. However, such vibrations can cause significant damage to the formation, which then degrades the formation's mechanical integrity and compound wellbore instabilities. An appreciable body of evidence exists documenting that higher RPM (revolutions per minute), higher WOB (weight on bit), and pendulum BHA (bottom hole assembly) can cause enhanced agitation of the wellbore wall rock which may result in formation damage – often accompanied with an increased ROP (rate of penetration). The primary objective of this work is to review the state of modelling vibrations as documented in the literature and then advance the development and impact of vibrations and rock failure due to the aforementioned drilling parameters using numerical methods. Methods, Procedures, Process: The proposed research will analyze the complex dynamic interaction between drill-string and borehole using a commercially available finite element package (e.g., COMSOL or ABAQUS etc.). These packages have the capability to create multiphysics-based models and simulate engineering and industry applications. A drill-string borehole assembly will be modelled with geometric and associated environmental and material boundary conditions (e.g., stress state, pore-fluid pressure, mud pressure etc.). The results will then be calibrated with experimental and field data. Results, Observations, Conclusions: Subsurface drilling parameters are often undetermined or not measurable during drilling. The proposed model will aim at developing a drill-string-dynamics model to simulate the interactions between the wellbore and the drill string. This includes (i) estimation of drill string vibrations for different drill string and BHA designs, (ii) the resulting impact forces from drill string vibrations for different wellbore designs, and (iii) the influence of the estimated impact forces on the wellbore wall of particular (chosen) rock types at its current stress or yielding state. Novel/Additive information: The research project proposed herein will equip the well-planning engineers with advanced and robust tools to predict and mitigate wellbore instabilities resulting from drill-string vibrations. We develop new models that incorporate the dynamics mentioned above in order to evaluate quantitatively the effects of drill string vibrations on wellbore instabilities. We anticipate contributions along the following themes: Improved predictive capability of wellbore instabilities for specific drill string and well designs, leading to minimal wellbore rock failure trends for different trajectories, lithology, geologic structure, mud weights (i.e., overbalances) and more. Establish recommendations for optimal drilling parameters (WOB, RPM, and ROP) for specific drill strings and well designs as a function of input parameter uncertainties and variabilities.
-
Fully Coupled Hydromechanical Approach for Flow in Fractured Rocks Using Darcy-Brinkman-Biot(SPE, 2023-03-07) [Conference Paper]Coupling flow with geomechanical processes at the pore scale in fractured rocks is essential in understanding the macroscopic fluid flow processes of interest, such as geothermal energy extraction, CO2 sequestration, and hydrocarbon production from naturally and hydraulically fractured reservoirs. To investigate the microscopic (pore-scale) phenomena, we present a fully coupled mathematical formulation of fluid flow and geomechanical deformation to model the fluid flow in fractured rocks. In this work, we employ a Darcy-Brinkman-Biot approach to describe the fully coupled flow and geomechanical processes in fractured rocks at the pore scale. Darcy-Brinkman-Stokes (DBS) model is used to model multi-scale flow in the fractured rocks, in which fracture flow is described by Navier-Stokes equations and flow in the surrounding matrix is modeled by Darcy's law. With this approach, a unified conservation equation for flow in both media (fracture and matrix) is applied. We then apply Biot's poroelasticity theory and Terzaghi's effective stress theory to capture the geomechanical deformation. The continuity of the fluid pressure is imposed to connect the DBS equation and the stress-seepage equation. This coupled model is employed to determine the permeability within the microfracture. Numerical results show that this coupled approach can capture the permeability under the effects of solid deformation and multi-scale formation. We develop a fully coupled model to capture the pore-scale flow-geomechanically process in fractured rocks. To our knowledge, the fully coupled framework is developed and applied to characterize fracture permeability at the pore scale in fractured rocks for the first time.
-
Drilling Monitoring System: Mud Motor Condition and Performance Evaluation(SPE, 2023-03-07) [Conference Paper]Condition monitoring of bottom hole assembly (BHA) is essential during the different lifecycle stages of the drilling process, whether during the planning, implementation, or post-job failure analysis. Mud motor condition evaluation can assist in preventing mud motor damage and increasing drilling efficiency. This paper aims to develop a monitoring system that combines field data, data analytics and physics-based modelling to evaluate mud motor condition and performance. The drilling monitoring system is a set of modelling and analysis tools that utilize actual drilling data, power section performance data and drillstring design to recreate the drilling process. An unprocessed drilling dataset is required to assess the drilling operations (rotating or sliding mode, rotating off-bottom, backreaming, connections) and reconstruct the borehole trajectory from the measured survey and duty cycle (rotating and sliding mode). Interaction of the BHA and the borehole generate side forces and bending moments along the length of the BHA that are evaluated at each depth increment during the drilling process. Generated power and efficiency of the mud motor are calculated and incorporated into the dynamic simulation. The case study investigates two motor runs in vertical and inclined sections. Dynamic modelling and extensive data analytics assist in visualizing and correlating the input and output variables during the drilling process. The continuous evaluation of the differential pressure on the motor is the primary parameter that is investigated. The motor condition is established with a continuous wear-off test while drilling and correlation matrices to indicate a constant motor state for 135 hours. The system accurately monitors the motor's operating efficiency, with the additional advantage that the mud motor and drill bit performance are differentiated. Precise adjustments of the drilling parameters for the optimum depth of cut positively impact motor efficiency. In addition, an interesting observation shows that accurate modelling of downhole and surface torque provides significant insights regarding bit state. The results demonstrate that with the current methodology decrease in drilling efficiency is detected and is associated with bit wear. The work enables the evaluation of the mud motor condition and performance by utilizing a monitoring system and actual drilling data. The drilling monitoring system is a modelling analysis tool that can provide continuous feedback to the drilling operators about the condition of the BHA. Hence, it enables a real-time optimization process to manage mud motor condition and enhance drilling efficiency.
-
THz-band, Tbps MIMO Communications: A Joint Data Detection and Decoding Framework(IEEE, 2023-03-07) [Conference Paper]Efficient data detection and decoding are addressed under terahertz (THz)-band channel conditions and terabit-persecond (Tbps) baseband processing constraints. We investigate the performance and complexity tradeoffs of candidate data detectors in correlated ultra-massive multiple-input multiple-output (UM-MIMO) THz channels. Under high correlation, channel-matrix puncturing in subspace detectors can significantly reduce computational complexity and introduce much-needed parallelizability. Simulation results demonstrate that subspace detectors outperform conventional detectors in typical line-of-sight-dominated THz channel conditions. We advocate for a joint data detection and decoding framework that does not parallelize channel-code decoders to satisfy stringent Tbps baseband constraints but parallelizes data sources through channel puncturing and adopts short codes instead.
-
Advanced Hole Cleaning in Horizontal Wells: Experimental Investigation Supported by a Downhole Clamp-On Tool(SPE, 2023-03-07) [Conference Paper]Up to 80% of stuck pipe events are hole cleaning related in the case of high-angle wells. Therefore, significant attention should be given to understanding hole cleaning as it is crucial to restricting stuck pipe-related non-productive time (NPT). In order to optimize hole cleaning efficiency, the fundamental objective of the proposed paper is to experimentally investigate cuttings transport supported by a downhole clamp-on tool. This work approaches existing challenges by designing and building a custom flow loop that recreates the drilling environment of horizontal wells. The study provides additional steps and new ideas in developing a reliable experimental setup for a proper hole cleaning investigation. Accordingly, the process includes comprehensive dimensional analysis, detailed design, and building a desired experimental flow loop setup. A unique mechanical design allows pipe rotation while achieving a closed-loop system. A clamp-on tool assists in agitating the cuttings to reduce accumulation at the bottom of the borehole. Experimental performance with various cuttings compares scenarios with and without pipe rotation. Among the key factors influencing cuttings transport in horizontal wells are drill pipe rotation (RPM), flow rate (Q), mud rheology, cuttings size, flow regime, and penetration rate (ROP). This research focuses on the mechanical removal of solid cuttings. Experimental work emphasizes cuttings' behavior showing different patterns for their movement in deviated wells by utilizing image processing. Drillstring rotation proves to be a crucial factor for efficient hole cleaning. The specific shape and dimensions of the clamp- on tool affect the efficiency of the hole cleaning process and impact the distance covered by the agitated cuttings downstream of the tool. The concept of the tool depends on blades that agitate cuttings as it rotates. Optimum tool design considers the physical properties of the fluid and the cuttings. The results show that as the tool agitates cuttings and moves them into the higher velocity region, the cuttings advance with the flow, which improves cuttings transport and reduces bedding formation. Assuming low flow rates, tool application increased average particle velocity within the tool more than four times (372%) and twice after the tool. In addition, differential pressure (Δp) shows a significant decrease while the tool operates, indicating improvement in hole cleaning. Lab-scaled flow loop development aims to simulate drilling conditions with drillpipe rotation and different downhole clamp-on tool geometries. The results show different flow patterns from experimental observations of liquid-particles flow in the horizontal wellbore, assisted by the proposed downhole clamp- on tool. The innovative tool design is a promising step in reducing hole cleaning issues with mechanical- assisted tools.
-
Locating CO2 Leakage in Subsurface Traps Using Bayesian Inversion and Deep Learning(SPE, 2023-03-07) [Conference Paper]Geologic CO2 sequestration (GCS) is a promising engineering measure to reduce global greenhouse emissions. However, accurate detection of CO2 leakage locations from underground traps remains a challenging problem. This study proposes a workflow that combines Bayesian inversion and deep learning algorithms to detect the sites of CO2 leakage. There are four main steps in the workflow. Step 1: we identify the key uncertainty parameters. Here we mean the CO2 leakage location. Then we get the training set using Latin Hypercube Sampling (LHS) method and perform the high-fidelity simulation using CMG. Step 2: we train the surrogate model using the data set collected from the last step, in which the Bayesian optimization is used to tune the hyperparameters automatically. Step 3: we perform the Bayesian inversion to invert the CO2 leakage location, in which the surrogate serves as the forward model to reduce the computational expense. Step 4: we feed the inverted CO2 leakage location into the high-fidelity model to produce the pressure response. If the error between the pressure response between the surrogate and the high-fidelity model is small enough, the solution is accepted. Otherwise, the accuracy of the surrogate model and the convergence of the Bayesian inversion process are revisited. We validate this method using a synthetic model of CO2 injection. Results show that the proposed Bayesian inversion assisted by the deep learning algorithm can accurately detect the CO2 leakage location with narrow uncertainties. This approach provides an accurate and efficient way to detect CO2 leakage locations in real-time applications.
-
Investigating Separation Efficiency of Oil-in-Water Emulsions Subjected to an Acoustic Field(SPE, 2023-03-07) [Conference Paper]The demand for improved technologies that enhance the separation of low concentrations of oil from water and thus reduce the increasing retention times in surface separation facilities is increasing. Acoustophoresis is a promising technique where acoustic direct radiation forces can aid to demulsify O/W emulsions and enhance oil separation of the produced water stream. Herein, we explored the coalescence behavior of oil-in-water emulsion in a stationary acoustophoretic setup. To quantify the acoustic O/W separation efficiency and identify the proper acoustic parameters for various O/W emulsion compositions, a series of experiments were conducted in large-scale (4-in × 4in × 12-in) acoustic resonator using O/W emulsions of different compositions. The separation efficiency is assessed by comparing the oil layer thickness of an emulsion separated by gravity alone with that separated after being subjected to the standing wave field. The mixing time and speed was 15 minutes and 2000 rpm respectively which was optimized experimentally for stable emulsion. Rushton impeller was used for high turbulence mixing. Results demonstrated the importance of optimizing the acoustic parameters (frequency, power) with respect to the emulsion droplet size distribution for improving the separation efficiency. The formation of bands and the accelerated separation of oil droplets are facilitated by the application of sufficient energy to the proper standing wave. This study showed that when ultrasonic was applied to the emulsion under limited frequency and power, coalescence was shown on the surface, meaning in a standing wave field, oil droplets aggregate and collide in the anti-nodal planes, where their coalescence and buoyancy occur when direct radiation force (Aggregating oil droplet), and secondary acoustic force (when coalescence takes place due to the causing of attractive or repulsive forces), therefore separation will happen. Results showed that the oil layer thickness recovered from the O/W emulsion subjected to the acoustic field was ~70% higher than that of gravitational separation alone. Results also showed that deviating from the standing wave frequency or delivering excessive acoustic power can result in random droplet motion, secondary emulsification, and a decrease in separation efficiency. An observation experiment was done using stable emulsions with droplet sizes ranging from 50 to 100 micron subjected at ultrasonic radiation with varying amount of acoustic power starting with small amount of powerand increased gradually until reach the maximum power; 111 to 280 watts, showedemulsion separation is progressing with maximum efficiency or standing wave is creating the maximum capacity for an emulsion dispersed content. This regime of system depends critically on the oil content of an emulsion. Also, computational work has been conducted using COMSOL Multiphysics for illustration of frequency influence on oil separation. This work provides novel information to direct the field implementation of in-line acoustic oil-water separation tool by identifying the key parameters that influence oil coalescence hence the separation efficiency of the tool.
-
Deep Learning Solves Complex Physics With an Example of CO2 Mineralization(2023-03) [Conference Paper]
-
Numerical Investigations on the Doublet Huff and Puff Technology to Extract Heat from the Geothermal Reservoirs and Storing of CO2(IPTC, 2023-02-28) [Conference Paper]In this work, we studied the implementation of huff and puff technology to extract heat from the geothermal reservoir. Two-dimensional numerical investigations were carried out using a fully coupled two-phase thermo-hydro-mechanical model with dynamic rock and fluid properties. COMSOL Multiphysics (a finite element solver) was utilized to build the model. The CO2 geofluid is injected in a supercritical state in a water-saturated geothermal reservoir. The results were showing promising for the extraction of heat and storing of CO2. In the simulation model, we designed a well pair (two-vertical wells) system with two different operating perforations in the same well with huff and puff cycle operation, and this technology is named as Doublet Huff and Puff (DHP). Injection wells operating at the top of the formation and production wells are operating at the bottom. The injection well-1 and production well-1 are operating at same time (i.e., 2 years). During this period, injection well-2 and production well-2 are ideal, and injection well-1 and production well-1 are ideal while operating injection well-2 and production well-2. This process is continued till the whole reservoir is saturated with the injected CO2 and/or the reservoir temperature reaches 60 % (i.e., geothermal reservoir life) of its original temperature. The CO2 plume expanding throughout the reservoir effectively while extracting heat from the reservoir. The sensitivity of well distance, injection temperature, injection velocity, and perforation length on the production temperature was investigated. The production temperature stays stable and high for a long time and no influence on the production temperature. Thus, the proposed technique (DHP) can be implemented for sequestering large amounts of CO2 along with heat extraction in geothermal reservoirs.
-
Telecommunication Traffic Forecasting via Multi-task Learning(ACM, 2023-02-27) [Conference Paper]Accurate telecommunication time series forecasting is critical for smart management systems of cellular networks, and has a special challenge in predicting different types of time series simultaneously at one base station (BS), e.g., the SMS, Calls, and Internet. Unlike the well-studied single target forecasting problem for one BS, this distributed multi-target forecasting problem should take advantage of both the intra-BS dependence of different types of time series at the same BS and the inter-BS dependence of time series at different BS. To this end, we first propose a model to learn the inter-BS dependence by aggregating the multi-view dependence, e.g., from the viewpoint of SMS, Calls, and Internet. To incorporate the interBS dependence in time series forecasting, we then propose a Graph Gate LSTM (GGLSTM) model that includes a graph-based gate mechanism to unite those base stations with a strong dependence on learning a collaboratively strengthened prediction model. We also extract the intra-BS dependence by an attention network and use it in the final prediction. Our proposed approach is evaluated on two real-world datasets. Experiment results demonstrate the effectiveness of our model in predicting multiple types of telecom traffic at the distributed base stations.
-
EZInterviewer: To Improve Job Interview Performance with Mock Interview Generator(ACM, 2023-02-27) [Conference Paper]Interview has been regarded as one of the most crucial step for recruitment. To fully prepare for the interview with the recruiters, job seekers usually practice with mock interviews between each other. However, such a mock interview with peers is generally far away from the real interview experience: the mock interviewers are not guaranteed to be professional and are not likely to behave like a real interviewer. Due to the rapid growth of online recruitment in recent years, recruiters tend to have online interviews, which make it possible to collect real interview data from real interviewers. In this paper, we propose a novel application named EZInterviewer, which aims to learn from the online interview data and provides mock interview services to the job seekers. The task is challenging in two ways: (1) the interview data are now available but still of low-resource; (2) to generate meaningful and relevant interview dialogs requires thorough understanding of both resumes and job descriptions. To address the low-resource challenge, EZInterviewer is trained on a very small set of interview dialogs. The key idea is to reduce the number of parameters that rely on interview dialogs by disentangling the knowledge selector and dialog generator so that most parameters can be trained with ungrounded dialogs as well as the resume data that are not low-resource. Specifically, to keep the dialog on track for professional interviews, we pre-train a knowledge selector module to extract information from resume in the job-resume matching. A dialog generator is also pre-trained with ungrounded dialogs, learning to generate fluent responses. Then, a decoding manager is finetuned to combine information from the two pre-trained modules to generate the interview question. Evaluation results on a real-world job interview dialog dataset indicate that we achieve promising results to generate mock interviews. With the help of EZInterviewer, we hope to make mock interview practice become easier for job seekers.