Study on the Decrease in Air Dielectric Strength of DC Voltage According to the Ignition Properties of Combustibles

Abstract

Most electrical transmission lines are located in forests, and currently in South Korea, deregulation has allowed various structures to be built on the ground below transmission lines. Events of fires occurring below high-voltage transmission lines can lead to large-scale electrical accidents. To prevent such accidents, this study examined the ignition properties of combustible materials and their relationship with the reduction in air dielectric strength. Experiments were performed on two types of synthetic resins and lumbers, namely cypress and pine. A cone calorimeter was used to measure ignition properties such as effective heat of combustion, heat release rate, and soot yield. In addition, we built a dielectric strength testing device to measure the reduction in air dielectric strength caused by flames. These measurement results can serve as a basis for revising fire safety standards.

1. Introduction

Transmission lines are exposed to various forms of combustible materials such as forests, structures, and plains. As such, if a fire occurs on the ground below a transmission line, the transmission line may be exposed to high temperatures, leading to its physical damage such as a decrease in its tensile strength [1]. In addition, electrical damage may occur between the transmission line and the ground or between several transmission lines.

Table 1. Number of forest fires and the number of transmission line breakdowns by forest fires.

Variable	2005	2006	2007	2008	2009	2010
Number of forest fires	516	405	418	389	570	282
Number of transmission line breakdowns by forest fires	6	6	5	2	4	4
Variable	2011	2012	2013	2014	Average (2005–2014)
Number of forest fires	277	197	296	492	384.2
Number of transmission line breakdowns by forest fires	5	-	-	-	3.2

details the annual number of forest fires and the number of transmission line breakdowns caused by them. An annual average of 3 transmission line breakdowns occurs due to forest fires. Moreover, in 2000, the entire 345 kV East Sea substation temporarily suspended operations due to a forest fire that occurred in the Donghae and Samcheok districts of Gangwon-do [2,3].

In 2014, the electrical facility technology standards were revised and structures on the ground beneath transmission lines of 400 kV or more were allowed, thereby creating an environment in which various combustible materials such as forests and structures could exist on the ground beneath transmission lines in South Korea. If a fire occurs on the ground beneath a DC transmission line, a line short-circuit between phases or a short-circuit between the transmission line and the land may occur due to the flames and combustion products. This may lead to widespread blackouts due to the reduced reliability of electrical grids and breakdowns of high-voltage transmission lines [4].

Robledo-Martinez investigated the combustion of gasoline, lumber, sugarcane leaves, and ash in the air. The dielectric strength of air during the burning of each material was compared to the dielectric strength of normal air, and it was found that dielectric strength was reduced by 49% for gasoline, 27% for lumber, 37% for sugarcane leaves, and 42% for ash. This finding confirmed that the dielectric strength of air varied according to the combustible materials [5].

Peng Li measured the dielectric strength of air gaps by measuring the amount of conductive particles created during forest fires. The experiments were conducted assuming that the fires occurred in fir tree branches, bamboo, or straw. It was found that as the length of the conductive particles increased, their effect on the dielectric strength of air also increased. In particular, a difference of up to 40% was noted in the presence of conductive particles. Therefore, to study the reduction in dielectric strength caused by flames, it is necessary to select materials that generate fewer conductive particles [6].

Assuming a forest fire, You Fei used 24 cm × 24 cm lumber with a length of 6 m to create wood cribs of 23.3 kg, 46.6 kg, and 69.9 kg, and used them as fire sources. Voltage levels of 110 kV, 220 kV, or 500 kV were applied. The experimental results indicated that fire duration, flame height, and fire load were influential and that flame height was the most influential variable [7].

Tian Wu et al. analyzed CCTV footage in which transmission lines had short-circuited due to forest fires. They performed direct experiments to analyze the dielectric strength of air during similar forest fires. The experiments were performed at a voltage of 250 kV, and the fire source was a wood crib formed by stacking 2 cm × 3 cm × 45 cm wood blocks. KCI alkali metal salt was added to the fire source to study the effect of its presence and absence in combustible materials. In addition, the space between poles was measured at 50% and 100% flame presence then compared. The dielectric strength of air was lower for the wood fire to which alkali salt had been added than for the normal wood fire, and dielectric strength was lower when more flames were present in the space between poles. The dielectric strength of the wood fire to which KCI was added was reduced by 30% compared to a normal wood fire. This can be attributed to the effect of thermal ionization on dielectric strength [8].

Zi-heng Pu et al. analyzed the properties of air dielectric strength under transmission lines during forest fires. The study examined the frequency of short-circuits between phases and short-circuits to the ground below transmission lines caused by forest fires. Since 2010, a total of 128 short-circuit incidents have been recorded, including 37 short-circuits between phases (28.9%). It was determined that the reduction in the dielectric strength of air below transmission lines occurs mainly in the flame area, and the flame height does not reach the region between phases [9].

Kwak Dongsoon measured the dielectric strengths of liquid nitrogen, GN₂, and CGN₂, and described the differences in dielectric strengths according to the electrode shape. In addition, gas temperature, vapor, and mist density influenced dielectric strength [10].

Park Changgi discovered that in flames without smoke, dielectric strength gradually decreased as the flame height increased. This was a result of the increase in relative air density when the flame was farther from the electrode [11].

Kim Insik examined the dielectric properties of air when a flame was between the cylinder-shaped rods and plane electrodes. There was a reduction of up to 32.1% in the flashover voltage when a flame was present compared to when it was absent. Furthermore, the corona wind extinguished the flame as the gap length and horizontal distance increased [12].

Chrzan proposed guidelines for firefighting activities for fires occurring below transmission lines. Firefighting activities are proscribed when the flame approaches transmission lines. They are permitted when there is a distance of 3 m or more between the bottom of the transmission line and the top of the flame (only for transmission lines of 400 kV or less). In addition, it is recommended to cut off the power source when a flame is detected at the bottom of a transmission line [13]. Therefore, there is a need to study the possibility of physical damage due to the dielectric breakdown voltage of the air and flames when fires occur on the ground below DC transmission lines.

This study was conducted according to the procedure in Figure 1. First, we measured the fire properties of each material using a cone calorimeter. Second, we analyzed the breakdown of insulation that occurred when a material was burned using the dielectric strength test. Finally, we confirmed the correlation between the two experimental results through correlation analysis.

2. Materials and Methods

2.1. Selecting Experiment Materials

In these experiments, the materials selected as fire sources included two types of plastic resins because most inside structures of combustible materials are plastics, as well as cypress and Douglas fir, which are two types of lumber that are often used as exterior materials. Materials such as polystyrene, which can generate numerous conductive particles and affect experiment results, were excluded. Polyethylene (PE) and polypropylene, which generate fewer conductive particles, were selected as the plastic materials.

The flame retardancy displayed in

Table 2. Properties of polymer.

Material	$\mathbf{Density} (\mathbf{kg}/{\mathbf{m}}^3)$	Flammability	Flash Point (°C)
PP	900	HB	-
PE	920	HB	221

is the result of a UL94 plastic material combustibility test used by the United States National Institute of Standards and Technology. When the combustion speed is 3.0 mm or less, it is assigned the horizontal burning grade. The cypress was cut into 10 cm × 10 cm × 2.5 cm pieces before burning. The experiments were performed five times on each specimen to obtain sufficient reproducibility.

Table 3. Properties of wood.

Material	$\mathbf{Density} (\mathbf{kg}/{\mathbf{m}}^3)$	Moisture Content
Cypress	480	8%
Douglas fir	450	8%

shows the physical properties of the wood used in the experiment.

2.2. Cone Calorimeter Experiments and Analysis Method

Figure 2 is a schematic layout of the cone calorimeter, and Figure 3 shows the components of the cone calorimeter. The cone calorimeter used the oxygen consumption method to calculate caloric values. The oxygen consumption method is based on the theory that the amount of heat and oxygen that is consumed from the air during combustion is such that a fixed caloric value of 13.1 MJ is emitted per 1 kg of oxygen. In addition, the gas types and yield values that were generated during the combustion process were compared to the mass reduction rate, and the production rate was quantified. The KS F ISO 5660 standard experimental device was used to measure and analyze the ignition properties of experimental materials, including three types of polymers, cypress, medium density fiberboard, and heptane [14].

2.3. Dielectric Strength Experimental Method

Figure 4 is a schematic layout of the experimental equipment used in the study. The experiments were performed in the following format. A power supply capable of supplying up to 60 kV of DC voltage was connected to the electrode to simulate a scenario in which dielectric breakdown occurs due to the flames of a grounded burner.

The experiments were designed using the similarity law. According to South Korea’s electrical equipment technology standards, a DC 500 kV transmission line must be at least 22 m away from the top of a structure or other combustible materials. Wires used in overhead transmission lines in Korea are stipulated as 480 mm². The separation distance of 22 m from the transmission line and the separation distance of 5 cm from the electric wire in the experiment was calculated by applying the formula below. According to the calculation, the area of the electric wire should be 1.1mm², and this can be converted to a thickness of about 1.2 mm. Therefore, 2 mm was used, as this is the smallest thickness for aluminum round bars available in the market.

$${\mathrm{k}}_{\mathrm{l}}=\frac{{\mathrm{r}}_{\mathrm{s}}}{{\mathrm{r}}_{\mathrm{a}}}=\frac{{\mathrm{h}}_{\mathrm{s}}}{{\mathrm{h}}_{\mathrm{a}}}=\frac{{\mathrm{S}}_{\mathrm{s}}}{{\mathrm{S}}_{\mathrm{a}}}$$

${\mathrm{k}}_{\mathrm{l}}$: dimensional ratio for geometrical similarity.

${\mathrm{r}}_{\mathrm{s}}$: distance of the scale model.

${\mathrm{r}}_{\mathrm{r}}$: actual distance.

${\mathrm{h}}_{\mathrm{s}}$: scaled model separation distance between transmission line and flame.

${\mathrm{h}}_{\mathrm{r}}$: actual separation distance between transmission line and flame.

${\mathrm{S}}_{\mathrm{S}}$: scaled model cross-sectional area.

${\mathrm{S}}_{\mathrm{r}}$: actual cross-sectional area.

$$\begin{gathered} \frac{5\mathrm{cm}}{22\mathrm{cm}}\left(\mathrm{electrical}\mathrm{line}’\mathrm{s}\mathrm{height}\mathrm{ratio}\right) \\ =\frac{{\mathrm{S}}_{\mathrm{S}}}{480 {\mathrm{mm}}^2}\left(\mathrm{electrical}\mathrm{line}’\mathrm{s}\mathrm{area}\mathrm{ratio}\right) \\ {\mathrm{S}}_{\mathrm{S}}=1.1 {\mathrm{mm}}^2 \end{gathered}$$

The experiments were performed with three conductors including two 99.9% pure aluminum bars that were 2 and 6 mm thick, as well as an aluminum-conductor steel-reinforced (ACSR) cable. The cypress and Douglas fir materials were stacked up for ignition because they are difficult to ignite. As such, the heat release rate was high, and in the case of the 2 mm electrical line, the experiments could not be performed due to the large dip. Therefore, the experiments were performed with only two conductors: the 6 mm line and the ACSR cable. Figure 5 is a real photo of the experimental equipment.

3. Results

3.1. Ignition Property Measurement Results

Table 4. Fire property measurement result.

	PE	PP	Cypress	Douglas Fir
Average HRR (heat release rate) (kW)	2.25	2.65	0.82	0.96
Average HRRPUA (heat release rate per unit area) (kW/m2)	198.97	234.15	82.19	95.65
Max HRR (kW)	4.56	4.87	2.66	3.92
Max HRRPUA (kW/m2)	296.21	316.22	266.02	392.13
E.HOC (effective heat of combustion) (kJ/kg)	38,730	37,909	13,107	10,235
CO_Y (CO yield) (g/g)	0.022	0.320	0.046	0.048
CO2_Y (CO2 yield) (g/g)	2.356	2.304	1.159	0.747
Soot_Y (soot yield) (g/g)	0.046	0.061	0.000	0.002

summarizes the combustion test results for each material, and Figure 6 shows the heat release rate of each material. The heat release rate and effective heat of combustion of plastics was higher than that of wood, and there was a large amount of combustion products. The heat release rate properties, in the case of the two types of polymers, indicated that the heat release rates rapidly decreased after reaching their maximum. For synthetic resin, the carbide remained in the specimen holder as combustion proceeded for a certain time, and the heat release rate decreased due to the residue. For cypress and Douglas fir, we observed a smoldering combustion, and the heat release rate remained steady. For cypress, ignition occurred at 200 s and combustion continued until around 1600 s.

3.2. DC Voltage Air Dielectric Strength Measurement Results

In the dielectric strength experiments, the separation distance between the conductor and the material was set at 5 cm, and five rounds of experiments were performed by varying the material, type of voltage, and electrical line thickness. For cypress and Douglas fir, the heat release rate was high. Moreover, as aforementioned, the experiments could not be performed with the 2 mm electrical line due to the large dip; therefore, the experiments were performed only with the 6 mm line and ACSR conductors.

The dielectric breakdown voltage measurement results obtained from the dielectric strength experiments are shown in

Table 5. Dielectric breakdown voltage and average value of dielectric strength test.

Material	Thickness	Breakdown Voltage(kV)
		1st	2nd	3rd	4th	5th	Average
PE	2 mm	11.6	11.4	11.3	11.3	11.8	11.48
	6 mm	10.07	11.2	10.9	10.8	10.9	10.9
	480 mm²	11.6	11.7	11.4	11.7	11.2	11.52
PP	2 mm	11.6	11.6	11.6	12.4	12.5	11.94
	6 mm	12.5	12.7	12.8	12.6	12	12.52
	480 mm²	12.1	11.4	11.5	11.7	12.2	11.78
Cypress	6 mm	14.4	14.9	14.6	14.7	15	14.72
	480 mm²	14	14.7	14.3	14	14.6	14.32
Douglas fir	6 mm	12.9	12.7	12.9	12.7	13.3	12.9
	480 mm²	13.5	12.9	13.3	12.8	12.7	13.04

. There was no significant difference between the experimental results for the 2 mm line, 6 mm line, and other electrical lines used in the experiments. It was hence deduced that the distance from the bottom of the electrical line to the surface of the combustible material was more important than the thickness of the electrical line.

3.3. Analysis of Correlation between Ignition Properties and Dielectric Strength Experiment Results

To determine which variables affected the dielectric strength experiment results, a correlation analysis was performed between ignition properties and measurement results for the dielectric strength of air during combustion of the 6 mm materials.

Table 6. Relationship between dielectric strength test results in vertical direction and fire properties.

Division		Dielectric Strength
HRR	Pearson correlation coefficient	−0.704 **
	Significance probability	0.001
	N	20
HRRPUA	Pearson correlation coefficient	−0.702 **
	Significance probability	0.001
	N	20
Max.HRR	Pearson correlation coefficient	−0.833 **
	Significance probability	0
	N	20
Max.HRRPUA	Pearson correlation coefficient	−0.197
	Significance probability	0.405
	N	20
E.HOC	Pearson correlation coefficient	−0.734 **
	Significance probability	0
	N	20
CO_Y	Pearson correlation coefficient	−0.031
	Significance probability	0.897
	N	20
CO2_Y	Pearson correlation coefficient	−0.660 **
	Significance probability	0.002
	N	20
Soot_Y	Pearson correlation coefficient	−0.677 **
	Significance probability	0.001
	N	20

presents the results of the analysis showing the relationships between ignition properties and the vertical dielectric strength experimental results. Most factors had a negative correlation with dielectric strength. Other than CO yield, the variables showed correlations with significance probabilities of 0.01 or less. The dielectric strength of the flame showed strong negative correlations with the average heat release rate, average heat release rate per unit area, and effective heat of combustion. These results are statistically significant with a significance probability of 0.01 or more. However, if effective heat of combustion is high, the heat release rate is usually high as well. Therefore, it is not certain that all variables affect the electric field strength. High effective heat of combustion when the material is combusted indicates a high heat release rate; therefore, the problem of multicollinearity must be considered [15].

4. Conclusions

In this study, we confirmed the tendency of dielectric strength to change according to ignition properties. It was found that dielectric strength was related to ignition properties such as effective heat of combustion and the heat release rate of the materials. Dielectric strength decreased as the energy emission from the fire source increased due to the generation of more ions through thermal ionization within the flame. Several variables were found through combustion experiments. However, determining their influence on dielectric strength was difficult because of multicollinearity between these variables. Moreover, if only one variable is examined rather than all variables, this variable becomes significant as it can be used to determine whether a given material is more dangerous than other materials.

For PE, which had the lowest measured dielectric strength, and cypress, which had the highest measured dielectric strength, there was a 25.6% difference in dielectric strength during a fire. We conclude that additional measures are required for ensuring the safety of transmission lines according to the combustible materials that exist in the structures.

Various combustible materials may exist on the ground below transmission lines, and the experimental results indicate that dielectric strength may vary according to the combustible material. Based on this finding, we conclude that additional measures are required for ensuring the safety of transmission lines according to the combustible materials used in the structures. Different tendencies may be observed in the flames of a full-scale fire; however, we believe that these tendencies can be predicted by studying the ignition properties of combustibles in future research.