The Effect of Air Humidity on the Performance of DC Link Capacitor Components

Abstract

DC link capacitors (DCLCs) are key devices in converters. The relative humidity affects the performance of the elements of a DCLC. Currently, there are relatively few studies on the effect of relative humidity on the electrical characteristics of the elements of DCLCs during the production process. This study describes the effects and the control of the relative humidity in the production process. For this purpose, a DCLC component withstand voltage test platform and a DCLC component aging test platform were established. The voltage withstand tests were conducted during different processes and at different relative humidities, and life aging tests were conducted at different relative humidities and different storage times. The results show that after the winding and metal-spraying processes, the voltage withstand levels of the components stored at 30% RH, 60% RH, and 80% RH were very close to that of the components at 3000 VDC; after the heat-setting process, the voltage withstand levels of the components stored at 30% RH, 60% RH, and 80% RH showed a decreasing tendency compared with the level of the components at 3000 VDC. Regarding the samples, after the heat-setting process, stored at 30% RH and 60% RH for 24 h, 96 h, 168 h, 240 h, and 336 h and 2000 h of the aging tests, the capacitance decreases monotonically as the storage time is increased but never exceeds 3%, and the capacitance change decreases as the relative humidity increases. This study can provide important guidance for the humidity control of various processes in the production of DCLCs.

1. Introduction

A DC link capacitor (DCLC) uses metallized films (MFs) with self-healing properties as a dielectric [1,2]. DCLCs have advantages in terms of withstanding high-temperature and their voltage, large ripple current, and lifespan [3,4]. DCLCs play a role in voltage equalization, energy storage and filtering in converters [5]. Currently, DCLCs are widely used in power transmission, clean energy and electric locomotives [6,7].

There are many studies on the performance of DCLCs, focusing on issues such as their insulation and breakdown characteristics, lifetime, and reliability [8,9,10,11]. Humidity and heat resistance affect the life and reliability of DCLCs. Air humidity oxidizes and corrodes the zinc–aluminum composite film, affecting the service life of DCLCs [12].

Currently, studies on the effect of air humidity on the performance of metalized film capacitors (MFCs) have focused on the failure mechanism, reliability assessment, and life prediction of MFCs based on the humidity, temperature, and voltage conditions [13,14,15,16]. In reference [17], the degradation of a plastic box DCLC under different relative humidity (RH) conditions was investigated based on accelerated testing and post-failure analysis for a total of 8700 h. Decay curves for the capacitance reduction were derived at 85% RH, 70% RH, and 55% RH. In reference [18], the moisture diffusion process of encapsulated film capacitors was analyzed according to Fick’s second law, revealing the effect of moisture diffusion on the capacitance loss. In reference [19], an accelerated aging test and failure mechanism analyses were conducted, and lifetime predictions were made. The results show that at a relative humidity above 69%, water molecules and oxygen cause galvanic corrosion and damage the aluminum–zinc layer on the MF. In reference [20], based on a 24 h alternating-humidity and heat aging test that was suitable for DCLCs for photovoltaic and wind power, the results showed an exponentially nonlinear capacitance degradation and an increase in the equivalent series resistance under alternating-humidity and high-temperature conditions.

It is worth noting that the results of these studies on the effect of air humidity on the capacitance loss (lifetime performance) of MFCs in accelerated aging tests were based on semi-closed structure capacitors packaged in plastic boxes as the research object. As for dry DCLCs used in smart grids and rail transportation, their cases are made of stainless steel and aluminum, they are all sealed structures, and the effect of the air humidity on their life performance in practical applications is much smaller. However, these dry DCLCs have more production processes and long production cycles, and the components of the dry DCLCs are in contact with air during the production process, meaning that some of the processes need to be carried out while controlling air humidity. However, there is little research on this production process. Therefore, it is necessary to study the effect of air humidity on the performance of the components of dry DCLCs.

Based on the above analysis, this study takes the components of a DCLC as the research object and firstly introduces the production process of dry DCLCs and their main generation process, as well as the air humidity regulations in the production process. Then, a DCLC component withstand voltage test platform and a DCLC component aging test platform were established. Next, the DCLC components after winding, metal-spraying and heat-setting were stored in three different relative humidity environments, and the withstand voltage tests were conducted on these components to analyze the effects of different relative humidities on the withstand voltage performance of the DCLC components during different production processes. Then, the DCLC components after the heat-setting process were stored in two different relative humidity environments for five different periods of time, and then aging tests were conducted on these components to analyze the effects of different relative humidity environments and their storage times on the lifetime performance of the DCLC components. Finally, future research directions are summarized and considered. It is hoped that this study will provide important guidance for the humidity control of various processes in the production of DCLCs and become a reference for the study of the effect of relative humidity on the performance of DCLCs.

2. Air Humidity Control in the Production Process

2.1. Production Processes of Dry DCLCs

The production processes of dry DCLCs include winding, metal-spraying, heat-setting, enabling, welding, capacitor assembling, vacuum-casting, routine testing, and packaging for storage [21].

Winding is the process of winding two layers of MF onto a mandrel using a winding machine to form an element.

Metal-spraying refers to the spraying of two end surfaces of an element with materials such as zinc, tin, etc., in a molten state with the aid of a dry compressed air jet.

During heat-setting, the dry DCLC components usually need to be processed together. Polypropylene film heat shrinkage is used to modify the physical properties of the components, and heat treatment is applied so that the components in the moisture evaporate and become denser. The process is usually temperature-controlled at a maximum of 105 °C.

Enabling refers to the application of voltage to both ends of the element, using the self-healing properties of the MF to remove undesirable parts of the internal insulation of the element.

During welding, brazing is used to solder the coupled copper foil or wire to the two gold-painted end surfaces of the component in order to assist the two electrodes of the component. The electrodes of the different components are welded to form a group of components according to the design of the structure.

2.2. Air Humidity Regulations in the Production Process

Usually, there are air humidity regulations in the operational instructions for the winding and powering processes. When the components are winding, the humidity of the air is ≤60% RH; the wound components are placed in a drying room with ≤30% RH and must be sprayed with metal within 24 h. If the powering components cannot be produced on the same day, the components should be put into a drying room with ≤30% RH.

During the production process, after the winding, metal-spraying, heat-setting, powering, and welding processes, the end face of the component is in contact with the air. In case of coiled components, there is a gap between the two end surfaces of the components, and air enters the two end surfaces, causing them to inhale the moisture in the air. After the metal-spraying, heat-setting, enabling, and welding processes, the components are sprayed with metal particles at both ends, and when they are placed in the air, moisture will be adsorbed on the surface of the metal layer.

Usually, capacitor manufacturers will install dehumidifiers to control the air humidity in the winding workshop, drying room, welding workshop, and main assembly workshop.

3. Experimental Method

3.1. Samples

The DCLC elements selected for this experiment are of a single-string structure. The manufacturing flowchart of the DCLC element sample is shown in Figure 1.

Each element was wound with two layers of MF with a thickness of 6 μm. The mean square resistance of MF is 35 Ω/□. The capacitance of the sample element was controlled to 200 ± 5% μF. The rated voltage (U_NDC) of the sample is 1200 V DC and its rated operating field strength (E₀) is of 250 V/μm. Each sample was filled with polyurethane.

After the winding process, three groups of samples were stored in 30% RH, 60% RH, and 80% RH for 24 h. These samples were numbered as C_1.1, C_1.2, and C_1.3, respectively, and were used for the metal-spraying, heat-setting, and enabling processes.

After the metal-spraying process, the three groups of samples were stored under 30% RH, 60% RH, and 80% RH for 24 h. These samples were numbered as C_2.1, C_2.2, and C_2.3, respectively, and then used for the heat-setting and enabling processes.

After the heat-setting process, the three groups of samples were stored under 30% RH, 60% RH, and 80% RH for 24 h. The samples were numbered as C_3.1, C_3.2, and C_3.3, respectively, and were used for the enabling process.

After heat-setting process, five groups of samples were stored at 30% RH for 24 h, 96 h, 168 h, 240 h, and 336 h and were numbered as C_4.11, C_4.12, C_4.13, C_4.14, and C_4.15. Meanwhile, five groups of samples were stored at 60% RH for 24 h, 96 h, 168 h, 240 h, and 336 h and were numbered as C_4.21, C_4.22, C_4.23, C_4.24, and C_4.25.

These conditions of relative humidity were based on the output of the humidity in our high-/low-temperature test chamber. The brand used was SUWEI, model GDJS-225.

The processes for the above samples change the relative humidity after the winding, metal-spraying, and heat-setting processes, respectively, assuming that the relative humidity of the other processes is the same, thus ignoring the effect of the relative humidity of the subsequent processes on the performance.

The samples for the withstand voltage tests were identified as C_1.1, C_1.2, C_1.3, C_2.1, C_2.2, C_2.3, C_3.1, C_3.2, and C_3.3. The samples used for the aging tests were identified as C_4.11, C_4.12, C_4.13, C_4.14, C_4.15, C_4.21, C_4.22, C_4.23, C_4.24, and C_4.25.

Five samples were tested for each number (C_1.1, C_1.2, …, C_4.25) for a total of ninety-five test samples used.

3.2. Withstand Voltage Test

In order to test the withstand voltage of the samples, a suitable test platform was set up including equipment and measuring devices. The measurement circuit is illustrated in Figure 2. A photo of the withstand voltage test platform is illustrated in Figure 3. In Figure 2 and Figure 3, C is the sample, U_DC is the power supply, S is the discharging switch, and R is the discharging resistance. The test frequency is set to 100 Hz when the LCR meter measures capacitance. The power supply comprises a power control console, voltage regulator, and power supply unit, which is capable of providing 0–10,000 V DC. The temperature control range of the thermostat is between 20 °C and 150 °C (with an accuracy of ±1 °C).

The withstand voltage test method is in accordance with the self-healing test method in IEC 61071:2017 Capacitors for Power Electronics [22].

We place the samples in the thermostat, which is set to a test temperature of 20 °C. The initial test voltage is 1.5 U_NDC for 5 min, then the value of voltage test is increased sequentially by 100 VDC until the sample fails. The capacitance of the sample was measured with an LCR meter after each test voltage, and the capacitance change rate (ΔC) was calculated according to Equation (1).

$$\mathsf{\Delta }C={\displaystyle \frac{C_1-C_0}{C_0}}\times 100\%$$

(1)

where C₁ is the tested capacitance, and C₀ is the capacitance before the test.

The mean values and standard deviations (SDs) of ΔC are, respectively, calculated by [18]

$$\mathsf{\Delta }{C}_{\mathrm{mean}}={\displaystyle \frac{{\displaystyle \sum _i^N\mathsf{\Delta }{C}_{\mathrm{i}}}}{N}}$$

(2)

$$SD=\sqrt{{\displaystyle \frac{{\displaystyle \sum _i^N{(\mathsf{\Delta }{C}_{\mathrm{i}}-\mathsf{\Delta }{C}_{\mathrm{mean}})}^2}}{N-1}}}$$

(3)

3.3. Life Aging Test

In order to carry out the life aging test of the samples, a test platform was set up including equipment and measuring devices, as presented in Figure 4. The temperature control range of the thermostat is between 20 °C and 150 °C. The electric power source can provide 0–8000 V DC. The test frequency is set to 100 Hz when the LCR meter measures capacitance.

The life aging test method is in accordance with the method of the durability test in IEC 61071: 2017 Capacitors for Power Electronics [8,22].

The test temperature is set on the thermostat to 70 °C. The sample is applied at a voltage of 1.3 U_NDC. We remove the sample after every 100 h and immediately measure its capacitance. The same procedure was then repeated but for a longer duration. The temperature of the measured sample before starting the test and the measured temperature at the end of the final test are room temperature, and the temperature difference between the two does not exceed 2 °C. The rate of change in capacitance before and after the test needed to be no more than 3% [22]. The test time corresponding to a capacitance change of −3% indicates the lifetime of the sample [8].

The capacitance measured at high temperature (C_t) is converted to capacitance at room temperature (C_T) as shown in Equation (4) [8]:

$$C_{\mathrm{T}}=C_{\mathrm{t}}+k_{\mathrm{c}}(T_{\mathrm{t}}-T_0)C_{\mathrm{t}}$$

(4)

where T_t is the value of high temperature, T₀ is the value of room temperature, and k_c is the temperature coefficient of MPF (commonly set as −0.0002/K).

The capacitance change rate ΔC is also calculated according to Equation (1).

4. Results and Discussion

4.1. Withstand Voltage Test

Usually, there are air humidity regulations in the operational instructions for winding and powering processes, which are not to exceed 60% RH and 30% RH, respectively. In case of the failure of the equipment controlling the humidity, the air humidity may reach 80% RH. Therefore, the sample humidity settings for the withstand voltage test are 30% RH, 60% RH, and 80% RH.

Voltage withstand tests were carried out on DCLC components after three production steps and in three different relative humidity environments. After each step of the withstand voltage test, the capacitance of the samples was measured. The mean value of the samples was then used to evaluate the curve of the rate of change in capacitance versus the test voltage. The mean capacitance change rate ΔC_mean versus the test voltage curves are illustrated in Figure 5 and

Table 1. Capacitance change rates of various samples.

Sample	C1.1	C1.2	C1.3	C2.1	C2.2	C2.3	C3.1	C3.2	C3.3
Voltage (V)	3000	3000	3000	3000	3000	3000	3000	3000	3000
∆Cmean (%)	−3.405	−3.385	−3.481	−3.558	−3.505	−3.538	−3.964	−4.305	−4.506
SD	0.078	0.088	0.106	0.042	0.028	0.064	0.02	0.059	0.028

, where the error bar along the curve corresponds to one SD.

The SD in Figure 5 and Table 1 shows that these values of capacitance change rate are closer to the mean and the test data are more consistent.

From Figure 5 and Table 1, the following can be observed:

(1)

The capacitance of the components in different relative humidity environments decreases significantly at a test voltage of 3000 V DC. The mean capacitance change rate ΔC_mean of the components in environments of 30% RH, 60% RH, and 80% RH after the winding process are very close and −3.405%, −3.385%, and −3.481%, respectively. The mean capacitance change rate ΔC_mean of the components in environments of 30% RH, 60% RH, and 80% RH after the metal-spraying process are very close and −3.558%, −3.505%, and −3.538%, respectively. The mean capacitance change rate ΔC_mean of the components in environments of 30% RH, 60% RH, and 80% RH after the heat-setting process are −3.964%, −4.305%, and −4.506%, respectively.

(2)

The capacitance change in the DCLC element shows an increasing and then decreasing trend with increasing voltages. This is due to the process of the withstand voltage test: due to the electrodynamic power, the electric stress leads the DCLC components of the MPF to contract [23], so the capacitance of the first element is high; in the process of testing the weak points of the MPF after the breakdown of the self-healing process, the area of the pole plate—one of the components of the capacitance test—decreases [8].

(3)

The voltage withstand capability of the components is very close in the different humidity environments after the winding process and metal-spraying process. This is due to the fact that the components are sequentially subjected to metal-spraying and then heat-setting after the winding process. In the metal-spraying process, the component ends are sprayed with the metal material in a molten state, and the high temperature can remove moisture from the element ends [14]. In the heat-setting process, the heat-setting temperature is controlled at 105 °C or below, and this high temperature also leads to the evaporation of moisture in the components [8]. Therefore, the relative humidity after the winding and metal-spraying processes has a small effect on the voltage withstand capability of the components.

(4)

The voltage withstand capability of the components after the heat-setting process decreases in the different humidity environments, and under the same humidity environment, the voltage withstand capability of the components after the winding, metal-spraying, and heat-setting processes decreases in this order. This is due to the components being placed in different environments after the heat-setting process, and since no high-temperature treatment takes place in the subsequent process, the moisture that comes into contact with the ends of the components will always exist, which in turn affects the voltage performance of the components. After the heat-setting process, the relative humidity has a more obvious effect on the voltage withstand capability of the components.

Currently, breakdown voltage testing is mainly carried out on non-MPFC bodies, which are on metalized or polypropylene films, and components were tested in reference [8]. Reference [8] reports breakdown voltages of 7000 V. In addition, the corresponding breakdown field strengths were 583.3 V/μm. In the current study, the breakdown voltage is 3000 V, and the corresponding breakdown field strength is 625 V/μm. These results are superior to existing test values. Thus, the results of this test are informative.

4.2. Life Aging Test

Life aging tests were conducted on the components that were stored for five different time periods in two relative humidity environments after the heat-setting process. The capacitance values were measured several times, and the average values of the samples were used to derive a curve of the rate of change in capacitance versus time.

Figure 6a and

Table 2. Capacitance change rates for various storage times at 30% RH.

Sample	C4.11		C4.12		C4.13		C4.14		C4.15
Time (h)	1000	2000	1000	2000	1000	2000	1000	2000	1000	2000
Capacitance change rate (%)	−0.111	−1.373	−0.126	−1.539	−0.132	−1.644	−0.144	−1.705	−0.184	−1.794

show the results of five samples at 30% RH for different storage times. Figure 6b and

Table 3. Capacitance change rates for various storage times at 60% RH.

Sample	C4.21		C4.22		C4.23		C4.24		C4.25
Time (h)	1000	2000	1000	2000	1000	2000	1000	2000	1000	2000
Capacitance change rate (%)	−0.117	−1.375	−0.156	−1.602	−0.182	−1.728	−0.224	−1.817	−0.306	−1.943

show the results of five samples at 60% relative humidity for different storage times.

The effect of the humidity and other non-temperature factors on the failure of film capacitors obeys the Aneurysm law [20,24]. The inverse power law model can be expressed as follows [25]:

$$AF={\displaystyle \frac{L(R{H}_{\mathrm{U}})}{L(R{H}_{\mathrm{A}})}}={\left({\displaystyle \frac{R{H}_{\mathrm{A}}}{R{H}_{\mathrm{U}}}}\right)}^n$$

(5)

where AF is the life acceleration factor, L is the life of the capacitor, RH_U is the relative humidity under working conditions; RH_A is the accelerated relative humidity; and n is the humidity coefficient, a dimensionless quantity related to materials.

Consequently, Figure 6 shows the results of the life aging tests and the fitted curves according to their exponential forms. In Figure 6a,b, the fitted curves for the experimental data at different times are in good agreement with the fitted curves, and the goodness of fit (the coefficient of determination) is >0.99 and >0.993, respectively.

The following can be seen from Figure 6 and Table 2 and Table 3:

(1)

After the heat-setting process, the capacitance decreases monotonically as the storage time is increased but never exceeds 3%. This is due to the component being placed under the humidity environmental condition, and the longer it is stored, the more moisture is adsorbed on the component ends from the air. The component ends are sprayed with an about 1 mm thick zinc alloy layer (electrode), and the moisture is adsorbed on the component ends by slowly penetrating into the zinc alloy layer and the metalized layer of the metalized film upon contact [19]; after a period of time, the metalized layer is corroded, resulting in corrosion spots and the formation of a large corrosion area, which results in a decrease in the capacitance value of the sample [26].

(2)

After the heat-setting process, the capacitance change rate ΔC of the aging test of the samples stored at 60% relative humidity is slightly higher than that of the aging test of the samples stored at 30% relative humidity; after 1000 h of testing, the capacitance change rates ΔC of the samples stored for 24 h, 96 h, 168 h, 240 h, and 336 h are 0.006%, 0.03%, 0.05%, 0.08%, and 0.122%, respectively; and after 2000 h of testing, the differences in the capacitance change rates ΔC of the aging tests of the samples stored for 24 h, 96 h, 168 h, 240 h, and 336 h are 0.002%, 0.063%, 0.078%, 0.112%, and 0.149%, respectively. The capacitance change rates ΔC of the aging tests when the samples are stored for the same time under two different ambient relative humidity conditions are relatively close, and the capacitance change rate of the aging test where the samples are stored for 24 h is almost the same. This is due to the components being placed under different humidity environmental conditions, where the sample components’ ends undergo the adsorption of different densities of moisture in the air. The adsorption of water under conditions of high relative humidity are slightly higher. When the components are stored for a long time, the component end’s adsorption of water increase; this adsorption of water also occurs through the zinc alloy layer, as the water slowly penetrates into the metalized layer of the metalized film upon contact. In the initial stage, the adsorption of water by the metalized film will increase upon contact. In the initial stage, the metalized layer (Zn-Al) and the electrode (Zn) of the metalized film have a certain degree of corrosion resistance [27], and the capacitance of the samples is not significantly reduced over a short period of time; after a period of time, the metalized layer is corroded, and corrosion spots are produced, forming a large corrosive area and resulting in a rapid reduction in the capacitance value of the samples.

According to the above analysis, the following can be seen:

(1)

After the thermal heat-setting process, the long-term durability of the components at 30% RH is better than that at 60% RH, but the difference is not significant.

(2)

Under the same relative humidity conditions, the longer the storage time of the components, the worse the long-term durability, and the shorter the life span.

(3)

The component storage time affects component performance.

At present, many researchers have attempted to predict the lifespan of DCLCs under different humidity and temperature conditions [17,18,19,20,26]. However, these studies were mainly focused on DCLCs based on semi-closed structures encapsulated in plastic boxes, and temperature and humidity values were mainly varied, while the test voltage was the rated voltage in the life aging tests. In reference [17], the lifetime of the sample element aged at 1.0 U_N (1100 V DC) was reported to be 2000 h at 85 °C and 55% RH (the capacitance change rate needed to be no more than 5%). In reference [19], the lifetime of the sample element aged at 1.3 U_N (1560 V DC) reached 2000 h at 70 °C (the capacitance change rate needed to be no more than 2%). The object of this study is the DCLC elements in the production process, mainly aiming to study the effect of humidity on the performance of DCLC elements in the production process. In the aging test, this process is not related to the conditions of humidity, mainly to observe the effect on the components’ lifespan of the two humidity environments when the components are stored for different time periods. The results of this test are informative.

5. Conclusions

In the present study, DCLC components with a rated operating voltage of 1200 V and an operating field strength (E₀) of 250 V/m were used as research objects. The differences in the performance of DCLC components under different relative humidities in the production process were analyzed through tests, and a voltage withstand test of the DCLC components was carried out, while a life aging test was carried out to analyze the effects of air humidity on the lifetime and voltage withstand performance of DCLC components. The main results obtained can be summarized as follows: the relative humidity affects the voltage withstand capability and service life of DCLC components; controlling the relative humidity to 60% of the conditions can ensure the quality of the performance of DCLC components; and the longer the components are stored under the same relative humidity conditions, the shorter their service life will be.

The conditions of relative humidity used in this study were based on the humidity of the high-/low-temperature test chamber. The relevant test results show that the adopted method is feasible and effective in practical engineering and that it is useful for analyzing and solving the effect of relative humidity on the performance of DCLC components. This humidity condition is very different from the natural air humidity that occurs in real life. However, analyzing the differences in the performance of DCLC components at different relative humidities during the production process through testing is helpful in analyzing and solving quality problems in the production process of DCLCs. In the future, it is necessary to continue to study the effect of relative humidity on the performance of DCLC components in the production process in depth.