Dynamic Thermal Model Research Of ONAN Transformer Based On Distributed Optical Fiber Sensing

With the development of science and technology,electric energy gradually occupies a central position in the entire energy system.As a key node of electric energy transmission and conversion,the importance of transformers is self-evident.During the operation of the transformer,various losses are generated in the internal winding,iron core and structural parts,which increases its temperature.Excessive temperature will bring a series of serious consequences such as insulation degradation,so real-time monitoring of internal hot spot temperature is very important for the safe and economical operation of transformers.In the existing operation and research,due to the influence of the complex electromagnetic environment inside the transformer,it is difficult to achieve accurate sensing of the winding temperature field through traditional sensors,and it is also difficult to conduct detailed research on the internal heat generation and heat dissipation characteristics of the winding.In this paper,the built-in optical sensing fiber is used together with the distributed sensing method to realize the real-time monitoring of the temperature field of the winding in the operating state.On this basis.the heat dissipation of winding is studied,and a novel hot spot temperature calculation model is proposed,and its effectiveness is verified by the actual transformer operation data on site.The main work is as follows:The key technology of compositing sensing optical fiber with transformer is studied.Using the optical fiber laying on the surface of the winding,the real-time monitoring of temperature distribution of entire length of the winding can be realized.The prototype manufacturing of 35kV transformer and 110kV transformer with built-in sensing optical fiber has been completed,and all tests of the two prototypes meet the requirements of the national standard.The key point location technology based on fiber optic sensing and the temperature-strain decoupling technology of Brillouin scattering are studied.The temperature rise test of the transformer under various operating conditions is carried out,and the winding temperature field under different load ratios is obtained.The data were analyzed from the perspectives of routine test monitoring quantity comparison,winding temperature distribution under different load rates,hot spot location,and refined temperature field.Taking the 35kV transformer prototype as a sample,the dynamic thermal model of layer winding is studied.According to the shape of the temperature curve and the rate of temperature rise,the heating process is divided into three stages:conduction period,convection period and quasi-steady state period,and the heat dissipation model corresponding to each stage is analyzed.The state in which the winding and the local oil domain reach a thermal equilibrium state is called a quasi-steady state,and this concept is used to reveal the difference in the temperature rise rate of the front and rear sections during the temperature rise of the hot spot of the winding.Based on the characteristic analysis of the quasi-steady state,the conclusion that there is a definite mapping relationship between the winding and the oil,which is independent of time and space,is verified by the DTS data.According to this mapping,the real-time inversion of the temperature of the external heat radiator to the temperature of the hot spot of the internal winding is realized,and the relative error is less than 6%.Taking the modified 110kV transformer prototype as a sample,the heat dissipation law of the disk-type winding is studied.Three typical heat exchange models are established.namely,winding itself,between winding and oil,and radiator and external environment.Based on the DTS data,the spatio-temporal variation of heat dissipation of one single disk is analyzed,the cause of the axial temperature gradient of the winding is studied,and the corresponding empirical model is established.The average convective heat transfer coefficient of the winding surface is obtained based on the measured.data,and compared with the local convective heat transfer coefficient calculated by dimensionless numbers.Taking the temperature of the first disk as the intermediate variable and the convective heat transfer coefficient as the parameter,the local heat path model of the first disk of the winding is established.Combined with the temperature gradient model,the accurate prediction of the hot spot temperature of the winding is realized.Aiming at the currently commonly used IEC load guideline model and other hot spot prediction models,the error characteristics and error sources in actual operation are analyzed.Based on the thermal circuit model,the correction factors of the top oil time constant and the winding time constant are deduced,so that they can be updated in real time according to the change of the winding load rate and the oil temperature.The more accurate steady-state temperature rise of top oil is determined by exponential fitting method to the temperature rise test data.By analyzing the cooling curve of the winding after power off,the oil temperature distribution in the oil channel is obtained,and based on this,an improved scheme of using the oil temperature at the inlet and outlet of the heat radiator to calculate the average oil temperature is proposed.Based on the above research,an improved model of the IEC loading guide model is formed.Through the test of 4 days operation data,compared with the traditional model,the top oil prediction error of this model is reduced by 17.6%,and the hot spot prediction error is reduced by 33.1%.Using this model to predict the actual operation data of two 110kV main transformers at a station in Cangzhou,Hebei Province and a station in Jiaxing,Zhejiang Province,the results show that compared with the IEC load guideline model,the improved model proposed in this paper can reduce the prediction error of top oil temperature Respectively decreased by 27.0%and 30.6%.The accuracy and better transplantability of this improved model are proved.The accuracy and better transplantability of this improved model are proved.