A Bayesian approach to the dynamic modeling of ESP-lifted oil well systems: An experimental validation on an ESP prototype

Costa, Erbet A.; Abreu, Odilon S.L. de; Silva, T.de O.; Ribeiro, Marcos Pellegrini; Schnitman, Leizer

doi:10.1016/j.petrol.2021.108880

Cited by 11 publications

(25 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This research's primary contribution lies in presenting a novel approach for constructing and validating machine learning (ML) models specifically tailored for ESP systems. The key concept involves calculating the uncertainties associated with an experimentally validated nonlinear model proposed by Costa et al [11] and subsequently propagating these uncertainties to ML-based models.…”

Section: Introductionmentioning

confidence: 99%

“…Regarding the problem of modeling ESP systems, when it comes to modeling the process variables of ESP systems, the focus has been on finding linear or nonlinear phenomenological dynamic models [11] , [16] , [23] , [24] . These models are stiff nonlinear ordinary differential equations (ODEs), which are costly and possibly unstable to simulate and can be troublesome for real-time applications, e.g., control and real-time optimization (RTO).…”

Section: Introductionmentioning

confidence: 99%

“…Thus, Haykin [13] presents several learning models and points out that nonlinear autoregressive predictors with exogenous inputs (NARX) are suitable for representing dynamic systems. Therefore, this work proposes to group the operating variables of the ESP system through the NARX predictor, hence compiling a dynamic predictor that emulates mechanistic models such as the models of Costa et al [11] , Krishnamoorthy et al [16] , Pavlov et al [23] and Rønning [24] . Then, the DNN model can be trained using the predictor structure.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

An uncertainty approach for Electric Submersible Pump modeling through Deep Neural Network

Almeida Costa,

de Menezes Rebello,

Viena Santana

et al. 2024

Heliyon

Self Cite

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An uncertainty approach for Electric Submersible Pump modeling through Deep Neural Network

Almeida Costa,

de Menezes Rebello,

Viena Santana

et al. 2024

Heliyon

Self Cite

View full text Add to dashboard Cite

“…Another recent study Costa et al (2021) applied a similar method to an electric submersible pump system. With the inputs as prior, Bayesian inference and the MCMC methods were adopted for parameter estimation.…”

Section: Introductionmentioning

confidence: 99%

Parameter Estimation for a Gas Lifting Oil Well Model Using Bayes' Rule and the Metropolis–Hastings Algorithm

Ban¹,

Ghaderi²,

Janatian³

et al. 2022

MIC

View full text Add to dashboard Cite

Oil well models are frequently used in the oil production process. Estimation of unknown parameters of these models has long been a question of great interest in the oil industry field. Data collected from an oil well system can be useful for identifying and characterizing the parameters in the corresponding model. In this article, we present a solution to estimate the parameters and uncertainty of a gas lifting oil well model by designing Bayesian inference and using the Metropolis-Hastings algorithm. To present and evaluate the estimation, the performance of the chains and the distributions of the parameters were shown, followed by posterior predictive distributions and sensitivity analysis. Compared with the conventional maximum likelihood estimation methods that tried to identify one optimum value for each parameter, more information of the parameters is obtained by using the proposed model. The insights gained from this study can benefit the optimization and advanced control for the oil well operation.

show abstract

“…The convergence of the chains, acceptance ratio, posterior distributions, correlation between posterior parameters, the reliability plot and posterior predictive checks were presented to evaluate the approach. Kang et al [9] applied a similar method to an electric submersible pump system. With given prior, Bayesian inference and the MCMC methods were adopted for parameter estimation.…”

Section: Introductionmentioning

confidence: 99%

Parameter and State Estimation for an Oil Production Model using Julia

Ban¹,

Pfeiffer²,

Lie³

2022

Linköping Electronic Conference Proceedings

View full text Add to dashboard Cite

Dynamic models of industrial processes play an instrumental role in the operation of such processes from smart sensors, data reconciliation, to advanced control. For good performance, a precise model is normally required. The issue of improving models has received considerable critical attention. In this work, we consider the estimation of model parameters and initial states of a gas lifting oil well model, followed by filtering of its states. By utilizing information from both first-principle model and data, the results are presented to show the estimated values and their uncertainties. Julia is the main programming language used in this study. This research study provided an opportunity to advance the understanding of the optimization and estimation for the oil well operation.

show abstract

A Bayesian approach to the dynamic modeling of ESP-lifted oil well systems: An experimental validation on an ESP prototype

Cited by 11 publications

References 17 publications

An uncertainty approach for Electric Submersible Pump modeling through Deep Neural Network

An uncertainty approach for Electric Submersible Pump modeling through Deep Neural Network

Parameter Estimation for a Gas Lifting Oil Well Model Using Bayes' Rule and the Metropolis–Hastings Algorithm

Parameter and State Estimation for an Oil Production Model using Julia

Contact Info

Product

Resources

About