Theoretical And Experimental Study On Inversion Of Cryogenic Two Phase Flow Based On Multi-electrode Capacitive Sensor

The filling processes of cryogenic propellant,cryogenic air separation,hydrogen liquefaction and liquefied natural gas production all involve the transportation of cryiogenic fluids.The phase fraction and phase distribution are of great significance for the monitoring of these industrial processes,and are also one of the basic parameters for the study of heat transfer and flow characteristics of cryogenic two-phase flow.Compared with normal temperature fluids such as water and ethanol,the physical properties of cryogenic fluids,such as density,thermal conductivity and dielectric constant of liquid phases,are different by order of magnitude,which often leads to more measurement difficulties of physical quantity,higher measurement algorithms' requirements for accuracy and noise resistance,and stricter requirements for signal acquisition hardware system in resolution and accuracy.In addition,the cryogenic cavitation,the cryogenic heat pipe and other cryogenic two-phase flow and heat exchange experiments require minimizing the interference to the flow.Therefore,the ideal technical means for the measurement of phase fraction and phase distribution of cryogenic two-phase flow should have the following characteristics:1)Non-intrusive and non-invasive.Ensure that the terminal of measuring equipment does not directly form heat leakage to the cryogenic fluids and does not interfere with flow fields;2)The phase fractions and phase distributions should be measured simultaneously;3)The hardware should have high compatibility with different measurement environments,and the cost is controllable.In order to apply to industrial applications involving cryogenic fluids transportation,the hardware should have a high fault tolerance capability to the measurement environment,and the manufacturing cost should not be too high;4)Environmentally friendly and high safety.The measurement processes should not produce pollution to the ecological environment and should not damage the health of operators.In the traditional cryogenic two-phase flow measurements,it is difficult to achieve a good balance among the above four requirements.In this paper,a measurement scheme of phase fraction and phase distribution of cryogenic fluids two-phase flow based on the multi-electrode capacitive sensor is proposed.The measurement algorithms that correlate the capacitance vector with void fraction and phase distribution information are studied.In this paper,the electric field characteristics of capacitive sensor,phase fraction measurement algorithm and phase distribution inversion algorithm are studied.Main works are as follows:1)The characteristics of electric field distribution in the measurement regime of capacitive sensor are revealed.The mathematical relationship between the capacitance between pairs of electrodes and the permittivities of any elements in the two-phase flow field are obtained.When one electrode of the capacitive sensor is excited by high potential,the excited electric field lines will pass through the whole measurement area with different density,so the change of dielectric at any position in the measurement area will affects the capacitances between electrodes.By discretizing the fluid pipeline cross section into mesh elements.The first order linear relationship between the boundary capacitance variation(dCp,q)and the change of dielectric constant of different mesh elements(d?e)is obtained.And the sensitive fields(Spe,qe=dCp,q/d?e)corresponding to different electrode pairs are obtained.When the sensor forms a uniform field in the measurement area,the sensitive field strength is a constant,and the phase fraction of fluids has a linear relationship with the capacitance change at any position.As a result,the capacitances are independent of the phase distribution in the measurement area.However,this also leads to the loss of phase distribution information,thus only the phase fraction can be measured.When the electric field is not uniform in the measurement area,the sensitivity is higher at the position with larger electric field density,and the phase distribution will significantly affect the capacitance value.Consquently,it's difficult to monitor the phase fraction of cryogenic fluids in different flow patterns by the traditional concave two-electrode type capacitive sensor.2)The nonlinear mapping between the capacitance vector and the void fraction is fitted in the high dimensional space based on the least square support vector regression(LSSVR).And the fuzzy regression algorithm(FLSSVR)is developed for noisy samples.Numerical experiments preliminarily verify the feasibility of the multi-electrode capacitive sensor combined with these algorithms to measure the void fraction of cryogenic fluids.Taking LN2-VN2 as the working fluid pair,the feasibility of measuring the void fraction using multi electrode electrode sensor based on the support vector regression is verified by the numerical experiments.Six typical phase distributions are given as templates for training sample generation.The use of irregular phase distribution greatly enriches the coverage of training samples for each flow pattern.The influences of inputs of different normalization methods on the calculation results is studied.It is found that taking the linear normalized capacitance vector as input can obtain the highest accuracy calculation result of void fraction,of which the relative error is within 10%and the absolute error is within 0.02.For the noise-contaminated training data,the FLSSVR is applied.And the fuzzy membership degree is defined according to the distance from the sample point to the first fitting hyperplane in the high-dimensional space.The results of numerical experiments show that,compared with LSSVR,for FLSSVR,the average relative error and root mean square error of the proposed method are reduced by 7.41%and 9.80%respectively.3)Based on the electrical capacitance tomography(ECT)technique,the phase distributions of cryogenic fluids are obtained.An improved algorithm that the linearization error is eliminated before invesion calculation is proposed.The effectiveness of the traditional and improved algorithms for cryogenic fluids mreasurement are verified by simulations and experiments for the substituting working material pairs at room temperature.ECT uses the inversion algorithms to solve the inverse problem that calculating the dielectric distribution when the independent capacitances are known,and obtains the phase distribution information in the measurement area,which is outputted in the form of images.LN2-VN2 applied as the two-phase working fluid pair,the inversion quality and solving speed of five different phase distributions by five traditional linear algorithms and one traditional nonlinear algorithm are compared.The suitable traditional algorithms for the application of the 8-electrode capacitive sensor in the cryogenic two-phase flow inversions are determined.Furthermore,the relationship between the normalized capacitance vector and the linearization error is fitted by the LSSVR.Combining with the traditional algorithms,the linearization error is eliminated before the inversion calculations,and the LSSVR coupled inversion algorithm is obtained.In the simulations,the LSSVR coupled algorithm can eliminate the semi-convergence characteristic of the traditional iterative algorithms.The improved algorithm greatly improves the inversion quality,reduces the image error by 27.95%-63.77%,and increases the correlation coefficient by 4.63%?106.39%.An device for inversion imaging of normal temperature dielectric pairs,which is alomost same as to the dielectric characteristic of liquid methane and gaseous methane,is built The experimental results show that the LSSVR coupled algorithm can effectively reconstruct the dielectric distribution information,and also has the good performance of generalization ability.