the characterisation of hail and fraudulent impacts to vehicle body panels

The characterisation of hail and fraudulent impacts to vehicle body panels Visvarajah Somasundaram

Abstract On the 25th December 2011 there was a hail storm in the state of Victoria, Australia,

which caused approximately AU $712 million worth of damage. Some of this damage was caused to passenger vehicles. Through the investigation of damaged vehicles its shows that some of the damages are created intentionally by the vehicles owners with the use of different tools and/or objects'. Hence, a robust method is required in conclusively differentiating genuine hail damage from a fraudulent damage on vehicle body panels. The aim of this project to characterise the hail and fraudulent impacts to vehicle body panels with respect to deformation size and shape as well as investigate FEA modelling for hail impact.

Due to the availability of resources and time constraints only hail impact experiments to a passenger vehicle bonnet (from a late model Proton Satria) using 40mm spherical hail were permitted. FEA simulations using Abaqus/Explicit assumed a smooth and spherical shaped hail impacting a flat thin panel.

It was observed that 40mm hail impacting a bonnet at around terminal velocity gave dent diameters ranging from 9mm to 28mm and dent depths ranging from 1.49 mm to 0.20 mm. The simulated results showed that a 40mm hail impact on mild steel plate of sizes 100mm2, 150mm2, 200mm2, and 250mm2 yielded dent depths from 1.79 mm to 1.59 mm and dent diameters from 25.6mm to 24.9 mm.

Keywords Hail, Abaqus/Explicit, terminal velocity,

1 Introduction The need to understand the full

impact of hail damage to vehicles has been an ongoing topic of concern to both owners and insurance industry since there is a recent rise in fraudulent insurance claims. According to historical disaster statistics published by insurance council of Australia, 11 hail storms were recorded in the past decade. On Christmas day 2011, a large hail storm hit Melbourne

and country Victoria causing $712 million worth property damage and also resulted in a large number of fraudulent insurance claims.

The Magistrate Court of Victoria held a dispute case concerning a Ford utility vehicle damaged in the Christmas day hail storm. The insured, AP Carpentry Pty Ltd had been denied indemnity by the insurer, Insurance Manufacturers of Australia Pry Ltd. Through forensic investigation, the insurer found that

1School of Aerospace, Mechanical, and Manufacturing Engineering, RMIT University, AUSTRALIA

approximately 90% of damages were inconsistent with hail damage and had been manufactured with tools (“The inconsistent indentations exhibited scratch marks and often appeared in linear or clustered patterns”2).

The insured’s expert witness conceded that approximately 16% to 36.6% of damages were inconsistent with hail damage. The court found that in addition to some genuine hail damage, the insured had intentionally manufactured a significant amount of damage over the vehicle in an attempt to render the vehicle a total loss and make a claim for the agreed value of the vehicle. Consequently, the insurer was entitled to deny indemnity. Previous experiments conducted at Delta-V Experts concluded on cosmetic observations of damaged surfaces on vehicle body panels. These observations alone could not conclusively distinguish between genuine and fraudulent hail damages in all cases. In contribution to these observations, additional investigation into hail impacts were required to provide numerical backing. The objective of this paper is to characterise hail and fraudulent damages to vehicle body panels with respect to deformation size and shape and investigate mathematical simulations.

2 Hail Hail is a naturally occurring and often highly localised phenomenon where small balls, or irregular lump clusters are precipitated from an ice, water, and air mixture. It is generally opaque and has a layered structure. Common hail sizes can range between 5-100 mm in diameter.

For sizes between 5-10 mm, they generally appear spherical or conical in shape. 10-20 mm sized hailstones tend to be ellipsoidal or conical. Larger 10-50 mm hailstones take on ellipsoidal shapes with lobes, while still larger hailstones between 40-100 mm appear irregular (including disk shapes) with protuberances5.

The Australia Bureau of Meteorology defines a large damaging hail as requiring Ø20mm ($2 coin size) or greater. This is also a defining parameter for a severe thunderstorm.

Large Hail commonly occurs in mid latitude between 30˚ and 50˚ in both northern and southern hemispheres (between the Arctic Circle and Tropic of Cancer and between the Antarctic Circle and Tropic of Capricorn) during late Spring and early Summer. During this period, the surface temperatures are sufficiently warm to promote instability associated with strong thunderstorms, while the upper atmosphere is still cool enough to support ice3.

2.1 Hail Formation Hail is formed in thunderstorm clouds called cumulonimbus consisting of ice crystals and liquid water droplets below 0°C. Collisions with supercooled water droplets caused by powerful updrafts of air within the storm promote hail growth. The supercooled water droplets freeze over the surface of an ice crystal, frozen raindrop, dust or other nuclei. The cycle of freezing supercooled water droplets forms the layered structure of hail. The cycling updrafts cause continual hail growth until it can no longer support the weight of the hail or when the hail is pushed out of the draft and falls to Earth (illustrated in Figure 1).


Figure 1. Hail Formation7. There are two forms of hail growth; dry and wet growth. In dry hail growth, the air temperature is well below freezing, causing the water droplets to freeze upon contact with the nucleus. This sudden freezing process traps air bubbles in the hail, which gives a cloudy layer of ice.

In wet growth, the air temperature is below freezing but not enough to promote sudden freezing upon contact. The water droplets spreads and slowly freezes over the surface of the hailstone allowing air bubbles to escape leaving a translucent layer of ice.

2.2 Density of Hail The density of hail can vary significantly and is dependent on numerous factors such the formation process (e.g. wet or dry growth), temperature and size. Field et al. shows that hail with sizes smaller than 20 mm can have its density significantly varying between 50 to 890 kg m-3, while hail with larger sizes can have its density varying between 810 to 915 kg m-3 (Listed in Table 1).

Table 1. Experimentally determined densities of graupel and hail (>5mm) and size ranges5. Size Range [mm] Density Range [kg m-3] Source 0.5 - 3.0 50 - 450 Locatelli & Hobbs (1974) 0.5 - 1.0 450 - 700 Zikamunda and Vali (1972) 1.0 - 2.0 250 - 450 Zikamunda and Vali (1972) 0.4 - 3.0 80 - 350 Bashkirova and Pershina (1964) 0.5 - 3.0 850 - 890 Braham (1963) 0.5 - 6.0 500 - 700 List (1985) 0.8 - 3.0 130 - 130 Magono (1953) 8.0 - 19.0 310 - 610 Knight and Heymsfield (1983) 1.0 - 7.0 200 - 700 Heymsfield (1978) 26.0 - 36.0 834 - 856 Prodi (1970) 11.0 - 31.0 810 - 900 Vittori and Di Caporiacco (1959) 9.0 - 39.0 870 - 915 Macklin et al. (1960)

2.3 Hail Fall Speed The fall speed of hail can be highly variable and can be affected by environmental factors such as updrafts and downdrafts. A simplification is made by assuming the hail to be a smooth

spherical object and falling at terminal velocity for worst case scenario. The theoretical terminal velocity (VT) of hail is expressed in equation (1).


(1) Where VT is terminal velocity (m/s), g is acceleration due to gravity (9.81 m/s2), ρHail is the density of hail (kg/m3), DHail is the diameter of hail (m), CdSphere is the drag coefficient of a sphere, and ρAir is the density of air (kg/m3). The expression shows that as the size and density of hail increases, the terminal velocity increases. However, as the drag coefficient and air density increases, the terminal velocity decreases.

The drag coefficient of hail is a function of Reynold’s number, which generally exceeds 102. Figure 2 shows drag characteristics with increasing Reynold’s number for a smooth and rough sphere. A Cd range from 0.5 and 1.0 can be expected for hail impacts to vehicle body panels.

Figure 2. Drag Coefficient of a sphere9. Figure 3 shows the fall speed range of different hail sizes bounded by terminal velocities for Cd of 0.5 (fastest fall speed) and 1.0 (slowest fall speed) calculated using hail density of 900 kg/m3, 9.81 m/s for acceleration due to gravity, and air density of 1.225 kg/m3 at 15˚C, which is

the approximate ambient temperature during a hail storm in Australia.

Figure 3. Fall speed range.

2.4 Economic Impact of Hail Historical Disaster Statistics published by the Insurance Council of Australia detail at least 11 recorded hail related events within the past decade. The figures are only an approximation of the insured losses (e.g. home, content, vehicle damage) based upon reported data. Only events with potential losses exceeding AUD$10 million are recorded. Therefore, actual hail related events may be greater than recorded and the associated costs of recorded events may be an underestimate of actual figures.

From Table 2, the total 2011 normalised cost of all 11 events is approximately AUD$3.2 billion, averaging to about AUD$291 million per event. Though actual figures concerning vehicle damages are not specified, even if 20% is attributed to vehicle damages in each event, the normalised cost per event concerning vehicle damages would approximately be AUD$58 million.


Table 2. Historical disaster statistics from the Insurance Council of Australia for hail related events over the past decade4.

Event Catastrophe Number

Date (dd/mm/yyyy) State Original Cost

(AUD$million)

2011 Normalised

Cost (AUD$million)

VIC Christmas Day Storms*

CAT 118 25/12/2011 VIC 712 N/A

Melbourne Storms*

CAT 102 6/03/2010 VIC 1044 1160

Severe Hail storms

CAT NSW 07/6 9/12/2007 NSW 415 486

Hail Storm CAT NSW 07/4 9/10/2007 NSW/QLD 97 109 Hail CAT NSW 06/1 31/10/2006 NSW 51 60 Hail CAT QLD 05/2 12/10/2005 QLD 61 89 Hail N/A 19/05/2005 QLD 17.6 28

Hail, Storm CAT VIC 05/1 16/05/2005 NSW/TAS/VIC 216.7 304 Hail, Storm CAT NSW 04/1 13/12/2004 NSW 32.3 46

Hail CAT QLD 04/1 24/01/2004 QLD 28.5 54 Hail N/A 31/12/2003 VIC 100 156

*denote events known to have contained hail, N/A - Not Available

3 Literature Review Published literature regarding hail damage to vehicle body panels are very limited. However, hail impact studies are mainly regarding aerospace applications (e.g. aircraft skin) at high impact speeds (excess of 80 m/s) with large deflections. This is excessive compared to hail impacts to vehicle body panels, which generally involve relatively low impact speeds (less than 50 m/s) and small deflections. Though there may be some theoretical differences between low and high impact speeds, the same foundation can be used to develop a method to analyse low hail impact speeds. Through literature reviews, the expectation is to find and develop a mathematical model/simulation that can be applied to hail impacts to vehicle body panels. Hail impacts occur within a fraction of a second and are highly dynamic. Therefore, a transient dynamic

finite element analysis approach is used. Common commercially available finite element packages used in literature are LS-DYNA and Abaqus. Both offer different integration methods in which hail impacts can be analysed. These are Lagragian, arbitrary Lagrangian Eulerian (ALE), and smoothed particles hydrodynamics (SP). Lagragian correlates the mesh with the material properties that may result in excessive distortions and numerical instability with large material deformations. ALE is a combination of Lagragian and Eulerian integration methods, allowing large deformations because of the Eulerians method fixing the mesh in space and allowing the material to flow through it. SPH allows large material deformation and is a meshless integration method. The material property is represented as grouped particles rather than a mesh. The material property is distributed


throughout the model based on the distance between each particle. Previous finite element models include the use of an elastic-plastic model with plastic strain and pressure failure criterion, where the hail behaves like a fluid (only carrying hydrostatic stresses) after failure. Recent studies of hail impact try to account for the strain rate sensitivity of ice under compression, where ice is ductile under low strain rates and brittle under high strain rate. This observation usually occurs at strain rates greater than 10-2 s-1.

4 Finite Element Analysis of Hail Impacts using Abaqus/Explicit The impact of hail onto vehicle body panels was simulated using Abaqus explicit. The intent of this simulation to validate the experimentally measured depth and diameter of the dent, thus demonstrate that a suitable model has been developed. Though there are numerous factors influencing hail impacts, not all factors can be accounted for within this project. A simplification was made when developing the model where it is assumed that the hail is smooth, sphere shaped, and impacts normally to a flat thin plate (vehicle body panel).The set of units used in Abaqus are shown in Table 3. Table 3. Units used in Abaqus

Quantity SI Length m Force N Mass kg Time s Stress Pa (N/m2) Energy J

4.1 Symmetry The hail and thin flat plate was originally modelled as a ¼ scale using symmetry to reduced computational cost. However, the ¼ scale model encountered mesh distortion and instability problems (shown in Figure 4) and a decision was made to develop a full scale model.

Figure 4. Mesh distortion.

4.2 Material Characteristic of Hail The ice material model in Abaqus is composed of a simple elastic - plastic behaviour with failure criteria based on tensile hydrostatic pressure. The elastic material properties for hail are obtained from literature are shown in table 4 . Table 4. Abaqus material property input for hail10. Material Characteristics Values Density (kg/m3) 900 Young's Modulus (Pa) 9.38E+09 Poisson's Ratio 0.33 Yield stress at εp = 0 and 1 (MPa)

5.2E+06

Figure 5 shows three curves fitted to experimental data published by Jones and Kim and Kuene. These curves are defined as stress ratios scaling 5.2MPa yield strength over the range of strain rates.


Figure 5. Curve fits to compressive strength versus strain rate data10.

Based on simulations, Tippmann, J.10 found that the lower bound curve showed a better comparison to peak impact forces of experimental data than the average and higher bound curve. Therefore, the lower bound curve was used as the strain rate input for the hail material characteristic. The curve value is then scaled to stress ratios over the 5.2 MPa yield strength for the range of strain rates10. The exact values for the lower bound curve listed in Table 5. Table 5. Lower Yield Strength Ratio Input10.

Stress Ratio Strain Rate (S-1) 1 0

1.01 0.1 1.267017189 0.5 1.382015232 1 1.649032421 5 1.764030465 10 2.031047654 50 2.146045697 100 2.413062886 500 2.52086093 1000

2.795078118 5000 2.910076162 10000 3.177093351 50000 3.292091395 100000 3.559108583 500000 3.674106627 1000000

The tensile failure pressure setting used in the Abaqus model was to specify the deviatory stress failure as brittle and the pressure stress as brittle. The deviatory stress of the failed material is set to zero when the failure threshold pressure is reached, while the hydrostatic stresses (both compression and tension) in the material remain up to the cut-off stress.

Tippmann, J.10 showed that a tensile failure pressure of 517kPa gave good correlation with experimental data. The tensile failure pressure criterion was used and inputted into the hail material characteristic in Abaqus via keyword.


4.3 Hail Size and Velocity Data obtained from the Bureau of Meteorology Australia show damaging hail sizes commonly found range between Ø40 mm to Ø100 mm. Taking into account the previous results obtained at Delta-V Experts and limitations of the experimental equipment, the hail sizes originally chosen for FEA modelling were Ø40mm, Ø60mm and Ø80mm. The hail impacts were assumed to occur at terminal velocity with a drag coefficient (Cd) of 0.5 and with a density of 900kg/m3. Table 6 shows the terminal velocity associated with each hail size used in developing the FEA models. Table 6. Hail terminal fall speed for hail diameter sizes of 40 mm, 60 mm, and 80 mm.

Hail Size (mm)

Fall Speed (m/s)

Fall Speed (km/h)

40 19.8 71.3 60 24.2 87.1 80 28.0 101

4.4 Material Characteristic of Vehicle Panels Tensile testing is one of the most fundamental tests for engineering to determine the nonlinear properties of a material, therefore a tensile test was conducted to the specimen collected from vehicle body panel of Proton Satria.

Figure 6. Specimen collected from proton Satria.

the results obtained from tensile test are engineering stress and engineering strain, however Abaqus expects the stress strain data to be entered as true plastic strain and true stress. Hence an appropriate conversion were made and imported to Abaqus. Vehicle body panels can be very complex in shape. For the finite element model, the panel is assumed thin and flat.

Figure 7 True Stress Strain curve derived from tensile test for Proton Satria.

The Proton Satria bonnet obtained showed a general thickness of 1mm. Based on this, the flat plate was modelled with 1mm thickness using material properties derived from tensile test. The material characteristics of panel is listed in Table 7 and Table 8.

Table 7. Material Characteristics of Mild Steel

Material Characteristics Values Density (kg/m3) 7850 Young's Modulus (Pa) 180.56E+09 Poisson's Ratio 0.33

Table 8. Plastic Characteristic of Mild Steel

Yield Stress Plastic Strain 2.43E+08 0 2.50E+08 0.006816926 2.76E+08 0.018797865 3.00E+08 0.035155276


3.25E+08 0.056113752 3.50E+08 0.081851277 3.75E+08 0.115709227 4.00E+08 0.156856325 4.25E+08 0.202525738 4.50E+08 0.251981658 4.70E+08 0.300070023

Plate sizes 100mm2, 150mm2, 200mm2, and 250mm2 were modelled based on the effective sheet metal exposed on the bonnet with no bonnet frame reinforcing underneath observed on both the Holden Commodore VX sedan and Proton Satria bonnets.

4.5 Elements The solution procedure in Abaqus/Explicit uses an explicit integration rule and diagonal, lumped mass matrices, therefore, the Abaqus/Explicit element library is limited to linear elements with reduced integration. 3D8R elements were used to model the hail and flat plate.

4.6 Mesh Abaqus/Explicit is conditionally stable because the stability of the solution is dependent on the time increment (time increment must be less than the critical time increment). Equation (2) below expresses the stable time increment. This is determined only by Abaqus/Explicit and is influenced by the smallest element in the model. Therefore, it is desired to have uniform elements in the model for a faster computational time.

(2) Where, Le = Length of smallest element E = Young’s Modulus

ρ = density A partition scheme recommended by Tippman, J.10 was used to partition the hail as shown in Figure 6.

Figure 8. Sketch of key sphere dimensions with normalised unit diameter10.

Figure 9. Fully partitioned 40mm diameter hailmodel

A structured mesh seed size of 0.001 was used to model the 40mm hail. Bias mesh is used as shown in the figure 9 hence finer elements near the impact face.

Figure 10. applied mesh on 40mm diameter hail model


4 elements where used across the thickness (seed size of 0.00025 across plate thickness) to accurately visualise deflection of the plate using 3D8R elements. Since the FEA models aimed at visualising the deflection of the plate a finer mesh was required at the impact area. A seed bias was used on the plate with a bias ratio 4 and created 40 elements around the impact area.

Figure 11. Full Scale Ø40mm Hail and 10000mm2 plate Model Mesh.

4.7 Constraints Between hail and bonnet models a general contact with a hard and frictionless interaction property was specified as shown in figure 8.

Figure 12. general Interaction between Outer Surface of Hail and plate Surface.

Dynamic, Explicit step was created for the impact between the hail and the bonnet with a time period of 0.003 s, which was the average impact duration from the impact experiment. A Linear bulk viscosity of 1.2 and quadratic bulk viscosity of 0

was used based on Park, H.13 sensitivity studies.

Fixed boundaries with pinned joint were added to all four edges of the flat plate and a terminal velocity of 19.8m/s was specified for the hail via predefined fields as shown in Figure 9.

Figure 13. Hail Impact Model Constraints

5 Hail Impact Experiment Setup and Procedure Due to time constraints and availability of resources only a small-scale hail impact experiment was carried out on a late model Proton Satria bonnet to validate the FEA model using Ø40mm moulded hail. The Ø60mm and Ø80mm hail moulds made using a 3D printer encountered sealing problems and were abandoned due lack to time and resources. The equipments used during the hail impact experiment are detailed in Table 9.

Table 9 - Experimental Equipment

Experimental Equipment Ø19mm x 1.5m Powerband x 1 Clamp Hose Fit 13-25mm x 2 Table Tennis Balls x 24 Ø19mm x 2m Reinforced Pressure Hose Ratchet Crossbow trigger Delta-V Experts' crossbow frame Proton Satria Bonnet High speed camera


Camera tripod Nylon rope Steel bolt x4 Steel nut x 4 Steel plate x 2 Laptop with VLC Player LED lamp Black marker Digital depth micrometer Ruler Blu-tack A red bonnet from a late model Proton Satria was purchased for the impact experiment. All pre-existing damages were marked (marked in black in Figure 10) before the impact test.

Figure 14. A red late model Proton Satria bonnet with pre-existing damages marked.

The bonnet was placed upright against a wall facing the crossbow on plastic boxes at a measured distance using a tape measure (shown in Figure 11). A high speed camera was placed on a tripod perpendicular to the experimental setup to record the hail impact.

Figure 15. Experimental setup of hail impact test.

5.1 Modified Crossbow To save time and resource, the crossbow previously used at Delta-V Experts for their impact experiments was reused. However, existing problems with the current design had to be addressed. It was noted that the crossbow was not durable, requiring constant replacing of rubber cords and zip ties, which affected the precision and accuracy and was not able to launch larger hail sizes (e.g. Ø80mm) at terminal speeds.

The previous crossbow was a T-frame construction (a straight lathe and stock) with two rubber cords affixed to the lathe using nuts, bolts, and zip ties (shown in Figure 12).

Figure 16. Rubber cord affixed to lathe on previous crossbow using nuts, bolts and zip ties.


The two rubber cords were joined to single nylon rope using zip ties (shown in Figure 13).

Figure 17. String component on previous Delta-V Experts crossbow made from rubber cords, zip ties, nylon cord, and a cardboard pad.

The nylon rope had a cardboard pad attached and was used to pull back the rubber cords and mount on the trigger (shown in Figure 14).

Figure 18. String component on previous Delta-V Expert crossbow mounted on trigger in safety lock position. The new crossbow utilised the existing T-frame. A single Ø19mm x 1.5m spearfishing powerband was double knotted at both ends and mounted to the

lathe using bolts, nuts, and steel plates as shown in Figure 15.

Figure 19. Current crossbow with powerband mounted on lathe using nuts, bolts, and steel plate. The introduction of the thicker powerband eliminated the use of cable ties and provided greater durability and elastic potential energy. The cardboard pad was replaced with a reinforced pressure hose that was slotted on the powerband and fixed using 2 clamps shown in Figure 16. The reinforced hose ensured a fixed spacing between the clamps (approximately 80mm, which was the size of the largest hail planned for testing) by resisting the compression of the powerband as it was drawn back.

Figure 20. Current crossbow string component.


As a consequence of using the powerband, greater effort was required to draw back the powerband. Therefore, a ratchet was required to draw back the powerband, which was mounted to a bolt on the T-frame. Two nylon ropes were tied to the hose, one was used to draw back the powerband using a ratchet and the other was used as a trigger release rope (shown in Figure 17).

Figure 21. String component on current crossbow mounted on trigger in safety lock position.

5.2 Ø40mm Hail Impact Speed Validation Test Before commencing the hail impact test on the bonnet, a series of impacts against a wall using table tennis balls filled with water were conducted. This was done to determine the appropriate release rope length used to achieve an impact velocity of 19.8 m/s (the calculated terminal velocity for a Ø40mm hail). The launch speed of the hail was found using a high speed camera with a recording rate of 1000 fps (frames per second). The number of elapsed frames between the hail leaving the crossbow and impacting against the wall were counted using a laptop with VLC player. The velocity of the hail was calculated using the formula below.

(3)

Where smeasured is the measured distance between the crossbow and the impacted object, fcamera is the recording rate of the camera, and felapsed is the number of frames elapsed between the hail leaving the crossbow and contact with the impacted object. The appropriate rope length was found to be approximately 80cm accounting for some length lost due to tying. The rope gave a reach of 25cm between hose and the trigger and gave an impact velocity range between 17.8 and 21.8 m/s.

5.3 Ø40mm Hail Impact Testing and Analysis Procedure The procedure used during the impact test once the equipment was setup are as follows:

1. Measure and record distance between crossbow and bonnet.

2. Ratchet the powerband back using the nylon rope and mount release rope on trigger.

3. Place hail in front of hose. 4. Press record button on high speed

camera to start recording. 5. Pull trigger to release hail. 6. Press record button on high speed

camera to stop recording. 7. Mark and number impact point on

bonnet.

A total of 20 impacts were done following steps 1 to 7. The hail impact speeds were


calculated using the procedure described in the previous section. Measurements were taken on the bonnet concerning the diameter and depth of each hail dent. This was done in a dark room using an LED lamp to scan across the dented surface. This method was previously used at Delta-V Experts and provides better clarity of the dented surface. The procedure used is detailed below.

1. Find and mark dent center by using a LED lamp to scan over dented surface.

2. Use a calibrated digital depth micrometer to measure the depth of the dent. Record measurement.

3. Repeat step 2 twice. 4. Attach a ruler below the dented area

on the bonnet using Blu-tack. 5. Align camera with dented area using

the tripod. 6. Use the LED lamp to highlight the

dented circumference and take a photograph (shown in Figure 18).

7. Repeat step 5 twice at different light angles.

The photographs taken were analysed on a laptop to determine the diameter of each hail dent.

Figure 22. Hail dent diameter measured using a ruler.

6 Results 6.1 Experimental Results Figure 19 shows the location of each hail impact highlighted in yellow on the bonnet.

Figure 23. Hail Impact Points highlighted in yellow on Red Late Model Proton Satria Bonnet.

Figure 20 is an image of the underside of the bonnet rotated 180° about the vertical axis with the hail impact points highlight in yellow.

Figure 24. 180° rotated (about vertical axis) image of Red Late Model Proton Satria bonnet underside with hail impact points numbered in yellow and sheet metal section number in black.


Table 10 shows the dimensions and surface area of exposed sheet metal on the bonnet with respect to the numbered sections in Figure 20 in black. Table 10. Dimensions and Surface Area of Sheet Metal Sections on Bonnet

Table 11 shows the taken measurements (e.g. dent diameter, dent depth, frames, etc.) and calculated results (velocity) from the Ø40mm hail impact test on the Proton Satria bonnet. Highlighted in green are the impacts with little to no hail breakage (such as the hail shown in Figure 21), highlighted in yellow are impacts with no measureable panel damage (e.g. no dent), and highlight in grey are invalid results due to improper launching of the hail.

Figure 25. Hail from impact test H40I15 after bonnet impact.

Section Dimensions (m) Area (m2) 1 0.115 0.11 6.33E-03 2 0.31 0.155 4.81E-02 3 0.275 0.14 3.85E-02 4 0.19 0.17 3.23E-02 5 0.18 0.115 2.07E-02 6 0.17 0.16 2.72E-02 7 0.27 0.145 3.92E-02 8 0.17 0.13 2.21E-02 9 0.23 0.155 3.57E-02

10 0.34 0.18 6.12E-02 11 0.15 0.11 8.25E-03 12 0.3 0.14 4.20E-02 13 0.25 0.14 3.50E-02 14 0.19 0.15 2.85E-02


Table 11. Hail Impact Experimental Results

Hail test No.

Dent Diameter

(mm)

Dent Depth Measurement

1 (mm)


2 (mm)


3 (mm)

Average Dent

Depth (mm)

Camera FPS Frames

Time Taken from

crossbow to bonnet

Distance between crossbow

and bonnet

Velocity (m/s)

H40I01 21 0.44 0.46 0.45 0.45 1000 46 0.046 1.03 22.39 H40I02 26 0.97 1.05 0.99 1.00 1000 46 0.046 1.03 22.39 H40I03 25 1.21 1.24 1.23 1.23 1000 48 0.048 1.03 21.46 H40I04 16 0.21 0.19 0.19 0.20 1000 49 0.049 1.03 21.02 H40I05 18 0.22 0.23 0.21 0.22 1000 50 0.05 1.03 20.6 H40I06 24 1.5 1.48 1.5 1.49 1000 54 0.054 1.03 19.07 H40I07 24 0.63 0.66 0.65 0.65 1000 53 0.053 1.03 19.43 H40I08 16 0.23 0.23 0.25 0.24 1000 54 0.054 1.03 19.07 H40I09 9 0.52 0.57 0.56 0.55 1000 54 0.054 1.03 19.07 H40I10 0 0.01 0.00 0.01 0.00 1000 59 0.059 1.03 17.46 H40I11 7 0.1 0.09 0.12 0.10 1000 99 0.099 1.3 13.13 H40I12 25 0.77 0.79 0.78 0.78 1000 63 0.063 1.3 20.63 H40I13 16 0.25 0.26 0.21 0.24 1000 61 0.061 1.3 21.31 H40I14 12 0.23 0.22 0.24 0.23 1000 59 0.059 1.3 22.03 H40I15 28 1.08 1.04 1.02 1.05 1000 60 0.06 1.3 21.67 H40I16 22 0.46 0.47 0.5 0.48 1000 63 0.063 1.3 20.63 H40I17 19 0.5 0.55 0.54 0.53 1000 64 0.064 1.3 20.31 H40I18 22 1.22 1.18 1.17 1.19 1000 65 0.065 1.3 20 H40I19 19 0.48 0.51 0.49 0.49 1000 67 0.067 1.3 19.40 H40I20 23 0.54 0.58 0.54 0.55 1000 65 0.065 1.3 20 Green indicates little to no hail breakage Yellow indicates no measurable panel damage (e.g. no dent). Grey indicates invalid result.


6.2 Finite Element Analysis Result

Proton Satria body panel

Figure 26. Proton Satria Plate Hail Impact Deflection

Figure 27 Simulated Hail Impact on 10000mm2 Proton Satria

Table 12 Impact Depth and Diameter for different Size Plates of Proton Satria.

Plate Size (mm x mm) Impact Depth (mm) Impact Diameter (mm)

100 x 100 1.79 25.6

`150 x 150 1.64 25.1

200 x 200 1.59 24.9

250 x 250 1.59 24.9


Hail impact simulations for Different hail sizes

Figure 28. Proton Satria Plate Hail Impact Deflection for different hail sizes

Table 13. Impact Depth and Diameter for different Size Plates of Proton Satria.

Hail Size (mm) Fall Speed(m/s) Impact Depth (mm) Impact Diameter (mm)

40 mm 19.8 1.59 24.9

60 mm 24.2 3.6 47.9

80 mm 28.0 5.24 72.3

Figure 29. Simulated Hail Impact on 10000mm2 Flat Plate.


7 Discussion Hail impact experiment shows that most of the hail impacts occurred at or around the bonnet frame as shown in figure 20. The experimental results showed that a 40mm spherical hail impacting a bonnet at around terminal velocity caused dent diameters between 9mm to 28mm and dent depths between 0.20mm to 1.49mm. The experimental results also indicate that the diameter of the dents and the depth of the dents are independent of the presence of the bonnet frame. This is due to a sufficient gap between the bonnet sheet metal and the bonnet frame at the impacted areas preventing the bonnet frame from having influence on the dent resistance of the bonnet.

Figure 30: The bonnet Frame back of the Bonnet.

Based on the observation of the dents following findings were revealed : 1. hail impact will not scratch or mark the paint but the paint may chip ; 2. dents caused by hail will cause the light to move smoothly and continuously across the dent and the light will not "break" or crease; 3. folds and curves on the panels did not affect the shape of the (physical appearance) dent caused to the panel; and 4. for the same size hail the higher impact speed hail caused more damage.

The body panel sizes between (100mm)2 to (250mm)2 were simulated for impact. Results showed that the dent diameter varied between 25.6mm to 24.9mm between plate sizes and a depth ranged of 1.79mm to 1.59 mm with increasing plate size, which is up to 6.7% larger than the largest dent depth (1.49mm) observed during the experiments. The numerical results have good correlations with the experimental results, as its deviate only 6 .7 %.

The hail simulations show very little hail deformation similar to Figure 26. It can be seen in Figure 27 that as the plate becomes bigger, a greater deflection peak is reached but a smaller final deflection of the plate is observed. The greater deflection peak may be attributed to the increased distance between the fixed plate boundary and the centre of impact as the plate became larger, similar to increasing the length of a beam and observing a greater deflection under bending. The lower final deflection may be due to the spring back effect of the plate. The larger the plate, the greater this effect.

Factors which may be attributed to the differences between the simulated and experimental results include:

1. layered structure of the hail: The hail model used for experiment is 40mm moulded ice(monolithic ice). These were created using a spherical split mould having a filling hole. however the Actual hail ice has a spherically layered, or 'onion skin', construction. these layered structure provide an extra degree toughness to the hail. which is not considered during the experiment.


Figure 31. Comparison of moulded ice and Layered Hail Stone

2. Inconsistency of projection: the Projection of ice was achieved using a crossbow hence inconsistency in projecting speed and angle , which would be better if we used a nitrogen gas canon which can control the speed from 10m/s - 200m/s.

Figure 32. Nitrogen gas cannon Setup.

3. Inconsistent hail temperature: The mechanical properties of the ice are strongly depend on the temperature, however during the experiment ice was not launched with consistent temperature.

4. Density of the hail is not determined : during the experiment the weight of the hail is not recorded. hence it is not possible to compare the density between experimental and numerical hail models.

5. The material characteristic of the hail: It would be assumed that if hail breakage can be included into the hail model, the expected plate deflection would be less due to the portion of the impact energy that would be dissipated

through crack propagation and breakage of the hail.

6. Curvature and edges on the bonnet: These would aid in the dent resistance of the bonnet. When an impact occurs normally to an edge or curvature the effective cross-sectional thickness of the bonnet is greater than a flat plate, therefore is more dent resistant.

7. Paint and clear coat: The paint and clear coat of a bonnet and other body panels may also influence in providing some dent resistance.

8. Sources of error: These include the way the depth of the dent was measured, which can significantly vary when measuring on a curvature or edge to the crossbow not launching the hail normal to the bonnet.

Data obtained from the Bureau of Meteorology Australia show damaging hail sizes commonly found range between Ø40 mm to Ø100 mm. The Table 13 shows the Depth and the Diameter of the dents for different sizes of the hail. It is observed that the damage is larger as the hail size increases. this is due to the momentum increases as the size of the hail increases. The developed numerical model will be a great tool to investigate the hail damage in the future. Finite element analysis methods are frequently used in forensic engineering. It gives more detailed results than a physical experiment (quantities can be measured). Since experimental tests are expensive, difficult to perform and time consuming it is obvious to understand the importance of developing numerical models.


8 Conclusion The aim of this paper is to characterise hail and fraudulent damages to vehicle body panels with respect to deformation size and shape. To achieve this aim two different methods were implemented. Initially hail impact experiment was conducted to a Proton Satria bonnet using 40mm moulded hail. Then the experimental results were compared with numerical results. For the numerical method a 40mm hail and bonnet models were made, refined and compared with the experimental data. It was observed that 40mm hail impacting a bonnet at around terminal velocity gave a dent diameter range of 9mm to 28mm and a dent depth range of 0.20mm to 1.49mm. The simulated results showed that a 40mm hail impact on mild steel plate of sizes (100mm)2, (150mm)2, (200mm)2, and (250mm)2 yielded dent depths between 1.79mm to 1.59mm and dent diameters between 25.6mm to 24.9mm. As the good correlation between experimental and numerical results , the current numerical model could be used to verify the real hail damages. Only by changing the material properties according to the vehicle model the maximum depth and diameter of the dent could be determined. Materials properties of the panel is vary according to the vehicle, hence the non linear material properties of the panel needs to be determined through the tensile test and the imported to the model.

Acknowledgements The author acknowledge the guidance and feedback throughout the project from the academic supervisor, Dr. Xu Wang, the industrial supervisor from Delta-V Experts, Dr. Shane Richardson and Mr. Andreas Sandvik for their guidance and advices throughout the project, and to Dr. Toh Yen Pang for his consultation times and guidance with Abaqus FEA modelling.

References 1. National Weather Service Southern

Region Headquarters (2011) Thunderstorm Hazards - Hail, Available at: http://www.srh.noaa.gov/jetstream/tstorms/hail.htm(Accessed: 14th March 2013).

2. Liam Campion (2012) Hail the Fraud Finding, Available at: http://www.hallandwilcox.com.au/news/Pages/Insurable-Interest.aspx#Hail (Accessed: 14th March 2013).

3. Geoscience Australia (2011) Where does Severe Weather occur?, Available at: http://www.ga.gov.au/hazards/severe-weather/severe-weather-basics/where.html (Accessed: 25th March 2013).

4. Insurance Council of Australia (2012) Historical Disaster Statistics, Available at: http://www.insurancecouncil.com.au/assets/statistic/current%20and%20historical%20disaster%20statistics%20aug%2012.pdf (Accessed: 28th March 2013).

5. Field P.R., Hand W., Cappelluti G., McMillan A., Foreman A., Stubbs D. and Willows M. – Hail Threat Standardisation – FINAL report for EASA.2008.OP.25, Date 14 November 2009


6. Bureau of Meteorology (2011) Thunderstorm and Strong Wind Confirmation Reports Overview, Available at: http://www.bom.gov.au/climate/storms/overview.shtml (Accessed: 29th March 2013).

7. Scijinks () What makes it rain?, Available at: http://scijinks.nasa.gov/rain (Accessed: 5th June 2013).

8. Schulson, E.M. (2001) 'Brittle Failure of Ice', Engineering Fracture Mechanics, 68(), pp. 1839-1887.

9. Tom Benson (2010) Drag of a Sphere, Available at: http://www.grc.nasa.gov/WWW/k-12/airplane/dragsphere.html (Accessed: 5th June 2013).

10. Tippmann, Jeffery Dwayne. (2011). Development of a strain rate sensitive ice material model for hail ice impact simulation. UC San Diego: b6994352. Retrieved from: http://www.escholarship.org/uc/item/294018cv

11. Anghileri, M., L. Castelletti, F. Invernizzi, and M. Mascheroni. "A Survey of Numerical Models for Hail Impact Analysis Using Explicit Finite Element Codes." International Journal of Impact Engineering 31, no. 8 (2005): 929-44.

12. Pernas-Sánchez, J. (2012) 'Numerical modeling of ice behavior under high velocity impacts', International Journal of Solids and Structures, 49(), pp. 1919–1927.

13. Park, H. (2006) ‘Resistance of Adhesively Bonded Composite Lap Joints to Damage by Transverse Ice Impact. Thesis, Purdue University


the characterisation of hail and fraudulent impacts to vehicle body panels

Documents