Title : Accurate prediction of bottom-hole pressure in vertical wells of algerian fields using a new hybrid intelligent system based on neural network -fuzzy logic and PSO algorithm
Abstract:
The evaluation of the pressure drop due to multiphase flow in vertical pipes is an old problem in oil and gas industry. Correct prediction of pressure drop between the edges of production wells is a crucial task as it needed for the viable and low-cost design of tubing, also for the optimization of production strategy, which is a key target for oil production maximization and operational price discount. But the prediction of this parameter is complicated due to the variation in fluid flow rate through the two phase flow stream; otherwise it's not feasible economically to deploy a pressure gauge in each well to measure the Bottom-hole Pressure (BHP) directly. To overcome these difficulties numerous correlations and mechanistic models were advanced on the grounds since 1950.
Notwithstanding, the relevance of every single existing correlation is extremely constrained and likewise with mechanistic models, they all require longer computations and have significant error. The error in predicting Bottom-hole Pressure engender a big error in predicting the well potential during its life-cycle which leads to making bad decisions on fateful operational tasks such as tubing design, artificial-lift system design and well production monitoring. Accordingly, precise and quick computation/estimation of the Bottom-hole Pressure is of incredible significance and is one of the fundamental difficulties in petroleum engineering.
Therefore, introducing a more powerful, fast and accurate method than the traditional ones to determine (BHP) it becomes a necessity. The main focus of this study is to establish an appropriate novel hybrid intelligent system to predict the bottom-hole flowing pressure in a multiphase vertical flow with higher accuracy than the existing methods using a sum of 150 field data sets amassed from Algerian fields. This method which combines two approaches, artificial neural networks (ANN) and fuzzy-logic (FL) is called ‘‘Adaptive Neuro-Fuzzy Inference System’’ (ANFIS). The goal of their combination is to amplify their strengths and compliments their weaknesses. After filtering the data and building the ANFIS model using a hybrid (Last Square and Back Propagation) algorithm for learning, a comparison has been made between this technique and the most generally utilized correlations and mechanistic models, using graphical and statistical error analysis.
The comparison indicates that the proposed method is more accurate, reliable and efficient than all the other correlations used in this work. For more improving the performance of this model, a particle swarm optimization (PSO), Algorithm as well-known a famous method to solve complex optimization problems is employed for the surest layout of both vector of linear coefficients of consequents and the Gaussian membership functions of antecedents in such network. Finally the results of (PSO), Algorithm are compared with the primary hybrid algorithm and showed pretty pleasant outcomes.
Audience Take Away Notes:
- The public can apply the Adaptive Neuro-Fuzzy Inference System (ANFIS) to predict bottom-hole pressure (BHP) in their own projects, improving the accuracy of their pressure-drop estimates in multiphase vertical flows.
- By understanding the advantages of ANFIS over traditional methods, they can adopt this hybrid approach to improve the reliability and efficiency of their BHP predictions, leading to better decision-making and better operational results.
- Participants will be able to use particle swarm optimization (PSO) to refine their ANFIS models, to achieve better performance and greater accuracy in their predictive analyses.
- By using ANFIS and PSO, the public will get more accurate predictions of bottom-hole pressure (BHP), which will reduce errors in estimating well potential and lead to better-informed operational decisions.
- Yes, faculty can use this study as a basis for exploring other advances in hybrid intelligent systems, such as combining ANFIS with other optimization algorithms or applying the model to different types of flows and multiphase conditions.
- Yes, implementing ANFIS and PSO to predict bottom-hole pressure (BHP) offers several practical benefits such as increased accuracy and reliability that can greatly simplify and improve a designer's work.
- Indeed, ANFIS integrates the strengths of artificial neural networks and fuzzy logic, enabling it to capture complex relationships in multiphase flow dynamics more accurately than traditional empirical correlations or mechanistic models. The result is more accurate predictions of bottom-hole pressure (BHP) under varying conditions.
- PSO optimization fine-tunes the ANFIS model parameters, such as membership functions and coefficients, to minimize prediction errors. This optimization ensures that the model adapts to specific field conditions, thereby improving its accuracy in real-world applications.
