Multiphase Flow Metering Research Articles

Deep learning models for improved accuracy of a multiphase flowmeter

Measurement of oil and gas two-phase flow with variable flow regimes relies to a large extent on flow patterns and their transitions. Using multiphase flowmeters in flows with high gas volume fractions is therefore usually associated with large uncertainties. This work presents a dynamic neural network method to measure the flow rate using a nonlinear autoregressive network with exogenous inputs (NARX). Total temperature and total pressure are used as network inputs and the obtained results are compared with a multilayer perceptron (MLP). Comparison between modeling results and the experimental data shows that the NARX network can predict oil and gas flow with variable flow regimes with less error compared to the MLP model, e.g. an absolute average percentage deviation (AAPD) of 0.68% instead of 1.02%. The present work can hence be seen as a proof-of-concept study that should motivate further applications of deep learning models to facilitate enhanced accuracy in flow metering.

The Investigation of Gas Distribution Asymmetry Effect on Coriolis Flowmeter Accuracy at Multiphase Metering.

Multiphase flows are encountered in various industries, and the Coriolis flowmeter (CFM) is considered a high potential flowmeter for the metering of these flows. However, the decoupling effect and asymmetrical gas distribution in a CFM might decrease the accuracy of its multiphase flow metering The asymmetry of gas distribution in a CFM and its influence on the metering accuracy have only been qualitatively investigated in a few studies. The present paper quantitatively describes the gas distribution asymmetry in several CFMs under different flow conditions by numerical simulation. The simulation methodology is developed and validated by a results comparison with a conducted experiment and published data for bubbly, stratified and transitional flow regimes. U-shaped and triangle-shaped CFMs of different diameters are investigated at different gas volume fractions and flow rates. It is shown that the increase in the gas volume fraction and the reduction in the mixture flow rate lead to the increase in the gas distribution asymmetry. The strong correlation between the gas distribution asymmetry and the experimentally observed CFM error is demonstrated. The correction of the CFM error is proposed based on this correlation allowing the metering error to be decreased from 34% to 10% for the investigated conditions.

Artificial Neural Network Model to Predict Production Rate of Electrical Submersible Pump Wells

Summary Production data are essential for designing and operating electrical submersible pump (ESP) systems. This study aims to develop artificial neural network (ANN) models to predict flow rates of ESP artificially lifted wells. The ANN models were developed using 31,652 data points randomly split into 80% (25,744 data points) for training and 20% (5,625 data points) for testing. Each data set included measurements for wellhead parameters, fluid properties, ESP downhole sensor measurements, and variable speed drive (VSD) sensors parameters. The models consisted of four separate neural networks to predict oil, water, gas, and liquid flow rates. Sensitivity analyses were performed to determine the optimum number of input parameters that can be used in the model. The best performance was achieved with ANN models of 16 input parameters that are readily available in ESP wells. The results of the best ANN configuration indicate that the mean absolute percent error (MAPE) between the predicted flow rates and the actual measurements for the testing data points of the oil, water, gas, and liquid networks is 3.7, 5.2, 6.4, and 4.1%, respectively. In addition, the correlation coefficients (R2) are 0.991, 0.992, 0.983, and 0.979 for the estimated oil, water, gas, and liquid flow rates for the testing data points, respectively. The performance of the ANN models was compared against performance of published physics-based models and the results were comparable. Unlike the physics-based methods, the ANN models have the advantage that they do not require periodic calibration. The ANN models were used to predict the flow rate curves of an oilwell in the Western Desert of Egypt. The results were compared to the actual separator test data. It was clear that the model results matched the actual test data. The ANN model is useful for predicting individual well production rates within wide variety of pumping conditions and completion configurations. This should allow for continuous monitoring, optimization, and performance analysis of ESP wells as well as quicker response to operational issues. In comparison to traditional separators and multiphase flowmeters (MPFMs), the use of the developed ANN models is simple, quick, and inexpensive.

Microwave probe sensing location for Venturi-based real-time multiphase flowmeter

Real-time in-line interpretation of liquid properties is important for multiphase flow measurements; for example, many phase fraction measurement principles have a dependency on characteristics of salinity. Therefore, it would be desirable to have a sensor, such as a microwave sensor, which can continuously measure the salinity of a flow. In addition to salinity measurement, the microwave sensor can also measure water fraction, which is required for a multiphase flowmeter based on single-energy gamma-ray attenuation; however, choosing a suitable probe sensing location for a microwave salinity sensor in a multiphase flowmeter can be challenging, as the sensor needs to be located at near-wall liquid-rich region to accommodate a wide range of flow conditions. Currently, a microwave sensor is installed in the lower area of a horizontal blind-tee inlet spool of a multiphase flowmeter for salinity measurement. Integrating the microwave sensor into the vertically mounted multiphase flowmeter can reduce the flowmeter (carbon) footprint and manufacturing costs and can improve water-to-liquid ratio measurement due to faster local oil-water mixing. The associated challenge is that the sensor needs to be located at near-wall liquid-rich region to accommodate a wide range of flow conditions, including high gas-volume-fraction flows, where the near-wall liquid layer present in the vertical pipe is usually very thin. In this study, computational fluid dynamics modeling is used to evaluate the suitability of four different sensing locations along the vertical cross-section of a Venturi-based multiphase flowmeter based on near-wall liquid-richness. The results show that the Venturi inlet is the most suitable location for microwave sensor measurement, compared to the mid-convergence section, the mid-divergence section, and the Venturi outlet for a range of inlet liquid-volume-fractions. The findings have been validated by experimental microwave sensor measurements in a multiphase flow loop facility.

Gas Condensate Wells: Challenges of Sampling, Testing and Production Optimization

The main problem of fluid sampling during well testing of reservoirs with near-critical fluids (gas condensate and volatile oil) is due to the fact that even a small pressure drawdown usually leads to the formation of a two-phase mixture in the bottom hole area, and it is almost impossible to take representative samples with downhole samplers or a formation tester. Sampling via test-separator and the current non-separation methods are also imperfect. An alternative method—MIKS (Multiphase IsoKinetic Sampling)—of gas condensate well testing was proposed, which is based on emulsifying a multiphase flow to particles of about 1–10 μm. Thereby MIKS would eliminate the problem of particle slippage in a homogeneous flow and enables high-quality sampling directly from the flowmeter line. The initial formation fluid is characterized by the maximum value of the condensate-gas ratio (CGR). Therefore, first, the well effluent would be adjusted to the mode with the maximum CGR using a choke manifold and a multiphase flow meter. Then the flow mixture is transferred to a by-pass line with an emulsifier to achieve an isokinetic flow. Thereafter, pressure samples can be taken into pressurized sampling bottles, in which thermodynamic conditions are preset according to the flow line. The efficiency of sampling and recombining procedures allows for conducting a study of reservoir samples in the field laboratory directly on the rig and obtaining a complete PVT report even before the completion of drilling and abandonment of the well. An additional economic effect is achieved by reducing the costs of transporting and samples storage. Well test equipment setup becomes much more compact and less weight; the costs of drilling time are reduced, which is viably important for well testing on the Arctic conditions. Another major problem in the development of gas condensate reservoirs is avoiding the condensate banking around producing wells. Optimization of condensate production can be achieved by maintaining the well operation mode at maximum CGR level by means of multiphase flowmeters. The formed condensate bank can be destroyed by a combination of methods—hydraulic fracturing, followed by cycling process—purging the formation with dried gas and/or injection of methanol into the formation. Methanol can be obtained from synthesis gas as a by-product in the utilization of associated gas also at the field. The specified set of measures will allow to revive the GC wells that are losing productivity, as well as to extend the period of high productivity of new wells.

Development of the concept of a multiphase flow meter

In the oil and gas industry, several problems can be identified that affect the subsequent rates of development of the relevant industry: low quality of oil products and low rates of application of new technologies, which directly slows down the country's economic component. Currently, the development of multiphase flow meters is underway, however, the creation of a universal device is still an unsolved problem, and the existing developments require calibrations. The main problem in calculating the flow rate of the fluid is the determination of the density of the components of the passing mixture. The use of X-ray radiation allows you to solve most of the technical problems, as well as to determine the density with a high accuracy. Flow meters using this concept are of little use. The purpose of the study of our work is to update the applicability of the use of the X-ray range in determining the density of the passing flow. In the course of the work, a search and analysis of tabular data characterizing the interaction of radiation with matter was carried out, according to the results of which acceptable sources of characteristic radiation were determined. In the next step, we modeled the structure of the fluid and derived a system of equations that allows one to determine the densities using existing instruments, as well as constants that are determined experimentally. The result of the research is a solvable system of equations, as well as primary modeling and determination of the component, quantitative composition of the gas-liquid mixture under consideration in laminar flow. The result of this work is to obtain a theoretical confirmation of the relevance of using the device based on this concept, the subsequent addition of characteristic radiation detectors will also help to determine the chemical composition of the fluid.

A new correlation for calculating wellhead oil flow rate using artificial neural network

A separator and multiphase flow meters are considered the most accurate tools used to measure the surface oil flow rates. However, these tools are expensive and time consuming. Thus, this study aims to develop a correlation for accurate and quick evaluation of well surface flow rates and consequently the well inflow performance relationship. In order to achieve the abovementioned aim, this study uses artificial neural network (ANN) for flow rates prediction particularly in artificial lifted wells especially in the absence of wellhead pressure data. The ANN model is developed and validated by utilizing 350 data points collected from numerous fields located in Nile Delta and Western Desert of Egypt with inputs include; wellhead temperature, gas liquid ratio, water cut, surface and bottomhole temperatures, water cut, surface production rates, tubing cross section area, and well depth. The results of this study show that, the collected data are distributed as follows; 60% for training, 30% for testing and 10% for the validation processes with R2 of 0.96 and mean square error (MSE) of 0.02. A comparison study is implemented between the new ANN correlation and other published correlations (Gilbert, Ros and Achong correlations) to show the robustness of the developed correlation. The results show that the developed correlation able to predict oil flow rates accurately with the lowest mean square error.

Workflow to predict wellhead choke performance during multiphase flow using machine learning

Multiphase flow metering in oil and gas production wells is essential for monitoring the production performance from oil and gas reservoirs. Accurate measurements of multiphase flow rate that passes through the chock at the wellhead assembly is highly important as it provides a crucial information to estimate production to asses and optimize the reservoir performance. The current practice is to conduct production rate tests monthly for the entire wells that are connected through the same manifold. Another way to estimate the flow rate at the surface is using any of the several correlations to calculate the flow rate using production and wellhead chock data. The high uncertainty in production rate predictions is very well expected and the main source of this uncertainty is the reliance on sporadic well test data and empirical multiphase flow correlations. The objective of this study is to develop a machine learning model that outperforms the industry's widely used correlations in predicting the flow rate of critical and subcritical multiphase flow through the wellhead choke. This objective is achieved by developing a detailed workflow to generate a choke performance prediction model. Validating the accuracy and reliability of the developed model is done using actual data from more than 4000 production rate tests from different fields to investigate the model's accuracy. Testing the developed models showed that it provides a higher accuracy than the several correlations it was compared against in critical and subcritical flow. The average accuracy (correlation coefficient) of prediction was around 0.92 for critical and subcritical flow models. The developed models using machine learning provided a reliable method to predict flow rate through the wellhead chock by utilizing readily available surface data.

Multiphase Flowmeter Health Monitoring Strategy: Maximizing the Value of Real-Time Sensors and Automation for Industrial Revolution 4.0

Summary Metering with multiphase flowmeters (MPFMs) is a well-proven intelligent field technology, which utilizes complex flow models to accurately measure multiphase flow (oil, gas, and water) on a real-time basis. The major realized benefits of utilizing MPFM real-time sensor data technology are reduced maintenance costs, optimized maintenance schedules, and improved safety via minimizing visits and predicting well performance. MPFMs also have the capability to perform comprehensive real-time test validation in a short period of time, making them a valuable tool. Leveraging their capabilities, engineers can perform various production analysis scenarios remotely using continuous data streaming to reduce and ultimately eliminate reliance on operators and vendors in remote fields. Nevertheless, such MPFMs require frequent maintenance, which could be quite difficult to manage when having a large number of meters. An in-house application was developed to support those objectives and to remotely monitor the health of all MPFMs on a real-time basis. The in-house analytics system comprises four modules: Field Overview, MPFM Overview, Well Test Management, and Alarm Management. The steps used to avail this kind of solution cover infrastructure requirements, visualization requirements, optimization, and prediction. The solution was designed as the major interface for engineers to proactively detect and prevent major MPFM component failures and highlight all types of projected MPFM component failures. It proactively harvests preset alarms, which leads to preventing internal MPFM parts from failing. These alarms are generated based on critical parameters measured by the MPFM such as rates, differential pressure (DP), and radioactive measurements. Annual savings are realized through use of the monitoring solution by minimizing costly offshore visits and optimizing the visits based on problematic MPFMs. The authors will discuss a newly developed MPFM health monitoring strategy based on real-time solutions, driving migration from a preventative to a predictive maintenance concept to reduce costs and optimize manpower time by eliminating unnecessary visits.

Virtual flow predictor using deep neural networks

Multiphase flowmeters are important to monitor oil wells, as they allow operators to obtain a real-time estimate of the production. However, due to its high installation cost, uncertainty, and reading errors, the oil and gas industry started to invest in virtual meters, mathematical models that can obtain the oil and gas flow forecast using the instrumentation already available in the well, such as pressure and temperature sensors. The purpose of this article is to develop a virtual multiphase flow predictor using artificial intelligence models such as Long Short-Term Memory (LSTM), Gate Recurrent Units (GRU), Multi-Layer Perceptron (MLP) neural networks and deep learning. As input data, this paper uses a public database available on the Oil and Gas Authority (UK) to train the models and demonstrate the possibility of predicting multiphase flow with a reasonable error margin, proving that the method is efficient in estimating well production rates.

Intelligent Production Monitoring with Continuous Deep Learning Models

Summary Monitoring of production rates is essential for reservoir management, history matching, and production optimization. Traditionally, such information is provided by multiphase flowmeters or test separators. The growth of the availability of data, combined with the rapid development of computational resources, enabled the inception of digital techniques, which estimate oil, gas, and water rates indirectly. This paper discusses the application of continuous deep learning models, capable of reproducing multiphase flow dynamics for production monitoring purposes. This technique combines the time evolution properties of a dynamical system and the ability of neural networks to quantitively describe poorly understood multiphase phenomena and can be considered as a hybrid solution between data-driven and mechanistic approaches. The continuous latent ordinary differential equation (Latent ODE) approach is compared with other known machine learning methods, such as linear regression, ensemble-based model, and recurrent neural network (RNN). In addition, the framework of virtual flowmetering (VFM) is analyzed from the point of multiphase measurement technology. In this work, the application of Latent ODEs for the problem of multiphase flow rate estimation is introduced. The considered example refers to a scenario in which the topside oil, gas, and water flow rates are estimated using the data from several downhole pressure sensors. The predictive capabilities of different types of machine learning and deep learning instruments are explored using simulated production data from a multiphase flow simulator. The results demonstrate the satisfactory performance of the continuous deep learning models in comparison with other machine learning methods in terms of accuracy, where the normalized root-mean-squared error (RMSE) and mean absolute error (MAE) of prediction below 5% were achieved. While Latent ODE demonstrates the significant time required to train the model, it outperforms other methods for irregularly sampled time series, which makes it especially attractive to forecast values of multiphase rates.

Flow Regime detection using gamma-ray-based multiphase flowmeter: a machine learning approach

Flow Regime detection using gamma-ray-based multiphase flowmeter: a machine learning approach

Virtual Multiphase Flow Meter using combination of Ensemble Learning and first principle physics based

Abstract This paper describes a Virtual Flow Meter (VFM) to estimate oil, gas and water flow rate by combining two distinct approaches i.e., data-driven Ensemble Learning algorithm and first principle physics-based transient multiphase flow simulator. The VFM uses a common real-time sensor readings and the estimated flow rates were then combined using a new combiner approach which provides confidence decay and historical performance factors to assign confidence and contribution weights to the base estimators, and then aggregates their estimates to deliver more accurate flow rate estimates. This technique was tested for over 6 months at an offshore oil facility having two oil wells. The technique successfully delivered a 50% improvement in measurement performance compared to stand-alone VFMs. This combiner technique will be of great benefit to surveillance engineers by providing additional real-time production monitoring in addition to acting as a verification tool for physical multiphase flow meters (MPFMs).

МОДЕЛИРОВАНИЕ ОБНАРУЖЕНИЯ ГАММАИЗЛУЧЕНИЯ В КОНТЕКСТЕ БЕСКОНТАКТНОГО ИЗМЕРЕНИЯ МНОГОФАЗНОГО ПОТОКА В НЕФТЯНЫХ СКВАЖИНАХ

The interaction of gamma rays with matter is the crucial factor in the detection of these rays. Scintillation detectors respond to gamma rays by producing a small flash of light, or scintillation. The detector crystal is mounted on a photomultiplier tube that converts the scintillation into an electrical pulse which is processed to generate a signal. A class of multiphase flowmeters is based on this principle to measure the different fractions contained in oil well flow. In this paper, different modes of gammaray interactions were described, the statistical distribution for gammaray generation and detection was presented, and a simulation of detection of gammaray passing through a tube containing oil and gas was performed. The simulation led to the generation of data which was analysed and a conclusion was reached that the detected gamma rays have a different statistical distribution than the generated rays due to the gas bubbles' presence and absence in the flow.

Monitoring the composition of gas condensate well streams based on samples taken from multiphase flow meters

The article highlights the relevance of reliable estimation of the composition and properties of reservoir gas during the development of gas condensate fields and the complexity of the task for reservoirs containing zones of varying condensate content. The authors have developed a methodology that allows monitoring the composition of gas condensate well streams of similar reservoirs. There are successful examples of the approach applied in Achimov gas condensate reservoirs at the Urengoy oil and gas condensate field. The proposed approach is based on the use of the so-called fluid factors, which are calculated on the basis of the known component compositions of various flows of the studied hydrocarbon system. The correlation between certain "fluid factors" and the properties of reservoir gas (usually determined by more labor-consuming methods) allows one to quickly obtain important information necessary to solve various development control tasks.

Application of Wavelet Feature Extraction and Artificial Neural Networks for Improving the Performance of Gas-Liquid Two-Phase Flow Meters Used in Oil and Petrochemical Industries.

Measuring fluid characteristics is of high importance in various industries such as the polymer, petroleum, and petrochemical industries, etc. Flow regime classification and void fraction measurement are essential for predicting the performance of many systems. The efficiency of multiphase flow meters strongly depends on the flow parameters. In this study, MCNP (Monte Carlo N-Particle) code was employed to simulate annular, stratified, and homogeneous regimes. In this approach, two detectors (NaI) were utilized to detect the emitted photons from a cesium-137 source. The registered signals of both detectors were decomposed using a discrete wavelet transform (DWT). Following this, the low-frequency (approximation) and high-frequency (detail) components of the signals were calculated. Finally, various features of the approximation signals were extracted, using the average value, kurtosis, standard deviation (STD), and root mean square (RMS). The extracted features were thoroughly analyzed to find those features which could classify the flow regimes and be utilized as the inputs to a network for improving the efficiency of flow meters. Two different networks were implemented for flow regime classification and void fraction prediction. In the current study, using the wavelet transform and feature extraction approach, the considered flow regimes were classified correctly, and the void fraction percentages were calculated with a mean relative error (MRE) of 0.4%. Although the system presented in this study is proposed for measuring the characteristics of petroleum fluids, it can be easily used for other types of fluids such as polymeric fluids.

Machine Learning Applications to Predict Surface Oil Rates for High Gas Oil Ratio Reservoirs

AbstractWell-performance investigation highly depends on the accurate estimation of its oil and gas flowrates. Testing separators and multiphase flowmeters (MPFMs) are associated with many technical and operational issues. Therefore, this study aims to implement the support vector machine (SVM), and random forests (RF) as machine learning (ML) methods to estimate the well production rate based on chokes parameters for high GOR reservoirs. Dataset of 1131 data points includes GOR, upstream and downstream pressures (PU and PD), choke size (D64), and actual data of oil and gas production rates. The data have GOR was up to 9265 SCF/STB, the oil rate varied from 1156 and 7982 BPD. SVM and RF models were built to estimate the production rates. The ML models were trained using seventy percent of the dataset, while the models were tested and validated using 30% of the dataset. The dataset was classified to 622 wells that were flowing at critical flow compared with 509 wells that were flowing at subcritical conditions based on a PD/PU ratio of 0.55. Four machine learning models were developed using SVM and RF for subcritical flow and critical flow conditions. Different performance indicators were applied to assess the developed models. SVM and RF models revealed average absolute percentage error (AAPE) of 1.3 and 0.7%, respectively, in the case of subcritical flow conditions. For critical flow conditions, the AAPE was found to be 1.7% in the SVM model, and 0.8% in the RF model. The developed models showed a coefficient of determination (R2) higher than 0.93. All developed ML models perform better than empirical correlations. These results confirm the capabilities to predict the oil rates from the choke parameters in real-time without the requirement of instrument installation of wellsite intervention.

Uncertainty Estimation for Machine Learning Models in Multiphase Flow Applications

In oil and gas production, it is essential to monitor some performance indicators that are related to the composition of the extracted mixture, such as the liquid and gas content of the flow. These indicators cannot be directly measured and must be inferred with other measurements by using soft sensor approaches that model the target quantity. For the purpose of production monitoring, point estimation alone is not enough, and a confidence interval is required in order to assess the uncertainty in the provided measure. Decisions based on these estimations can have a large impact on production costs; therefore, providing a quantification of uncertainty can help operators make the most correct choices. This paper focuses on the estimation of the performance indicator called the water-in-liquid ratio by using data-driven tools: firstly, anomaly detection techniques are employed to find data that can alter the performance of the subsequent model; then, different machine learning models, such as Gaussian processes, random forests, linear local forests, and neural networks, are tested and employed to perform uncertainty-aware predictions on data coming from an industrial tool, the multiphase flow meter, which collects multiple signals from the flow mixture. The reported results show the differences between the discussed approaches and the advantages of the uncertainty estimation; in particular, they show that methods such as the Gaussian process and linear local forest are capable of reaching competitive performance in terms of both RMSE (1.9–2.1) and estimated uncertainty (1.6–2.6).

Technology Focus: Subsea Systems (August 2021)

Flow assurance in subsea oil and gas fields often presents significant challenges. Every field has its own combination of difficulties, and no universal process or system can be used to mitigate these. Detailed knowledge across a broad range of competencies, therefore, is required to find solutions that can minimize the risk of not getting the hydrocarbons safely to the process facilities. Many subsea fields that are being developed today are long tiebacks, taking advantage of existing offshore infrastructure or producing directly to shore. These developments must deal with the long-distance transport of hydrocarbons in deep cold water, commonly increasing the risk of hydrate formation and wax deposition, for example. In addition, large elevation changes from deep water to surface and topographical challenges along the pipeline can create flow-regime effects that can hinder production. The loss of temperature in a long subsea pipeline also creates challenges for fields that produce heavy oil because the oil viscosity in some cases increases dramatically at low temperatures, in addition to effective viscosities increasing because of oil and water emulsions. Other phenomena such as scale deposition, foaming, sand production, erosion, and corrosion must be considered and dealt with as well. Various smart-technology innovations for subsea oil and gas production contribute to reducing the risk of these flow-assurance issues. Some of them are described in this month’s selected SPE papers. A good example is as follows: When wells start to produce water, the operator needs to understand where the water is coming from and quantify volumes in order to start a mitigation program to avoid hydrate formation. This is one of the reasons why subsea multiphase flowmeters have become an essential feature in all new subsea fields. The most common remedy for flow-assurance problems is probably the use of chemical additives. A sensor technology that can directly determine the ratio between produced water and chemicals such as monoethylene glycol has been recently introduced in subsea production systems. This measurement enables the optimization of chemical-injection rates, thereby contributing to significant savings in capital expenditure (reduced design margins) and operational expenditure (reduced overdosage margins). Another effective way to prevent hydrates and wax is to keep the process temperature above critical limits by applying active flowline heating. New technologies for highly reliable and efficient subsea electrically heat-traced flowlines have recently been qualified, industrialized, and installed. Technologies as described here can play an important role in future subsea field developments. The recommended readings for this feature date back further back in time than usual, but are relevant to the theme of this year’s main selections. Recommended additional reading at OnePetro: www.onepetro.org. OTC 29232 Real-Time Subsea Hydrate Management in the World’s Longest Subsea Tieback by Christophe Vielliard, OneSubsea, a Schlumberger Company, et al. OTC 31078 Electrically Heated Trace Flowline on the Ærfugl Project—A Journey From Product Qualification to Offshore Campaign by Guy Mencarelli, Subsea 7, et al. SPE 195784 A New Flow-Assurance Strategy for the Vega Asset: Managing Hydrate and Integrity Risks on a Long Multiphase Flowline of a Norwegian Subsea Asset by Stephan Hatscher, Wintershall Norge, et al.

Applications of Artificial Intelligence to Predict Oil Rate for High Gas-Oil Ratio and Water-Cut Wells.

Measuring oil production rates of individual wells is important to evaluate a well’s performance. Multiphase flow meters (MPFMs) and test separators have been used to estimate well production rates. Due to economic and technical issues with MPFMs, especially for high gas–oil ratio (GOR) reservoirs, the use of a choke formula for estimating well production rate is still popular. The objective of this study is to implement different artificial intelligence (AI) techniques to predict the oil rate through wellhead chokes. Support-vector machine (SVM) and random forests (RF) were used to generate different models to predict the production rates for high GOR and WC wells. A set of data (548 wells) was obtained from oil fields in the Middle East. GOR varied from 1000 to 9351 scf/stb, and WC ranged from 1 to 60%. Around 300 wells were flowing under critical flow conditions, while the rest were subcritical. Hence, two cases were studied using each AI model. Seventy percent of the data was used to train both RF and SVM models, while 30% of the data was used to test and validate these models. The developed RF and SVM models were then compared against the previous empirical formulas. The RF model in both critical and subcritical flow conditions was able to perfectly match the actual oil rates. SVM was able to predict the general trend for the oil rates but missed some of the sharp changes in the oil rate trend. The average absolute percent error (AAPE) values in the subcritical flow for SVM and RF were 1.7 and 0.7%, respectively, while in the critical flow, the AAPE values were 1.4 and 0.75% for SVM and RF models, respectively. SVM and RF models outperform the published formulas by 34%. The results from this study will help to estimate the real-time oil and gas rates based on the available data from wellhead chokes without the need for field intervention.