Bayesian neural networks for virtual flow metering: An empirical study

Grimstad, Bjarne; Hotvedt, Mathilde; Sandnes, Anders T.; Kolbjørnsen, Odd; Imsland, Lars

doi:10.1016/j.asoc.2021.107776

Cited by 17 publications

(7 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When applied to VFM, for which the data is generated by a nonstationary process, both virtual and real concept drift will negatively influence the long-term predictive performance. A VFM performance that diminishes with time, has been documented in several publications (Grimstad et al, 2021;Hotvedt et al, 2022;Sandnes et al, 2021). In the following section, passive learning methods are discussed.…”

Section: Parameter Estimation Of Steady-state Modelsmentioning

confidence: 94%

“…On the other hand, due to the slow dynamics of the reservoir, steady-state reservoir conditions for a certain time interval can often be assumed (Shippen, 2012). Furthermore, considering the inherent complex multiphase flow characteristics, which make it challenging to develop and solve nonstationary VFMs, steadystate VFMs are the most common approach in literature (Bikmukhametov and Jäschke, 2019), both for physics-based models (Shippen, 2012;Varyan et al, 2015) and machine learning (ML) models (AL-Qutami et al, 2017a,b,c, 2018Bikmukhametov and Jäschke, 2020;Grimstad et al, 2021). Nevertheless, studies show that steady-state VFM models should be updated or recalibrated in time to provide adequate long-term prediction accuracy (Sandnes et al, 2021;Hotvedt et al, 2022).…”

Section: Test Separatormentioning

confidence: 99%

“…and not improve the model performance significantly (AL-Qutami et al, 2018;Grimstad et al, 2021). Thus, a windowing strategy can be applied to discard redundant data (Ditzler et al, 2015).…”

Section: Comparison Of Periodic Batch and Online Learningmentioning

confidence: 99%

See 2 more Smart Citations

Passive learning to address nonstationarity in virtual flow metering applications

Hotvedt¹,

Grimstad²,

Imsland³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Steady-state process models are common in virtual flow meter applications due to low computational complexity, and low model development and maintenance cost. Nevertheless, the prediction performance of steady-state models typically degrades with time due to the inherent nonstationarity of the underlying process being modeled. Few studies have investigated how learning methods can be applied to sustain the prediction accuracy of steady-state virtual flow meters. This paper explores passive learning, where the model is frequently calibrated to new data, as a way to address nonstationarity and improve long-term performance. An advantage with passive learning is that it is compatible with models used in the industry. Two passive learning methods, periodic batch learning and online learning, are applied with varying calibration frequency to train virtual flow meters. Six different model types, ranging from data-driven to first-principles, are trained on historical production data from 10 petroleum wells. The results are two-fold: first, in the presence of frequently arriving measurements, frequent model updating sustains an excellent prediction performance over time; second, in the presence of intermittent and infrequently arriving measurements, frequent updating in addition to the utilization of expert knowledge is essential to increase the performance accuracy. The investigation may be of interest to experts developing soft-sensors for nonstationary processes, such as virtual flow meters.

show abstract

Section: Parameter Estimation Of Steady-state Modelsmentioning

confidence: 94%

Section: Test Separatormentioning

confidence: 99%

See 1 more Smart Citation

Passive learning to address nonstationarity in virtual flow metering applications

Hotvedt¹,

Grimstad²,

Imsland³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Neural network based approaches are prevalent and a common feature of earlier works is that wells are modeled individually from well tests or MPFM measurements. Several studies report impressive results, achieving average prediction errors below 5%, albeit with limited sample sizes [9,7,10].…”

Section: Introductionmentioning

confidence: 99%

“…The recent study in [10] applied Bayesian neural networks to data-driven VFM. The authors questioned if a robust data-driven method can be obtained by individually modeling wells from historical observations.…”

Section: Introductionmentioning

confidence: 99%

Multi-task learning for virtual flow metering

Sandnes¹,

Grimstad²,

Kolbjørnsen³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Virtual flow metering (VFM) is a cost-effective and non-intrusive technology for inferring multi-phase flow rates in petroleum assets. Inferences about flow rates are fundamental to decision support systems which operators extensively rely on. Data-driven VFM, where mechanistic models are replaced with machine learning models, has recently gained attention due to its promise of lower maintenance costs. While excellent performance in small sample studies have been reported in the literature, there is still considerable doubt towards the robustness of data-driven VFM. In this paper we propose a new multi-task learning (MTL) architecture for data-driven VFM. Our method differs from previous methods in that it enables learning across oil and gas wells. We study the method by modeling 55 wells from four petroleum assets. Our findings show that MTL improves robustness over single task methods, without sacrificing performance. MTL yields a 25-50% error reduction on average for the assets where single task architectures are struggling.

show abstract

Artificial neural networks

Ahmadi

2024

Artificial Intelligence for a More Sustainable Oil and Gas Industry and the Energy Transition

View full text Add to dashboard Cite

Bayesian neural networks for virtual flow metering: An empirical study

Cited by 17 publications

References 18 publications

Passive learning to address nonstationarity in virtual flow metering applications

Passive learning to address nonstationarity in virtual flow metering applications

Multi-task learning for virtual flow metering

Artificial neural networks

Contact Info

Product

Resources

About