Combining Physics-Based and Data-Driven Models for Estimation of WOB During Ultra-Deep Ocean Drilling

2018 ◽  
Author(s):  
Tatsuya Kaneko ◽  
Ryota Wada ◽  
Masahiko Ozaki ◽  
Tomoya Inoue
Keyword(s):  
Hybrid Model ◽  
Numerical Models ◽  
Drill String ◽  
Operating Conditions ◽  
Data Driven ◽  
High Frequency Oscillation ◽  
Lumped Mass ◽  
Unknown Parameters ◽  
Mass Model ◽  
Time Operation

Offshore drilling with drill string over 10,000m long has many technical challenges. Among them, the challenge to control the weight on bit (WOB) between a certain range is inevitable for the integrity of drill pipes and the efficiency of the drilling operation. Since WOB cannot be monitored directly during drilling, the tension at the top of the drill string is used as an indicator of the WOB. However, WOB and the surface measured tension are known to show different features. The deviation among the two is due to the dynamic longitudinal behavior of the drill string, which becomes stronger as the drill string gets longer and more elastic. One feature of the difference is related to the occurrence of high-frequency oscillation. We have analyzed the longitudinal behavior of drill string with lumped-mass model and captured the descriptive behavior of such phenomena. However, such physics-based models are not sufficient for real-time operation. There are many unknown parameters that need to be tuned to fit the actual operating conditions. In addition, the huge and complex drilling system will have non-linear behavior, especially near the drilling annulus. These features will only be captured in the data obtained during operation. The proposed hybrid model is a combination of physics-based models and data-driven models. The basic idea is to utilize data-driven techniques to integrate the obtained data during operation into the physics-based model. There are many options on how far we integrate the data-driven techniques to the physics-based model. For example, we have been successful in estimating the WOB from the surface measured tension and the displacement of the drill string top with only recurrent neural networks (RNNs), provided we have enough data of WOB. Lack of WOB measurement cannot be avoided, so the amount of data needs to be increased by utilizing results from physics-based numerical models. The aim of the research is to find a good combination of the two models. In this paper, we will discuss several hybrid model configurations and its performance.

Dynamic analysis of a helical gear reduction by experimental and numerical methods

2020 ◽  
Vol 68 (1) ◽  
pp. 48-58
Author(s):  
Chao Liu ◽  
Zongde Fang ◽  
Fang Guo ◽  
Long Xiang ◽  
Yabin Guan ◽  
...  
Keyword(s):  
Numerical Methods ◽  
Dynamic Analysis ◽  
Dynamic Behavior ◽  
Fourth Order ◽  
Numerical Models ◽  
Helical Gear ◽  
Operating Conditions ◽  
Dynamic Responses ◽  
Lumped Mass ◽  
Gear Mesh

Presented in this study is investigation of dynamic behavior of a helical gear reduction by experimental and numerical methods. A closed-loop test rig is designed to measure vibrations of the example system, and the basic principle as well as relevant signal processing method is introduced. A hybrid user-defined element model is established to predict relative vibration acceleration at the gear mesh in a direction normal to contact surfaces. The other two numerical models are also constructed by lumped mass method and contact FEM to compare with the previous model in terms of dynamic responses of the system. First, the experiment data demonstrate that the loaded transmission error calculated by LTCA method is generally acceptable and that the assumption ignoring the tooth backlash is valid under the conditions of large loads. Second, under the common operating conditions, the system vibrations obtained by the experimental and numerical methods primarily occur at the first fourth-order meshing frequencies and that the maximum vibration amplitude, for each method, appears on the fourth-order meshing frequency. Moreover, root-mean-square (RMS) value of the acceleration increases with the increasing loads. Finally, according to the comparison of the simulation results, the variation tendencies of the RMS value along with input rotational speed agree well and that the frequencies where the resonances occur keep coincident generally. With summaries of merit and demerit, application of each numerical method is suggested for dynamic analysis of cylindrical gear system, which aids designers for desirable dynamic behavior of the system and better solutions to engineering problems.

Truncation Design and Model Testing of a Deepwater FPSO Mooring System

2016 ◽  
Vol 138 (2) ◽  
Author(s):  
Hongwei Wang ◽  
Gang Ma ◽  
Liping Sun ◽  
Zhuang Kang
Keyword(s):  
Hybrid Model ◽  
Dynamic Properties ◽  
Model Testing ◽  
Model Tests ◽  
Mooring System ◽  
Lumped Mass ◽  
Mass Model ◽  
Numerical Reconstruction ◽  
Lumped Mass Model ◽  
System Configurations

Limitation of offshore basin dimensions is a great challenge for model tests of deepwater mooring system. The mooring system cannot be modeled entirely in the basin with a reasonable model scale. A classical solution is based on hybrid model tests for the truncated mooring system. An efficient truncation method is proposed in this paper taking advantage of the mechanical characteristics of catenary mooring system. Truncation procedures are presented both in vertical and horizontal directions. A turret moored floating production storage and offloading (FPSO) is analyzed, and its mooring system is truncated from the original 914 m water depth to 736 m and 460 m, respectively. Numerical simulations are performed based on catenary theory and lumped mass model to these three systems, including the original untruncated system and two truncated systems. The static characteristics and dynamic response are investigated, and the results are compared between the truncated and untruncated system, and good agreements are obtained, verifying the preliminary truncation design. Model tests are conducted to the three mooring system configurations in the deepwater basin of the Harbin Engineering University. The static and dynamic properties are found to be mostly consistent between the untruncated system and two truncated systems, except for some discrepancy in 460 m system. It indicates that the truncation design is successful when the truncation factor is large, while difference still exists when the truncation factor is small. Numerical reconstruction to the model test in 460 m and extrapolation to 914 m are also implemented. The results are found to be consistent with those in 914 m, verifying the robustness and necessity of the hybrid model testing, especially for the mooring system with large truncation.

scholarly journals Structural System Identification in the Time Domain using Evolutionary and Behaviorally Inspired Algorithms and their Hybrids

2009 ◽  
Vol 6 (2) ◽  
pp. 64
Author(s):  
S. Sandesh ◽  
Abhishek Kumar Sahu ◽  
K. Shankar
Keyword(s):  
System Identification ◽  
Time Domain ◽  
Numerical Models ◽  
Parametric Identification ◽  
Lumped Mass ◽  
Unknown Parameters ◽  
Mass System ◽  
The Time Domain ◽  
Structural System Identification ◽  
Speed And Accuracy

 In this study, parametric identification of structural properties such as stiffness and damping is carried out using acceleration responses in the time domain. The process consists of minimizing the difference between the experimentally measured and theoretically predicted acceleration responses. The unknown parameters of certain numerical models, viz., a ten degree of freedom lumped mass system, a nine member truss and a non-uniform simply supported beam are thus identified. Evolutionary and behaviorally inspired optimization algorithms are used for minimization operations. The performance of their hybrid combinations is also investigated. Genetic Algorithm (GA) is a well known evolutionary algorithm used in system identification. Recently Particle Swarm Optimization (PSO), a behaviorally inspired algorithm, has emerged as a strong contender to GA in speed and accuracy. The discrete Ant Colony Optimization (ACO) method is yet another behaviorally inspired method studied here. The performance (speed and accuracy) of each algorithm alone and in their hybrid combinations such as GA with PSO, ACO with PSO and ACO with GA are extensively investigated using the numerical examples with effects of noise added for realism. The GA+PSO hybrid algorithm was found to give the best performance in speed and accuracy compared to all others. The next best in performance was pure PSO followed by pure GA. ACO performed poorly in all the cases. 

Data-driven parametrizations in numerical models using data assimilation and machine learning.

2020 ◽  
Author(s):  
Julien Brajard ◽  
Alberto Carrassi ◽  
Marc Bocquet ◽  
Laurent Bertino
Keyword(s):  
Machine Learning ◽  
High Resolution ◽  
Data Assimilation ◽  
Physical Model ◽  
Hybrid Model ◽  
Numerical Models ◽  
Coarse Graining ◽  
Model Error ◽  
Data Driven ◽  
Model Combining

<p>Can we build a machine learning parametrization in a numerical model using sparse and noisy observations?</p><p>In recent years, machine learning (ML) has been proposed to devise data-driven parametrizations of unresolved processes in dynamical numerical models. In most of the cases, ML is trained by coarse-graining high-resolution simulations to provide a dense, unnoisy target state (or even the tendency of the model).</p><p>Our goal is to go beyond the use of high-resolution simulations and train ML-based parametrization using direct data. Furthermore, we intentionally place ourselves in the realistic scenario of noisy and sparse observations.</p><p>The algorithm proposed in this work derives from the algorithm presented by the same authors in https://arxiv.org/abs/2001.01520.The principle is to first apply data assimilation (DA) techniques to estimate the full state of the system from a non-parametrized model, referred hereafter as the physical model. The parametrization term to be estimated is viewed as a model error in the DA system. In a second step, ML is used to define the parametrization, e.g., a predictor of the model error given the state of the system. Finally, the ML system is incorporated within the physical model to produce a hybrid model, combining a physical core with a ML-based parametrization.</p><p>The approach is applied to dynamical systems from low to intermediate complexity. The DA component of the proposed approach relies on an ensemble Kalman filter/smoother while the parametrization is represented by a convolutional neural network.  </p><p>We show that the hybrid model yields better performance than the physical model in terms of both short-term (forecast skill) and long-term (power spectrum, Lyapunov exponents) properties. Sensitivity to the noise and density of observation is also assessed.</p>

scholarly journals Validation of Numerical Models of a Rotorcraft Crashworthy Seat and Subfloor

Aerospace ◽  
2020 ◽  
Vol 7 (12) ◽  
pp. 174
Author(s):  
Paolo Astori ◽  
Mauro Zanella ◽  
Matteo Bernardini
Keyword(s):  
Numerical Models ◽  
Element Model ◽  
Drop Test ◽  
Reference Scenario ◽  
Lumped Mass ◽  
Mass Model ◽  
Seat Cushion ◽  
Lumped Mass Model ◽  
Energy Absorbers ◽  
Energy Absorbing

The present work explores some critical aspects of the numerical modeling of a rotorcraft seat and subfloor equipped with energy-absorbing stages, which are paramount in crash landing conditions. To limit the vast complexity of the problem, a purely vertical impact is considered as a reference scenario for an assembly made of a crashworthy helicopter seat and a subfloor section, including an anthropomorphic dummy. A preliminary lumped mass model is used to drive the design of the experimental drop test. Some additional static and dynamic tests are carried out at the coupon and sub-component levels to characterize the seat cushion, the seat pan and the honeycomb elements that were introduced in the structure as energy absorbers. The subfloor section is designed and manufactured with a simplified technique, yet representative of this structural component. Eventually, a finite element model representing the full drop test was created and, together with the original lumped mass model, finally validated against the experimental test, outlining the advantage of using both the numerical techniques for design assistance.

scholarly journals Data-driven components in a model of inner-shelf sorted bedforms: a new hybrid model

2014 ◽  
Vol 2 (1) ◽  
pp. 67-82 ◽  
Author(s):  
E. B. Goldstein ◽  
G. Coco ◽  
A. B. Murray ◽  
M. O. Green
Keyword(s):  
Machine Learning ◽  
Numerical Model ◽  
Hybrid Model ◽  
Learning Algorithm ◽  
Numerical Models ◽  
Data Driven ◽  
Data Sets ◽  
Inner Shelf ◽  
Reference Concentration ◽  
Deterministic Description

Abstract. Numerical models rely on the parameterization of processes that often lack a deterministic description. In this contribution we demonstrate the applicability of using machine learning, a class of optimization tools from the discipline of computer science, to develop parameterizations when extensive data sets exist. We develop a new predictor for near-bed suspended sediment reference concentration under unbroken waves using genetic programming, a machine learning technique. We demonstrate that this newly developed parameterization performs as well or better than existing empirical predictors, depending on the chosen error metric. We add this new predictor into an established model for inner-shelf sorted bedforms. Additionally we incorporate a previously reported machine-learning-derived predictor for oscillatory flow ripples into the sorted bedform model. This new "hybrid" sorted bedform model, whereby machine learning components are integrated into a numerical model, demonstrates a method of incorporating observational data (filtered through a machine learning algorithm) directly into a numerical model. Results suggest that the new hybrid model is able to capture dynamics previously absent from the model – specifically, two observed pattern modes of sorted bedforms. Lastly we discuss the challenge of integrating data-driven components into morphodynamic models and the future of hybrid modeling.

scholarly journals Data driven components in a model of inner shelf sorted bedforms: a new hybrid model

2013 ◽  
Vol 1 (1) ◽  
pp. 531-569 ◽  
Author(s):  
E. B. Goldstein ◽  
G. Coco ◽  
A. B. Murray ◽  
M. O. Green
Keyword(s):  
Machine Learning ◽  
Numerical Model ◽  
Hybrid Model ◽  
Learning Algorithm ◽  
Numerical Models ◽  
Data Driven ◽  
Data Sets ◽  
Inner Shelf ◽  
Reference Concentration ◽  
Deterministic Description

Abstract. Numerical models rely on the parameterization of processes that often lack a deterministic description. In this contribution we demonstrate the applicability of using machine learning, optimization tools from the discipline of computer science, to develop parameterizations when extensive data sets exist. We develop a new predictor for near bed suspended sediment reference concentration under unbroken waves using genetic programming, a machine learning technique. This newly developed parameterization performs better than existing empirical predictors. We add this new predictor into an established model for inner shelf sorted bedforms. Additionally we incorporate a previously reported machine learning derived predictor for oscillatory flow ripples into the sorted bedform model. This new "hybrid" sorted bedform model, whereby machine learning components are integrated into a numerical model, demonstrates a method of incorporating observational data (filtered through a machine learning algorithm) directly into a numerical model. Results suggest that the new hybrid model is able to capture dynamics previously absent from the model, specifically, the two observed pattern modes of sorted bedforms. However, caveats exist when data driven components do not have parity with traditional theoretical components of morphodynamic models, and we discuss the challenges of integrating these disparate pieces and the future of this type of modeling.

Second-Law Analysis in a Steam Power Plant for Minimization of Avoidable Exergy Destruction

2010 ◽  
Author(s):  
Tapan K. Ray ◽  
Pankaj Ekbote ◽  
Ranjan Ganguly ◽  
Amitava Gupta
Keyword(s):  
Work Performance ◽  
Operating Conditions ◽  
Performance Criteria ◽  
Second Law ◽  
Feed Water ◽  
Exergy Destruction ◽  
Actual Performance ◽  
Time Operation ◽  
Major Equipment ◽  
Cycle Efficiency

Performance analysis of a 500 MWe steam turbine cycle is performed combining the thermodynamic first and second-law constraints to identify the potential avenues for significant enhancement in efficiency. The efficiency of certain plant components, e.g. condenser, feed water heaters etc., is not readily defined in the gamut of the first law, since their output do not involve any thermodynamic work. Performance criteria for such components are defined in a way which can easily be translated to the overall influence of the cycle input and output, and can be used to assess performances under different operating conditions. A performance calculation software has been developed that computes the energy and exergy flows using thermodynamic property values with the real time operation parameters at the terminal points of each system/equipment and evaluates the relevant rational performance parameters for them. Exergy-based analysis of the turbine cycle under different strategic conditions with different degrees of superheat and reheat sprays exhibit the extent of performance deterioration of the major equipment and its impact to the overall cycle efficiency. For example, during a unit operation with attemperation flow, a traditional energy analysis alone would wrongly indicate an improved thermal performance of HP heater 5, since the feed water temperature rise across it increases. However, the actual performance degradation is reflected as an exergy analysis indicates an increased exergy destruction within the HP heater 5 under reheat spray. These results corroborate to the deterioration of overall cycle efficiency and rightly assist operational optimization. The exergy-based analysis is found to offer a more direct tool for evaluating the commercial implication of the off-design operation of an individual component of a turbine cycle. The exergy destruction is also translated in terms of its environmental impact, since the irretrievable loss of useful work eventually leads to thermal pollution. The technique can be effectively used by practicing engineers in order to improve efficiency by reducing the avoidable exergy destruction, directly assisting the saving of energy resources and decreasing environmental pollution.

Analysis of Mistuned Blade Vibrations Based on Normally Distributed Blade Individual Natural Frequencies

2015 ◽  
Author(s):  
Felix Figaschewsky ◽  
Arnold Kühhorn
Keyword(s):  
Probability Distribution ◽  
Development Process ◽  
Structural Model ◽  
Natural Frequencies ◽  
Rotor Blades ◽  
Lumped Mass ◽  
Mass Model ◽  
Influence Coefficients ◽  
Random Mistuning ◽  
Rotor Assembly

With increasing demands for reliability of modern turbomachinery blades the quantification of uncertainty and its impact on the designed product has become an important part of the development process. This paper aims to contribute to an improved approximation of expected vibration amplitudes of a mistuned rotor assembly under certain assumptions on the probability distribution of the blade’s natural frequencies. A previously widely used lumped mass model is employed to represent the vibrational behavior of a cyclic symmetric structure. Aerodynamic coupling of the blades is considered based on the concept of influence coefficients leading to individual damping of the traveling wave modes. The natural frequencies of individual rotor blades are assumed to be normal distributed and the required variance could be estimated due to experiences with the applied manufacturing process. Under these conditions it is possible to derive the probability distribution of the off-diagonal terms in the mistuned equations of motions, that are responsible for the coupling of different circumferential modes. Knowing these distributions recent limits on the maximum attainable mistuned vibration amplitude are improved. The improvement is achieved due to the fact, that the maximum amplification depends on the mistuning strength. This improved limit can be used in the development process, as it could partly replace probabilistic studies with surrogate models of reduced order. The obtained results are verified with numerical simulations of the underlying structural model with random mistuning patterns based on a normal distribution of individual blade frequencies.

Combustion Tuning for a Gas Turbine Power Plant Using Data-Driven and Machine Learning Approach

2021 ◽  
Vol 143 (3) ◽  
Author(s):  
Suhui Li ◽  
Huaxin Zhu ◽  
Min Zhu ◽  
Gang Zhao ◽  
Xiaofeng Wei
Keyword(s):  
Machine Learning ◽  
Power Plant ◽  
Gas Turbine ◽  
Operating Conditions ◽  
Data Driven ◽  
Ann Model ◽  
Promising Alternative ◽  
Combustion Performance ◽  
Wide Range ◽  
Gas Turbine Power Plant

Abstract Conventional physics-based or experimental-based approaches for gas turbine combustion tuning are time consuming and cost intensive. Recent advances in data analytics provide an alternative method. In this paper, we present a cross-disciplinary study on the combustion tuning of an F-class gas turbine that combines machine learning with physics understanding. An artificial-neural-network-based (ANN) model is developed to predict the combustion performance (outputs), including NOx emissions, combustion dynamics, combustor vibrational acceleration, and turbine exhaust temperature. The inputs of the ANN model are identified by analyzing the key operating variables that impact the combustion performance, such as the pilot and the premixed fuel flow, and the inlet guide vane angle. The ANN model is trained by field data from an F-class gas turbine power plant. The trained model is able to describe the combustion performance at an acceptable accuracy in a wide range of operating conditions. In combination with the genetic algorithm, the model is applied to optimize the combustion performance of the gas turbine. Results demonstrate that the data-driven method offers a promising alternative for combustion tuning at a low cost and fast turn-around.

Sign in / Sign up

Export Citation Format

Share Document