Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Results
Abstract Statistical and machine learning approaches to pipeline leak detection can benefit from augmentation with simple physical models that can predict line pack, particularly in cases where the fluids involved change density strongly with temperature and pressure. Determining the system wide temperature changes needed for these kinds of compensation models is not trivial, as the temperature measurements taken on the system are often only valid local to the sensors. This paper presents a machine learning approach to obtaining the temperature differences needed for compensation and then applying them. The compensation method is tested against the uncompensated method in three different cases with NGLs, one of which is simulated and two of which come from real pipeline data. The compensated imbalance shows marked improvement in reducing false alarms while increasing sensitivity to leaks.
Abstract In liquid lines with significant elevation changes, the presence of slack line flow can be a significant operational problem. Firstly, slack line flow conditions make control downstream of the slack condition difficult. Secondly, and more importantly, when the vapor bubble from slack flow is decreasing in size, the downstream flow rate will be less than upstream flow rate, which will appear as a leak to an online model. Via an uncertainty analysis, it is shown that it is not possible for an online model to predict slack flow accurately, particularly if the pipeline is batched and/or utilizes DRA. Both false positive and false negative predictions of slack line conditions are common. Instead of a deterministic model, a statistical approach is proposed. Monte Carlo analysis was used to determine the distribution of conditions around the slack flow conditions including a detailed analysis of the pipeline at the high points of the pipeline where slack conditions will first appear. The derived distribution was used to determine probability of slack conditions at any time. The new, statistical approach is illustrated with multiple examples from operating pipelines. Over time it is expected that a sufficient number of events will be detected and analyzed such that a pattern-matching algorithm could be applied to enhance the detection of slack conditions.
Abstract Product mixing occurs often in batched liquid-phase pipelines. If not successfully mitigated, this can result in a product being assigned as a less valuable one at the terminal end, leading to a significant financial impact. There is also risk of contamination of one product with an impurity the other product is carrying, especially in cases where the product thermodynamic properties are very similar. Current methods for calculating interface volume, for example, Austin & Palfrey 1963, do not appear to scale-up accurately in a consistent manner from the lab to the pipeline. This paper presents a technique for determining the diffusion coefficient in the differential equation for diffusion transport in a pipeline, based on actual measured data. The transport equation was made non-dimensional and the Laplace Transform solution approach was applied; the inverse Laplace Transform was determined using an analytical method. The diffusion coefficient was then deduced from actual pipeline densitometer measurements by varying the coefficient to obtain an acceptable fit to the data. Continued application of this approach should result in a database of coefficients that can be applied to a particular flowing scenario.
- North America > United States (0.46)
- North America > Canada > Alberta (0.29)
Abstract Natural Gas (NG) is odorless and therefore requires an odorizer to be injected into its flow to ensure detection when a gas leak occurs and thus provide satisfactory safety levels. The odorisation process is a delicate step that takes place within the City Gate Stations (CGS), which are key elements of the NG network infrastructure. This work aims to develop a method for offline monitoring of the odorization process within a CGS located in central Italy, based on the exploiting of the odorization station dataset through several machine learning models development, to evaluate the odorization process performance. An unsupervised machine learning method based on two different algorithms, the LOF (Local Outlier Factor) algorithm and the K-Means clustering, is developed, and then data mining is carried out on the dataset to extract useful information. The results show that the use of the algorithm made it possible to identify anomalous points in the dataset and their dependence on the main operating parameters of the CGS, as well as some clusters of under-odor and over-odor tendencies for the system under consideration. Introduction and Background Natural gas, pumped by compressors in the high-pressure network, arrives at CGS stations with known composition, tending to consist of about 95 % methane and the rest of a mixture of various hydrocarbons and other gases [1]. This mixture is odorless, colorless, and flammable, so it is important to ensure the addition of an odorant into gas that makes the mixture identifiable in the event of a leak [2], before it is sent to regional and local distribution networks. From a technical point of view, odorizing systems can be divided into injection systems and vaporization systems, depending on the different types of equipment and physical principles used to insert the odorizing substance into the gas flow. The former injects directly an odorant into the flowing NG stream, while the latter is based on the diffusion of odorant into the gas stream [1], [3], [4]. In a study by MARCOGAZ [5], the Technical Association of the European Natural Gas Industry, the requirements, characteristics of the various odorants, processes, and regulations related to NG odorization of 19 different European countries, including Italy, were summarized.
- Europe > Italy (0.70)
- North America > United States (0.68)
Abstract Governments, organizations, industries, and individuals across the world recognize that energy transition is required to meet climate change goals. Early and pro-active consideration and out of the box thinking are needed to successfully adapt to the energy transition future. Utilities and pipeline companies have infrastructure that is well positioned to adapt to energy transition. Topics of interest for facility planners currently related to energy transition include: Hydrogen blending into natural gas results in the need to transport increased volumes of gas due to hydrogenโs lower energy content. The equations of state, gas composition and gas properties should be carefully considered. Hydrogen blended gas will impact the capacity of the pipeline system, regulating and compressor stations as well as operational items such as pigging velocities. Impacts to customers bills should also be considered as customers may have variability in energy content. Renewable natural gas injection is limited by the takeaway market of the system. Utilities can increase this market by employing reduced and remote operated regulator set pressures, installing compression to move gas into upstream higher-pressure systems, installing tanks to store excess gas or trucking to CNG injection locations. RNG can also have lower heat content which could require additional modelling to determine heat content contour maps for billing purposes. Advanced metering will be required for utilities to effectively measure the effectiveness of energy conservation programs and to employ demand response programs. There are other benefits including reduced meter reading costs and improved peak hour and peak day demand estimation. Governmentโs policy of restricting or banning natural gas may cause a spiral of increasing costs for a reduced number of customers. Utilities must employ an โall or nothingโ approach such that pipes can be retired to reduce O&M costs. Focus on old pipelines and work from extremities towards sources. Integrated Resource Planning is focused on peak hour and peak day demand reduction to reduce, defer, or avoid pipeline infrastructure using demand side and supply side alternatives. Demand side alternatives include geo-targeted DSM. ICF International Inc.9 was hired by Enbridge to study the impact to peak hour and peak day demand using theoretical modelling. Some energy efficiency options such as upgraded windows, insulation and appliances reduce the peak period demand while smart or programable thermostats increase the peak period demand. Planning teams considerations and challenges need to consider that demand forecasting, and energy efficiency estimations are not an exact science, and are difficult to estimate with certainty. The demand reduction may be well within the forecast and measurement error. Facilities planning errs on the side of conservatism. Forecasting the need for a facility project far enough ahead to be able to create an energy conservation program and measure its effectiveness needs to happen years before a pipeline project is required considering it takes 5 years to plan a large-scale reinforcement project. Forecast development for energy transition says that energy consumption cannot be zero as buildings and equipment can only be so efficient. It will take time to transition existing natural gas customers to alternatives. It is unknown how long it will take. The modellers generally do not forecast and are not experts in energy transition.
Abstract Reduced order models, commonly referred to as ROMs, have been used in many areas of engineering due to their reduced complexity and corresponding speed of computation and solution. In transient pipeline simulation, transfer function-based ROMs have experienced widespread adoption spanning multiple decades due to their efficiency compared with higher order computational fluid dynamics. Many examples can be traced back to the 1980s Krรกlik et. al. paper that considers simplified transfer functions. The choice of order of the simplified transfer functions then becomes the primary parameter that other authors amend compared with Krรกlik. However, changing the order of the simplified transfer functions can have profound effects on the response of the system. Incidentally, it is necessary to understand the difference between true pipeline behaviors versus effects from the modelerโs choice in further approximating the simplified transfer functions. In this paper, we explore various simplified transfer function models and compare them with both a commercial grade modeling software and more importantly, with field data obtained from a large US based gas transmission pipeline. The field data validates certain model approximations and provides the modeling community a new set of high-resolution data that can be used for validating other models. We end the paper with an analysis and conclusion of how accurate the transfer function models were given the observed field data. Introduction and Background Pipelines are the most efficient method to transport energy over large distances. They produce the lowest amount of Greenhouse Gases over their life cycles when compared with Railway or Road based transportation [1] as well as being the safest method when comparing incident number and magnitude [2].
Abstract Erosion damage caused by solid particles in multiphase flows can affect the operation and integrity of fluid transport pipeline systems. Amongst available approaches for predicting erosion, mechanistic models provide reliable predictions under various conditions with a low calculation requirement time. A novel mechanistic model for predicting erosion in multiphase flows has been developed which is based on the trajectory of representative particles and the characteristics of various upward vertical gas-liquid flow regimes. Computational fluid dynamics (CFD) models are used to construct a trajectory-based model in a three-dimensional domain of the elbow geometry which describes the flow around the particles. In addition to gas-sand and liquid sand flows, multiphase flow regimes such as bubbly flow, churn flow, annular flow, and mist flow are characterized and applied to the trajectory-based model. The validity of the proposed model is examined using a large number of experimental erosion data from previous studies and comparing with the available mechanistic models from the literature suggests that the new model can provide more accurate predictions for both gas-dominated flow and high liquid rate multiphase flow conditions. Introduction and Background Multiphase flows are commonly found in various industries including oil and gas and there are many applications where particle-induced erosion should be monitored. Over the years, many approaches have been developed and introduced to predict erosion including multiphase fluid and solid particle flows. These approaches include one-equation models, mechanistic models, computational methods, and machine learning algorithms. While the one-equation, empirical models only account for the bulk fluid velocity [1, 2], mechanistic models consider particle velocities using a simplified fluid velocity [3, 4] and computational methods utilize two or three-dimensional Computational Fluid Dynamics (CFD) flow results, along with the particle motion equation [5, 6]. Moreover, the one-equation, empirical models account for limited parameters and can be highly conservative in predicting the erosion rate especially for multiphase flows as they only account for fluid mixture velocity and density, and sometimes pipe diameter and particle size. On the other hand, computational models using CFD codes describe the process in more detail and are more consistent with the erosion data for a wide range of flow conditions. Mechanistic models aim at predicting erosion more accurately than the empirical one-equation models and provide predictions for many cases quickly and at a lower computational cost compared to CFD simulations. Specifically, transient multiphase flow CFD simulations require substantial computational resources.
Abstract Accumulation of sand within pipelines can pose significant problems in the oil and gas industry. Sand particles tend to settle out of suspension leading to the formation of stationary or moving beds along the bottom of pipelines. Such beds provide an environment that result in flow assurance problems such as increased corrosion rates, increased pressure drop in pipes and pigging blockage. Thus, the prediction of sand transport velocities is of great importance for flow assurance in petroleum as well as mining industries. The processes in these applications must be designed and operated at a sufficient fluid velocity to avoid solid deposition. Recent studies at the Tulsa University Sand Management Projects (TUSMP) have shown that Artificial Intelligence โ Machine Learning (ML) methods can be effectively and accurately used in predicting minimum particle transport velocities in pipelines. However, these methods have not been rigorously developed and tested for multiphase air-water flows with particles. The purpose of this work is to investigate the use of several ML models to predict the critical velocities in horizontal and inclined multiphase flow pipelines. In this study, three machine learning algorithms, including Support Vector Machine, Random Forest, and Extreme Gradient Boosting, are utilized to predict minimum flow rates required to transport particles successfully in intermittent and stratified gas-liquid flow regimes. The models predict the value of critical velocities in pipes via ML using accessible parameters as inputs, namely, sand concentration, pipe inclination, pipe size, liquid density, liquid viscosity, particle density, and particle size. First, these models are trained with a set of 1640 data points. After hyper-parameters of each model are optimized, the results are verified with a test data set and their predictive abilities are cross-compared. After the final set of models is constructed, an error analysis was performed by evaluating the results when input parameters, such as superficial velocities and fluid properties, were changed. Later, the predictive performance of the method is also validated using out-of-sample data available from the literature. Finally, the predictive abilities of the best model are further validated by comparing its performance with well-established mechanistic models based on empirical correlations. The Random Forest results reveal a better training performance and prediction. The results also indicate that the proposed method gives comparable or even higher scores by contrast to correlations and mechanistic models for multiphase flow, and could be easily employed for industrial applications. The application of the above-mentioned ML algorithms and the large database used for their training allowed extending the proposed methodology to a wider applicability range of input parameters as compared to standard accessible techniques. The ML results present competitive and even more accurate predictions as compared with the existing mechanistic models, indicating a great potential of utilizing the data-driven machine learning methodology for applications in flow assurance.
Development of a Simple Statistical Model for the Prediction of Gas Hydrate Formation Conditions
Sayani, Jai Krishna Sahith (University College Dublin) | Kamireddi, Venkateswara Rao (Jawaharlal Nehru Technological University Kakinada) | Pedapati, Srinivasa Rao (Universiti Teknologi PETRONAS) | Kolliopoulos, Georgios (Universitรฉ Laval)
Abstract Gas hydrates are a major flow assurance problem in the oil & gas industry. If not treated well, they pose a major threat to pipeline failure. In this work, a simple statistical correlation is developed for the prediction of gas hydrate formation temperature. For this, initially, the thermodynamic equilibrium conditions for various gas mixtures are gathered from the literature. This is done to maintain the accuracy with the real-time conditions that could encounter in the pipelines. These data points are used for the development of a statistical model. This model is validated with the latest literature data and its reliability for the prediction of gas hydrate formation temperature is confirmed. Further, the performance of the improved model is compared with some of the well-known statistical models to evaluate its efficiency. Introduction and Background Gas hydrates are crystalline compounds formed when gas molecules get trapped inside the hydrogen-bonded cages formed by water molecules [1]. These gas hydrates are formed at low temperatures and high pressure considering the suitable size of the guest molecules that matches the cavity's size [2]. Water from the reservoirs is usually used to make petroleum reservoir fluids. As a result of the combination of water and hydrocarbons, gas hydrate formation can occur, causing flow restriction and blockage, as well as major operational, economic, and safety issues. The ongoing development of many petroleum fields raises the danger of encountering gas hydrates, which might pose operational challenges. Low seabed temperatures and high operation pressures enhance the likelihood of blockage due to gas hydrate formation in multiphase transfer lines from the wellhead to the production platform. Hydrate formation can also occur in other facilities, such as wells and process equipment [3]. Oil and gas corporations are constantly setting new water depth records in their search for hydrocarbon riches in deep waters. Water-based drilling fluids are generally preferred over oil-based drilling fluids due to environmental concerns and regulations, particularly in offshore exploration. However, the development of gas hydrates in the event of a gas kick is a well-known concern in deep water offshore drilling employing water-based fluids. The hydrostatic pressure of the drilling fluid column, together with the comparatively low bottom temperature, could create ideal thermodynamic conditions for the formation of gas hydrates in deep-water drilling. During the confinement of the kick, this might generate major well safety and control issues.
- Asia > India (0.46)
- North America > United States (0.46)
- North America > Canada (0.28)
- Reservoir Description and Dynamics > Non-Traditional Resources > Gas hydrates (1.00)
- Production and Well Operations > Production Chemistry, Metallurgy and Biology > Inhibition and remediation of hydrates, scale, paraffin / wax and asphaltene (1.00)
- Facilities Design, Construction and Operation > Flow Assurance > Hydrates (1.00)
Abstract The paper demonstrates the construction, training, and uncertainty quantification analysis of an artificial neural network model for single phase liquid pipeline leak detection through pressure drop and flow rate monitoring. The demonstrated methods reached acceptable error levels efficiently using theoretical data produced by a theoretical physics model. The results demonstrate that the randomly simulated leakage can be efficiently detected using the trained ANN (Artificial Neural Network) model based on theoretical data derived from physical equations. However, complexity appears when simulated leakage with modeled uncertainty are used in the training of the AI models. The propagation and influence of the uncertainty in the input data on the ANN method are discussed. Randomness following certain probability distributions is introduced into the data to measure the influence on the efficiency and reliability of the training and results of the ANN model. The paper also discusses the influence of the range of input data on the predictability of the ANN model in leak detection. Introduction and Background Data driven Machine Learning technologies such as ANN (Artificial Neural Networks) provide great innovation opportunities towards the design and operation of oil and gas pipeline systems. ANN based models have proved to be efficient in predicting the pressure drop within a pipeline (Brkic, D., and Cojbasic, Z., 2016; Shayya, W.H., and Sablani, S.S, 1998; Salmasi, F., etc. 2012; Fadare, D.A., and Ofidhe, U.I., 2009) as well as solving the inverse problems such as detecting pipeline roughness progression, inner diameter change, and leakage (Cheng, D., Zeosky, D., 2019). The increasing interest towards machine learning models causes the need to highlight the concerns of uncertainty propagation into the engineering applications of the methods. Machine learning models are developed through analysis of data derived either from physics models or measurements.
- Research Report > New Finding (0.48)
- Research Report > Experimental Study (0.34)