Virtual Flow Meter Research Articles

Использование алгоритма «виртуального расходомера» при выводе нефтяных скважин на режим

Использование алгоритма «виртуального расходомера» при выводе нефтяных скважин на режим

Developing a Hybrid Data-Driven, Mechanistic Virtual Flow Meter - a Case Study

Virtual flow meters, mathematical models predicting production flow rates in petroleum assets, are useful aids in production monitoring and optimization. Mechanistic models based on first-principles are most common, however, data-driven models exploiting patterns in measurements are gaining popularity. This research investigates a hybrid modeling approach, utilizing techniques from both the aforementioned areas of expertise, to model a well production choke. The choke is represented with a simplified set of first-principle equations and a neural network to estimate the valve flow coefficient. Historical production data from the petroleum platform Edvard Grieg is used for model validation. Additionally, a mechanistic and a data-driven model are constructed for comparison of performance. A practical framework for development of models with varying degree of hybridity and stochastic optimization of its parameters is established. Results of the hybrid model performance are promising albeit with considerable room for improvements.

VERTUAL METERING AS A TOOL FOR ESP-EQUIPPED WELLS MONITORING

This paper describes an approach to creating a virtual flow metering system (VFM). Proposed method is based on electric submersible pump system’s monitoring. Whole VFM market was analyzed with taking method’s relevance, possible and real industry applications into consideration. Created algorithm uses data that is regularly collected from oilfields and stored in corporate’s databases. Mainly used parameters are well’s construction, telemetry data from ESP system, pressure and temperature from different parts of the well, and available oil production data, divided by phases, that is obtained rarely and used for calibration. Suggested solution considers electric part of an ESP system and so it reveals possibility to estimate ESP degradation dynamics for liquid-gas mixture production restoring by using correction coefficients. Presented algorithm was tested on the Company’s oilfield. One well is shown as an example of method’s work, including all steps from data preprocessing to liquid production dynamics recovering. Also results of algorithm usage on several dozen wells on various oilfields are presented. As one of the main results approving the possibility to increase liquid production measurements frequency should be mentioned. Such possibility is based on using indirect operating parameters and recovering dynamics of model’s inner coefficients (e.g., ESP degradation), that make possible to estimate effectiveness of pump exploitation and possible optimization necessity. As another important result implementing of the algorithm into Company’s IT systems should be mentioned. Such implementation reveals possibility for monitoring a big amount of wells in fast and effective way.

Multi-Phase Flow Metering in Offshore Oil and Gas Transportation Pipelines: Trends and Perspectives.

Multi-phase flow meters are of huge importance to the offshore oil and gas industry. Unreliable measurements can lead to many disadvantages and even wrong decision-making. It is especially important for mature reservoirs as the gas volume fraction and water cut is increasing during the lifetime of a well. Hence, it is essential to accurately monitor the multi-phase flow of oil, water and gas inside the transportation pipelines. The objective of this review paper is to present the current trends and technologies within multi-phase flow measurements and to introduce the most promising methods based on parameters such as accuracy, footprint, safety, maintenance and calibration. Typical meters, such as tomography, gamma densitometry and virtual flow meters are described and compared based on their performance with respect to multi-phase flow measurements. Both experimental prototypes and commercial solutions are presented and evaluated. For a non-intrusive, non-invasive and inexpensive meter solution, this review paper predicts a progress for virtual flow meters in the near future. The application of multi-phase flows meters are expected to further expand in the future as fields are maturing, thus, efficient utilization of existing fields are in focus, to decide if a field is still financially profitable.

Uncertainty analysis for different virtual pump water flow meters

The water flow rate through a pump can be calculated using other measured inputs, such as pump head, motor input power, and frequency, along with calibrated characteristic curves. Currently virtual pump flowmeters can be classified into three categories—pump head based, shaft power based, and pump efficiency based—based on which measured inputs and characteristic curves are applied. Thus, different virtual meters may have different flow measurement errors propagated from different measured inputs and curve fittings. This article aims to evaluate the measurement uncertainties of these three virtual meters. First, mathematical models of virtual flow measurement and uncertainty analysis are introduced for three virtual meters. Then through experiments on a 125 m3/h pump, three virtual meters are developed and the uncertainties of the developed virtual meters are evaluated. Finally, the water flow rates measured by three virtual meters are compared with those measured by an ultrasonic flowmeter. Both the uncertainty analysis and flow measurement comparison show that the pump head–based and shaft power–based virtual meters have unacceptable uncertainties, 13 m3/h and 17 m3/h, respectively, at lower frequencies due to the mismatch between the pump speed and motor input frequency. In contrast, the pump efficiency–based virtual meter always remains a robust accuracy with uncertainties less than 1.9 m3/h.

Accuracy improvement of virtual pump water flow meters using calibrated characteristics curves at various frequencies

The water flow rate of a pump can be calculated using other measurable inputs along with calibrated pump characteristic curves. Head-based and power-based virtual pump water flow meters apply the affinity laws to convert the calibrated pump head and power curves at the rated frequency to the curves at measured frequency and then calculate the water flow rate using measured head and power respectively. Due to the mismatched correlation between the motor speed and input frequency, these two flow meters cannot achieve acceptable accuracy. The purpose of this paper is to improve the accuracy of these two flow meters using the calibrated head and power curves at multiple frequencies rather than at the rated frequency only. First the mathematical models of flow measurements and uncertainty analysis are introduced, then two experiments are conducted, one for the virtual meter calibration and uncertainty analysis while another for the accuracy improvement validation. The results show that the uncertainty of water flow measurements decreases from 13 m3/h to 5 m3/h for the head-based flow meter and 17 m3/h to 6 m3/h for the power-based flow meter if the calibrated pump characteristic curves at multiple frequencies replace the curves at the rated frequency.

Machine Learning Improves Accuracy of Virtual Flowmetering and Back-Allocation

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 192819, “Improving the Accuracy of Virtual Flowmetering and Back-Allocation Through Machine Learning,” by Pejman Shoeibi Omrani, SPE, Iulian Dobrovolschi, and Stefan Belfroid, SPE, TNO, and Peter Kronberger and Esteban Munoz, Wintershall Noordzee, prepared for the 2018 Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, 12–15 November. The paper has not been peer reviewed. In this study, the authors investigated a fully data-driven approach using artificial neural networks (ANNs) for real-time virtual flowmetering and back-allocation in production wells. The main goal was to develop computationally efficient data-driven models to determine multiphase production rates of individual phases (gas and liquid) in wells using existing measured data in fields. The results showed that ANNs were capable of estimating multiphase flow rates accurately in both simulated and field data. Introduction Virtual-flowmetering (VFM) methods are categorized in two types generally: physics-based models (hydrodynamical approach) and data-driven models. In the physics-based approach, multiphase flow rates are estimated by simplified hydrodynamical models using sensor data (e.g., pressure and temperature) as input parameters. The second approach is solely dependent on the available data in the field, performing statistical analysis on this data and deriving relations between input features and quantities of interest (in this case, multiphase flow rates). Such an approach requires a sufficient data set to train the models. In the context of data-driven VFM, such a data set could be obtained from periodic test-separator data or well tests. In this study, two types of ANNs were assessed: feed-forward (multilayer-perceptron) and recurrent [long short-term memory (LSTM)] to capture temporal dependencies. ANNs were used for real-time gas flowmetering at an individual well level using total production rates measured downstream of the combined production separators. Methodology The feed-forward ANN is a suitable method for modeling the relations between a set of features and output parameters by functional mapping of the features in the input layer to the outputs. Fig. 1 shows a schematic of a feed-forward ANN. The relation of the input to the output parameters is found by calibrating the weights and biases in the hidden (middle) layers. Depending on the complexity of the input/output relations, the number of hidden layers, and neurons therein, could be varied. The disadvantage of feed-forward networks is their inability to identify temporal trends in the data set. Recurrent neural networks (RNNs) are suggested for prediction of systems in which states are changing continuously before reaching an equilibrium, or when the time-dependency of states is crucial for predicting these states. A common disadvantage of some RNNs is their inability to capture long-term dependencies in the data. Use of an LSTM network is suggested for predictions involving systems in which both short- and long-term dependencies are present. An LSTM network is composed of sever-al units of RNN, with each unit composed of a cell with an input, output, and forget gate.

Oil Production Monitoring using Gradient Boosting Machine Learning Algorithm

Abstract Data-driven solutions for multiphase flowrate estimation in oil and gas production systems are among the alternatives to first principles virtual flow metering systems and hardware flow metering installations. Some of the most popular data-driven methods in this area are based on artificial neural networks which have been proven to be good virtual flow metering tools. However, neural networks are known to be sensitive to the scaling of input data, difficult to tune and provide a black-box solution with occasionally unexplainable behavior under certain conditions. As an alternative, in this paper, we explore capabilities of the Gradient Boosting algorithm in predicting oil flowrates using available field measurements. To do this, we use an efficient implementation of the algorithm named XGBoost. In contrast to neural networks, this algorithm is insensible to data scaling, can be more intuitive in tuning as well as it provides an opportunity to analyze feature influence which is embedded in algorithm learning. We show that the algorithm provides accurate flowrate predictions under various conditions and can be used as a back-up as well as a standalone multiphase flow metering solution.

Virtual outdoor air flow meter for an existing HVAC system in heating mode

Abstract Measurements from the operation of Heating, Ventilation and Air conditioning (HVAC) system are collected by Building Automation System (BAS) mostly for control purpose. In many air handling units (AHUs) the airflow meter is not installed on the outdoor air stream to reduce the initial costs. To provide such a missing information needed for the ongoing commissioning of HVAC system, a virtual flow meter (VFM) should be implemented in the BAS. This paper presents VFMs for the heating season when the heat recovery coils are used. Two models are presented which predict the outdoor air ratio (factor α), in the absence of such an air flow meter, by using BAS trend data. Two energy balance equations, one for the air mixing box and another for the heat recovery coils, are coupled. The results from a case study building show that the two models predict the daily average outdoor air ratio of 0.89 and 0.87, respectively, compared with the reference daily average ratio of 0.86.

Integrated Surveillance Offshore Turkmenistan

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 188357, “Integrated Surveillance in Cheleken Block Offshore Turkmenistan by Use of Automated Steady-State and Transient Models,” by Roberto Espinoza, Dragon Oil, and Suraj Mohan Uniyal and Ivan Marcelo Jaramillo Rivadeneira, Schlumberger, prepared for the 2017 Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, 13–16 November. The paper has not been peer reviewed. The complete paper describes a work flow in which wells and production networks (15 platforms and 100 active wells located in the Cheleken Block offshore Turkmenistan) are automatically modeled daily with steady-state and transient tools and ultimately analyzed by a Web-based surveillance system, the Cheleken Block Central Data Gathering System (CDGS). These models are updated and run automatically on a daily basis by means of the CDGS. Introduction Well and network modeling are being used widely across the industry as a method of problem detection, production enhancement, and production optimization. However, well and network modeling require a large amount of data that needs to be systematically collected, cleaned, validated, and updated. They also require considerable manpower, because many time models are not calibrated adequately to ensure reliable results. An operator has implemented a robust system, the CDGS, that allows data capture, validation, and approval in a systemic manner. There was a need to capitalize on this ability and integrate it with steady-state and transient models to increase the efficiency of the overall process. Philosophy Applied The initial status of the project presented several critical challenges: Well models were executing sporadically. The network model did not consider the frequent well and network flow routing. Optimization of choke or flow routing was not possible with the current configuration. Visualization of day-to-day flow routing and other production key performance indicators of the field in a centralized system was required in order for the operators, supervisors, and reservoir teams to reduce their nonproductive time, solve day-to-day operational problems, and optimize field production. The project was designed and implemented on the basis of four fundamental processes. A production-data-management system (PDMS) was implemented to systematically acquire all well-operation data and process information generated during day-to-day activity from producing assets. These data needed to be validated, stored, and made available in a structured manner to different user groups. Various core-analytics processes such as back allocation, virtual flowmetering, volume correction, and software-platform integration were performed to automate data flow. A single, integrated platform for monitoring and surveillance of operational parameters across the entire field was implemented. This integrated surveillance tool provides a single platform that connects operations for a broad range of disciplines. Software from different vendors that performed steady- and transient-state well and network analysis and diagnostics was integrated to the core PDMS. The models can be run on a daily basis, and their results can be stored in the system to be visualized. Day-to-day optimization for choke and flow routing of the production network can be performed with these steady-state and transient models, with results provided to the operators to implement in the field.

A Machine Learning Approach for Virtual Flow Metering and Forecasting

We are concerned with robust and accurate forecasting of multiphase flow rates in wells and pipelines during oil and gas production. In practice, the possibility to physically measure the rates is often limited; besides, it is desirable to estimate future values of multiphase rates based on the previous behavior of the system. In this work, we demonstrate that a Long Short-Term Memory (LSTM) recurrent artificial network is able not only to accurately estimate the multiphase rates at current time (i.e., act as a virtual flow meter), but also to forecast the rates for a sequence of future time instants. For a synthetic severe slugging case, LSTM forecasts compare favorably with the results of hydrodynamical modeling. LSTM results for a synthetic noisy dataset of a variable rate well test show that the model can also successfully forecast multiphase rates for a system with changing flow patterns.

Virtual multiphase flow metering using diverse neural network ensemble and adaptive simulated annealing

Real-time production monitoring in oil and gas industry has become very significant particularly as fields become economically marginal and reservoirs deplete. Virtual flow meters (VFMs) are intelligent systems that infer multiphase flow rates from ancillary measurements and are attractive and cost-effective solutions to meet monitoring demands, reduce operational costs, and improve oil recovery efficiency. Current VFMs are very challenging to develop and very expensive to maintain, most of which were developed for wells with dedicated physical meters where there exists an abundance of well test data. This study proposes a VFM system based on ensemble learning for fields with common metering infrastructure where data generated is very limited. The proposed method generates diverse neural network (NN) learners by manipulating training data, NN architecture and learning trajectory. Adaptive simulated annealing optimization is proposed to select the best subset of learners and the optimal combining strategy. The proposed method was evaluated using actual well test data and managed to achieve a remarkable performance with average errors of 4.7% and 2.4% for liquid and gas flow rates respectively. The accuracy of the developed VFM was also analyzed using cumulative deviation plot where the predictions are within a maximum deviation of ± 15%. Furthermore, the proposed ensemble method was compared to standard bagging and stacking and remarkable improvements have been observed in both accuracy and ensemble size. The proposed VFM is expected to be easier to develop and maintain than model-driven VFMs since only well test samples are required to tune the model. It is hoped that the developed VFM can augment and backup physical meters, improve data reconciliation, and assist in reservoir management and flow assurance ultimately leading to a more efficient oil recovery and less operating and maintenance costs.

Experimental verification of a virtual water flowmeter applicable to air conditioning systems

The main purpose of this research is to suggest a virtual water flowmeter applicable in hydraulic systems with low-cost sensors. Practical procedures for realizing virtual water flowmeters for constant flow pumps (i.e., Cases 1 and 2) and variable flow pumps (i.e., Cases 3 and 4) are proposed in this paper. In Case 1, for measuring the water flow rate through a constant flow pump, the pump head was measured and the water flow rate was estimated using the pump characteristic curve obtained from the manufacturer or by deriving an empirical model based on the field measurement data. In Case 2, the water flow rate through a valve with a constant flow pump was estimated by measuring the pressure head between the valve and the valve opening position. The valve characteristic curve can be obtained by deriving an empirical model based on the initial field test results. The water flow rate through the variable flow pump was calculated, using the pump affinity laws and an empirical model, by measuring the frequency (i.e., Case 3) or input power (i.e., Case 4). The initial setup and test procedures for virtual flowmeters are also proposed in this research for practical use in the field. To verify the proposed procedures, experiments were conducted using constant and variable flow pumps in a pilot system. It was found that the virtual water flowmeters showed 2.9% to 6.7% of root mean square error from Case 1 to Case 4, compared with an ultrasonic water flowmeter, through long-term experimental verification.

Восстановление дебита на основе алгоритмов «виртуального расходомера» для проведения гидродинамических исследований скважин

Восстановление дебита на основе алгоритмов «виртуального расходомера» для проведения гидродинамических исследований скважин

Development of a virtual pump water flow meter with a flow rate function of motor power and pump head

Water flow rates are key operating variables in chilled and hot water systems. The water flow rate through a pump can be virtually measured using available motor power and pump head with projected motor and pump efficiencies. In general, motor efficiency is implicitly determined by motor power while pump efficiency is given as a function of water flow rate. As a result, the water flow rate has to be calculated through a numerical method, which is difficult to apply in building automation systems (BAS). The objective of this paper is to develop a virtual pump water flow meter, which can be implemented in BAS with an explicit expression of motor power and pump head. First, motor efficiency is regressed as a function of motor power by consolidating multiple dependent factors, then pump efficiency function is reconstructed with pump shaft power and head, and finally experiments are conducted to develop and validate a virtual pump water flow meter on a chilled water pump. The experimental results show that the virtual flow measurements agree well with the flow measurement by a physical meter. The measurement standard deviation is 0.5L/s for a pump with the design flow rate of 37.9L/s.

Development of a virtual pump water flow meter using power derived from comprehensive energy loss analysis

Water flow rates are key controlled variables in chilled and hot water systems. The water flow rate through a pump can be virtually determined using a measured pump head and input power. Available input power can be either motor input power obtained from the connected variable-frequency drive or variable-frequency drive input power measured by a power meter. The harmonics created by the variable-frequency drive induce additional energy loss in the motor and consequently degrade the motor efficiency. The objective of this article is to evaluate a virtual pump water flow meter using either the variable-frequency drive input power or motor input power with calibrated pump, motor and variable-frequency drive efficiencies along with the consideration of additional harmonic energy loss. First, the water flow correlation with the measured pump head and available power input is discussed, then the calibration procedure is introduced to identify three efficiencies, and finally the experiment is conducted to develop and validate a virtual pump water flow meter on a pump motor variable-frequency drive system. The experimental results show that the water flow rates determined by the virtual water flow meter agrees well with the results measured by a physical water flow meter with the coefficient of determination of 0.91and 0.97, respectively.

Virtual flow meter for chilled and condenser water for chillers: Estimates versus measurements

This article presents a virtual flow meter that estimates the chilled and condenser water mass flow rates for chillers using trend data commonly found in a building automation system. The quasi-steady-state thermodynamic mathematical model uses six different scenarios of available sensors that could be available in a building automation system. Two case studies of chillers installed in an ice rink and an institutional campus are used to demonstrate the use of the virtual flow meter tool. The estimates of virtual flow meter compared well with measurements, as the coefficient of variance of the root mean squared error and the normal mean bias error, two goodness-of-fit statistics of virtual flow meter models, are between 3.3% and 9.8% and between –6.9% and 7.8%, respectively, depending on the number of sensors available. The article also presents a discussion on the uncertainty of estimates and measurements.

Metrological analysis of a virtual flowmeter-based transducer for cryogenic helium.

The metrological performance of a virtual flowmeter-based transducer for monitoring helium under cryogenic conditions is assessed. At this aim, an uncertainty model of the transducer, mainly based on a valve model, exploiting finite-element approach, and a virtual flowmeter model, based on the Sereg-Schlumberger method, are presented. The models are validated experimentally on a case study for helium monitoring in cryogenic systems at the European Organization for Nuclear Research (CERN). The impact of uncertainty sources on the transducer metrological performance is assessed by a sensitivity analysis, based on statistical experiment design and analysis of variance. In this way, the uncertainty sources most influencing metrological performance of the transducer are singled out over the input range as a whole, at varying operating and setting conditions. This analysis turns out to be important for CERN cryogenics operation because the metrological design of the transducer is validated, and its components and working conditions with critical specifications for future improvements are identified.

Development and Testing of a Virtual Flow Meter Tool to Monitor the Performance of Cooling Plants

This paper introduces a graphical user interface (GUI) for a virtual flow meter (VFM) Tool that estimates the chilled and condenser water mass flow rates through the evaporator and condenser, respectively, of a chiller. The quasi steady-state thermodynamic mathematical model uses six different scenarios of available sensors, which could be available in a building automation system (BAS) to facilitate the use of the model for a large portion of the building stock. The developed model can be used for reciprocating and centrifugal chillers. A case study from an institutional building is used to demonstrate the use of the VFM Tool during the summer months of 2014. The VFM Tool was able to estimate the chilled and condenser water mass flow rates with a coefficient of variance of the root mean squared error (CV(RMSE)) ranging from 4.9% to 13.4% and the normal mean bias error (NMBE) ranging from -12.2% to -1.4%, depending on the number of sensors available.

Use of Trend Data from BEMS for the Ongoing Commissioning of HVAC Systems

The Building Energy Management Systems (BEMS) that are installed in commercial and institutional buildings contain a gold mine of data, which could be exploited for the purpose of ongoing commissioning of heating, ventilation and air-conditioning systems. Two applications are presented in this paper. First the paper discusses the feasibility of a virtual air flow meter to estimate the outdoor air flow rate brought in the air-handling units. Second the paper presents the development of benchmarking models of electric demand of chillers in terms of measurements of the following variables: the chilled water flow rate, the supply air flow rate at the air-handling unit, and the return chilled water temperature.