Liu, Wenyuan (China University of Petroleum – Beijing) | Hu, Jinqiu (China University of Petroleum – Beijing) | Sun, Fengrui (China University of Petroleum – Beijing) | Sun, Zheng (China University of Petroleum – Beijing) | Chu, Hongyang (China University of Petroleum – Beijing) | Li, Xiangfang (China University of Petroleum – Beijing)
Hydrates generation-blockage in submarine natural gas pipelines has always been related to the safety of deepwater natural gas production and transportation. However, the current hydrate formation risk prediction in subsea pipelines is still immature. In this paper, a model for evaluating the risk of hydrate formation in submarine natural gas pipelines has been established. The model has been applied and the sensitivity analysis of typical factors has been carried out. The results show that: (a) owing to the low temperature of the seabed, hydrate formation region (HFR) often exists in submarine pipelines. Avoiding HFR by injecting inhibitors is the key to ensure the safe transmission.
In this study, we aim to demonstrate how machine learning can empower computational models that can predict the flow rate of a given well. Given current real-time data and periodic well tests, this new method computes flow rates using data-driven model. The computational model is based on analyzing the relations and trends in historical data. Relational databases include huge amounts of data that have been accumulated throughout decades. In addition, there is a large number of incoming operational data points every second that gives a lot of insight about the current status, performance, and health of many wells. The project aims to utilize this data to predict the flow rate of a given well.
A variety of well attributes serve as inputs to the computational models that find the current flow rate. Artificial Neural Networks (ANN) were used in order to build these computational models. In addition, a grid search algorithm was used to fine-tune the parameters for the ANN for every single well. Building a single unique model for every well yielded the most accurate results. Wells that are data-rich performed better than wells with insufficient data. To further enhance the accuracy of the models, models are retrained after every incoming patch of real-time data. This retraining calibrates the models to constantly represent the true well performance and predict better. In practice, Flow rate prediction is used by production engineers to analyze the performance of a given well and to accelerate the process of well test verification. One of the main challenges in building unique models for every well is fine-tuning the parameters for the artificial neural networks, which can be a computationally intensive task. Parameter fine-tuning hasn't been discussed in previous literature regarding flow rate prediction. Therefore, our unique approach addresses the individuality of every well and builds models accordingly. This high-level of customization addresses the problem of under-fitting in ANN well models.
Core & Log Neural Network Modeling (CLONNE) has been initiated to utilize an ANN to optimize usage of available data to generate synthetic logs and core data which enable user to eliminate any special logs and core data acquisition in the future. This will reduce the well cost and time required for data acquisition and data analysis.
CLONNE process starts with data gathering of the available core and log data which then QC'ed and conditioned for bad hole, light hydrocarbon, thin lamination and normalized. Then pair of core and log data are combined as dummy well to generate the first CLONNE model that can be used to predict for the whole fields. Conventional data including density, neutron, sonic, GR logs and other parameters are used to generate output. A random well from the field is selected to test the predictability matching of CLONNE versus the real data acquired. Several calibration performed to provide the best predictability.
Currently a number of CLONNE models have been created for offshore fields in Malaysia. For CLONNE Synthetic logs, 4 models have been created to predict Porosity, Bulk Density, Neutron and Shear Slowness. For CLONNE Synthetic core, 3 models have been created to predict Grain Size, Permeability and Porosity. All of this models have managed to predict quite well in both thick sand and laminated sand. More models will come to predict other log curves and core parameters. The models established has been tested in one field, where a synthetic sonic log has been created. After the drilling and subsequent logging run, an actual sonic log has been deployed and compared which yield to 96% comparable. The data predicted from CLONNE can greatly save almost 15 months spend to acquire and analyze core data and also almost RM 6 Million total expenditure to acquire and analyze core data.
In 2018, CLONNE has achieved RM 6 Million cost avoidance from application in 3 fields in Malaysia. The CLONNE model generated can be implement to Basin wide prediction thus enable the sharing use of data. This will help to integrate the data available instead of data being utilize in the specific field only.
The Spraberry trend area is part of a larger oil-producing region within the Midland basin in United States. The main targets, Spraberry, Dean and Wolfcamp, are reservoirs of shales interbedded with clastic formations. Thus, the reservoirs exhibit TIV (transverse isotropy vertical) anisotropy due to thin laminations. A pilot well was drilled vertically in the complex lithology and logged with the advanced acoustics measurements. Shallow penetration of Stoneley energy into the formation raised concerns about the depth resolution of the inverted shear slowness derived from it. It is very difficult to get a reliable horizontal shear slowness from Stoneley when the borehole condition is rugose, there is a complex mud rheology and gas influx inside the borehole.
A machine learning based approach integrating the advanced acoustics measurements and petrophysical interpretation is adopted to provide the solution to get the lithology-based horizontal shear slowness. To eliminate the variability of getting the horizontal shear slowness from Stoneley wave, to process for an advanced geomechanics product like for TIV anisotropy analysis, two machine learning algorithms are used. First one is a very commonly used linear supervised learning algorithm multi-linear regression (MLR) and second is random forest (RF) a nonlinear supervised learning algorithm. These algorithms take inputs from formation evaluation and advanced acoustics to predict the horizontal shear slowness. The random forest algorithm being an ensemble learning method have greater predictive capabilities compared with any linear supervised learning models and many of the non-linear supervised learning algorithms. The inputs for RF and MLR regressions are values of dry weight fractions of calcite, dolomite, quartz, illite, total porosity, permeability, gamma ray, compressional slowness and fast shear slowness. These values are obtained for the entire depth of interest from advance logging tools and interpretation techniques. To check the performance of the model, standard machine learning techniques such as the error evaluation metrics of the mean squared error and the coefficient of determination (
Hyperparameter tuning of the RF model has been done to improve upon the prediction accuracy. After the parameters are tuned, the mean squared error and
Zhang, Jianbo (China University of Petroleum East China) | Wang, Zhiyuan (China University of Petroleum East China) | Duan, Wenguang (CNPC Xibu Drilling Engineering Company Limited and China University of Petroleum East China) | Fu, Weiqi (China University of Petroleum East China) | Tong, Shikun (China University of Petroleum East China) | Sun, Baojiang (China University of Petroleum East China)
Hydrate formation and deposition usually exist during deep-water gas well testing, which easily cause plugging accident in the testing tubing if it was not found and handled in time. A method to estimate and manage hydrate plugging risk in real-time during deep-water gas well testing is developed in this work. This method mainly includes the following steps: predicting hydrate stability region, calculating hydrate behaviors, analyzing the effect of hydrate behaviors on the variation of wellhead pressure, monitoring the variation of wellhead pressure and estimating hydrate plugging risk in real-time, and managing hydrate plugging risk in real-time. As hydrates continue to form and deposit, the effective inner diameter of the tubing decreases, and the wellhead pressure also decreases accordingly. The risk of hydrate plugging can be estimated by monitoring the variation of wellhead pressure. When the wellhead pressure decreases to the critical value for safety at a given gas production rate, it is indicated that hydrate plugging is likely to occur. Under this condition, hydrate inhibitor is needed to inject into the tubing to reduce the severity of hydrate plugging, and real-time monitoring of wellhead pressure variation is also needed to guarantee the risk of hydrate plugging in the testing tube is within safe range. By using this method, the real-time estimation and management of hydrate plugging during the testing process can be achieved, which can provide basis for the safe and efficient testing of deep-water gas wells.
Zhang, Feifei (Yangtze University) | Islam, Aminul (Equinor ASA) | Zeng, Hao (Sinopec) | Chen, Zengwei (Sinopec) | Zeng, Yijin (Sinopec) | Wang, Xi (Yangtze University) | Li, Siyang (Yangtze University)
Nearly one third of the drilling time lost is caused by stuck drill pipe. In many cases, stuck pipe is preventable if early signs are detected and timely measures are taken, particularly for stuck pipe events caused by solids (primarily drilled cuttings) in the wellbore. This paper presents a physics-based model and data analytics combined approach to predict stuck pipe caused during drilling.
The new proposed method combines the physics-based first principle models, including transient solid transport model, drill string model (torque and drag model) and the data-driven models. The proposed models will be worked based on the analysis of both field and experimental data. The physics-based models capture the basic rules of fluid mechanics, drill string mechanics, and multiphase flow during the drilling operations.
By analyzing field data from historical wells and experimental data, an EnKF based data-driven model is applied to provide parameters and coefficients needed by the physics-based models. The data-driven model improves the reliability of the results predicted by the first-principle based model and allows it to continuously improve itself.
Based on a transient approach, this development can use real-time drilling operational data as inputs, predict the stuck pipe risks, and provide warnings when a high risk for stuck pipe scenarios are encountered. Comparing to existing stuck pipe prediction approaches, the new proposed approach can distinguish the hole cleaning related stuck pipe risk and other reasons to create stuck pipe. This hybrid method in build technology will use as a supporting tool in decision make. Resulting to bring an opportunity for the drillers to avoid the potential stuck pipe incidents by taking a proper action in time.
Han, Chao (China University of Petroleum East China) | Guan, Zhichuan (China University of Petroleum East China) | Li, Jingjiao (China University of Petroleum East China) | Hu, Huaigang (China University of Petroleum East China) | Xu, Yuqiang (China University of Petroleum East China)
A hybrid methodology is presented to predict equivalent circulating density (ECD), which combines autoregressive integrated moving average (ARIMA) and back propagation neural network (BPNN) models. Research results are compared to previously published ECD prediction method that is based on theoretical calculation of hydraulic parameters.
The hybrid methodology is based on data analysis theory. It uses ARIMA model to capture the linear trends of ECD, and then the BP neural network is used to predict the nonlinear and stochastic change law of ECD. Finally BP neural network prediction results are used to correct the prediction error of ARIMA to get the ECD prediction results. With a deepwater well in the South China Sea, a simulation experiment is carried out to verify the comprehensiveness and accuracy of the method presented in this paper.
The prediction results are similar to those of the traditional hydraulics model, which considers the effect of temperature and pressure on drilling fluid density and rheological parameters. Comparisons are also provided for three classical time series prediction models, including the support vector machine, multiple linear regression and grey prediction. The root mean square error (RMSE) and the mean absolute deviation (MAD) were selected as evaluation indicators. The comparison results show that the hybrid model can reflect the variation law of ECD more accurately. Therefore, relative to the traditional methods of prediction, the hybrid methodology has the advantages of advanced modeling thought and simple operation, and it can be selected for prediction of ECD.
Because of the effects of high temperature, high pressure and uncertain factors, the accurate prediction of ECD is very difficult. This paper provides a novel idea for accurately predicting ECD by analyzing the implicit relationship of ECD series data through data mining.
Objective/Scope: The definition of the locations of new wells in mature fields is a challenging problem especially in contexts of high geological complexity and low data reliability, when running fluid-flow simulations can be extremely difficult. For this reason, we develop an innovative Surrogate Reservoir Model, based on a data-driven process, which combines Machine Learning algorithms with spatial interpolation techniques. We call our approach WIZARD (acronym for: Well Infilling optimiZAtion through Regression and Data analytics). Methods/Procedures/Process: WIZARD is a collection of different data-driven methods for the sake of definition of new infilling well locations on the basis of the expected cumulative oil productions (after a fixed target period) of unexploited areas of the reservoir. The first method, named COSMIC, is used to find a correlation between petrophysical well properties and well productivity through a regression algorithm. The second method, that we call WIZARDMAP, uses spatial interpolation methodologies like K-Nearest Neighbours to estimate input petrophysical well data far away the existing wells and the trained COSMIC model applied to these interpolated data to predict the expected cumulative oil productions in unexploited areas of the reservoir. Finally, predictions of WIZARMAP model are compared with the ones given by another method that we call WIZARDROC, that is a predictive model trained by using only the cumulative oil productions of the existing wells and their locations.
Ma, Kang (China University of Petroleum-Beijing) | Jiang, Hanqiao (China University of Petroleum-Beijing) | Li, Junjian (China University of Petroleum-Beijing) | Zhang, Rongda (China University of Petroleum-Beijing) | Zhang, Lufeng (China University of Petroleum-Beijing) | Fang, Wenchao (Sinopec Petroleum Exploration and Production Research Institute) | Shen, Kangqi (China University of Petroleum-Beijing) | Dong, Rencheng (University of Texas at Austin)
For mature oilfields which have entered into the high water cut stage, many stimulation measures are adopted in order to maintain the oil production. Those measures include drilling new wells, general measures, and strengthened measures. Even though the oil production increase when the measures conducted, it will cause different degrees of production decline in the next year. Due to the irrational composition of oil production in the matured field, abnormal production decline is becoming the primary problem for stable production. Establish an effective early warning system (EWS) is important to release production alarm and take necessary measures in advance. In this paper, the factors that can affect the abnormal decline are selected and the influence degree of different factors are compared by grey relational analysis. The machine learning was adopted to build the EWS. Three distinct forms of input data are considered to improve the prediction accuracy. By using the degree of deviation from normal as the input data for the prediction model have the highest accuracy. However basic machine learning model contains many input parameters which can't obtain easily. The number of input parameter is optimization based on the variation of accuracy under different input parameter number. In order to improve the prediction accuracy the artificial samples are added into the training process. The prediction accuracy of the final optimization model can reach 92%. According to the EWS the production condition of different reservoir is evaluated. The result reveals the possibility of the occurrence of anomalous decline in different reservoir which can effectively guide the oilfield production strategy. The EWS can be an effective tool in the oil production monitor in the mature oil field.
Greater Bongkot North is a gas field located in Gulf of Thailand and on production since 1993. Most of the old wellhead platforms (30%) lack remote well test facilities which requires personnel visits for any well test measurement. Often, well testing in these platforms get lower priority compared to other operations in a matured field. This project implemented artificial intelligent (AI) technique to estimate gas rate from other available engineering and geological parameters.
A new approach using machine learning was applied to estimate gas production rate where actual measurements are not available. Actual production well test data was used to train the model. Input parameters used were:
Surface facility information Fluid properties Production condition Geological setup
Surface facility information
A blind test on the subset of historical data showed a level of confidence (R2) value of 0.93. This provided confidence to proceed with a full field pilot. A pilot was conducted during January to May 2018. The area of pilot was spread across various geological, operating and surface condition setups to reduce sampling bias. The pilot demonstrated the following use cases:
Improved prediction accuracy in wells with no recent test, achieving primary object of model. Detection of well behavior changes: The model could detect changes in well behavior without human intervention much before the trends become obvious for engineers to detect. Improved potential estimation in wells with leaks in wellhead chokes where conventional analysis followed in Bongkot is not possible due to improper wellhead shut-in pressure measurement. Improved efficiency with production allocation: The conventional method requires significant time (40-80 person hours per month) to make the data available for production allocation. This can be shortened significantly by use of this method
Improved prediction accuracy in wells with no recent test, achieving primary object of model.
Detection of well behavior changes: The model could detect changes in well behavior without human intervention much before the trends become obvious for engineers to detect.
Improved potential estimation in wells with leaks in wellhead chokes where conventional analysis followed in Bongkot is not possible due to improper wellhead shut-in pressure measurement.
Improved efficiency with production allocation: The conventional method requires significant time (40-80 person hours per month) to make the data available for production allocation. This can be shortened significantly by use of this method
In essence, this project demonstrated the potential use of artificial intelligent to improve efficiency in a matured gas field operating under marginal conditions.