In the context of the global energy transition, photovoltaic systems and their core power conversion equipment, the solar inverter, have seen widespread deployment. However, thermal runaway faults within capacitors inside the solar inverter generate combustible gases such as hydrogen and propane, which can be easily ignited by electrical sparks or high-temperature sources, leading to cover plate rupture or ejection. To establish a reliable explosion resistance testing standard for solar inverters, we conducted a comparative study of hydrogen and propane deflagration characteristics in a custom-designed box-type explosion vessel that mimics the structural features of a typical residential solar inverter. This paper presents the experimental methodology, pressure measurement results, aluminum plate deformation analysis, and mathematical modeling to determine the dominant parameter affecting structural damage.
Our experimental system consisted of an explosion reaction chamber, a pressure acquisition system, and an automated gas mixing and concentration monitoring system. The steel box had internal dimensions of 496 mm × 431 mm × 131 mm, fabricated from 8 mm Q355 steel. Inside were two rows of eight metal tubes (60 mm long, 30 mm inner diameter) to simulate internal obstacles typical of solar inverters. The cover plate was made of 1 mm thick AL1070 aluminum alloy, fixed by ten M10 bolts at the two short sides with a tightening torque of 10 N·m, creating a clamped-short-side, free-long-side boundary condition. A high-energy spark igniter was mounted at the center of the bottom plate. Pressure was recorded using a CY-YD205 sensor, a YE5853 charge amplifier, and an HDO4000A digital oscilloscope. Gas mixtures were prepared using a dynamic gas mixing device and verified by two JFQ-1150E gas concentration analyzers (accuracy ≤0.1% for hydrogen, ≤0.01% for propane).
We tested hydrogen/air mixtures at concentrations of 20%, 25%, 30%, 36%, 45%, and 55% by volume, and propane/air mixtures at 3.0%, 3.5%, 4.0%, 4.6%, 5.0%, 5.5%, and 6.0% by volume. Each condition was repeated three times. After filling the chamber with the premixed gas using a seven-volume displacement method and verifying the concentration, we allowed the mixture to settle for three minutes before ignition. The pressure history and the permanent deflection of the aluminum plate were measured for each test.
The pressure–time curves revealed distinct characteristics between the two gases. For propane, a single pressure peak was observed, while hydrogen exhibited multiple peaks due to shock wave reflection and superposition within the confined space. Hydrogen also displayed Helmholtz oscillations during the venting phase, which were absent in propane tests. The peak overpressure for propane reached a maximum of 27.5 kPa at 4% concentration, while hydrogen reached 83.1 kPa at 36% concentration. The peak overpressure varied with concentration in a bell-shaped manner for both gases, with hydrogen values being about three times higher than those of propane at their respective optimum concentrations.
The rate of pressure rise was significantly higher for hydrogen. The maximum rate for propane was 1190.4 kPa·s⁻¹ at 4% concentration, whereas hydrogen reached 10387.2 kPa·s⁻¹ at 36% concentration, a factor of 8.7. The time to reach peak pressure was shortest at the optimum concentration for both gases, and the values for hydrogen were consistently smaller than those for propane.
We calculated the positive pressure impulse defined as
$$ I = \int_{t_1}^{t_2} \Delta P(t) \, dt $$
For propane, the impulse peaked at 4% concentration and followed the same bell-shaped trend as overpressure. For hydrogen, the impulse also reached a maximum at 36% concentration, but the relationship with concentration was less regular. Interestingly, although hydrogen produced higher overpressures, its impulse values were generally lower than those of propane due to the much shorter positive phase duration.
The damage aftereffect was quantified by measuring the maximum deflection of the aluminum plate along the two free long edges and averaging them. The deflection results mirrored the overpressure trends: maximum deflection occurred at 36% hydrogen (largest value) and 4% propane. The hydrogen deflections were significantly larger. To quantify the equivalence of damage capability, we defined a ratio
$$ \theta = \frac{H_a}{H_b} $$
where \(H\) is the plate deflection, \(a\) denotes hydrogen concentration, and \(b\) denotes propane concentration. We found that \(\theta \approx 1.06\) for 25% hydrogen and 4.6% propane, indicating similar destructive capabilities. For 36% hydrogen versus 4% propane, \(\theta = 2.82\), meaning hydrogen caused nearly three times the deformation.
To identify the dominant parameter governing the aluminum plate deformation, we performed linear regression analysis. For propane, both peak overpressure and impulse showed strong linear correlations with deflection (Pearson correlation coefficients: \(R_{P-H}=0.999\), \(R_{I-H}=0.972\)). For hydrogen, due to weaker coupling between overpressure and impulse, we used multiple linear regression:
$$ H = c + a \Delta P + b I $$
The regression results are summarized in Table 1.
| Parameter | Estimated value | Standard error | t-statistic | p-value |
|---|---|---|---|---|
| a (overpressure coefficient) | 0.872 | 0.058 | 14.85 | 6×10⁻⁴ |
| b (impulse coefficient) | -0.074 | 0.039 | -1.91 | 0.151 |
The overpressure coefficient was highly significant (p < 0.05), while the impulse coefficient was not (p = 0.151). This statistical evidence, together with the overpressure–impulse damage criterion, indicates that peak overpressure is the dominant parameter in this solar inverter cover plate deflection. The natural period of the aluminum plate (first mode \(T_1 = 0.107\) s) was comparable to the positive pressure duration, placing the response in the overpressure–impulse regime, but the higher-frequency modes and the clamping constraints shift the behavior closer to the overpressure-dominated region.
We employed dimensional analysis to establish a mathematical relationship between overpressure and deflection. The deflection \(H\) depends on pressure \(P\), plate length scale \(l\), density \(\rho\), and elastic modulus \(E\). Selecting \(\rho\), \(l\), and \(E\) as repeated variables, the Buckingham \(\Pi\) theorem yields two dimensionless groups:
$$ \Pi_1 = \frac{H}{l}, \quad \Pi_2 = \frac{P}{E} $$
Since all plates in our tests had the same material and dimensions, the model simplifies to
$$ H = \beta P^{\alpha} $$
Fitting the experimental data for each gas produced the empirical equations given in Table 2, with excellent goodness of fit.
| Combustible gas | Empirical equation | Goodness of fit (R²) |
|---|---|---|
| Propane | \(H = 2.6 \times P^{1.06}\) | 0.988 |
| Hydrogen | \(H = 0.2 \times P^{1.28}\) | 0.951 |
These equations provide a rapid prediction of the deformation caused by hydrogen or propane deflagration in a solar inverter structure. The material parameters of the aluminum plate are listed in Table 3 for reference.
| Property | Value |
|---|---|
| Elastic modulus \(E\) (GPa) | 69 |
| Density \(\rho\) (kg/m³) | 2700 |
| Free length \(l\) (m) | 0.496 |
| Thickness \(d\) (m) | 0.001 |
| Cross-sectional moment of inertia \(I\) (m⁴) | 8.33×10⁻¹¹ |
| Line density \(m\) (kg/m) | 2.7 |
| First natural period \(T_1\) (s) | 0.107 |
In conclusion, our comparative study of hydrogen and propane deflagration in a solar inverter-like enclosure revealed that hydrogen produces significantly higher peak overpressure, rate of pressure rise, and structural damage than propane at comparable concentrations. A 25% hydrogen mixture and a 4.6% propane mixture showed similar destructive effects on the aluminum cover plate. Regression analysis and the overpressure–impulse criterion demonstrated that peak overpressure is the dominant parameter controlling the deformation. Therefore, in selecting a surrogate gas for solar inverter explosion resistance testing, preference should be given to the gas whose deflagration overpressure matches that of the actual thermal runaway gas mixture. The proposed power-law relationships between overpressure and deflection provide a convenient tool for predicting the aftereffects of hydrogen or propane explosions in similar thin-plate structures. These findings support the development of international standards for solar inverter blast testing.