With the aim of improving the shortcomings of the traditional single hidden layer back propagation (BP) neural network structure and learning algorithm, this paper proposes a centrifugal pump performance prediction method based on the combination of the Levenberg–Marquardt (LM) training algorithm and double hidden layer BP neural network. MATLAB was used to establish a double hidden layer BP neural network prediction model to predict the head and efficiency of a centrifugal pump. The average relative error of the head between the experimental and prediction obtained by the double hidden layer BP neural network model was 4.35%, the average relative error of the model prediction efficiency and the experimental efficiency was 2.94%, and the convergence time was 1/42 of that of the single hidden layer. The double hidden layer BP neural network model effectively solves the problems of low learning efficiency and easy convergence into local minima—issues that were common in the traditional single hidden layer BP neural network training. Furthermore, the proposed model realizes hydraulic performance prediction during the design process of a centrifugal pump.
3 May 2018 in Science China Physics, Mechanics & Astronomy
To analyze the interaction between wind turbines and the atmospheric boundary layer, we integrated a large-eddy simulation with an actuator line model and examined the characteristics of wind-turbine loads and wakes with reference to a corresponding experiment in Gansu. In the simulation, we set the wind turbine to have a rotor diameter of 14.8 m and a tower height of 15.4 m in the center of an atmospheric boundary layer with a 10.6° yaw angle. The results reveal an obviously skewed wake structure behind the rotor due to the thrust component normal to the flow direction. The power spectra of the inflow fluctuation velocity exhibit a region of −5/3 slope, which confirms the ability of large-eddy simulations to reproduce the energy cascade from larger to smaller scales. We found there to be more energy in the power spectrum of the axial velocity, which shows that coherent turbulence structures have more energy in the horizontal direction. By the conjoint analysis of atmospheric turbulence and windturbine loads, we found that when the inflow wind direction changes rapidly, the turbulence kinetic energy and coherent turbulence kinetic energy in the atmospheric turbulence increase, which in turn causes fluctuations in the wind turbine load. Furthermore, anisotropic atmospheric turbulence causes an asymmetric load cycle, which imposes a strike by the turbine blade on the shaft, thereby increasing the fatigue load on the shaft. Our main conclusion is that the atmospheric boundary layer has a strong effect on the evolution of the wake and the structural response of the turbine.
24 December 2024 in International Journal of Fluid Engineering
Helical axial-flow multiphase pumps are core devices for developing deep-sea oil and gas, but the complex two-phase flow inside such pumps affects their transport stability. As reported here, for enhanced flow characteristics, splitter blades were added at the noncoincidence area of the impeller tail by using a multi-objective optimization method, and their impact on the pump performance was explored in comparison with the original model. The results indicate that Gaussian process regression combined with particle swarm optimization accurately predicts the pump external characteristics. Under an inlet gas volume fraction of 30%, the optimized model increases the head coefficient by 22.2% without reducing efficiency. Although splitter blades enhance the impeller tail, they intensify gas-phase accumulation at the tail. The optimized model promotes more-uniform two-phase flow in the channels, with the axial velocity uniformity of the gas phase and liquid phase improving by 15.1% and 9.5% and the average velocity angle increasing by 12.3° and 8.3° under an inlet gas volume fraction of 30%.
With the decrease of oil and gas resources on land, increased attention has been paid to multiphase oil–gas exploitation and the transportation technology represented by oil–gas multiphase pumps. The helical–axial multiphase pump has become the focus of research on oil and gas mixed transmission technology due to its relatively high operating efficiency and adaptability to a wide range of gas volume fraction changes. In order to investigate the thickness variation in the air foil from the hub to the shroud of the blade on the mixture transportation characteristics of the gas–liquid two-phase flow in a helical–axial pump, the thickness ratio coefficient ξ was introduced, and the hydraulic performance of the single compression unit with different thickness ratio coefficients was investigated. A single compression unit including an impeller, diffuser, inlet section and outlet section of a helical–axial multiphase pump. The hydraulic performance including the hydraulic head and efficiency was investigated by numerical simulation with the Eulerian multiphase model and the shear stress transport (SST) k-w turbulence model. In order to demonstrate the validity of the numerical simulation approach, the hydraulic head and efficiency of the basic model was measured based on a gas–liquid two-phase flow pump performance test bench. The simulation results agreed well with the experimental results; the error between the simulation results and experimental results of different inlet gas volume fractions was within 10% at the design point, which indicated the numerical simulation method can be used in the research. The thickness ratio coefficient ξ, which was taken as a variable, and the aggregation degree λ of the gas were introduced to analyze the gas–liquid mixture transportation characteristics of the pump. The thickness ratio coefficient was selected in a range from 0.8 to 1.8. The results showed that, for the same hub thickness, the head coefficient and efficiency increase, and the aggregation degree of gas decreases with the decreasing of the thickness ratio coefficient. The head coefficient of the modification multiphase pump was 5.8% higher in comparison to the base pump while the efficiency was 3.1% higher than that of the base pump, the aggregation degree of this model was the lowest, which was 30.3%; the optimal model in the research was the model of scheme 1 with ξ = 0.8. The accumulation of gas in the flow passage of the impeller could be delayed to the trailing edge of the blade by adjusting the thickness ratio coefficient, which produced a super-separated airfoil for helical–axial multiphase pumps and effectively ensured reliable operation under high gas volume fraction conditions. The accumulation area of gas was consistent with the area in which the gradient of turbulent kinetic energy was large.
31 May 2018 in Science China Physics, Mechanics & Astronomy
Experiments and numerical simulations of the wake field behind a horizontal-axis wind turbine are carried out to investigate the interaction between the atmospheric boundary layer and a stand-alone wind turbine. The tested wind turbine (33 kW) has a rotor diameter of 14.8 m and hub height of 15.4 m. An anti-icing digital Sonic wind meter, an atmospheric pressure sensor, and a temperature and humidity sensor are installed in the upstream wind measurement mast. Wake velocity is measured by three US CSAT3 ultrasonic anemometers. To reflect the characteristics of the whole flow field, numerical simulations are performed through large eddy simulation (LES) and with the actuator line model. The experimental results show that the axial velocity deficit rate ranges from 32.18% to 63.22% at the three measuring points. Meanwhile, the time-frequency characteristics of the axial velocities at the left and right measuring points are different. Moreover, the average axial and lateral velocity deficit of the right measuring point is greater than that of the left measuring point. The turbulent kinetic energy (TKE) at the middle and right measuring points exhibit a periodic variation, and the vortex sheet-pass frequency is mostly similar to the rotational frequency of the rotor. However, this feature is not obvious for the left measuring point. Meanwhile, the power spectra of the vertical velocity fluctuation show the slope of −1, and those of lateral and axial velocity fluctuations show slopes of −1 and −5/3, respectively. However, the inertial subranges of axial velocity fluctuation at the left, middle, and right measuring points occur at 4, 7, and 7 Hz, respectively. The above conclusion fully illustrates the asymmetry of the left and right measuring points. The experimental data and numerical simulation results collectively indicate that the wake is deflected to the right under the influence of lateral force. Therefore, wake asymmetry can be mainly attributed to the lateral force exerted by the wind turbine on the fluid.
To analyze the interrelations among impeller blade geometry, flow field fluctuation intensity and impeller-induced hydrodynamic noise of jet centrifugal pump (JCP), a Gappy proper orthogonal decomposition (POD) method combined with computational fluid dynamics/computational fluid acoustics (CFD/CFA) technique was proposed to reconstruct and predict the unsteady flow field fluctuation intensity and flow-induced noise in impeller region. The snapshot sets were composed of blade profile parameters, flow field fluctuation intensity data and the data of sound pressure level of hydrodynamic noise in the frequency domain. Similar mesh reconstruction and flow field interpolation were carried out to have the same number of flow field data. The snapshot sets were decomposed into a linear combination of orthogonal bases using the singular value decomposition (SVD) method. The orthogonal basis coefficients corresponding to the objective variables were fitted by the least square method. The results show that the proposed method has a good accuracy in predicting the flow field fluctuation intensity and flow-induced noise of the JCP impeller domain. The relative error of pressure fluctuation intensity field is less than 4.0%, relative velocity fluctuation intensity field is less than 3.0%, turbulent kinetic energy fluctuation intensity field is less than 4.5%, and impeller-induced hydrodynamic noise is less than 10%. Taking the method as a surrogate model to predict the flow field fluctuation intensity and the radiation level of hydrodynamic noise in the optimization process of centrifugal pump impeller, it could not only reduce the calculation amount and time significantly and improve optimization speed and efficiency greatly but could also provide a reference for vibration characteristics of the models.