A process for controlling the amount of transformer winding distortion

Abstract

The application discloses a kind of process methods of controlling transformer winding deformation amount, it is related to transformer manufacturing technical field, comprising the following steps: S1 the axial force distribution under multiple working conditions is calculated and weighted superposition is obtained comprehensive axial force field function, according to this, pre-tightening force demand distribution function is constructed, according to the distribution characteristics of pre-tightening force demand distribution function, winding is divided into several pre-tightening subareas along the axial direction;S2 according to each pre-tightening subarea, the gasket of different compression stiffness is configured, and in pre-tightening force attenuation sensitive area, the self-awareness compensation gasket with insulating packaging layer is configured;S3 sequentially carries out cyclic pre-compression processing, step loading stress redistribution, thermal coupling solidification and deadweight compensation differentiation pre-tightening processing;S4 actual pre-tightening force distribution is measured by embedded sensor, deviation with target value is evaluated and adjusted;S5 is completed fastening using subarea independent fastening system.

Description

Technical Field

[0001] This invention relates to the field of transformer manufacturing technology, specifically to a process method for controlling the deformation of transformer windings. Background Technology

[0002] As one of the most important pieces of equipment in a power system, the operational reliability of power transformers directly affects the safety and stability of the power grid. Mechanical deformation of transformer windings is one of the main causes of transformer failures. Especially under the impact of short-circuit faults, the windings will be subjected to enormous electromagnetic forces, among which axial electrodynamic forces can cause axial deformation of the windings, including failure modes such as axial compression, axial displacement, and end bending.

[0003] Existing research indicates that the axial stability of transformer windings primarily depends on winding preload control and structural support design. However, traditional winding processes suffer from the following problems: First, the preload distribution is uneven. Recent studies have found that due to the influence of the winding's own weight, the actual stress of the insulation pad is often significantly lower than the design target value. For example, the design target preload stress is 3 MPa, but the actual measurement is only about 1.92 MPa. This uneven distribution of preload will lead to a deterioration in the dynamic response characteristics of the winding under short-circuit impact.

[0004] Second, the cumulative effect of insulation materials is overlooked. Transformers will experience multiple short-circuit impacts during service, and the insulation pads and wire paper insulation will undergo cumulative deformation, causing the preload to gradually loosen. However, existing processes lack a compensation mechanism for this cumulative effect.

[0005] Third, the axial vibration distribution characteristics were not fully considered. Studies have shown that the vibration of the winding at the spacer intervals exhibits a clear stepped distribution characteristic, and the vibration amplitude of the winding between two spacers is greater than that at the end clamping position, but traditional processes have not optimized the design for this characteristic.

[0006] Fourth, the ampere-turn balance of high and low voltage windings is difficult to control precisely, which can easily generate unbalanced axial electromagnetic forces and increase the risk of axial deformation.

[0007] Therefore, there is an urgent need to develop a new winding process that can effectively prevent axial deformation of transformer windings. Summary of the Invention

[0008] To solve the above-mentioned technical problems, the present invention provides an adaptive gradient preload process for transformer windings based on axial force field distribution, comprising the following steps: S1. Based on the transformer's capacity and voltage level, establish a mapping relationship between the axial force field distribution and the preload requirement. The specific process includes: S1.1 Multi-condition axial force field superposition calculation: The axial force distribution of transformer windings under different operating conditions was calculated using the finite element method. Transformer windings are subjected to various types of axial forces during actual operation.

[0009] First, under the short-circuit impact condition, the transient axial electromagnetic force distribution under the rated short-circuit current is calculated. When a short circuit occurs, the current in the winding increases sharply, generating a strong axial electromagnetic force. This force exhibits a significant concentration effect in the winding end region, with the peak force reaching 2.5 to 4 times that in the middle region.

[0010] Secondly, the self-weight condition is considered, and the gravity distribution of the windings and accessories is calculated. The weight of the winding itself creates a cumulative effect in the axial direction, with the pressure at the bottom being greater than that at the top, causing the preload to naturally decrease along the axial direction.

[0011] Next is the thermal stress condition, calculating the thermal stress distribution under rated load temperature rise. The windings undergo thermal expansion at operating temperatures, generating thermal stress due to axial constraints. This stress distribution is closely related to the temperature field distribution.

[0012] Finally, the comprehensive axial force field function is established. This formula is based on the principle of force superposition, weighting and superimposing axial forces under various operating conditions to form a comprehensive force field function. Here, α, β, and γ are the weighting coefficients for each operating condition, and z is the axial position coordinate (with the bottom of the winding as the origin). The physical meaning of the weighting coefficients is to reflect the relative importance of the influence of various operating conditions on the axial stability of the winding. Their determination adopts the Analytic Hierarchy Process (AHP) based on fault mode analysis, and the specific steps are as follows: First, statistical analysis was conducted on transformer winding axial deformation fault cases over the past 10 years to identify the occurrence frequency fi and severity Si of three fault modes: short-circuit impact, self-weight accumulation, and thermal stress cycle.

[0013] Secondly, an evaluation matrix A is established based on the consequences of the failure (the degree of equipment damage D, the downtime loss L, and the safety risk R), where the element aij represents the importance ratio of the i-th type of failure to the j-th type of failure.

[0014] Then, the eigenvalue method was used to solve for the largest eigenvalue of the evaluation matrix A. And its corresponding eigenvector W, after normalization, yields the weight vector. .

[0015] Finally, calculate the consistency ratio. ,in RI is the average random consistency index. CR < 0.1 is required; otherwise, the evaluation matrix needs to be adjusted.

[0016] For transformers with voltage levels of 110kV and above, based on the analysis of numerous fault cases, we recommend α=0.55~0.65, β=0.20~0.25, and γ=0.15~0.20. Among these, the short-circuit impact condition has the largest weight, reflecting that short-circuit faults are the primary cause of axial deformation of the windings.

[0017] S1.2 Construction of the preload demand distribution function: Based on the axial force field distribution, construct the preload demand distribution function. Ks is the safety factor, ranging from 1.2 to 1.5. This factor is used to compensate for uncertainties in the calculation model, material property dispersion, and the conservatism of superimposed operating conditions, ensuring sufficient safety margin even under the most unfavorable combination of operating conditions. A(z) is the bearing area at this axial position, i.e., the effective contact area between the winding and the insulating pad at this position. Since the winding structure may vary along the axial direction (such as the end insulation reinforcement area), the bearing area is a function of position. This is the base preload, used to compensate for initial creep of the insulation material and assembly gaps, ensuring that the windings maintain basic axial constraint when there is no external load. A typical value is 15-25% of the design preload.

[0018] The physical meaning of this formula is to convert the axial force field into pressure demand, and then superimpose the foundation preload to form a preload demand curve that is non-uniformly distributed along the axial direction.

[0019] S1.3 Non-uniform partitioning strategy determined: according to Based on the distribution characteristics, the winding is divided into N preload zones (N≥5) along the axial direction. The target preload Pi of each zone is obtained by integrating and averaging the preload demand function within the zone. , where Li is the axial length of the i-th partition.

[0020] The zoning principle has two points. First, areas where the preload demand change rate is greater than 15% per 100mm are separated into separate zones to ensure precise control of the preload in those areas. Second, the difference in target preload between adjacent zones should not exceed 20% to avoid stress abrupt changes at the zone boundaries.

[0021] S2, Design an insulating pad system with an axial mechanical property gradient distribution, including variable stiffness pads, self-sensing compensation pads, and their spatial layout optimization. Specific steps include: S2.1 Variable stiffness pad design: Based on the preload requirements of each zone determined in step S1, three types of pads with different compressive stiffness are designed: High-rigidity pads for high preload zones employ a multi-layer composite structure, consisting of 3-5 layers of laminated cardboard of varying densities stacked alternately, with an overall compression modulus of 2500-3500 MPa. The density of each layer increases from the outside in, creating a stiffness gradient and preventing stress concentration. The design principle utilizes this gradient structure to achieve layer-by-layer stress transfer and dispersion, thereby improving the pad's load-bearing capacity and fatigue resistance.

[0022] Medium-stiffness pads for medium preload zones are made of homogeneous high-density laminated paperboard with a compression modulus of 1800~2500MPa, meeting medium preload requirements while maintaining good processability.

[0023] And flexible buffer pads for the transition zone, which adopt a composite structure of a main layer and a buffer layer. The main layer accounts for 60-70% of the thickness, and the buffer layer is made of modified silicone rubber or polyurethane elastomer, with an overall compression modulus of 800-1500 MPa. The function of the buffer layer is to absorb vibration energy and reduce stress transmission between adjacent zones.

[0024] S2.1 features a self-sensing compensation pad design with an insulating enclosure: In the preload decay sensitive area, specifically the top 30% of the winding, a self-sensing compensation pad is installed. This pad adopts a five-layer composite structure, from top to bottom: a pressure-bearing main body layer, providing the main load-bearing capacity and structural stiffness; an upper insulating encapsulation layer, used to isolate the SMA conductor; an SMA driving layer, using NiTi shape memory alloy sheets, which generates recovery stress upon temperature triggering; a lower insulating encapsulation layer, also used to isolate the SMA conductor; and an elastic energy storage layer, which stores and transmits the recovery force generated by the SMA. A single self-sensing compensation pad can generate an additional preload of 0.3~0.8MPa under temperature triggering conditions, with a compensation amount of 10~25% of the initial preload.

[0025] The main principle is that the SMA driving layer uses NiTi alloy sheets, utilizing the thermoelastic martensitic phase transformation characteristics of shape memory alloys to achieve active compensation. When the ambient temperature is below the phase transformation temperature, the SMA is in the martensitic phase, and the material is soft and easily deformable. When the temperature rises to above the phase transformation temperature (set to be 15~25℃ higher than the normal operating temperature of the winding), the SMA undergoes a martensite → austenite phase transformation, and the crystal structure rearranges to generate restoring stress. This restoring stress is transmitted through the elastic energy storage layer, actively compensating for the preload loss caused by the temperature rise.

[0026] Since NiTi alloy is a conductive material, direct embedding into the insulating system may lead to local electric field distortion and discharge risks. Therefore, two additional insulating encapsulation layers are provided to ensure electrical isolation from the surrounding insulating system.

[0027] Furthermore, to avoid false triggering or delayed triggering of the SMA, a graded phase change temperature design is adopted based on the characteristics of the winding axial temperature field distribution. First, computational fluid dynamics (CFD) is used to calculate the three-dimensional steady-state temperature field distribution of the winding under rated load to determine the hot spot temperature at each axial position. Secondly, the top 30% area is divided into 2-3 temperature zones based on the hot spot temperature distribution, with each zone configured with a different phase change temperature. SMA, such as the topmost region ; middle area Transition area The principle for setting the lower limit of the phase change temperature is to ensure that no false triggering occurs under the highest operating temperature condition (including overload), and the temperature margin is ≥10℃.

[0028] It is worth noting that the focus of this invention is on establishing an adaptive gradient pre-tightening process based on axial force field distribution, rather than the material preparation of the self-sensing compensation pad itself. Furthermore, shape memory alloys are a mature existing technology with decades of research and application history. NiTi alloys with specific phase transformation temperatures can be directly purchased commercially or obtained through methods known in the art, such as composition adjustment and heat treatment. Those skilled in the art can prepare a self-sensing compensation pad that meets the requirements without creative effort, based on the structural composition, layer function, thickness ratio, and material type disclosed in this specification, combined with common knowledge in the field. Therefore, the specification does not elaborate further on the preparation of the pad itself.

[0029] Furthermore, the pads are arranged in a non-uniform manner along the circumference. Specifically, the radial vibration node position of the winding is determined by finite element modal analysis, and the pads are densely arranged at this position to enhance the axial constraint at that location; the pads are appropriately sparsely arranged at the radial vibration antinode position to avoid stress concentration and insulation damage caused by excessive constraint.

[0030] S3 employs a multi-stage gradient pre-tightening strategy based on the viscoelastic properties of the material. Through a specific loading history, it fully activates the strain-strengthening effect of the insulating material, improving pre-tightening efficiency and long-term retention rate. This process consists of four stages: S4.1, Under normal temperature conditions, apply cyclic pre-compression treatment to the winding. The treatment includes cyclic loading with a triangular wave at a loading rate of 2~5MPa / min, cycling between 0.5MPa and 60% of the design preload, for a total of 20~50 cycles.

[0031] The purpose of this step is to address the issue that, due to the porous fibrous structure of insulating materials such as laminated cardboard, there are initially loose gaps between the fibers. Through cyclic loading, the fibers are gradually compacted, generating a frictional locking effect, and the material's microstructure tends to stabilize. This process eliminates the initial loose structure of the material, increases the initial compression modulus by 30-50%, and lays the foundation for subsequent pre-tightening treatment.

[0032] S4.2 employs a four-stage step loading method to achieve a gradual redistribution of stress. Specifically: Stage 1 involves rapid loading to 50% of the design preload at a rate of 10-15 MPa / min, held for 30-60 minutes, allowing initial creep of the material and releasing localized stress concentrations; Stage 2 continues loading to 75% of the design preload, held for 60-90 minutes, monitoring the compression differences in each zone and assessing the uniformity of stress distribution; Stage 3 loads to 90% of the design preload, held for 2-3 hours; Stage 4 loads to 105-115% of the design preload, held for 1-2 hours, and then unloads to 100%.

[0033] The fourth level of overload utilizes the viscoelastic hysteresis effect of the insulating material. When the material is unloaded after undergoing overload, due to viscoelastic hysteresis, the residual stress (i.e., preload) is higher than when directly loaded to the target value. This effect can increase the long-term retention rate of the preload by 15-25%.

[0034] S4.3 Heat treatment is performed while maintaining the final preload of the second stage to achieve thermal curing of the insulation material. In this process, the temperature is first increased to 80°C at a rate of 1-2°C / min and held for 4-6 hours to fully relax residual stress in the material. Then, the temperature is increased to the target temperature (typically 105-130°C) at a rate of 0.5-1°C / min and held for the specified time to complete the cross-linking and curing of the resin. During heat treatment, the preload will fluctuate due to differences in the thermal expansion coefficients of the materials. The preload fluctuation is maintained within ±3% by real-time monitoring and adjustment of the hydraulic system pressure. Subsequently, the material is slowly cooled to room temperature at a rate not exceeding 1°C / min to avoid sudden changes in preload or cracking of the insulation material due to thermal stress. Finally, self-weight compensated differential preload is performed. This mainly includes establishing a self-weight compensated preload model to address the axial gradient attenuation of the preload caused by the winding's own weight. ,in The design preload is the target value; H is the total height of the winding; z is the height from the bottom of the winding; k is the compensation coefficient, with a value of 0.08~0.15.

[0035] This formula reflects the compensation strategy where the pre-tightening force increases linearly along the axial direction from the bottom to the top. The physical meaning is that the bottom of the winding bears the additional compression of the self-weight of the upper winding, and the actual pre-tightening force is higher than the applied value; while at the top, there is no additional compression and the pre-tightening force is equal to the applied value. To make the final effective pre-tightening force evenly distributed along the axial direction, a higher initial pre-tightening force needs to be applied to the top. The compensation coefficient k is determined by the ratio of the weight per unit height of the winding to the designed pre-tightening force.

[0036] S4 measures the actual pre-tightening force distribution through an embedded sensor, evaluates the deviation from the target value, and makes adjustments: In this step, it is preferred to use a fiber Bragg grating (FBG) sensing system to achieve in-situ measurement, deviation evaluation, and adaptive adjustment of the pre-tightening force distribution. At the key axial positions of the winding, mainly the center positions of each pre-tightening zone and the junctions between zones, FBG pressure / temperature composite sensors are configured. The FBG sensor utilizes the reflection characteristics of specific wavelength light by the periodic refractive index modulation structure in the fiber core. When the grating undergoes strain or temperature changes, the reflected wavelength drifts, and the strain and temperature changes can be deduced by detecting the wavelength drift amount. Compared with traditional resistive or capacitive pressure sensors, the FBG sensor is a fully dielectric structure, which can eliminate the influence of conductive elements on the insulation system; has excellent anti-electromagnetic interference ability and is not affected by the operating electromagnetic field of the transformer; multiple gratings can be connected in series on a single fiber to achieve distributed measurement, reducing the number of leads; has excellent long-term stability and high service life.

[0037] The specific content of the adjustment includes: measuring the actual pre-tightening force at each measuring point , calculating the deviation from the target value , if all measuring points < 5%, it is judged as qualified, and the pre-tightening force distribution meets the design requirements; if there are measuring points where 5% ≤ < 10%, the local pre-tightening force of this zone is finely adjusted and then复测; if there are measuring points ≥ 10%, it is necessary to re-perform the stepped pre-tightening process in step S3.

[0038] Furthermore, based on the measurement data and the material creep model, a long-term attenuation prediction model of the pre-tightening force is established , this model is a two-mechanism saturated creep model, comprehensively considering the two mechanisms of transient creep and steady-state creep of the insulating material, and ensuring the physical rationality of long-term prediction. Among them, P0 is the initial pre-tightening force, and t is the operating time. The first term describes the transient creep process, based on the Kelvin-Voigt viscoelastic model, is the transient creep time constant, usually dozens to hundreds of hours, A is the transient attenuation coefficient (taking values of 0.05 - 0.15), and this term reflects the rapid stress relaxation caused by the rearrangement of molecular chain segments and the slip between fibers at the initial stage of loading. The second term The steady-state creep process is described using a modified Kohlrausch-Williams-Watts (KWW) relaxation function. The steady-state creep characteristic time is typically several thousand to tens of thousands of hours, B is the steady-state decay coefficient (value 0.05~0.20), and n is the creep exponent (value 0.1~0.35).

[0039] The creep index n is a key parameter characterizing the creep properties of a material. For laminated paperboard insulation materials used in transformers, n is mainly affected by the following factors: At the material level, n depends on the orientation and density of cellulose fibers, the degree of resin impregnation, and porosity, with a typical value range of 0.10~0.35; At the temperature level, creep is a thermally activated process, and increased temperature enhances the mobility of molecular chain segments, resulting in a corresponding increase in the n value. At room temperature, n is approximately 0.15~0.20, increasing to 0.20~0 at operating temperature (80℃). The value of n is 0.28, and it can reach 0.25~0.35 at the hot spot temperature (105℃). In terms of stress, n is small in the low stress range and the creep is approximately linear, while n increases and exhibits nonlinear characteristics in the high stress range. In terms of loading history, after cyclic pre-compression and overload pre-tightening treatment, the microstructure of the material tends to be stable, and frictional locking is formed between fibers. The value of n can be reduced from 0.25~0.35 in the untreated state to 0.12~0.20. This is the theoretical basis for the strain strengthening process of this invention to reduce the long-term creep rate.

[0040] The advantage of using a saturation model is that when At this time, the preload tends to a stable value. Instead of decaying indefinitely to zero or negative values, this aligns with the actual physical behavior of creep in insulating materials eventually stabilizing. For insulating pads processed using the method of this invention, typical parameter values ​​are A=0.08~0.12, B=0.08~0.15, n=0.15~0.25. =200~500 hours =8000~15000 hours. Within this parameter range, the model predicts that the preload retention rate after 30 years is P(30 years) / P0=(1-AB)≈75%~84%, which is in good agreement with the extrapolation results of accelerated aging tests.

[0041] Based on the prediction results, the SMA phase transition temperature of the self-sensing compensation pad is calibrated and adjusted before the transformer leaves the factory. When the model predicts that the preload decays to 85%~90% of the design value (corresponding to the trigger threshold of the SMA design compensation amount), it ensures that the winding temperature reaches the SMA phase transition temperature at this time, so as to achieve precise timing matching between preload decay and active compensation.

[0042] S5, Perform winding assembly and final tightening, including the following: An independent axial fastening system is adopted, and the traditional integral pressure plate structure is changed to a partitioned pressure plate structure. Each pre-tightening partition corresponds to an independent pressure plate unit, and radial guide grooves are set on the surface of the pressure plate to ensure smooth flow of transformer oil in the gaps between the partitioned pressure plates without affecting the winding heat dissipation. The partitioned pressure plates are flexibly connected to adapt to the different pre-tightening force requirements of each partition.

[0043] The fastening system uses servo hydraulic or electric cylinders as fastening actuators. Each actuator is equipped with high-precision force and displacement sensors. Force control ensures the accuracy of the preload, while displacement control ensures the consistency of the compression. During fastening, the zone with the highest preload requirement is tightened first to establish the main axial constraint. Then, the zones are tightened sequentially from high-demand zones to low-demand zones. Finally, overall equalization and fine-tuning are performed to eliminate mutual interference between zones during the fastening process.

[0044] Furthermore, it also includes S6, a reserved interface for monitoring preload conditions during operation, the main contents of which include: The FBG pressure sensor used for verification in S4 is retained as a permanent monitoring element inside the winding. The optical fiber is led out of the oil tank through a dedicated sealed wall-penetrating connector and connected to an external demodulator. A standardized monitoring data interface is reserved, supporting 4-20mA analog signal output and RS485 / Modbus digital communication protocol, allowing connection to a transformer online monitoring system or an independent preload monitoring device.

[0045] The software provided is for assessing the preload condition and has the following functions: real-time display of preload values ​​and historical trend curves at each measuring point; curve fitting based on the preload decay model and measured data to predict future preload change trends; assessment of the short-circuit withstand capability margin of the winding based on the degree of preload decay; and automatic warning when the preload at any measuring point decays to below 85% of the design value, prompting maintenance personnel to pay attention.

[0046] Compared with the prior art, the present invention has the following beneficial effects: 1. By establishing a comprehensive axial force field function and a preload demand distribution function, a quantitative match between the preload distribution and actual stress requirements is achieved. The weighting coefficients are determined using an analytic hierarchy process based on failure mode analysis, making the force field calculation results closer to engineering realities. Compared to traditional uniform preload processes, the safety margin in critical stress areas is significantly improved, while avoiding the risk of insulation damage caused by excessive preload in low-stress areas.

[0047] 2. The cyclic pre-compression stage involves 20-50 cycles of loading to activate the frictional locking effect between the fibers of the insulating material, increasing the initial compressive modulus of the material by 30-50%. The overload treatment in the step loading stage utilizes the viscoelastic hysteresis effect to significantly improve the long-term retention rate of the preload by 15-25%.

[0048] 3. The self-sensing compensation pad uses a NiTi shape memory alloy driving layer, which actively compensates for preload loss by utilizing the restoring stress generated by the martensitic-austenitic phase transformation. Compared with traditional passive elastic compensation elements such as disc springs, it has improved response speed and compensation accuracy, requires no external energy input, and the graded phase transformation temperature design ensures on-demand triggering at different axial positions, with a false triggering margin of ≥10℃.

[0049] Based on the above technological improvements, the process of this invention can reduce the risk of unplanned outages of transformers due to winding preload failure by more than 70%. Even with the increased process cost, it still brings significant economic benefits. Simultaneously, the real-time preload monitoring function can provide equipment health status information for power grid dispatching, improving the reliability of power grid operation.

[0050] The technical solution of the present invention will be further described in detail below through embodiments. Attached Figure Description

[0051] Figure 1 This is a flowchart of the process method of the present invention; Figure 2 This is a diagram showing the axial force field distribution curves of the winding under multiple operating conditions in Example 1; Figure 3 This is a process curve diagram of the multi-stage gradient pre-tightening process in Example 1. Detailed Implementation

[0052] To enable those skilled in the art to better understand the present application, the technical solutions in specific embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by those skilled in the art.

[0053] Example 1 This embodiment applies to a 110kV / 50MVA three-phase oil-immersed power transformer with connection group YNd11, high-voltage winding height 1200mm, high-voltage winding turns 786, short-circuit impedance 10.5%, design preload 2.5MPa, and normal operating hot spot temperature of the winding 78℃.

[0054] S1: An electromagnetic-structural coupled finite element model of the transformer was established using ANSYS Maxwell software to calculate the axial force distribution under various operating conditions. For the short-circuit impact condition, the peak value of the non-periodic component of the rated short-circuit current, approximately 12.5 times the rated current, was used to calculate the axial electromagnetic force distribution. It exhibits end-concentration characteristics, with peak force densities of 4.2 N / mm² at the top (z=1200 mm) and bottom (z=0). 2 and 3.8N / mm 2The value in the middle section (z=600mm) is 1.1 N / mm. 2 The cumulative gravity distribution was obtained from the calculation of the self-weight condition. The weight per unit height of the winding is approximately 0.85 kg / mm. Under thermal stress conditions with a rated load temperature rise of 65 K, the thermal stress distribution is as follows: Peak value approximately 0.3 N / mm 2 .like Figure 2 As shown, the solid line represents the distribution of short-circuit electromagnetic force. The long dashed line represents the weight distribution. The dotted lines represent the distribution of thermal stress. The dotted lines represent the combined axial force field. .

[0055] Based on nearly 10 years of fault statistics for 110kV transformers, an evaluation matrix was constructed using the analytic hierarchy process (AHP). Short-circuit impulse faults accounted for approximately 65% ​​and had the most severe consequences, often leading to permanent winding deformation. Weight accumulation faults accounted for approximately 22%, and thermal stress cycle faults accounted for approximately 13%. After calculation using the eigenvalue method and passing the consistency test (CR=0.032<0.1), the weighting coefficients were determined to be α=0.60, β=0.25, and γ=0.15.

[0056] The axial forces under each working condition are superimposed according to their weights to obtain the comprehensive axial force field function. Taking a safety factor of Ks = 1.3 and a bearing area of ​​A(z) = 28500 mm², 2 This is the effective contact area of ​​the winding and the foundation preload. =0.5MPa, the preload demand distribution function is calculated. Based on the rate of change of preload demand, the winding is divided into 6 preload zones from top to bottom. The axial range, target preload, and pad type for each zone are shown in the table below.

[0057] S2: The high-rigidity pad uses a 4-layer composite structure, with layers of density 1.1g / cm³ from the outside in. 3 1.25g / cm 3 1.35g / cm 3 1.25g / cm 3 The laminated cardboard has a single-layer thickness of 2mm and a total thickness of 8mm, with a tested compression modulus of 3150MPa. The medium-stiffness pads use materials with a density of 1.3g / cm³. 3 The homogeneous laminated cardboard is 6mm thick and has a tested compression modulus of 2180MPa. The main layer of the flexible cushioning pad uses a material with a density of 1.2g / cm³. 3 The laminated cardboard is 4mm thick, the cushioning layer is made of 2mm thick silicone rubber with a Shore hardness of 65A, and the overall compression modulus is 1050MPa.

[0058] In Zones I and II, in the top 30% area, self-sensing compensation pads are configured in an axial range of 840~1200mm. The total thickness of the pads is 10mm, and the thickness distribution of the five-layer structure is as follows: pressure-bearing main body layer 4.5mm (accounting for 45%), upper insulating encapsulation layer 0.6mm (accounting for 6%), SMA drive layer 1.5mm (accounting for 15%), lower insulating encapsulation layer 0.6mm (accounting for 6%), and elastic energy storage layer 2.8mm (accounting for 28%).

[0059] The SMA drive layer uses Ti-50.8at%Ni alloy sheets, and the phase transition temperature is adjusted through heat treatment. Based on CFD temperature field simulation results, the hot spot temperatures at the top of the winding are 78℃ (z=1200mm), 72℃ (z=1080mm), and 68℃ (z=840mm). The SMA phase transition temperature for section I is set to 98℃ (=78+20), and for section II it is set to 88℃ (=72+16). The insulating encapsulation layer uses an epoxy composite material with 40wt% Al2O3 filling.

[0060] S3: Under 25℃ conditions, a triangular wave loading mode was used, with a loading rate of 3MPa / min and a pressure ranging from 0.5MPa to 1.5MPa (60% of the design preload of 2.5MPa) for 30 cycles. After the cycles were completed, the compressive modulus of the test block increased by approximately 42% compared to the initial state.

[0061] Subsequently, the pressure was increased to 1.25 MPa (50%) at 12 MPa / min and held for 45 minutes; then increased to 1.875 MPa (75%) and held for 75 minutes, during which the compression deviation of each zone was monitored and found to be <0.15 mm; the pressure was increased to 2.25 MPa (90%) and held for 2.5 hours, with fine-tuning of the compression of each zone; finally, the pressure was increased to 2.875 MPa (115%) and held for 1.5 hours, then unloaded to 2.5 MPa (100%). Maintaining a preload of 2.5 MPa, the temperature was increased to 80℃ at 1.5℃ / min and held for 5 hours (pre-curing), then increased to 120℃ at 0.8℃ / min and held for 8 hours (primary curing). During the heat treatment process, real-time compensation was achieved using a hydraulic servo system, and the preload fluctuation was controlled within ±2.5%. The temperature was then cooled to room temperature at 0.8℃ / min. The overall process parameters are as follows: Figure 3 As shown, the horizontal axis represents time, the left vertical axis represents preload, and the right vertical axis represents temperature. The solid line reflects the change of preload over time, and the dotted line represents the change of temperature over time.

[0062] Take the compensation coefficient k=0.10, according to the formula The target preload after compensation for each zone was calculated: Zone I, with a center value z=1140mm, was calculated to be 2.74MPa; Zone II, with z=960mm, was calculated to be 2.70MPa; Zone III, with z=720mm, was calculated to be 2.65MPa; Zone IV, with z=480mm, was calculated to be 2.60MPa; Zone V, with z=240mm, was calculated to be 2.55MPa; and Zone VI, with z=60mm, was calculated to be 2.51MPa. Differential preload was applied to the top, upper-middle, lower-middle, and bottom zones respectively using four independent hydraulic cylinders.

[0063] S4: One FBG pressure sensor is placed at the center of each of the six zones, for a total of six measuring points. The sensors use polyimide-coated optical fibers with a grating length of 10 mm, a center wavelength interval of 5 nm, a measurement range of 0~10 MPa, and an accuracy of ±0.08 MPa (±0.8%FS). Measurement results show that the deviation ε between the actual preload and the target value at each measuring point is <4.2%, meeting the qualification criterion of ε<5%.

[0064] The parameters of the creep model fitted based on the measurement data are: A=0.10, B=0.12, n=0.18. =350 hours =12,000 hours. The model predicts that the preload retention rate will be 78.5% after 30 years, which is higher than the design lower limit of 75%.

[0065] S5-S6: Six partitioned pressure plates are used, each with radial guide grooves machined on its surface (groove depth 4mm, groove width 10mm, spacing 40mm). Adjacent pressure plates are connected by 0.5mm thick stainless steel corrugated sheets. The tightening sequence is: I→VI→II→V→III→IV, with final fine-tuning. The FBG sensor is led out through a fiber optic sealed wall-penetrator on the top of the tank and connected to an 8-channel FBG demodulator. Data is accessed to the substation integrated automation system via the Modbus protocol.

[0066] After the transformer in this embodiment was assembled, the following verification tests were conducted: Under a 1.5µm power frequency voltage, the partial discharge in the self-sensing pad area is 8.5pC, meeting the requirement of ≤10pC; the temperature rise of the winding hot spot under rated load is 63.5K, which is only 0.7K higher than that of the same model of traditional process product (62.8K); in the three-phase short circuit test according to GB 1094.5, the maximum axial displacement of the winding is 2.3mm, which is less than the design allowable value of 5mm, and the impedance change rate after short circuit is 0.8%; after aging at 130℃ for 1000 hours, the preload retention rate is 94.2%, while that of the traditional process control group is 81.5%.

[0067] Example 2 This embodiment applies to a 220kV / 180MVA three-phase, three-winding oil-immersed power transformer. The high-voltage winding height is 1850mm, the total winding weight is approximately 4200kg, the design preload is 3.2MPa, and the normal operating hot spot temperature is 85℃. Due to its large capacity and high winding height, the transformer's self-weight and short-circuit impact force are significantly greater than those in Embodiment 1.

[0068] The axial force distribution was calculated using the same finite element method as in Example 1. Due to the larger short-circuit capacity of the 220kV level, the peak short-circuit electromagnetic force at the end (z=1850) reached 6.8 N / mm. 2 The middle part is 1.5 N / mm 2 The weighting coefficients were determined based on statistical data of 220kV level faults as α=0.62, β=0.23, and γ=0.15.

[0069] Based on the distribution of preload requirements, the winding is divided into 8 preload zones, with finer divisions to accommodate taller windings and larger preload gradients. The top three zones, z=1295~1850mm, accounting for 30%, are equipped with self-sensing compensation pads.

[0070] Due to the increased winding height, the temperature field gradient is more pronounced within the top 30% range. CFD simulation shows the hot spot temperature distribution as follows: 85℃ at z=1850mm, 79℃ at z=1665mm, 74℃ at z=1480mm, and 70℃ at z=1295mm.

[0071] The transformer is divided into three zones from top to bottom: Zone A has an axial range of 1665~1850mm, a hot spot temperature of 82~85℃, and an SMA phase transition temperature of 107℃; Zone B has an axial range of 1480~1665mm, a hot spot temperature of 76~79℃, and an SMA phase transition temperature of 96℃; Zone C has an axial range of 1295~1480mm, a hot spot temperature of 70~74℃, and an SMA phase transition temperature of 86℃. When the transformer is overloaded, i.e., the hot spot temperature rises above 95℃, the SMA in Zone A will trigger compensation first; when the hot spot temperature further rises above 88℃, Zone B will trigger; during normal operation, all SMAs will not trigger.

[0072] Furthermore, due to the higher design preload, the preload processing parameters were adjusted accordingly. The cyclic precompression was adjusted to a pressure range of 0.5~1.92MPa (60%), with 40 cycles. The gradual loading stages were 1.6MPa→2.4MPa→2.88MPa→3.68MPa (115%), eventually stabilizing at 3.2MPa. The self-weight compensation was set to k=0.12, the top preload was 3.58MPa, which was 12% higher than the bottom, and the main curing temperature was increased to 125℃ and maintained for 10 hours.

[0073] Eight partitioned pressure plates are used, equipped with six independent servo hydraulic cylinders. The pressure plate guide grooves have a depth of 5mm, a width of 12mm, and a spacing of 35mm. CFD simulation verification shows that the improved oil flow distribution is uniform, and the hot spot temperature rise only increases by 1.2K.

[0074] Considering the importance of the 220kV transformer, a denser sensor network was configured, with one measuring point in each of the eight pre-tightened zones, plus one redundant measuring point each at the top and bottom, for a total of 10 FBG sensors. Wavelength division multiplexing technology was employed, with the sensors led out in series from two optical fibers.

[0075] After the transformer in this embodiment was assembled, the following verification tests were conducted: In the short-circuit dynamic stability test, the maximum axial displacement of the winding was 3.8 mm, which is less than the allowable value of 8 mm, and the impedance change rate after the short circuit was 1.1%.

[0076] The winding was manually heated to 110°C, and the SMA in area A was successfully triggered. The additional preload was measured to be 0.52 MPa, with a compensation rate of 16.3%.

[0077] After 2000 hours of accelerated aging, the preload retention rate was 92.8%, compared to 78.2% for the control group using traditional processes.

[0078] Example 3 This embodiment applies to a step-up transformer in a photovoltaic power plant. Unlike conventional grid transformers, photovoltaic step-up transformers face the following unique operating conditions: drastic daily load fluctuations; lower annual operating hours but more frequent start-ups and shutdowns; and increased risk of thermal stress fatigue due to frequent temperature cycling. The transformer parameters are 35kV / 20MVA, high-voltage winding height 850mm, design preload 2.0MPa, and operating hot spot temperature range 25~72℃.

[0079] Considering the characteristics of photovoltaic step-up transformer operation, the weighting coefficients were reassessed. Because the transformer's short-circuit current is relatively small but thermal cycling is frequent, the thermal stress weighting needs to be increased. The analytic hierarchy process (AHP) was used to recalculate the weighting coefficients, and the coefficients were adjusted to α=0.45, β=0.20, and γ=0.35.

[0080] To address frequent temperature cycling conditions, the pad material and structure were specifically designed. The buffer layer of the flexible buffer pad was upgraded from ordinary silicone rubber to fluorosilicone rubber, resulting in a lower coefficient of thermal expansion and superior resistance to heat aging. The multi-layer composite pad uses a silane coupling agent between each layer to improve interfacial bonding strength and prevent delamination caused by temperature cycling. The total pad thickness was increased from 8mm to 10mm, and the proportion of the elastic energy storage layer increased from 28% to 35%, improving its adaptability to thermal expansion and deformation.

[0081] Because photovoltaic step-up transformers experience large temperature fluctuations and need to activate compensation even at lower temperatures, the SMA phase change temperature design employs a two-way shape memory alloy strategy. The forward phase change (heat-triggered) temperature is set at 85℃, triggered under high load conditions to generate a positive additional preload. The reverse phase change (cooling-triggered) utilizes a trained two-way shape memory alloy, generating a small amount of reverse restoring force when cooled to below 40℃ to compensate for the increased preload caused by low-temperature shrinkage and prevent over-compression of the insulation.

[0082] The bidirectional shape memory effect can be achieved by thermomechanically training NiTi alloys, enabling them to undergo shape changes during cooling. The forward recovery stress is 0.4 MPa, and the reverse recovery stress is 0.15 MPa.

[0083] The windings were divided into 5 pre-tightening zones. Considering that the photovoltaic transformer experiences one temperature cycle per day, the design life of 25 years corresponds to approximately 9000 temperature cycles. The pad material underwent 9000 cycles of fatigue testing at -20℃ to +80℃, verifying that the compressive modulus attenuation was <8%, meeting the requirements for long-term use.

[0084] Due to the accelerated creep caused by frequent temperature cycling, the model parameters need to be adjusted. Based on the accelerated aging test data, the fitting values ​​were: A = 0.12 (slightly higher than 0.10 in Example 1), B = 0.15 (higher than 0.12 in Example 1), and n = 0.22 (higher than 0.18 in Example 1). =280 hours =9500 hours. The model predicts that the preload retention rate will be 73.5% after 25 years, close to the design lower limit of 70%.

[0085] Considering the economic requirements of 35kV transformers, a simplified monitoring scheme is adopted, with only one FBG sensor at the top and one at the bottom, and the state of the middle area is inferred by the pressure difference between the top and bottom.

[0086] After the transformer in this embodiment was assembled, the following verification tests were conducted: The preload was maintained at 96.8% after 1000 cycles of temperature cycling from -10℃ to +75℃ under simulated photovoltaic conditions (equivalent to about 3 years of operation).

[0087] When heated to 90℃, the positive compensation force is 0.38MPa; when cooled to 35℃, the reverse compensation force is 0.13MPa, which effectively suppresses the preload fluctuation caused by temperature cycling.

[0088] Finally, it should be noted that the described embodiments are merely some, not all, of the embodiments of the present invention. Those skilled in the art will understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents; that is, all other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

Claims

1. A process for controlling the amount of transformer winding distortion, characterized by, Includes the following steps: S1: Calculate the axial force distribution under various working conditions and weighted superposition to obtain the comprehensive axial force field function. Based on this, construct the preload demand distribution function. According to the distribution characteristics of the preload demand distribution function, divide the winding into several preload zones along the axial direction. S2: Configure pads with different compression stiffness according to the preload requirements of each preload zone, and configure self-sensing compensation pads with insulating encapsulation layers in the preload attenuation sensitive area. S3: Perform cyclic pre-compression treatment, step loading stress redistribution, thermo-coupling curing and self-weight compensation differential pre-tightening treatment in sequence; S4: Measure the actual preload distribution using embedded sensors, assess the deviation from the target value, and make adjustments accordingly; S5: Secure the device using a zoned independent fastening system.

2. The process method according to claim 1, characterized in that, In S1, the comprehensive axial force field function is: ,in The distribution of axial electromagnetic force under short-circuit conditions. For gravity distribution, The distribution of thermal stress is represented by z, which is the axial position coordinate. The weighting coefficients α, β, and γ are determined by the analytic hierarchy process based on fault mode analysis. By statistically analyzing axial deformation fault cases of transformer windings, the importance evaluation matrix of each fault mode under each working condition is established. The weight vector is obtained by solving the matrix using the eigenvalue method and performing a consistency check.

3. The process method according to claim 2, characterized in that, In S1, the preload demand distribution function is: in For safety factor; A(z) is the bearing area at this axial position; Based on the preload; according to The distribution characteristics divide the winding into N preloaded sections along the axial direction, where N≥5.

4. The process method according to claim 1, characterized in that, The S2 includes the following different compression stiffness pads: high stiffness pads, which are multi-layer composite structures made of 3 to 5 layers of laminated paperboard with different densities, with the density of each layer increasing from the outside to the inside to form a stiffness gradient, and an overall compression modulus of 2500 to 3500 MPa; medium stiffness pads, which are made of homogeneous high-density laminated paperboard, with a compression modulus of 1800 to 2500 MPa; and flexible buffer pads, which are composite structures of a main layer and a buffer layer, with the buffer layer being modified silicone rubber or polyurethane elastomer, and an overall compression modulus of 800 to 1500 MPa.

5. The process method according to claim 1, characterized in that, The self-sensing compensation pad in S2 is arranged in the range of 30% of the winding top, adopts a five-layer composite structure, and sequentially comprises a pressure-bearing main body layer, an upper insulation packaging layer, a shape memory alloy driving layer, a lower insulation packaging layer, and an elastic energy storage layer; the shape memory alloy driving layer adopts a NiTi alloy sheet; the upper insulation packaging layer and the lower insulation packaging layer have a volume resistivity ≥10 14 Ω·cm and a dielectric strength ≥25 kV / mm, and electrically isolate the shape memory alloy driving layer from the surrounding insulation system.

6. The process method according to claim 5, characterized in that, The shape memory alloy drive layer employs a graded phase transition temperature design: the axial hot spot temperature distribution of the winding is obtained through computational fluid dynamics. The top 30% area is divided into 2-3 temperature zones, each with a different phase change temperature. Shape memory alloy, phase transition temperature set to ΔT is graded according to the axial position, with 20~25℃ for the top region, 15~20℃ for the middle region, and 10~15℃ for the transition region; the temperature margin for preventing false triggering is ≥10℃.

7. The process method according to claim 1, characterized in that, The S3 cyclic pre-compression stage employs triangular wave cyclic loading at a loading rate of 2-5 MPa / min, with pressure ranging from 0.5 MPa to 60% of the design preload, and cyclically repeated 20-50 times. The step loading stress redistribution stage includes four loading levels: the first level loads to 50% of the design preload and holds for 30-60 minutes; the second level loads to 75% and holds for 60-90 minutes; the third level loads to 90% and holds for 2-3 hours; and the fourth level loads to 105%-115% and holds for 1-2 hours before unloading to 100%.

8. The process method according to claim 1, characterized in that, The self-weight compensation differential preload in S3 has the following compensation preload force model: in For designing the preload, H is the total height of the winding, z is the height from the bottom of the winding, and k is the compensation coefficient. Differentiated preload is implemented by setting at least 4 independently controllable preload actuators in the winding axial direction. The preload applied by the top actuator is 8% to 15% higher than that at the bottom.

9. The process method according to claim 1, characterized in that, In S4, a preload attenuation prediction model is established based on measurement data. Where P0 is the initial preload, A is the transient attenuation coefficient, and B is the steady-state attenuation coefficient. The transient creep time constant is... The steady-state creep characteristic time is given by n, which is the creep exponent and ranges from 0.1 to 0.

35.

10. The process method according to any one of claims 1 to 9, characterized in that, The partitioned independent fastening system in step S5 includes independent pressure plate units corresponding to each pre-tightening partition. The surface of the pressure plate unit is provided with a radial guide groove, and adjacent pressure plate units are connected by a flexible connector.

Images