Associazione Italiana per l’Analisi delle Sollecitazioni (AIAS)XXXI Convegno Nazionale –18-21 Settembre 2002, Parma

SESSIONE GPROGETTAZIONE MECCANICA

A STUDY ON STRUCTURAL BEHAVIOUR OF A CAR-BONNETDURING CHILD-HEAD IMPACT, USING PARAMETRIC-

VARIATIONAL CAD/CAE MODELLING TECHNIQUES ANDNUMERIC COMPUTATION BY EXPLICIT CODES

G. Monacellia, V. Petrellaa, A. Naddeoa

A. Donnarummab, N. Cappettib

a Elasis S.C.p.A. – Via ex-aeroporto S.N., 80038, Pomigliano d’Arco (NA) –Italy; e-mail: alessandro.naddeo@elasis.fiat.it

b Dept. of Mechanical Engineering - University of Salerno – Via Ponte DonMelillo, 84084, Fisciano (SA) – Italy; e-mail: Donnaru@diima.unisa.it

ABSTRACTAll over the world, and mainly in United States, since 1977 to 1991, automotiveindustries’ research centres and above all public bodies such as NTHSA and self-governing bodies e enti autonomi as regards automotive firms, also those connectedwith consumers’ associations had processed several statistical data on real accidentbetween vehicles and pedestrians, taking care, obviously, to pedestrians’ injury. Inreference papers of period of twenty years 1977-97 and in documents of EEVC,the scientist describes a proposed homologation test for child-head impact; it isrepresented by an impact simulation of some standardized impactors on car bonnet,in order to evaluate the child-head injury as deceleration of their centre of gravity.Impact time generally doesn’t exceed the 20 ms (milliseconds); in this time we canrecognize two different phases: the first phase, dependent mainly from the localbonnet stiffness, with a length never greater than 4 ms, and a second phase, untilthe end of the phenomenon, dependent from boundary conditions, impactingbodywork parts, impacting engine parts and so on… This is true especially whennear the bonnet we don’t find emerging, quite stiff, parts. In the first phase we canfind the maximum acceleration’s value, that heavily affects the final value of HPC.Our research is focused essentially on vehicle configuration and on material’scarachteristics in the predictable imapct zone of pedestrian and on the vehicle. Ouraim is to create, for child head impact, the same condition of first phase of impact,in which local stiffness play the main rule of the game, realizing a test-case by usinga CAD-CAE parametric/variational model of car-bonnet.

1. INTRODUCTION

All over the world and especially in United States, between 1977 and 1991, research centresconnected with automotive industry, and generally motor vehicles, took care of processingnumerous statistical studies about accident between vehicles and pedestrians with obviousregard to damage typology suffered by pedestrians[1].The awakening of public bodies and private business to this concern started systematic studieson the problem of pedestrians’ passive safety as shown by the numerous scientific articles andby the proposal draft of European Committee about new homologation criteria forautomotives of the future taking care of pedestrian safety[2].In such a context, this paper is intended to study the pedestrian injury at the impact withvehicles and methods of designing and constructing vehicles in order to the reduce the damageitself. This paper excludes the measures connected to the rules of the road.For determining the entity of damage on pedestrians we refer to the measurement scale calledAbbreviated Injury Scale (AIS) (see Tab.1) in the Association for the Advancement ofAutomotive Medicine (AAAM) manual; this association, in fact, during the last twenty years,dealt with processing statistical data in order to analyse the problem of the pedestrian safetyand recognize required changes on vehicles to improve their safety, especially passive safety..

AIS O “No Injury”

AIS l “Minor Injury”

AIS 2 “Moderate Injury”

AIS 3 “Serious Injury”

AIS 4 “Severe Injury”

AIS 5 “Critical Injury”

AIS 6 “Maximum Injury”

AIS 7-9 “Unknown”

Table 1: AIS - Abbreviated Injury Scale

Using such a scale accidents reported in statistics were evaluated and they came to anagreement in considering injury that exceeds the level AIS = 3 mortal for the pedestrian (seeFig.3).Studying case history we found predominance of impact on some body parts instead of others;we especially note a great incidence of “child” pedestrians’ death (see Fig.2) so that isnecessary giving absolute precedence to the analysis of the impact of “child” pedestrian’s headon car bonnet[3,4] (obviously the front bonnet, since statistics show how most of the crashesare frontal running down).

These problems inducedEuropean Community toplan out restrictiveregulations for vehiclehomologation; theseregulations have toconsider the effects ofpedestrian impact; the setof rules, that are not yeteffective, is exposed in aproposal draft whichdescribes conditions forperforming tests,impactors’ featuressimulating human body,measures to carry out andhomologation limit value Inthis paper we refer to thatproposal draft in order todefine the simulationmodel and use FEMimpactors’ models.

Fig. 1. Pedestrian AIS2+ occurrence….

Fig. 2. Age distribution.

2. CHILD-HEAD vs. BONNET IMPACT

Now, it’s useful to shortly describe the test simulating impact “child headform on bonnet top”,finely explained in EEVC WG 17 Report of December 1998[5,6,7,8].The collision with “child” and subsequent impact of the head on the car-bonnet is simulated bya laboratory test, which uses an impactor simulating “child” skull.The impactor Child headform (Fig.3) consists of an aluminium sphere (a) coated with a12.5±0.5 mm stratum of vinyl skin (b), whose total weight is 2.5±0.05 Kg.In the centre of the impactor is inserted a tri-axial accelerometer (c) with a positioningtolerance, as regards barycentre, of ±10mm. The inertia as regards the centre of gravity has tobe equal to 0.0036±0.0003 Kg m².The impactor is thrown in “free-flight” at the velocity of 11.1±0.2 m/s with an impact angle of50 ±2°; the test is repeated nine times in different parts of the bonnet inside an area delimitatedby some “reference lines”, that we aren’t going to describe because they are not relevant toour aim.Impact velocity is inferred from statistical analyses which, as shown in Fig. 4, point out afrequency of injury at the level AIS 3-4 higher in the 65% of cases examined at velocitieshigher than 40 Km/h.

A typical trend ofacceleration diagram[9]is shown in Fig.5(resulting vector and itscomponents); byanalysing such a diagramwe note two distinctphase of the impact:1st phase: during thestarting 4ms of theimpact we note a highacceleration degreecorresponding to thephase of first impact;during this phase theframeworks that had thecrash absorb a lot ofenergy, especially kineticenergy, while theabsorption asdeformation energy islimited since the materialwave of deformationhadn’t enough time topropagate trough theframework2nd phase: afterwards,acceleration growsagain, in the instantduring the which theframework begins todeform in all its parts, sothe energy is

fundamentally deformation energy, in

Fig. 3. Child Headform impactor

Fig. 4. Impact speed by AISproportional to velocity in deformation of different materials involved; this phase is, generally,of 15ms. During this period of time the impactor finishes the impact phase and afterwardsreceives the elastic back-strain of the bonnet material that will allow the subsequentrebound[10].The evaluation criteria is an energy one and uses an index called HPC - Head PerformanceCriteria which is defined from the formula below[8]:

( )max1

12

2

112

5.2

ttadttt

HPCt

t

−

−= ∫ (1)

(a)

(b)

(c)

In the formula, t1 and t2 are the initial and the final instants of a time range shorter or equal tothe impact duration, a is the acceleration vector calculated by accelerometer.

In this test, the evaluationcriteria of performances of thevehicle “profile”, corre-sponding to the 100% ofpedestrian’s survival chance, isthe not overcoming of the value1000 as regards the energeticparameter just defined above(Fig.4).Because of such a behaviourwe realized a test-case which,as we will explain below,analyses only the first phase ofthe impact.

Fig. 5. Acceleration curve

3. IMPACT TEST-CASE: ANALYSIS OF FRAMEWORK BEHAVIOURDEPENDING ON MOULD AND FRAMEWORK PARAMETERS

It exists a narrow dependence of results of our test, and so of vehicle behaviour, fromgeometrical-topological and mechanical-structural parameters of the only parts of componentsinvolved in the impact.The main components are:

• The bonnet and all elements it consists of (external skin, anti-vibration frame, structuralglue…)

• Wing tops• Engine parts just below the bonnet• All chassis and coachwork parts the bonnet can hit in consequence of a vertical impact

(high side spars, scuttle, headlamp cross-member, bonnet connecting hinges, etc.).Our research focuses attention on variability of results depending just on geometricalparameters, in order to point out a correlation, if it exists, between geometrical parametersgoverning on bonnet “shape” and output parameters in the test. This kind of information is veryrelevant as concerns designing a new vehicle since it allows to define “standards” useful todesigners; these standards seem not to be obstacles but a sort of challenge. Industrial design’saim, in fact, is to “dress” an idea, which, in our case, is the information we deduce fromanalysis!The analysis was carried out trough FEM simulation of impact test, using the ESI-groupsoftware called Pam-Crash, which, being an explicit code, allows solution of FEM problems

that are strongly not linear during the time and is recommended for crash analysis,characterized by events “speed”[11].Together with this software we acquired impactors’ models, provided with materialsproperties, impact interfaces, time and computing parameters, output queries, whichreproduce impactors described and standardized in the proposal draft. We are going to usethe Child-Headform model, tested by simulating real impactors’ homologation standards.

4. PARAMETRICAL MODEL OF BONNET PORTION, HYPOTHESIS ABOUTDEFORMATION WAVES PROPAGATION, IMPACT AND BOUNDARYCONDITIONS

We defined the virtual test-case modifying conditions indicated in “proposal draft” in order toemphasize the results:

• The impact angle of the impactor with the bonnet is fixed at 90° so that the wholeimpact energy can be transferred to the bonnet and won’t be wasted, as it happens in“standardized” impacts, as roto-translational energy of the head and friction betweenhead and bonnet.

• The impact occurs in correspondence of the bonnet stiffening framework, in presenceof structural glue, where, in addition to increasing parts’ bending stiffness, we havealso an increasing inertia in play, in order to have quicker energy absorption, bothinternal and kinetic, therefore a quicker deceleration of the head.

The following model represent the model used as the test-case.This paper limits the analysis to the first phase of the impact in which, as shown above, theconveyed energy is at the highest degree for each time unit, while the deceleration peak is ashigh as possible.In this stage of the impact, which goes on for 4ms, is possible to identify the greatestacceleration as an absolute value after about 2.5ms, which seems to be the critical time ofenergy transfer during the impact; obviously HPC parameter, as defined above, is stronglyinfluenced by the value of this peak; for this reason the aim of our analysis is to find, if it exists,a correlation between bonnet shape (in the whole) parameters and magnitude of the peak.To reach our objective we modelled a portion of the bonnet of such a width that theconstraining system won’t be reached by the deformation wave before the phase ofacceleration “peak” is finished. The width “empiric” value, due to practical and virtualexperience, is 600 mm, space that is run through by deformation wave in about 4ms, that isabout 1,5ms after the acceleration peak[12].As regards this hypothesis, we built the parametrical model of bonnet using Think3 CADThinkDesign[13].The bonnet portion is a square with the side of 600 mm, and consists of a double-curvature(main and secondary), under the which is mounted the framework represented by boxedframe with the shape of a finned U, placed according to the main curvature to which it isconnected using structural glue[11].Control parameters of our parametric model will be:

• main curvature of external bonnet (a)• secondary curvature of external bonnet (b)• height of internal anti-vibration frame-work (c)

• width of internal anti-vibration frame-work (d)• material thickness of external bonnet (e)• material thickness of anti-vibration frame-work (f)• external bonnet stiffness (g)• anti-vibration framework stiffness (h)

The last two parameters were calculated by varying the density, as well as thickness, so toleave stationary the mass of the object, varying instead the stiffness.In relation to each of parameters shown above we performed a variational analysis based on 7values, liable of variations of 10%.

Fig. 6. Parametric/variational model

Further hypothesis about the model are as follows:• Portion of bonnet constrained by clamps not elastically-pliable, which can be used

trough the hypothesis of not influence of boundary conditions on peak value, thatderives from the relation “deformation wave speed – vs. – width of bonnet cutting”

• Presence of just one reinforcement boxed-frame in the part involved in impact• impact on of the stiffest zone, that is the zone connecting the two frameworks, just in

the middle of the plate• impact perpendicular to surface, in free flight

Bonnet was modelled by Linear-Shell elements in which shearing stress in a directionperpendicular to the shell surface are computed too.Material related to framework is steel FeP04, was modelled as a curve σ−ε with a zone ofcompletely elastic behaviour, another of elastic not linear behaviour and another of plasticbehaviour; parameters of stress-strain curve come from experimental data [14]. The model ofmaterial used to simulate FeP04 behaviour is the Pam-Crash model “103”, for materials withelastic-plastic not-linear behaviour, which consider strain-hardening of materials as regardsstrain rate according to “Cowper-Symond” law.Material concerning external bonnet is the high strength Fe220BH, analogous to FeP04model. In that case, in Pam-Crash, the hardening curve doesn’t correspond to an exponential

law: at every cycle we take strain rate e from the “strain rate vs. strain” curve we take theinstantaneous stress of element by linear interpolation of values taken from four piecewise –linear curves given as input (coming from reference books).We put a great emphasis on structural glue modelling, simulated using springs withcharacteristic linear as regards deformation, which considers material hysteresis andanisotropic dumping coefficient[15].Material concerning structural glue has been modelled using non linear springs, Pam-crashmaterial-model N° 220, with parameters extracted from bibliography.Child-headform model has been bought form, and corresponds perfectly to the one used forexperimental tests belonging to the group of tests prescribed by WG17 draft[8].These impactors consist of an aluminium core, to which is added a weight behind (to reach theweight necessary to the test), and is covered by a PVC skin in the entire zone of the possibleimpact.Since “massive” feature of the object examined, modelling of the parts was done using solidlinear elements.The impactor’s aluminium “core” was modelled using the special material N° 99 called “nullmaterial”, assigning it features of a rigid body, assuming that deformability value of impactor’sinternal part is very low.Mass, necessary to make calculation in order to obtain numeric convergence on nodes energy(based on nodal mass), is placed in centre of gravity; therefore all of the parameters, such asinitial velocity, constraints, sensors and accelerometers, have always to be referred to this onlynode, to which, as convenience, we’ve associate also an output local frame in which axis Xcoincides with impactor’s symmetry axis and consequently with impact direction.This “output local frame” in the centre of gravity allows to take head impactor’s decelerationsduring the impact.

5. RESULTS ANALYSIS: CHARTS AND TABLES

The table below shows, as regards parameters in analysis, values we have chosen forcalculation:

a Main curvature 3000 3500 4000 4500 5000 5500 6000

b Secondary curvature 5000 5500 6000 6500 7000 7500 8000

c Height of internal anti-vibration frame-work 10 12 14 16 18 20 22

d Width of internal anti-vibration frame-work 20 24 28 32 36 40 44

e Bonnet thickness 0.5 0.55 0.6 0.65 0.7 0.75 0.8

f Frame-work thickness 0.5 0.55 0.6 0.65 0.7 0.75 0.8

g Bonnet stiffness 10.205 9.2773 8.5042 7.85 7.32665 6.8033 6.3781

h Frame-work stiffness 10.205 9.2773 8.5042 7.85 7.32665 6.8033 6.3781

Table 2: value of parameters

Results, as regards the taken peak (by calculation), and filtered by the standard “SAE CFC1000” filter are shown in the following chart and table.

a(%) calculated filtered

66.66 3.58476 3.47540

77.77 3.62939 3.49270

88.88 3.86045 3.56467

100.00 3.87758 3.53790

111.11 3.85928 3.56575

122.22 3.69718 3.58284

133.33 3.76158 3.60630

b(%) calculated filtered

76.92 3.85916 3.46528

84.62 3.74290 3.50317

92.31 3.73926 3.54572

100.00 3.78251 3.49537

107.69 3.68573 3.56546

115.38 3.68797 3.56203

123.08 3.54565 3.49379

c(%) calculated filtered

62.50 3.10073 3.09627

75.00 4.68351 3.66901

87.50 4.31659 3.57357

100.00 3.78251 3.49937

112.50 3.61456 3.60660

125.00 3.81589 3.63679

137.50 3.79265 3.67877

d(%) calculated filtered

62.50 3.66339 3.50542

75.00 3.39316 3.32029

87.50 4.73665 3.78022

100.00 3.78251 3.49537

112.50 3.53577 3.50234

125.00 3.54799 3.38977

137.50 3.50065 3.46545

e(%) calculated filtered

76.92 3.52076 3.38339

84.62 3.47608 3.47551

92.31 3.64013 3.39561

100.00 3.78251 3.49537

107.69 3.60864 3.55424

115.38 3.73302 3.58074

123.08 3.70542 3.63604

f(%) calculated filtered

76.92 3.41762 3.23221

84.62 3.40863 3.30821

92.31 3.51548 3.39397

100.00 3.78251 3.49537

107.69 3.89952 3.61144

115.38 3.90836 3.69967

123.08 3.88328 3.76998

g(%) calculated filtered

130.00 3.52311 3.50128

118.18 3.75172 3.58114

108.33 3.81945 3.44459

100.00 3.78251 3.49537

93.33 3.60920 3.52979

86.67 3.67515 3.49137

81.25 3.55090 3.47518

h(%) calculated filtered

130.00 3.47629 3.33765

118.18 3.40309 3.38693

108.33 3.59473 3.42959

100.00 3.78251 3.49537

93.33 3.88148 3.57773

86.67 3.94556 3.62737

81.25 3.80509 3.67704

6. CONSIDERATION

Because of peculiar test-case conformation, and for the special test, it’s possible to drawsome conclusions from our analyses:• as regards the increasing of main curvature (curve a), we noted that it grows in an almost

linear manner the acceleration taken on the head (impactor), while is lower the influenceof secondary curve’s growth. Assuming that during the first phase of impact energeticexchange is mainly of the kinetic type, we can explain this behaviour with the increasing ofimpacting surface and therefore of corresponding mass; this trend is not taken as regardssecondary curvature (curve b), and this fact can be easily explained: in fact, the growth ofthe mass involved in impact, which is relative only to bonnet and not to the frame as ithappens for main curvature, is reduced.

Fig. 6. Comparison of results

• Geometrical parameters which rule dimensions of anti-vibration frame are correlated insuch a manner that they do not influence regularly acceleration peak (curves c and d ).However it’s important to note that also the minimal variation in this parameters involvesacceleration peak’s variations that reach values of 20% (the highest in percentage)compared to the basic model. Such a behaviour depends on the complex spatialdistribution of masses which influences the kinetic energy exchange between impactor andbonnet.

• We found out a not linear relation between the growth of the acceleration peak and theincreasing thickness and therefore mass as well as stiffness but it is not very clear asregards bonnet thickness, while it’s evident for frame thickness. So we note that thehigher mass (of surface) run down, as regards frame, is more relevant during the energeticabsorption phase compared with bonnet skin only.

• The increasing of framework stiffness, that we reached incrementing thickness andreducing, in the same time, the mass in inverse proportionally manner, clearly shows a notcontrollable bonnet parameters behaviour (curve g) and not linear growth of accelerationpeak with, even if not so relevant (a few percentage as we see in curve h), framework

stiffness growth. This fact confirms hypothesis asserting that during impact first phaseenergetic absorption is chiefly of kinetic type while energetic absorption of thedeformational type, present as well, does not affect the phenomenon in a relevant manner.

7. CONCLUSION

The phenomenon, during the initial 3ms and therefore in deceleration period during which thepeak appears, is ruled by energy transfer as kinetic energy, so it depends on masses in playand on their position in the impact zone. Deformation energy, present as well, has a lowervalue than kinetic energy.Some parameters an “organically” control the acceleration peak, while others, on the contrary,cannot be systematized as regards their behaviour.Finally it’s convenient to present some suggestions concerning design, that can be deducedfrom our analysis, useful to those who will design a new vehicle, also from the viewpoint ofpedestrian safety:• It’s certainly necessary to minimize the mass in impact points; so, for example, can be

useful deducing the mass of some zones of bonnet’s skin and frame which could improveperformances during pedestrian impact. Such a datum becomes relevant also as regardsdesign because it can pave the way for solutions such as Tailored Blanks (sheets ofdifferent thickness laser welded and then moulded) and “reject” alternatives solutions suchas a bonnet provided with an instant lifting device suggested for racing-cars in order tosolve the problem concerning bonnet proximity to mechanical parts.

• Importance of glue features could have implications concerning design, in selecting glueand in the applying it.

Finally it can be useful to point out how current vehicles behaviour during the pedestrianimpact test usually is lacking; with regard to this concern a prove is provided by test resultsEuroNCAP shown on web site: www.euroncap.com from which originate the need forstudying problems and identifying design criteria as soon as possible, since they will probablyinfluence design choices in early phases, so that benefits will involve vehicles on market in quitea lot of years.

REFERENCES

[1] K.L. Jarrett, R.A. Saul - Pedestrian Injury-Analysis of the PCDS Field collision1998 (pap n. 98-S6-O-04).

[2] K.C. Clemo, R. G. Davis - The practicalities of pedestrian protection (pap n. 98-S10-P-16).

[3] J.Y. Foret-Bruno, G. Reverjon, J.Y. Le Coz – Injury pattern of pedestrian hit by car ofrecent design, 1998 (pap. N° 98-s10-o-02).

[4] C.A. Hobbs, P.J. McDonough – Development of the European New Car AssessmentProgramme (Euro NCAP), TRL – UK, 1998 (paper N° 98-s11-o-06).

[5] M. Case, M. Griffiths, J. Haley, M. Paine – International NCAP programs review -(pap n. 98-S11-O-03).

[6] EEVC WG1O, “Proposals for methods to evaluate pedestrian protection for passengercars”. Final Report. November 1994.

[7] EC DG III, “Draft proposal for a European Parliament and Council Directive relating tothe protection of pedestrians and other road users in the event of a collision with amotor vehicle and amending Directive 70/156/EEC”. February 1996.

[8] EEVC Working Group 17 Report, Improved test methods to evaluate pedestrianprotection afforded by passenger cars. December 1998.

[9] Konosu, H. Ishikawa, A. Sasaki - A study on pedestrian impact test procdure bycomputer simulation, (pap n. 98-S10-W-19).

[10] G.J.L. Lawrence, B.J. Hardy - Pedestrian safety testing using the EEVC pedestrianimpactors – (pap n. 98-S10-O-03), pp2131-2144.

[11] A. Naddeo, Progettazione componenti veicolo con tecniche di CADparametrico/variazionale. Integrazione CAD-CAE. Applicazione al caso particolare delsottosistema pianale di un veicolo con piattaforma unificata per propulsori alternativi,Tesi di Laurea, Università di Salerno, (1999).

[12] T. D. Gillespie, Fundamentals of vehicle design, SAE (1992).[13] G. Monacelli and A. Quaglione, L'impatto delle nuove tecnologie CAD nell'ambito della

progettazione automobilistica, Proc. ADM IX, (1995).[14] A. Donnarumma, Disegno di macchine, Masson (1996).[15] J. Fenton, Handbook of vehicle design analysis, MEP Mechanical Engineering

Publication Limited.[16] Euro NCAP website: www.euroncap.com[17] NHTSA website: www.nhtsa.org.