Research on the Heavy Gas Setting Method of Oil-Immersed Transformer Based on Oil Flow Acceleration Characteristics

Abstract

As the key non-electric protection equipment of an oil-immersed transformer, the gas relay plays an important role in ensuring the safe operation of the transformer. To further enhance the sensitivity of gas relays for the heavy gas alarm, this paper takes the BF type double float gas relay as the research object and proposes a new method for heavy gas setting, which is based on the internal oil flow acceleration characteristics of the gas relay. Firstly, the analytical derivation of the force acting on the gas relay baffle is carried out, and through theoretical analysis, the internal mechanism of heavy gas action under transient oil flow excitation is revealed. Then, the numerical simulation and experimental research on the variation of oil flow velocity and acceleration under different fault energies are carried out. The results show that with the increase of fault energy, the oil flow velocity fluctuates up and down during heavy gas action, but the oil flow acceleration shows a linear correlation. The oil flow acceleration can be set as the threshold of heavy gas action, and the severity of the fault can be judged. At the same time, the alarm time of the heavy gas setting method based on the oil flow acceleration characteristics is greatly shortened, which can reflect the internal fault of the transformer in time and significantly improve the sensitivity of the heavy gas alarm.

1. Introduction

The gas relay is an indispensable non-electric protection component of oil-immersed transformers. Once the transformer encounters abnormal conditions, the gas relay promptly responds by triggering an alarm signal. In more severe situations, it can further disconnect the transformer from the electric power system, thereby achieving effective protection for the transformer.

In recent years, numerous electric power enterprises have experienced transformer failures, with abnormal electrical, magnetic, and thermal parameters identified through algorithm-based monitoring [1,2,3,4]. However, cases of maloperation in transformer gas relay protection have also been documented, ultimately triggering unnecessary tripping [5,6,7,8]. Li, Z [6] studied that poor sealing caused internal short circuits in gas relays, leading to heavy gas alarm tripping. Beyond sealing defects, Cai, W and Dong, J.B [7] indicated that damaged main transformer cables inducing multi-conductor ground faults may activate switch tripping. Additionally, malfunctions such as shattered reed contact glass tubes in gas relays and oil level drops in oil-immersed transformers have been identified as potential causes of heavy gas alarm maloperation [8].

Regarding the issue of heavy gas misoperation in gas relays, relevant scholars have conducted extensive research. Literature [9,10,11,12,13,14,15,16,17,18] has delved into the mechanism of heavy gas misoperation in gas relays. Among them, literature [9,10] conducted in-depth research on the changes and distribution patterns of oil pressure and temperature driven by internal arc during transformer faults, which is helpful for the study of heavy gas action of gas relays. He, W.H. et al. [11] revealed the maloperation mechanism of converter transformer gas relays under inrush current excitation, concluding that “when the amplitude of the inrush current exceeds 6.37 kA, the gas relay will have the risk of maloperation,” establishing crucial reference criteria for diagnosing inrush current-induced malfunctions. Li, B. et al. [12] conducted response characteristic tests on a 25 MVA/110 kV oil-immersed power transformer, demonstrating that “the gas accumulation relay has high sensitivity but insufficient speediness,” while further identifying that “the action threshold is affected by static pressure difference.”

Literature [13,14] revealed the mechanism of vibration-induced misoperation of heavy gas. Pan, Z.C. et al. [13] indicated that transformer vibration can cause the floating ball body of the gas relay to vibrate, leading to misoperation. Zhou, D. et al. [14] pointed out that the excitation of winding vibration by short-circuit current may lead to the possibility of gas protection misoperation. Lins, I.X. et al. [15] focused on the motion law of the internal flow field of the double float gas relay. Li, S.N. et al. [16] further simulated and analyzed the misoperation caused by the impact of reverse oil flow inside the relay, and concluded that there is a small possibility that reverse oil flow inside the relay may cause misoperation.

Yan, C.G. et al. [17] identified inherent inertia in the baffle-floater-spring mechanism of mechanical gas relays, attributing operational delays during arc faults to delayed baffle movements. Ding, G.C. et al. [18] experimentally tested the operating characteristics of gas relays for light and heavy gas under different degrees of transformer faults, concluding that baffle actuation results from the combined effects of oil flow-induced dynamic pressure and static pressure in insulating oil. These findings explained maloperations in on-load tap changers’ gas protection even below preset flow thresholds, providing theoretical and experimental foundations for setting and calibrating gas relay setting values in such devices.

Literature [19,20,21,22,23,24,25] further proposed novel heavy gas protection setting methods and anti-maloperation countermeasures based on gas relay malfunction mechanism analyses. Notably, since gas relays in European and American power systems primarily operate with light gas protection while rarely activating heavy gas protection, these research efforts were predominantly led by Chinese scholars. Yan, D. et al. [19] introduced a protection delay enhancement method to prevent false alarms caused by abnormal baffle disturbances. Literature [20,21,22,23,24] conducted experimental investigations on the heavy gas operation characteristics of QJ-type gas relays under transient oil flow impacts. They developed next-generation gas relays capable of real-time baffle angle monitoring, establishing multiple transformer condition evaluation methods through the following experimental-numerical hybrid approaches: oil-immersed transformer condition monitoring based on baffle angle characteristics [21], heavy gas fault severity assessment based on baffle actuation duration [22], novel heavy gas alarm criteria using baffle rotational angular velocity [23], and non-electric quantity protection methods based on initial baffle rotational acceleration [24]. Building upon these foundations, Xue, B. et al. [25] proposed a comprehensive anti-maloperation strategy integrating gas relay baffle angle sensor data with differential current signals, ultimately designing an intelligent anti-maloperation gas relay prototype.

Current research on gas relay setting values by relevant scholars primarily focused on oil flow velocity and heavy gas characteristic parameters. However, these criteria exhibit instability in timely reflecting internal transformer faults and may contribute to heavy gas alarm maloperation. Based on the physical principle of transient oil flow generated during transformer faults, this study takes the BF type double float gas relay as the research object. Through structural and mechanical analysis, combined with experimental and simulation approaches, the heavy gas actuation characteristics and internal flow field of the gas relay are investigated. A novel setting strategy for heavy gas protection in gas relays is established based on oil flow acceleration characteristics. This method effectively demonstrates both the severity of internal transformer faults and the transient oil flow properties.

2. Structural and Mechanical Analysis of Gas Relay

2.1. Structural Analysis

The internal structure of gas relays is shown in Figure 1, mainly including the upper floating ball, lower floating ball, baffle, permanent magnet, reed switch, etc. Please refer to literature [26] for details.

Figure 2 is a schematic diagram of the gas relay performing the light gas action, and Figure 3 is a schematic diagram of the gas relay performing the heavy gas action. Please refer to literature [16] for details.

2.2. Mechanical Analysis

Figure 4 is a schematic diagram of the forces acting on the baffle and the lower floating ball, mainly including four parts of forces. Please refer to literature [26] for details.

During heavy gas action, the baffle and the lower floating ball rotate around the O axis, and the calculation formula for the total moment is:

$$M=-M_1+M_2+M_3-M_4$$

(1)

The magnetic moment M₂ generated by the permanent magnet against the baffle is expressed as follows:

$$M_2={\displaystyle \frac{{\mu }_0{M_s}^2{V}^2}{4\pi l^3{\sin }^3\alpha }}$$

(2)

In the formula, μ₀ represents the vacuum magnetic permeability, M_s represents the saturation magnetization, and V represents the volume of the magnet.

The expression for the impact torque M₄ generated by the transient oil flow acting on the baffle is:

$$M_4={\displaystyle {\int }_{A}l\cos \alpha }\left(\rho V{\displaystyle \frac{d{V}_2}{dt}}+0.5\rho C_d{\left({\displaystyle \frac{d{V}_2}{dt}}\right)}^{2}A\right)dA$$

(3)

In the formula, V₂ is the fluid volume, and ${\displaystyle \frac{d{V}_2}{dt}}$ is the volumetric flow rate. The first part $\rho V{\displaystyle \frac{d{V}_2}{dt}}$ refers to the inertial force generated by changes in fluid volume, while the second part $0.5\rho C_d{\left({\displaystyle \frac{d{V}_2}{dt}}\right)}^{2}A$ represents the drag force caused by fluid flow.

Since the impact torque generated by the steady-state oil flow impact is smaller than that of the transient oil flow impact, choosing to conduct a test analysis on the transient oil flow impact can more accurately reflect the actual force characteristics during the heavy gas action process.

3. Experimental Analysis of Oil Flow Acceleration Characteristics During Heavy Gas Actuation of the Gas Relay

3.1. Heavy Gas Actuation Test Bench for Gas Relays

The construction of the test bench is shown in Figure 5, mainly composed of the test bench body, measurement system, and control system.

The main body of the test bench includes eight parts: air compressor, air cannon, pulsating flow generating cavity, pipeline to be tested, gas relay, sylphon bellows, butterfly bamper, and capsule oil conservator. The air cannon provides the external excitation source required for the test platform and is a key link in energy excitation throughout the experiment. The pulsating flow generating cavity creates space for compressed air to expand and do work, and the pipeline to be tested is used to install test sensors for measuring and collecting data such as flow velocity and pressure. The gas relay is responsible for outputting heavy gas signals, while the sylphon bellows plays a dual role in pipeline correction and vibration suppression, ensuring the stability and accuracy of the entire pipeline system during the testing process. The butterfly bamper can effectively control the on/off of the pipeline, achieving flexible control of the testing process and facilitating operation in different testing stages. The main function of the capsule oil conservator is to store and replenish oil, ensuring stable oil levels in the system during the testing process and maintaining a normal testing environment.

Air compressors and air cannons are used to apply varying degrees of external excitation sources to the oil pipeline, generating surging oil flow, impacting the gas relay baffle, and simulating pipeline surging oil flow caused by internal faults in transformers. By using pressure sensors and ultrasonic flow velocity sensors to accurately measure pipeline pressure, flow velocity, and heavy gas signals, comprehensive and accurate data support is provided for subsequent research.

The measurement system includes temperature sensors, pressure sensors, ultrasonic flow velocity sensors, gas relays, flap level gauges, data acquisition instruments, and upper computer software, as shown in Figure 6.

Temperature sensors are used to measure the temperature of insulation oil inside pipelines; Pressure sensors are used to measure pressure changes in pipeline oil flow during testing; Ultrasonic flow velocity sensors are used to measure changes in flow velocity inside pipelines during testing; Gas relay is used to output heavy gas signals; The flap level gauge is used to measure the liquid level height of the pulsating flow generation chamber; The data acquisition device synchronously collects signals from various sensors, and combines with the upper computer software to achieve synchronous acquisition and processing of a large amount of measured data, thereby ensuring the integrity and accuracy of the data. The selection and specific technical parameters of the equipment used in the measurement system can be found in the literature [26].

To investigate the heavy gas action characteristics under different transient oil flow impacts, multiple repeated experiments were conducted, and relevant curves were drawn as shown in Figure 7 and Figure 8.

After a comprehensive comparison of the data from two experiments conducted, it was found that the correlation coefficients exceeded 99%, which verified the stability of the test bench.

3.2. Analysis of Oil Flow Acceleration Test During Heavy Gas Action

To further investigate the actuation status of the heavy gas alarm in the gas relay under varying transient oil flow impacts, experimental tests were conducted with air cannon outlet pressures set at 0.108 MPa, 0.109 MPa, 0.110 MPa, 0.115 MPa, 0.120 MPa, 0.125 MPa, 0.130 MPa, 0.135 MPa, and 0.140 MPa as excitation pressures. To ensure comprehensive signal acquisition during the baffle movement process, the sampling frequency was set to 1 kHz. The test results are shown in Figure 9.

During the heavy gas action of the gas relay, the opening of the baffle is the result of the combined action of pressure and flow velocity. As shown in Figure 9, the pressure reaches its maximum value faster than the flow velocity, and the transmission speed of pressure is faster than the flow velocity. The pressure acts on the baffle before the flow velocity. As the excitation pressure increases, the pressure inside the pipeline and the flow velocity of the oil flow will also continue to increase; that is, the impact torque on the baffle increases, and the duration of heavy gas action increases.

The pressure and flow velocity corresponding to the heavy gas action, and the pressure corresponding to the end of the heavy gas action, were extracted from Figure 9. The oil flow acceleration was obtained by differentiating the velocity curve in the figure, and the data are shown in

Table 1. Heavy gas action test data.

Excitation Pressure (MPa)	The Pressure Corresponding to Heavy Gas Action (kPa)	Oil Flow Velocity at Heavy Gas Alarm Actuation Initiation (m/s)	The Pressure Corresponding to the End of Heavy Gas Action (kPa)	Maximum Oil Flow Acceleration (m/s2)
0.108	-	-	-	0.727
0.109	1.380	0.807	−1.432	1.226
0.110	2.746	0.859	−1.552	1.374
0.115	3.345	0.890	−1.723	1.571
0.120	6.600	0.959	−1.621	2.829
0.125	8.271	0.977	−1.627	2.986
0.130	10.630	0.908	0.848	5.092
0.135	11.387	0.930	−0.13	5.379
0.140	14.443	0.820	2.153	8.286

.

As indicated by the data in Table 1, at an excitation pressure of 0.108 MPa, the peak oil flow velocity measures approximately 0.5 m/s, and no heavy gas alarm is triggered. When the excitation pressure increases to 0.109 MPa, the heavy gas alarm is activated, with the corresponding initial action velocity recorded at 0.807 m/s. As the pressure continues to rise, the initial velocity for heavy gas action fluctuates within the range of 0.8 m/s to 1.0 m/s. At a pressure of 0.140 MPa, this velocity fluctuates to 0.820 m/s, representing an 18% decrease from the 1.0 m/s setting value specified on the nameplate of the tested gas relay, and a 16% reduction from the experimental maximum flow velocity setting of 0.977 m/s.

In industrial applications, the oil flow velocity setting values for gas relays are calibrated under steady-state oil flow impacts, where the oil flow acceleration approaches zero. However, under transient oil flow impacts, the oil flow velocity exhibits significant fluctuations. Therefore, this study focuses on the heavy gas alarm actuation oil flow acceleration to investigate its relationship with excitation pressure (which reflects fault severity) and explore whether this acceleration can serve as a characteristic parameter for heavy gas alarms. As shown in Table 1, the heavy gas alarm actuation oil flow acceleration is positively correlated with excitation pressure. For example, at an excitation pressure of 0.109 MPa, the maximum oil flow acceleration reaches 1.226 m/s², which is significantly higher than the 0.727 m/s² observed at 0.108 MPa.

Thus, using the heavy gas alarm actuation oil flow acceleration as a characteristic parameter for heavy gas alarms is feasible. When the acceleration reaches 1.226 m/s², it indicates severe internal faults in the transformer, necessitating immediate shutdown and maintenance.

4. Simulation Analysis of Oil Flow Acceleration Characteristics During Heavy Gas Alarm Actuation of Gas Relay

To further explore the action mechanism of the gas relay baffle under transient oil flow and validate the accuracy of experimental results, numerical simulations of the baffle’s action process are conducted using the experimental flow velocity curves under different excitation pressures as simulation boundary conditions. This approach enables the acquisition of oil flow velocity and acceleration data at the moments when the baffle begins to rotate and reaches the heavy gas alarm angle.

4.1. Simulation Model

The fluid simulation model of the gas relay is constructed based on the actual parameters of the gas relay structure. SolidWorks2022 software is used to complete 1:1 3D solid modeling, accurately restoring the spatial layout of floating ball components, baffle mechanism, and oil passage structure. The physical and 3D models are shown in Figure 10.

Figure 11 is a schematic diagram of the important internal components of the gas relay. The floating ball mechanism consists of two floating balls, upper and lower, and its function is based on the principle of buoyancy, responding to internal faults of the transformer in stages. The baffle is a key executing component for heavy gas protection. When there is a serious fault inside the transformer, the oil flow in the oil tank flows rapidly due to thermal expansion and gas pressure, and the impact baffle rotates around the rotation center (O axis). When the rotation angle (α) of the baffle reaches the set threshold, the reed switch action is triggered, and a heavy gas alarm is issued.

The multi-level structure inside the gas relay poses geometric processing challenges for the simulation modeling of the flow field. Although the microstructures, such as the main body shell and the fastening screws, have little impact on the macroscopic movement of the fluid, they will form complex assembly interfaces and surface topologies during the geometric pre-processing. Therefore, when extracting the computational fluid domain, it is necessary to simplify the internal components and retain the core components, such as the baffle, the support frame, the double float balls, and the retaining pieces, as shown in Figure 12. Among them, the baffle is sleeved on the baffle shaft and can rotate around the shaft.

Accurately depicting the fluid computational domain is crucial for ensuring the precision of fluid simulations. When establishing a geometric model, it is necessary to align with actual working conditions to avoid local analysis deviations. Due to the issue of insufficient geometric precision in models constructed by SolidWorks2022 software, repair and optimization are required before importing the fluid domain into meshing software. In this paper, the volume extraction function of the ANSYS2023 Space Claim module is used to obtain the fluid domain. This method realistically reproduces the internal flow field of the gas relay but generates unnecessary regions, such as those at end-cover connections, which increases the mesh computation workload and consumes resources. Additionally, overly short inlets and outlets can easily lead to flow pattern distortion, affecting the accuracy of the simulation. Therefore, the automatically generated computational domain is optimized. The models before and after optimization are shown in Figure 13.

The accuracy, calculation efficiency, and convergence characteristics of the simulation results are closely related to the mesh quality and density distribution. Mesh planning requires coordinating the balance between the overall accuracy and the calculation load. When dividing the mesh of the fluid domain, an automatic method is introduced, and the surface size is adjusted. The mesh computational domain after encryption is shown in Figure 14. High-precision mesh encryption is implemented in the core motion area, while sparse meshes are used in the low-curvature area to reduce the calculation cost. It contains a total of 284,420 nodes and 1,527,271 mesh elements.

In numerical calculations, the quantity and quality of meshes directly affect the calculation time and accuracy. When the meshes are too dense, it will increase the rounding error and calculation duration, while when they are too sparse, it is difficult to ensure the accuracy of the results. Therefore, it is necessary to select the optimal number of meshes through mesh independence verification.

In this study, three mesh divisions with 1,156,754, 1,527,271, 1,836,948, 2,103,695, and 2,313,939 meshes are applied to the model, and numerical simulations are carried out under the excitation pressure of 0.109 MPa. The results are shown in

Table 2. Calculation results with different numbers of meshes.

Number of Grids	Heavy Gas Action Oil Flow Velocity (m/s)	Maximum Oil Flow Acceleration (m/s2)
1,156,754	0.782	0.985
1,527,271	0.852	1.231
1,836,948	0.848	1.256
2,103,695	0.849	1.257
2,313,939	0.850	1.259

. When the number of grids is 1836948, as the number of grids increases, the difference in results is within 5%. By comprehensively considering the calculation efficiency and accuracy, 1,836,948 is finally selected as the number of meshes for the numerical simulation of the model.

As an evaluation indicator, the influencing factors of unit quality mainly involve the relativity of quality standards and the approximation of evaluation methods. In the finite element parametric coordinate system, the degree of matching between physical coordinates and parametric coordinates determines the quality level of the element (0–1 scale), where 1 represents the ideal state and 0 represents the presence of anomalous elements with zero or negative volume.

The quality statistics of the grid cells in

Table 3. Quality of grid elements in the fluid computing domain.

Mesh Metric	Minimum Value	Maximum Value	Mean Value	Standard Deviation
Grid cell quality	0.13536	0.99999	0.83489	0.00963

show that the overall average quality of the model is in the excellent range, which meets the requirements of numerical calculation accuracy. The grid morphology of the local profile verifies this conclusion (see Figure 15).

In numerical simulation, the accuracy of boundary conditions directly affects the credibility of the calculation results. Especially in the study of the dynamic characteristics of the baffle of the gas relay, different boundary parameters will lead to significantly different simulation results. In this study, the oil flow velocity in the pipeline under multiple sets of excitation pressures is used as the dynamic boundary condition, and the flow velocity control is achieved in the inlet area through the PROFILE macro.

The excitation pressures were set to 0.108 MPa, 0.109 MPa, 0.110 MPa, 0.115 MPa, 0.120 MPa, 0.125 MPa, 0.130 MPa, 0.135 MPa, and 0.140 MPa. Flow velocity data extracted from the aforementioned experimental tests, as shown in

Table 4. Velocity excitation functions under different excitation pressures.

Excitation Pressure (MPa)	Velocity Excitation Function
0.108	0.01634 + 0.43069 t + 0.58513 t2 − 0.13706 t3 − 0.37002 t4
0.109	0.04993 + 0.44627 t + 2.77415 t2 − 3.83364 t3 + 1.42266 t4
0.110	0.03877 + 0.30228 t + 3.89745 t2 − 5.5609 t3 + 2.30505 t4
0.115	0.03738 + 0.43533 t + 4.02812 t2 − 5.69261 t3 + 2.29437 t4
0.120	0.0415 + 0.65216 t + 8.21292 t2 − 12.18082 t3 + 5.35491 t4
0.125	0.02616 + 0.45503 t + 11.86521 t2 − 21.41371 t3 + 11.04012 t4
0.130	0.04049 + 0.00423 t + 21.16775 t2 − 36.16894 t3 + 19.05476 t4
0.135	0.02057 + 0.33261 t + 20.90007 t2 − 34.11282 t3 + 16.01169 t4
0.140	0.04082 + 0.79846 t + 30.76397 t2 − 50.22432 t3 + 26.56221 t4

, were applied as dynamic boundary conditions for the simulations. To enhance computational efficiency, the simulation timeframe was limited to the baffle actuation phase, thereby avoiding redundant calculations from static flow fields.

The gas relay model involves the compound motion of baffle opening and floating ball rotation induced by oil flow impacts, requiring the use of dynamic mesh technology for dynamic simulation. The Fluent dynamic mesh model includes two modes: active rigid body motion and passive rigid body motion. In this study, the baffle opening/closing and floating ball movement fall under the category of passive rigid body dynamics. The differential equations of motion were formulated based on mechanical equilibrium equations and integrated into the DEFINE_CG_MOTION macro within the UDF (User Defined Functions) environment to achieve dynamic simulation of oil flow-driven mechanical motion.

The implementation of dynamic mesh technology comprises two core modules: inlet flow velocity regulation by the DEFINE_PROFILE macro, and baffle motion trajectory control governed by the CG_MOTION macro. For the UDF loading strategy, the compiled mode was selected to configure inlet conditions for complex boundary motion characteristics, and the interpreted mode was applied for baffle motion control to optimize memory usage. Time step determination adheres to the s/v criterion (where s is the minimum mesh size and v is the maximum flow velocity). This approach prevents negative volume mesh phenomena caused by excessive step sizes in dynamic mesh computations.

The transient solver employs a pressure-based algorithm, with the system gravity parameter disabled (compensated by gravitational torque compensation embedded in the UDF). The velocity field adopts an absolute velocity formulation. The SST k-ω turbulence model was selected, and the fluid medium was defined as insulating oil with a density of 960 kg/m³ and a viscosity of 0.012 Pa·s. Boundary configurations include velocity-inlet, pressure-outlet, and no-slip walls. The dynamic mesh zones are confined to the baffle and floating ball boundaries, with smoothing and re-meshing techniques applied to ensure computational stability. Solver parameters utilize the SIMPLE coupling algorithm, and spatial discretization is uniformly set to a second-order upwind scheme to maintain accuracy.

4.2. Simulation Analysis of Heavy Gas Alarm Actuation Characteristics

Figure 16 and Figure 17 show the velocity cloud diagrams of the baffle angle at 0°, 5°, 10°, and maximum angle under excitation pressures of 0.109 MPa and 0.140 MPa.

Figure 18 and Figure 19 show the pressure cloud diagrams of the baffle angle at 0°, 5°, 10°, and maximum angle when the excitation pressures are 0.109 MPa and 0.140 MPa.

Through an in-depth analysis of Figure 16, Figure 17, Figure 18 and Figure 19, the following conclusions are drawn:

• Before the baffle opens, peak flow velocity zones are concentrated in the gaps between the baffle and the frame, as well as near the oil bypass holes, due to the presence of those holes on the baffle. As the baffle opens, the resistance torque of the baffle drops sharply. The baffle and the lower floating ball have a relatively large angular velocity, so the peak flow velocity zone is concentrated near the baffle and the lower floating ball.
• Before the baffle opening, high-pressure regions are concentrated at the front of the baffle and the inlet zone. After the baffle opening, the pressure differential across the baffle is significantly reduced. Pressure peaks relocate below the lower floating ball. This redistribution occurs because the lower floating ball and baffle attain high angular velocity during rotation. The rigid-body motion exerts substantial pressure on the oil flow, amplifying localized pressure effects near the floating ball.
• The oil holes on the baffle serve a dual role: Under normal operating conditions, they provide additional flow paths for the oil. During fault conditions, they reduce pressure distribution on the baffle surface and mitigate the pressure differential across the baffle, and enhance relay stability by balancing transient forces. By redistributing flow and pressure during faults (e.g., sudden oil surges), the oil holes counteract the high angular velocity of the baffle and floating ball, preventing excessive stress concentrations that could trigger false alarms or mechanical failures.

The oil flow velocity corresponding to the heavy gas action obtained from the test and simulation, the maximum value of the oil flow acceleration, and the relative error between the two are organized in

Table 5. Heavy gas action test simulation error comparison.

Excitation Pressure (MPa)	Experimental Oil Flow Velocity at Heavy Gas Alarm Actuation (m/s)	Simulated Oil Flow Velocity at Heavy Gas Alarm Actuation (m/s)	Relative Error of Oil Flow Velocity at Heavy Gas Alarm Actuation	Maximum Oil Flow Acceleration in Experiments (m/s2)	Simulated Maximum Oil Flow Acceleration (m/s2)	Relative Error of Maximum Oil Flow Acceleration
0.109	0.807	0.852	2.758%	1.226	1.231	1.439%
0.110	0.859	0.894		1.374	1.391
0.115	0.89	0.883		1.571	1.565
0.120	0.959	0.974		2.829	2.871
0.125	0.977	0.988		2.986	3.063
0.130	0.908	0.894		5.092	5.026
0.135	0.93	0.950		5.379	5.420
0.140	0.820	0.863		8.286	8.565

.

As shown in Table 5, under the aforementioned operating conditions, relative errors between experimental and simulation results are as follows: Oil flow velocity at heavy gas alarm actuation initiation, 2.758%, and Oil flow acceleration at heavy gas alarm actuation initiation, 1.439%. Both relative errors are maintained within 5%, confirming the validity of the computational methodology. The numerical simulation results accurately predict the heavy gas alarm actuation characteristics of the gas relay. The simulation demonstrates high computational precision, supporting its reliability for engineering applications.

4.3. Result Analysis

Based on the specific law of rapid magnetic force decay, during the operation of the gas relay, once the baffle begins to rotate, it will inevitably reach its maximum angle. Therefore, in simulation studies, it is sufficient to investigate only the oil flow velocity at the moment the baffle begins rotating and the maximum oil flow acceleration. Accordingly, the setting value of oil flow velocity at baffle rotation initiation is defined as v₀, and the setting value of maximum oil flow acceleration is defined as a₀. The time elapsed from the application of excitation pressure until the oil flow velocity reaches v₀ is termed the oil flow velocity alarm time (t₁). The time elapsed from the application of excitation pressure until the oil flow acceleration reaches a₀ is termed the oil flow acceleration alarm time (t₂).

The parameters, including the time required from excitation pressure application to baffle rotation initiation, oil flow velocity at baffle rotation initiation, maximum oil flow acceleration, oil flow velocity alarm time (t₁), and oil flow acceleration alarm time (t₂), are summarized in

Table 6. Heavy gas action simulation data.

Excitation Pressure (MPa)	Time Required from Excitation Pressure Application to Baffle Rotation Initiation (s)	Oil Flow Velocity at Baffle Rotation Initiation (m/s)	Maximum Oil Flow Acceleration (m/s2)	Oil Flow Velocity Alarm Time t1 (s)	Oil Flow Acceleration Alarm Time t2 (s)
0.108	-	-	0.740	-	-
0.109	0.947	0.849	1.231	0.947	0.314
0.110	0.827	0.887	1.391	0.763	0.185
0.115	0.652	0.870	1.565	0.631	0.135
0.120	0.388	0.941	2.871	0.356	0.039
0.125	0.376	0.957	3.063	0.337	0.037
0.130	0.242	0.834	5.026	0.246	0.032
0.135	0.240	0.897	5.420	0.234	0.023
0.140	0.161	0.775	8.565	0.172	0.008

.

The oil flow velocity and maximum oil flow acceleration at baffle rotation initiation are plotted in Figure 20.

As shown in Figure 20, the oil flow velocity at baffle rotation initiation exhibits an initial rise followed by a decline, with significant fluctuations. In contrast, the oil flow acceleration at baffle rotation initiation shows a monotonically increasing trend. When the oil flow acceleration setting value is defined as 1.231 m/s² under an excitation pressure of 0.109 MPa, the maximum oil flow acceleration under 0.108 MPa excitation pressure is only 0.740 m/s², while all subsequent excitation pressures yield higher maximum accelerations. Thus, the oil flow acceleration at baffle rotation initiation can serve as a characteristic parameter for the heavy gas alarm in the gas relay, providing a reliable basis for precise alarm triggering and ensuring the safety and stability of related systems.

The oil flow velocity alarm time and oil flow acceleration alarm time under different fault energy levels are plotted as a bar chart in Figure 21.

Analysis of data from Table 6 and Figure 21 reveals the following conclusions regarding the heavy gas alarm mechanism of the gas relay: Whether using the oil flow velocity at baffle rotation initiation or the maximum oil flow acceleration as the characteristic parameter for alarm triggering, the alarm time decreases progressively as the excitation pressure increases.

Further comparative analysis of the two alarm methods reveals that when using the maximum oil flow acceleration as the characteristic parameter for the heavy gas alarm in the gas relay, the alarm time is significantly shorter compared to using the oil flow velocity at baffle rotation initiation. As shown in the data, across the entire excitation pressure range (0.109–0.140 MPa), the oil flow acceleration alarm time is consistently shorter than the oil flow velocity alarm time. For example, at 0.109 MPa excitation pressure, the acceleration-based alarm is 0.314 s, and the velocity-based alarm is 0.947 s. Response efficiency improved more than threefold.

This discrepancy arises from the physical nature of acceleration. As the first derivative of velocity, acceleration directly reflects the instantaneous dynamic changes in oil flow. Therefore, in practical applications, using oil flow acceleration as the characteristic parameter for alarm triggering enables faster and more timely activation of the heavy gas alarm, significantly improving its accuracy and reliability. This provides stronger assurance for the safe and stable operation of related equipment.

5. Conclusions

Using oil flow acceleration as the characteristic parameter, this study systematically investigated the heavy gas alarm actuation setting method for the gas relay. The feasibility and superiority of this method were validated through experiments and simulations, providing a theoretical foundation for enhancing the response speed and precision of gas relay alarms. The key research contents and findings are summarized as follows:

• Experimental results indicate that in the gas relay under transient oil flow impacts, the oil flow velocity corresponding to heavy gas alarm actuation initiation varied significantly with increasing excitation pressure, showing fluctuations of up to 16%. The oil flow acceleration exhibited a significant positive correlation with excitation pressure, with stark contrasts between actuated and non-actuated states.
• Simulated velocity and pressure cloud diagrams reveal that the oil bypass holes at the baffle of the gas relay provide additional flow paths for oil under normal (non-fault) conditions. Mitigate pressure buildup on the baffle surface and reduce the pressure differential across the baffle during faults, thereby enhancing gas relay stability.
• When the excitation pressure reaches the critical threshold, the acceleration threshold can accurately trigger the heavy gas alarm, serving as the actuation setting value for the gas relay. Compared to the velocity-based heavy gas alarm actuation method, the acceleration-based method significantly reduces alarm times. For example, at 0.109 MPa excitation pressure, the acceleration-based alarm is 0.314 s. The velocity-based alarm is 0.947 s. This demonstrates a threefold improvement in response efficiency.
• Oil flow acceleration, serving as the characteristic parameter for transient oil flow impacts, offers the advantages of rapid response and high precision, effectively enhancing the actuation reliability of the gas relay.

By using the pipeline oil flow acceleration when the BF type double float gas relay starts to operate with heavy gas as the setting value, the gas relay can respond faster to transformer faults, thereby improving the stability of the power system.

Shortcoming of this research work: The designed gas relay heavy gas action experimental platform differs from the actual transformer to some extent. The experimental platform uses an air cannon to simulate transformer faults and does not affect the temperature of the oil flow in the pipeline. In actual transformer faults, the temperature of the oil will increase, thereby changing its viscosity. In the future, we will conduct experiments on real transformers to study the influence of oil temperature and viscosity on the heavy gas setting of gas relays.