On the Relationship between Lightning Strike Parameters and Measured Free Surface Velocities in Artificial Lightning Strike Tests on Composite Panels

Abstract

In this manuscript, Current Component A lightning strike tests on three different types of carbon fiber reinforced composite panels are analyzed. The panels feature different levels of lightning strike protection: no protection, medium protection and heavy protection. In particular it was analyzed if there were any direct correlations between the peak electric current of the artificial lighting strike and the recorded velocities at the back surface of the composite panels. The existence of a master curve correlating the peak electric current, the mass of the composite panels and the measured back surface velocity was demonstrated. This finding implies that the back surface velocity correlates linearly to the inertia of the panel and the peak current of the lightning strike.

1. Introduction

Carbon fiber reinforced composite materials (CFRP) are known for excellent weight-specific material properties—such as weight-specific strength and weight-specific stiffness—which makes them attractive for application I primary and secondary structures in modern wide-body aircraft such as the Airbus A350 or the Boeing 787 [1]. During service life, aircraft may be subjected to extreme environmental conditions [2] such as hail storms or thunderstorms [3]. On average each aircraft is struck by lightning every 1000 flight hours [3]. Due to the combination of carbon fibers with high electric and thermal conductivity and polymer resin systems with low electric and thermal conductivity, composite structures are, unlike metals, sensitive to lightning strike events. Therefore, a lightning strike may cause severe structural damage in the vicinity of the impact and exit zones. The characteristic structural damage considered here is the separation of adjacent composite plies, so-called delamination [4,5]. Undetected delaminations can grow under cyclic compressive loading and subsequently result in catastrophic failure of the composite structure [6,7]. For the case of lightning impact, this structural damage is a result of complex interactions between several multi-physical mechanisms, such as thermal effects due to Joule heating, the generation of a plasma channel, transient mechanical forces due to magnetic volume forces, shock waves from the supersonic expansion of the hot plasma channel and shock waves caused by exploding material [8,9]. In order to understand these complex phenomena in more detail, several experimental studies have been performed [10,11,12,13,14]. Typically, this involves the generation of an artificial lightning strike as defined in SAE ARP5412B [15]. Figure 1 illustrates the four components of the artificial lightning strike defined in the standard. Most artificial lightning strike tests reported in the literature focus on the first part of this pulse, the Current-Component-A-Norm-Pulse, for direct effect testing with the following parameters: peak current 200 kA ± 10%, action integral 2 × 10⁶ A²s ± 20%.

The data typically reported for such experiments include in situ and ex situ data. The in situ data include the electric current vs. time response and the resulting velocity measured at the back face of the composite panel by means of interferometry (for example VISAR or PDV measurement systems). The reported ex situ data focuse on the extent of the damage zone by optical analyses, C-scan and X-ray. The ex situ data are subsequently used for verification and validation of numerical simulation models for describing the damage induced in composite materials by the artificial lightning strike. As previously mentioned and discussed in detail by Karch et al. [8,9], the primary structural effects in composite structures caused by lightning strikes are caused by several competing mechanisms, such as thermal effects, transient mechanical forces and shock waves. A numerical simulation model should take into account all of these multi-physical phenomena in order to have predictive capability—in particular with respect to the observed damage. In some cases, due to the aforementioned ease of experimental accessibility, the recorded back-surface velocity is also used as means of model validation. However, it remains unclear which of the effects of the lightning strike affect the back-surface velocity, and which affect only the damage. Also, it remains unclear if the back-surface velocity can be correlated to the extent of damage observed in the composite panels. This manuscript therefore analyzes the relationship between the peak current of the artificial lightning and the measured back-surface velocity based on experimental data presented in [16]. If a correlation can be found, this could be a first indicator that the mechanical forces affect the back surface velocity.

2. Material and Methods

2.1. Material

The material used in [16] was the Aerospace grade CFRP prepreg material AS4/8552 distributed by Hexcel. Several types of panels were produced with quasi-isotropic stacking sequence (0/45/90/-45)_2s. The curing process followed the supplier’s recommended curing cycle. The first type of panel was produced from CFRP prepreg only (type A). In real aerospace structures, some sort of lightning strike protection system is integrated into the composite structure in order to avoid severe structural damage in the event of a lightning strike. Typically, this lightning strike protection consists of metallic meshes or foils (typically made from aluminum or copper) [17,18,19,20]. In order to account for this industrial practice, additional panels are produced featuring a layer of copper mesh protection. Panels of type B feature a medium lightning strike protection of type AE 195 R, having an area weight of 195 g/m² and a rhombus shaped mesh with diagonals of length 1.40 mm (±6%) and 2.54 mm (±5%) and a mesh thickness of 0.152 mm (±10%). Panels of type C featured heavy lightning strike protection of type AE 815 R, having an area weight of 815 g/m² and a rhombus-shaped mesh with diagonals of length 1.50 mm (±6%) and 3.00 mm (±5%) and a mesh thickness of 0.254 mm (±10%). All test panels (type A, B, C) had dimensions of 400 mm × 400 mm. Prior to testing all panels were inspected using a SAM TEC EVOLUTION II Scanning Acoustic Microscope in order to identify pre-existing damage. For this purpose, the specimens were placed in a water bath so that the scanning probe was located at the back side of the plate (with no physical contact). An excitation frequency of 10 MHz was used during the scans. Initial damage was not observed for any of the panels tested.

2.2. Artificial Lighnting Strike Tests

A total of eight artificial current component A (see Figure 1) lightning strike tests were performed at Fraunhofer EMI’s custom-built lightning strike test facility, located in Efringen–Kirchen: two for the type A panels, two for the type B panels and four for the type C panels. The test specimens were mounted between two aluminum AW-6060 frames resulting in a free test section of 340 mm x 340 mm. The cathode is centrally positioned on the protected side of the panel. The peak current in the tests was varied in the range of 133 kA and 212 kA. The desired lightning strike properties were achieved by connecting a total of four capacitors, with a capacity of 650 µF and a maximum energy of 55 kJ each, in parallel. A spark gap was used as a fast high-power switch. The wiring was optimized in order to reduce the inductance allowing for short rise times for the lightning strike. A Photonic-Doppler Velocimeter (PDV) was used to measure the back-surface velocities of the CFRP panels. These back surface velocities, which are in the range of several tens to several hundred of meters per second, are a result of wave propagation through the thickness of the composite panel. Each time the wave reaches the free surface, the wave is reflected and the free surface velocity changes. The PDV is an established measurement technique from shock physics. The Doppler effect is used for measuring velocities using laser interferometry. Details on the PDV are described in the review by Dolan [21]. Following standard procedures, the evolution of electric current with time was recorded during the artificial lightning strike tests. This data can be directly extracted from the electric circuit. The test setup is shown in Figure 2. After testing, the composites were checked for internal damage (delamination) using ultrasonic inspection (C-scan) using the same instrumentation used before testing.

2.3. Analysis Method

In a first step, the recorded in situ data was plotted in a chart, where the abscissa was the recorded peak current in kA, recorded directly from the electric circuit, and the ordinate was the recorded maximum back-surface velocity in m/s, recorded from the PDV. In a second step, the abscissa was the pre-test weight normalized with the electric current. It is noted that the post-test weight of the panels differs from the pre-test weight of the panels due to evaporation of parts of the copper mesh and the composite resin under artificial lightning strike conditions. This normalization step was not initially anticipated in the design of the experiments. Therefore, the pre-test weight of the panels had to be estimated using the volume of the panels, the density of the composite and the area mass of the lightning protection. Two kinds of data are analyzed: the total pre-test weights of the panels and the pre-test weights of the protection. This form of data analysis differs from [16], where the electric current was normalized with the total panel weight only. This approach was selected in order to separate the effects of total panel weight and the weight of the copper mesh. If the same normalization had been chosen as in [16], data for the unprotected panels would go towards infinity for the case of normalization with the weight of the copper mesh (division by zero). By inversion of the fraction, this problem is suppressed.

3. Results

3.1. Phenomenology

The measured peak currents and maximum back-surface velocities reported in [16] are summarized in

Table 1. Recorded ex situ data for artificial lightning strike tests, extract from [16].

Test No.	Panel Type	Peak Current [kA]	Back-Surface Velocity [m/s]
1	Medium protection (B)	180	51.8
2	Medium protection (B)	185	54.6
3	Strong protection (C)	200	54.6
4	Strong protection (C)	212	60.8
5	Strong protection (C)	183	47.5
6	Strong protection (C)	175	43.7
7	No protection (A)	133	35.6
8	No protection (A)	212	86.0

. The recorded peak velocities are on the order of several tens of meters per second (35.6 m/s to 86 m/s), which is in line with test reports by Karch et al. [20]. Photographs of the panels after testing are shown in Figure 3. Tests 1 and 2 were performed on panels with medium protection, tests 3–6 were performed on specimens with heavy protection, and tests 7–8 were performed on specimens without protection. It is highlighted that for all panels featuring lightning strike protection (types B and C) no delamination was observed. The unprotected panels, type A, suffered severed levels of delamination (372 mm² and 743 mm², respectively). The protected panels analyzed within this manuscript therefore did not suffer from any relevant structural damage due to the artificial lighting strikes. The only damage that occurred for these tests was some minor damage due to local evaporation of the to the copper mesh. However, this limited local damage is not critical. The interested reader can refer to [16] for a more detailed analysis of the extent of the damage, including local evaporation of the copper mesh.

3.2. Relationships between Lightning Strike Parameters and Back-Surface Velocity

The data reported in Table 1 is graphically displayed in Figure 4. The abscissa reflects the applied peak current in kA, the ordinate reflects the resulting peak velocities measured during at the back face of the composite panel. Orange triangles mark the tests carried out on unprotected composite panels (type A), blue diamonds mark the tests carried out on composite panels with medium protection, green circles mark the tests carried out on composite panels with heavy protection.

Considering the composite panels with heavy protection only, there seems to be a linear correlation between the peak electric current and the resulting back surface velocity. Similar analyses cannot be performed for the composite panels of type A and B due to an insufficient amount of data points (only two each). Looking at the data in general it seems obvious that

• the peak velocity increases with increasing peak current for constant levels of protection
• and the peak velocity reduces with increasing levels of protection for constant levels of applied peak current.

However, there does not seem to be a master curve allowing direct correlation of the peak back surface velocity to the applied peak electric current. In [16], the authors proposed normalizing the peak electric current with the total weight of the panel. Following a similar approach, Figure 5 plots the peak back surface velocity on the ordinate and the weight normalized with the peak electric current on the abscissa. Two types of data are displayed here. The solid symbols indicate only the weight of the protection, normalized with the peak electric current. The hollow symbols indicate the estimated total panel weight normalized with the peak electric current. The reason for the two forms of data analysis is to identify if any potential correlation between the peak back surface velocity and panel weight are only dominated by the weight of the protection or also by the total panel weight. If a linear master curve can be observed only for the protection, this is a clear indication that the added mass of the protection is the key factor responsible for a potential correlation. If a master curve can be found for the total weight, this implies that the peak surface velocity is dominated by inertia effects. This is a rather important finding of this work as it has practical implications on the common practice of validating numerical simulation models—see for example the work by Karch et al. [20]—by comparing the predicted surface velocities against experimentally measured velocities. Under the assumption that the surface velocity is mainly driven by inertia effects, an accurate prediction of these velocities can be used as verification of a correct implementation of momentum transfer. However, no information can be extracted which would validate the other relevant components of the multi-physical modeling approaches.

Analyzing the data obtained for the case considering only the weight of the copper mesh protection does not provide clear evidence of a master curve that describes all data points. However, a similar linear trend, as seen in Figure 4, is observed for the panels with heavy protection levels (type C). Analyzing the data for the case considering the total weight of the panel, there seems to be a direct, linear correlation between the panel mass normalized with the peak electric current and the measured peak velocity at the back surface of the composite. Furthermore, it is noted that the measured back surface velocity is not an indicator for damage suffered by the composite panel. The same analysis cannot be justified for the unprotected case and the case with low levels of protection as only two data points are available for those cases.

4. Conclusions

In this manuscript we analyzed available experimental artificial lightning strike test data on carbon fiber reinforced composite panels with different levels of protection, taken from [16]. The question to be answered was if there were any correlations between the measured back surface velocities of the composite panels and relevant parameters of the artificial lightning strike. The following observations were made:

• The peak velocity increases with increasing peak current for constant levels of protection.
• The peak velocity decreases with increasing levels of protection for constant levels of applied peak current.
• Both the medium (area weight 195 g/m²) and the heavy (area weight 815 g/m²) copper mesh successfully protected the composite from the artificial current component A lighting strike. In consequence, all protected panels did not suffer any delamination. The medium level of protection therefore seems to be sufficient for protecting quasi-isotropic AS4/8552 carbon/epoxy composites from lightning strike events.
• For the tested composite panels with protection, no severe structural damage (e.g., delamination) was observed. The back surface velocity can therefore not be taken as a measure of threat for the composite panels.
• The relationship between the peak velocity and the total panel weight normalized with the peak electric current can be described by a linear master curve, which is independent of the level of protection.

The last observation leads to the main conclusion of this manuscript: As there is a linear relationship between the measured peak velocity at the back face of the composite panels and the total panel weight normalized with the peak electric current, the back surface velocity correlates to the mechanical impulse induced by the lightning strike and the inertia of the plate. In consequence, a correct numerical prediction of the back surface velocity only implies that the momentum transfer of the finite element code is working correctly. In our opinion the back surface velocity is therefore not an adequate measure for simulation code validation. Instead, the simulation community should work on predicting the extent of damage in the copper mesh (evaporation) and in the composite (delamination).