The Development of a Converter Transformer Fire Model Based on the Fire Dynamics Simulator and the Analysis of Cooling Mechanisms of Spraying and Coating

Abstract

This research develops a numerical fire model for a converter transformer utilizing the Fire Dynamics Simulator (FDS). The model’s accuracy was validated through comprehensive evaluations of temperature distribution, radiative heat transfer, and mass burning rate. Additionally, the cooling efficacy of fire-resistant coating and fine water mist with varying droplet sizes was investigated. The results indicate that fireproof coating significantly reduces the surface temperature of the transformer, thereby enhancing its fire resistance. Specifically, temperature reductions of 57.68%, 45.63%, 37.78%, and 36.78% were recorded at different facade heights. Furthermore, the cooling performance of fine water mist is strongly influenced by droplet size, primarily due to thermal buoyancy effects. Larger droplets (400 μm) exhibited the most efficient cooling effect directly beneath the spray, achieving temperature reductions of up to 67%. In contrast, smaller droplets (100 μm) showed diminished cooling performance in certain regions, owing to the compensatory buoyancy of hot air, even resulting in an 11% temperature increase in some cases. During the flame stabilization phase, the mass burning rate stabilized between 0.056 kg/(m²·s) and 0.070 kg/(m²·s), with the inhibitory effect of small particle mist becoming pronounced only after 450 s. These findings offer critical insights for optimizing fire protection strategies for converter transformers, highlighting the significance of cooling mechanisms and material properties.

1. Introduction

The Ultra High-Voltage Direct Current (UHV DC) transmission project plays a crucial role in the modern power system, and the converter transformer, as its core equipment, bears the important tasks of voltage conversion and power transmission [1,2,3]. However, due to its high-load operation and complex electromagnetic environment, the converter transformer faces many hazards in actual operation, and fire is one of the most serious risks [4]. Under the influence of external ignition sources or internal failures, the converter transformer may experience structural failure due to temperature rise, leading to more serious fire accidents [5,6,7].

Table 1. Summary of converter transformer fire incidents.

Year	Location	Consequence
2019	Yinan, China	The Yinan Converter Station, Pole II, low-end Y/Y-C phase converter transformer experienced a sudden fault and caught fire, with no casualties reported.
2020	Jinan, China	The converter transformer burst into flames and later caused a fire. The accident has resulted in one death and two injuries.
2022	Nevada, USA	The converter transformer exploded and caught fire at the Hoover Dam hydroelectric plant. There were no injuries or fatalities as a result of the incident.
2024	Ishikawa, Japan	The converter transformer exploded and burned, causing a loss of 500,000 volts of power and shutting down the nuclear reaction.

lists a number of accidents due to converter transformer fires in recent years, which not only caused serious damage to the equipment but also posed a significant threat to the stable operation of the power system [8]. Therefore, it has become imperative to research and develop effective fire retardants to improve the fire safety of converter transformers [9].

At present, the fire protection methods for converter transformers are mainly divided into two categories: active and passive. Among active fire protection, the water spray fire extinguishing system has become one of the most widely used fire extinguishing methods due to its efficient fire extinguishing performance and good environmental adaptability [10]. By atomizing water into tiny droplets, water spray quickly covers the fire source and reduces the temperature of the surrounding environment, thus achieving the effect of extinguishing the fire and preventing re-ignition. On the other hand, passive fire protection methods mainly rely on the fire-insulating properties of materials that can effectively delay the transmission of flame and high temperature to the interior of the equipment, thus protecting the core components of the equipment [11].

In recent years, a large number of studies have been carried out on the fire protection of converter transformers. In terms of fire combustion characteristics, converter transformer fires are generally considered to be insulating oil pool fire types [12], and experiments have been conducted to investigate the effects of various factors (wind speed [13,14], multiple liquid pools [15], height of enclosure walls, and initial temperature of the liquid pools [16]) on the fire burning rate [17,18,19,20], geometry [17], and the laws of thermal radiation [21,22]. In water spray fire extinguishing, the ability of fine water mist to extinguish electrical fires has been demonstrated [23], and studies have been carried out on the cooling effect of different fine water mist fire extinguishing systems and thermal radiation blocking [6,24,25,26,27,28,29,30,31]. In the study of thermal insulation materials, the failure mechanism of casing blocking structure on the valve side of the converter has been investigated [8,11], confirming its important role in slowing the spread of fire. In order to carry out in-depth research on converter transformer fire protection, the shortcomings of existing experimental studies can be compensated by numerical simulation methods [32]. A number of studies have been conducted to simulate the development of electrical fires based on CFD (Computational Fluid Dynamics) [33], to evaluate the fire extinguishing effect of fine water mist [34,35], and to study the thermal insulation effect of materials [8]. Simulation studies not only save experimental costs but also accelerate the development and application of technology and are therefore of great importance in studying fire protection means for converter transformers.

Existing experimental research provides more realistic data but is costly and poses safety risks. Additionally, some data are difficult to obtain during experiments. These limitations can be addressed through validated simulations, which reduce costs and provide more comprehensive data. Current fire simulations for converter transformers typically assume a fixed fire source power, which does not change with the addition of fire suppression methods. This assumption overlooks the dynamic characteristics of fire source power, which varies over time in real fires and fails to couple fire evolution with the effect of water mist spray of different particle sizes. Furthermore, the shell structure of the converter transformer is assumed to be steel, which may fail in a fire [36]. However, there is a lack of simulation studies on the performance of thermal insulation coatings applied to the converter transformer’s shell [25,26,27,28,29,30].

Therefore, this study will systematically investigate the effect of water spray with different particle sizes on the surface temperature and fire development of the converter transformer, and at the same time, comparatively analyze the temperature barrier effect of the thermal insulation coating to provide the theoretical basis and technical support for the optimization of fire protection technology of the converter transformer.

2. Converter Transformer Fire Calculation Model and Operating Condition Settings

2.1. Introduction of Software

In this paper, we use the FDS V6.0 (Fires Dynamics Simulator V6.0) software developed by NIST (National Institute of Standards and Technology) to solve the N-S equations using numerical methods for analyzing the fire evolution, heat radiation, and heat transfer of the converter transformer. The heat transfer model, thermal radiation model, and conservation model are described in the FDS 6th edition technical manual [37].

2.2. Model Configuration

The basic parameters and overall view of the converter transformer in the fire scene selected in this paper are shown in Figure 1, and the model component materials are mainly steel materials for the purpose of accurately simulating the heat conduction situation.

To address model simplification needs, the fire source fuel material for the insulating oil, generally the esters formed by C6–14 straight-chain or branched-chain saturated or unsaturated fatty acids and glycerol, uses heptane approximation to simplify the settings for the combustion model of the calculation model, simulating the pool fire on the converter transformer; due to the FDS grid using an orthogonal mesh, the liquid pool is set up as a square liquid pool with an equivalent circular area. The combustion model and boundary conditions are set up as shown in Figure 1 in addition to the lower layer for the inert boundary for simulation. The remaining five boundary conditions are set as an open boundary to ensure material exchange.

In the selected fire scenario of this article, the basic parameters and overall view of the transformer are shown in Figure 1. The material of the model components is primarily steel for the purpose of accurately simulating heat conduction.

In this study, a fixed heat release rate (HRR) was not chosen as the input condition for the fire source. Instead, the liquid evaporation model in FDS was used to calculate the mass loss rate of the liquid pool through the heat of evaporation, from which the HRR was derived. This approach not only allows for obtaining the cooling effect of the fine water mist spray but also its extinguishing effect on the pool fire. Two protection measures were applied in the simulation; the parameters of the fireproof coating are shown in

Table 2. Thermal insulation coating parameters.

Coating Area	Type	Density [kg/m3]	Specific Heat Capacity [J/(kg·°C)]	Thermal Conductivity [W/(m·°C)]	Thickness [mm]
Transformer Surface	Tick Coating	500	1000	0.102	7

, and the parameters of the water spray system are based on the default settings in FDS.

2.3. Fire Scene Simulation

In order to investigate the impact of different fire protection measures on the converter transformer, several groups of measurement points were set up for simulation. The specific locations and numbers of the measurement points are shown in the data in Figure 1. The working condition design takes into account both the fireproof coating and the particle size of the water mist. Five simulation conditions were designed to cover different particle sizes of water mist and the application of fireproof coating, with the specific working conditions listed in the table in Figure 1.

3. Validation of the Numerical Model

3.1. Validation of Pool Fire Mass Burning Rate

HRR is the heat release rate, denoted by Q, which is commonly used to refer to the amount of heat released from the fuel per unit of time in W, kW, or MW; the power of the fire source can be calculated using the following equation considering the ideal condition of a fuel efficiency factor of 1 [38]:

$$Q=A_P\times \dot{m}\times \mathsf{\Delta }H$$

(1)

In the equation, $A_P$ is the combustion surface area in square meters, m²; $\dot{m}$ is the mass burning rate in kilograms per square meter per second, kg/(m²·s); $\mathsf{\Delta }H$ is the heat of fuel combustion, which is taken as 48,060 kJ/kg for n-octane. The output from FDS is the HRR value, and through this equation, HRR can be used to calculate $\dot{m}$.

The parameters for n-octane in this combustion model are shown in

Table 3. Heptane key parameter table.

Parameter	Value	Unit
Density	675	kg/m3
Specific Heat	2.24	kJ/(kg·K)
Thermal Conductivity	0.14	W/(m·K)
Evaporation Heat	317	kJ/kg
Boiling Point	98.35	°C

. The experimental data are derived from the 1.2 m diameter (~1.1 m²) n-octane pool fire statistics compiled by Barauskas [39]. In this model, the pool fire is set as a square liquid pool of 1 m². Grid independence calculations were performed using grid sizes of 0.05 m, 0.10 m, and 0.20 m, with results shown in Figure 2. In the stable combustion phase of the flame, the mass burning rate $\dot{m}$ calculated with the 0.05 m and 0.10 m grids are very close, indicating that further grid refinement is unnecessary.

Table 4. Experimental comparison results.

Mesh Size	Experimental Value [kg/(m2·s)]	Analog Value [kg/(m2·s)]	Inaccuracies
0.20 m	0.067	0.088	31.0%
0.10 m	0.067	0.074	10.0%
0.05 m	0.067	0.068	1.5%

shows that the 0.05 m grid provides the most accurate prediction of the experimental results, with an error of only 1.5%, while the 0.1 m grid has an error of 10%. However, in practical simulations, the 0.1 m grid could theoretically improve computational efficiency by eight times, and a 10% error is also acceptable. This grid size also complies with the grid selection method recommended by NIST, which states that the grid size should be between 1/4 and 1/16 of the characteristic flame diameter $D^{*}$ [40]. The calculation of $D^{*}$ is shown in Equation (2):

$$D^{*}={\left({\displaystyle \frac{\dot{Q}}{{\rho }_{\infty }{c}_p{T}_{\infty }\sqrt{g}}}\right)}^{2/5}$$

(2)

In the equation, $\dot{Q}$ is the heat release rate, kW; $c_p$ is the specific heat of air, taken as 1 kJ/(kg·K); $T_{\infty }$ is the ambient temperature, taken as 293 K; $g$ is the gravitational acceleration, taken as 9.81 m/s²; and ${\rho }_{\infty }$ is the air density, taken as 1.2 kg/m³.

In the validation experiment, the mass burning rate of 0.067 kg/(m²·s) (HRR = 3600 kW) indicates that a grid size between 0.1 m and 0.4 m should be selected, consistent with the grid independence verification results. Therefore, it can be concluded that the predicted results of the mass burning rate in this model are accurate, and the model grid should be selected according to Equation (2).

3.2. Validation of Temperature Field

Temperature field validation was carried out using a pool fire experiment [41], where temperature measurements were taken at different heights above the pool fire. The temperature-versus-height curves obtained are shown as black lines in Figure 3, with the simulated data represented by red lines. The basic trend and values of the two sets of data are consistent, with differences appearing only near the flame core. This discrepancy is attributed to the fact that the FDS software does not account for the lower temperature in the flame core. However, this error does not affect the calculation of the temperature field outside the flame and can be treated as an outlier. In summary, the model’s temperature field prediction can be considered accurate.

3.3. Validation of Thermal Radiation

Thermal radiation validation refers to the thermal radiation calculation results of pool fire in the literature [36], and three sets of thermal radiation meter values at 5 m were measured in the experiment. According to the thermal radiation injury threshold provided by the national standard GB/T 37243-2019 [42], 4.7 kW/m² is the threshold of skin pain for personnel exposure, and 1.58 kW/m² is the threshold of no discomfort for prolonged exposure.

Table 5. Comparison of average thermal radiation values.

Measuring Point	Experimental Value [kW/m2]	Simulation Value [kW/m2]
Point 1	3.6	1.95
Point 2	3.3	2.09
Point 3	2.7	1.82

shows the simulation results and experimental values, the simulation results and the experimental values were calculated to be within the same injury threshold, and both were safe thresholds. It shows the consistency between the experiment and simulation in determining injury level and model prediction accuracy.

3.4. Validation of Heat Transfer Calculations

In order to ensure the accuracy of the heat transfer model, this paper verifies the heat transfer experiments by replicating the stone wool board of the converter transformer blocking structure in the literature [8].

In the experiment, the temperature change curve of the backboard of 0.13 m thick rock wool board under the influence of the fire source from 401 °C to 747 °C was measured, the maximum temperature of the outer frame of the backboard sheet was about 75 °C, and the maximum temperature of the back of the board was about 65 °C after 4 h of heating. In the simulation, the properties of the stone wool board are set as the same as in the experiment, the heating temperature is taken as 747 °C, and the temperature cloud of the back of the board in Figure 4 is finally obtained. The black area represents the area with a temperature of 74 °C (±22.5 °C), indicating that a high temperature of 74 °C has appeared on the back of the plate at the 4th hour (14,400 s), and the phenomenon that the temperature of the outer frame is higher than that of the center is the same as that of the experiment. The calculation results show that the model can accurately calculate the heat transfer of the insulation material.

4. Results and Discussion

4.1. Influence of Coating on the Refractory Properties of Converter Transformers

Based on validated models for combustion, temperature field, thermal radiation, and heat transfer, this paper performs a simulation of a conventional converter transformer fire to study the protective effect of the fireproof coating.

Figure 5 shows the fire development at different times. Before 200 s, the fire is primarily in the flame development stage. Due to incomplete combustion of long carbon chains, a large amount of black smoke is generated, and the flame height is shorter, not yet wrapping the sides of the box. After 200 s, the fire grows larger, and by 400 s, the smoke mixes with the flame and wraps the entire box. The fire develops steadily without any external mitigation measures.

Figure 6a shows that before and after applying the fireproof coating, the coating acts on the object’s surface without affecting the amount of thermal radiation absorbed. As a result, the heat radiation on the surface of the converter transformer does not change significantly and is primarily concentrated in the lower area of the side of the box and along its edges.

Figure 6b shows a significant temperature difference between the box’s surface with and without fireproof coating. The surface temperature of the box without the coating is much lower than that of the coated surface.

This difference is due to the fact that both surfaces receive the same amount of thermal radiation. However, the low heat transfer coefficient of the fireproof coating (0.102 W/m·°C) prevents effective heat transfer to the interior, causing heat to accumulate on the surface. As a result, the surface temperature of the coated area is higher than that of the uncoated box.

However, the smaller heat transfer coefficient increases the surface temperature of the coating while hindering the internal conduction of temperature, which can reduce the temperature of the covered box, and this result is demonstrated in Figure 7. For both sets of conditions, four temperature measurement points in the middle vertical region were selected on one side of the box and placed at a depth of 7 mm within the surface to measure the insulation effect of the fireproofing coating on the steel structure over a period of ten minutes.

Before 170 s of the fire, the fire had not fully developed, and due to heat transfer lag in the object, the temperature at the measurement points did not change significantly. After 170 s, all measurement points showed a temperature rise. The closer the measurement point was to the fire source, the faster the temperature rose and the higher the temperature was. After the application of the fireproof coating, the temperature at the measurement points was relatively lower. Over a span of ten minutes, the maximum temperature dropped by 57.68% at measurement point 60, 45.63% at measurement point 49, 37.78% at measurement point 27, and 36.78% at measurement point 05.

The results show that the fireproof coating can effectively reduce the surface temperature of the covered object. This is due to the coating’s lower heat transfer coefficient, which creates a higher temperature zone on the surface while preventing heat from spreading to the internal structure.

4.2. Effect of Different Particle Sizes on the Refractory Properties of Converter Transformers

The effects and mechanisms of fine water mist with different particle sizes on the fire parameters of the converter transformer are analyzed in this section. Figure 8 shows that as the particle size of the water mist decreases, the rate of water vapor generation in the calculated space increases to 1.1 kg/s, 1.3 kg/s, and 1.5 kg/s, respectively. A higher rate of water vapor generation in a semi-enclosed or enclosed space can effectively reduce the oxygen content, thereby lowering the fire power. Additionally, the heat carried away by evaporation helps reduce the fire temperature. In this case, however, the space was relatively open, and the increase in water vapor had a limited effect on the mass burning rate of the fire.

During the flame stabilization stage, the mass burning rate for the three particle sizes remained between 0.056 kg/m²·s and 0.070 kg/m²·s, showing little difference across particle sizes. Only after 450 s did the mass burning rate for the 100 μm (blue triangle) particle size show a slight decrease. This was due to the higher water evaporation rate, which reduced the oxygen and fuel concentration in the air.

In the previous section, it was observed that the temperature of the bottom measurement point where measurement point 05 is located is the highest, so in this section, measurement point 05 (in the middle of the two sprays) and measurement point 08 (directly below the spay) were selected to study the effect of different particle sizes on the surface temperature of the converter transformer, and the specific location is shown in Figure 9. The results show that the cooling effect of the water mist of different particle sizes at measurement point 08 is obvious, and with the increase in the particle size, the cooling effect is improved, and the temperature is reduced by 58%, 63%, and 67%, respectively, with respect to the non-sprinkled without spraying maximum temperature of 289 °C, the temperature reduction was 58%, 63%, and 67%, respectively, and the temperatures were all kept near the boiling point of 100 °C.

According to the above, the evaporation rate of smaller particles is higher, which results in more heat being removed. Therefore, the cooling effect should increase as the particle size decreases, which contradicts the calculation results. A reasonable explanation can be found by referring to Figure 10. Figure 10 shows the temperature field distribution: lower temperatures below the spray, higher temperatures in the spray interval, and as the particle size decreases, the high-temperature region expands. This distribution can be explained by the droplet distribution shown in Figure 10. For larger droplets (400 μm), the spray distribution is more regular, with droplets falling downward. For smaller droplets (200 μm and 100 μm), the droplet distribution is more chaotic, with not all droplets falling straight down. This phenomenon is caused by thermal buoyancy. Referring to the mechanism diagram in Figure 10, the droplets fall from the top and attach to the object’s surface, where the high-temperature vaporization reduces the surface temperature. However, thermal buoyancy from the flame at the bottom impedes the droplets from falling further, thus reducing the cooling effect on the surface. The effect of buoyancy is more pronounced on smaller droplets, which is why the cooling effect improves as particle size increases.

A similar pattern was observed at measurement point 05, where the cooling effect increased as the particle size increased, based on the same principle. This point is located in the area between the two sprays, so there are fewer droplets and the cooling effect is slightly weaker—43% for 400 μm droplets, and only 3% for 200 μm droplets, compared to a maximum temperature of 363 °C without spraying. For 100 μm droplets, the temperature increased by 11%. This phenomenon can also be explained by Figure 10. In regions with more droplets, the droplets fall vertically, and although thermal buoyancy is present, its effect is small, so the temperature is lower (around 100 °C). However, these droplets also hinder the upward movement of hot air, compensating for areas with fewer droplets. This results in greater thermal buoyancy, causing smaller droplets to blow upward more easily, which reduces the cooling effect. At the same time, more gas enters, and the temperature increases. By comparing the temperature field for 100 μm droplets with the temperature field without spraying in Figure 10, it can be seen that the high-temperature region on the surface is larger, supporting the above explanation.

In practical applications, it is important to consider the impact of environmental factors on the actual cooling effectiveness. For example, wind can disperse the water mist, reducing the cooling effect of the droplets, with smaller droplet sizes, being more susceptible to being blown away due to their lower density. Higher air humidity generally reduces the combustion rate of pool fires, so in arid regions, the potential fire hazards should be given more attention. The situation in a confined structure is more complex; insufficient oxygen can interrupt or reduce combustion, which may have less impact on the converter transformer. However, if the internal temperature remains high and there is a large amount of combustible gas, opening the confined space, such as during an inspection or other actions that disturb the seal, can lead to a sudden influx of air and, combined with abundant combustible gasses, create violent combustion or deflagration under high-temperature ignition, causing greater harm. In such cases, it is recommended to use water spraying for cooling and fire control, as fireproof coatings cannot lower the temperature in a sealed space, but water mist can. The absence of high temperatures after combustible gas contacts the air reduces the likelihood of deflagration.

5. Conclusions

This study develops a numerical model using FDS to evaluate fire protection measures for converter transformers. After using the fireproof coating, the surface temperature of the converter transformer is significantly reduced, and the maximum temperature of the facade from top to bottom decreases by 57.68%, 45.63%, 37.78%, and 36.78%, respectively, which significantly improves the fire-resistant performance. There are significant differences in the cooling effect of fine water mist of different particle sizes at different locations of the converter transformer. In the area below the spray, 400 μm and 100 μm particles reduce the temperature by 67% and 58%, respectively. In the area between the two sprays, 400 μm particles reduce the temperature by 43%, while 100 μm particles increase it by 11%. This occurs because larger droplets hinder hot gas movement while smaller droplets are blown away, reducing cooling efficiency and allowing more hot gasses to enter, thereby raising the temperature. The mass burning rate under the three groups of particle size water mist conditions showed slight differences, with smaller particle sizes resulting in a lower mass burning rate, indicating that smaller particle sizes, due to the expansion of the evaporation volume, can reduce the mass burning rate.

This research can be used to optimize the spray design for converter transformers, reducing the probability of structural failure caused by pool fires due to insulating oil leakage. Additionally, future experimental research can focus on fire protection measures for specific parts of converter transformers, such as connection points, to minimize potential hazards.