Numerical Optimization of Crash Pulses
Dr. W. J. Witteman – University of Eindhoven
EURO-PAM’99
1
Numerical Optimization of Crash Pulses
Numerical Optimization of Crash Pulses
Dr.Ir. W.J.Witteman
Prof.dr.ir. R.F.C.Kriens
Eindhoven University of Technology
Laboratory for Automotive Engineering & Product Design
EUROPAM ’99
9th User Conference
October 7-8 1999
Table of contents
Numerical Optimization of Crash Pulses
Abstract… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .. 2
1. Introduction....................................................................................................
3
2. The Overall Severity Index.............................................................................
4
3. Numerical model of the optimal restrained dummy inside a vehicle interior...
8
4. Optimizing the deceleration pulse at 56 km/h................................................
17
5. Optimizing the deceleration pulse for other velocities.................................... 32
5.1. Optimizing the deceleration pulse at 32 km/h...................................... 34
5.2. Optimizing the deceleration pulse at 64 km/h...................................... 41
6. Structural design specifications for different crash velocities......................... 47
7. Conclusions.................................................................................................... 48
8. References..................................................................................................... 49
Numerical Optimization of Crash Pulses
2
ABSTRACT
Numerical Optimization of Crash Pulses
Optimization of occupant restraint systems is nowadays a popular tool to minimize
further the injury level of a frontal vehicle impact. Especially for smaller cars, it gives
the opportunity to reach an acceptable injury level, despite short crumple zones or a
less stable structure. In this way, a real deceleration pulse of a full -scale crash test
can be used as input for a numerical model of a dummy inside an interior. The
deceleration pulse is dependent on the velocity, crash situation and the vehicle
stiffness. Adapting the airbag and belt parameters (airbag delay time, out stream
surface, mass flow, belt feed, spool effect, pretensioner, smart system et cetera) can
give the optimal combination for minimal injuries.
However, the injury level becomes never lower as can be reached at that crash
pulse. In this paper the more interesting case of the reverse question is answered :
which crash pulse gives the lowest injury levels with an already optimized restraint
system. After the optimal pulses are found for several crash velocities at a specific
deformation length, a structure must be built which realizes such an optimal pulse as
closely as possible.
For this research, a numeric model of an interior is built and a FEM dummy is
incorporated to find the levels of the injury criteria. To compare the results of different
crash pulses, an overall severity index is developed. At a c rash speed of 56 km/h, an
optimal pulse is created after several deliberated pulse variations. This pulse
deviates much from a traditional pulse, which shows normally an increasing stiffness
of the structure near the end of the crash, but gives much lower injuries. During the
first 18 cm deformation length the deceleration level must be high, then a low
deceleration interval is required, and at the end (dummy is restrained by belt and
airbag) the deceleration must be high again. Also a review is given of ot her crash
velocities, necessitating adaptation of the pulse characteristics for optimal results.
This could be realized in the same car with an intelligent structure. Finally the
unexpected conclusion is discussed that an airbag deployment is already usefu l at a
crash speed of 10 km/h.
Numerical Optimization of Crash Pulses
3
Numerical Optimization of Crash Pulses
1. Introduction
The aim of this research is to minimize the injury level of the occupant in several
frontal collisions. Therefore, it must be clear which parameters influence t he injury
level. If an undeformable passenger compartment and no intrusion of vehicle parts
like steering wheel, dashboard and pedals are assumed, the injury level is only
influenced by means of g forces of the deceleration pulse generated by the vehicle
front. Of course a perfectly balanced restraint (limitation of dummy movement)
system must be available for minimal risk. Also the type of occupant and its sitting
position are kept constant in this study. To be sure that the injury level is the lowest
possible, a numerical model is necessary to calculate the expected injury level by
variation of the deceleration pulse. If the optimal deceleration pulse for a specific
crash velocity is known, the structure must be designed to generate such a desired
crash behavior.
First a method is given to calculate an overall severity index based on biomechanical
injury criterions. After this, an integrated numerical model of dummy and car interior
will be described with corresponding restraint parameters yielding the lo west overall
severity index. With an ideal not deforming passenger compartment, it is acceptable
to use an uncoupled model of the dummy and the frontal deforming structure. A
common method is, to predefine a deceleration pulse as input on the passenger
cage. A full frontal coupled model has a longer calculation time, also because the
dummy movement has a longer crash duration time while the frontal structure is
already fully deformed. The usual real time interactions between the occupant and
the vehicle structure during a crash (Khalil 1995), which influence the vehicle
deceleration a little (Seiffert 1992) by means of the restrained dummy mass, can be
compensated in the input pulse. Of course for exactly determining the deceleration
pulse of a vehicle structure (not for determining the pulse that causes the lowest
injury level) the dummy masses must be added to the vehicle model with restraint
characteristics. Of course in case of a side impact an uncoupled method is not
allowed, the dummy mass and its close position to the door have a not negligible
influence on the deformation behavior of the relatively low mass of the side structure
(Landheer 1996).
With the aid of the interior model, variations of the deceleration pulse are compared
on basis of a calculated injury level and an optimal pulse is obtained for several crash
4
Numerical Optimization of Crash Pulses
velocities. Finally, structural design specifications are presented to realize such an
optimal pulse.
2. The Overall Severity Index
To compare the injury severity for different vehicle collisions, some sort of index or
formula is needed. The regulations for vehicle crashes only prescribe a maximum
value not to be exceeded for several different injury criteria. Because the used
vehicle model has no intrusion, only the injury criteria as mentioned in Table 1 with
their limiting values (Levine 1994, Mertz 1993, Seiffert 1997) are useful. Also the
relative importance to an overall severity index is given, expressed by a weight factor
(Viano 1990).
Table 1.
Relevant injury criteria with their by legislation limited values.
Injury Criterion
Limit value
Weight factor
HIC
1000
8
CHEST-G
60 g
2
CHEST-D
50 mm
2
FEMUR-F
10000 N
1
NECK-M
189 Nm
2
The Head Injury Criterion (HIC) is calculated on a specific time interval around a
deceleration peak of the dummy head to reach a maximum value as shown in next
formula, where t1 and t2 are the start and end time of the considered deceleration
interval in seconds and a(t) is the head deceleration in g as function of time:
2.5
t2


 1

HIC = max 
⋅∫a( t ) dt  ⋅( t 2 − t1 )


( t 2 − t1 ) t1

(1).
CHEST-G is the peak deceleration in g of the dummy chest.
CHEST-D is the peak compression of the dummy chest, mostly a result of the belt
force.
FEMUR-F is the maximal longitudinal force in the upper leg caused by the
dashboard.
Numerical Optimization of Crash Pulses
5
NECK-M is the flexion bending moment of the dummy neck by forward head rotation.
The demands for a realistic crash index are formulated as follows (Bakker 1997):
1. Some injury types are more important than others.
2. Improvements in high valued injury criteria count more than improvements in
already low valued criteria.
3. If one injury type has a much higher value with respect to its limited value than
the other criteria, then an improvement on this higher value must be count
additionally.
4. The improvement in one injury type counts less than the improvement of several
injuries types.
5.
A large improvement in an injury type counts more than a small one.
In literature a lot of different severity indexes as objective function for injury
assessment are proposed. Magazines of consumer groups (Auto Motor und Sport
1992) use colors for the head, the chest and the extremities to indicate the
percentage of the limit values of the corresponding injury criterions. NHTSA uses as
crash index a simple normalized summation with weight factors of the HIC value, the
Chest-G value and both femur loads (Seiffert 1992). Yang (1996) uses an injury
criterion index that is a not weighted summation of a normalized HIC, Chest-G,
Chest-D, Neck-M and Femur-F value, divided by 5. Another possibility is to look at
the societal harm. Injury costs from road trauma are quantified, involving both a
frequency and a unit cost component (Sparke 1998). In its most general form, it is a
measure of the total cost to the community of road trauma, and includes hospital
costs, rehabilitation costs, lost income and some value on lost ability and quality of
life. It can be used to evaluate the contribution made by vehicle design for occupant
protection. Also an injury risk function which is based on probability analysis of
biomechanical data is possible (Viano 1990, Wismans 1992). This enables injury risk
to be assessed in each body region. Whole body injury is determined by summing
the individual risk from each body region. The risk function can be coupled to the
Abbreviated Injury Scale (AIS), so that a probability on an AIS value (1= minor, 6=
maximum injury) can be calculated (Stolinski 1996). Individual probabilities of injury
for head and chest for each dummy may be combined into a single probability value,
which is the sum of both minus the product of both. Combination of risk and cost
models is used by Kebschull (1998) yielding a Risk/Benefit ratio. Although the last
mentioned indexes are usable for special objectives on injury costs and risk, they use
additional functions which are not directly related with the simulated injury criterions.
Also further evaluation of the actual distribution of human injury and the adequacy of
biomechanical data is required (Viano 1990). In our research for optimal crash pulses
a clear and understandable independent relation in injury criterion changes on the
6
Numerical Optimization of Crash Pulses
final severity index is desired for deliberated pulse variations. Moreover, the
mentioned overall severity indexes from literature do not fully meet demands 2 and 3.
Therefore, a new index is proposed in this paper. This new developed overall
severity index uses the limit values of Table 1 (X lim), a reference value (X ref) which is
the old level of an injury criterion in a reference situation that is used to compare with
(mostly the first calculation or the values of the best combination of a preceding table
with results), and a modified value (X mod) which is the new level of an injury criterion
after a change of the model parameters or a change of the crash pulse.
The Overall Severity Index (OSI) is given by the following formulae:
(HICmod − HICref) HICref
HICref

OSI = 
*
*
* 8+

HIClim
HIClim HIClim * TOTREF 
CHEST.Gref
(CHEST.Gmod − CHEST.Gref) CHEST.Gref

*
*

*2 +

CHEST.Glim CHEST.Glim * TOTREF 
CHEST.Glim
CHEST.Dref
(CHEST.Dmod − CHEST.Dref) CHEST.Dref

*
*

*2+

CHEST.Dlim CHEST.Dlim * TOTREF 
CHEST.Dlim
FEMUR.Fref
(FEMUR.Fmod − FEMUR.Fref) FEMUR.Fref

*
*

 * 1+

FEMUR.Flim FEMUR.Flim * TOTREF 
FEMUR.Flim
NECK.Mref
(NECK.Mmod − NECK.Mref) NECK.Mref

*
*

*2

NECK.Mlim NECK.Mlim * TOTREF 
NECK.Mlim
 100
+ 100
*
 0.75
(2).
where
TOTREF =
HICref CHEST.Gref CHEST.Dref FEMUR.Fref NECK.Mref
+
+
+
+
HIClim CHEST.Glim CHEST.Dlim FEMUR.Flim NECK.Mlim
(3).
The total is divided by 0.75 and multiplied with 100 to reach an OSI value of 110 if
each injury criterion value is increased by 10 per cent as compared to 50 per cent of
the maximum value.
The OSI depends on which injuries improve, which deteriorate, how much and at
what level. An improvement in high injury levels weighs more than an improvement in
low injury levels. Likewise, an improvement in one type of injury has less influence
than improvements in all injury types. If one injury type has large values with respect
to the others, then an improvement of this injury type weighs even more. Also some
injury types are more important than others, with weight factors this difference is
reflected. Brain damage for example is much more dangerous than a leg fracture.
The summation of 5 terms (modifications in injury types) in the OSI formula reflects
demand 4.
Numerical Optimization of Crash Pulses
7
Within the formula of the OSI several parts can be distinguished, which are based on
the mentioned demands and will be explained with the HIC as example:
HICmod − HICref
:
HIClim
The absolute change relative to the limit, which is the basis (reflects de mand 5).
HICref
:
HIClim
This gives a change at high values more importance than a change at low values
(reflects demand 2).
HICref
:
HIClim * TOTREF
The TOTREF is just a help function to keep the OSI formula small and this total term
stands for the current importance of this injury relative to the other four (reflects
demand 3).
8
:
A weight factor, which indicates the permanent injury importance with respect to the
other injury criteria (reflects demand 1).
In Table 2 some examples of modifications of the injury types show the influence on
the calculated OSI value.
8
Numerical Optimization of Crash Pulses
Table 2.
Examples of OSI calculations.
Injury Criterion
Limit value
Example 1
Reference
25 % of limit
Modified
+ 20 %
Example 2
Reference
50 % of limit
Modified
+ 10 %
Example 3
Reference
100 % of limit
Modified
+5%
Example 4
Reference
100 % of limit
Modified
+ 10 %
HIC
1000
CHEST-G CHEST-D FEMUR-F NECK-M
60 g
50 mm
10000 N
189 Nm
OSI
250
15
12.5
2500
47.25
100
300
18
15
3000
56.7
105
500
30
25
5000
94.5
100
550
33
27.5
5500
103.95
110
1000
60
50
10000
189
100
1050
63
52.5
10500
198.45
120
1000
60
50
10000
189
100
1100
66
55
11000
207.9
140
If for example only the HIC varies (while other injury criteria are constant) from 500 to
550, or from 600 to 650, or from 600 to 660, the OSI becomes successively 101.6,
102.3, 102.8.
With this index it is possible to compare different crash configurations and to find the
optimal parameter setting leading to the lowest OSI value.
3. Numerical model of the optimal restrained dummy inside a vehicle
interior
To simulate the movement behavior of an occupant and to measure the forces
working on the body, use is made of a modern deformable frontal finite element
dummy Hybrid III (ESI 1996). This dummy consists of rigid body elements and a full
Numerical Optimization of Crash Pulses
9
deformable thorax and pelvis and is developed by the safety group of ESI SA
(Rungis, France) in collaboration with the Biomechanics Department of Wayne State
University (Detroit, USA). This numerical dummy simulates the crash dummy of a fullscale frontal crash. In literature a good correlation is reported between computed
accelerations of the basic rigid body dummy and measured accelerations in a sled
test (Ni 1991). Also the new dummy with deformable chest and pelvis, shows good
correlation’s with low and high speed pendulum impact tests and with a sled test (ESI
1996, Schlosser 1995). Although it is better to use a human like dummy (which are
under development see Figure 1) to simulate a more realistic behavior, until now
legislation is based on measurements at full-scale dummies in crash tests. As soon
as reliable numerical human like dummies are available, car manufacturers should
use these dummies to design their vehicles with more reliable injury prediction
resulting in really safer cars. In Figure 2 the full scale 50 per cent test dummy Hybrid
III is shown together with its finite element equivalent.
Figure 1. Human body model (ESI).
10
Numerical Optimization of Crash Pulses
A
B
Figure 2. 50 per cent Hybrid III dummy (A) and the numerical model (B).
The dummy model must be positioned inside a realistic vehicle interior model (Bosch
1993, Relou 1995, Seiffert 1989, Wijntuin 1995). To this aim, a seat, a dashboard
with steering wheel, a floor plane and a restraint system have been modeled, see
Figure 3. The restraint system consists of a belt with automatic lock and a retractor,
and a folded (as far as necessary) airbag (Hoffman 1989). The folding FEM airbag
has a good interaction with a dummy and shows a good agreement with
experimental data (Hoffmann 1990, Lasry 1991). The seat has a so-called anti
submarining plate, which prevents forward moving of the dummy pelvis under the lap
belt.
Numerical Optimization of Crash Pulses
11
Figure 3. Dummy positioning within the interior with restraint systems.
The following restraint parameters with current values must be optimized for minimal
OSI value:
Belt feed (pull out distance of belt before locking), 6% (200 mm).
Belt pretensioner (pyrotechnic tightening of the belt), pull in distance vs. time curve
(150 mm pull in at 12 ms).
Spool effect, belt force vs. outlet curve (12 kN = 120 mm).
Belt strain, belt force vs. strain curve (30 kN = 18.3 % elongation).
Airbag delay time (activation time after collision start), 15 ms.
Airbag stream out surface, 3848.5 mm2.
Airbag mass flow, curve of mass flow dependent on inflation time.
In order to find the optimal tuning of the restraint system for the optimal deceleration
pulse (Relou 1995, Witteman 1993), a parameter study has been performed. The
optimization has been done by activation or deactivation of a restraint option and with
changes in the parameter values of maximal 25 per cent or 10 ms. This has been
done to find a direction for the optimal tuning and to gain insight in the influence of
the separate parameters.
The optimal tuning of the restraint parameters is valid for the prescribed deceleration
pulse. These optimal restraint parameter values are used in the next section to find a
new optimal deceleration pulse. Because, both are dependent on each other, the
optimal combination is an iterative process. This iterative process is not taken into
consideration because the optimal deceleration pulse found will just be an indication
Numerical Optimization of Crash Pulses
12
for the direction in which to search. First, structural solutions must show if this
“optimal” pulse can be obtained in different collisions. Optimization for a single
collision velocity, a single collision direction, and a single overlap percentage seams
not realistic. But with a new designed longitudinal structure (Witteman 1998) it is
possible to generate one, not yet optimized, deceleration pulse that is quite similar for
different overlap percentages and crash directions. A speed of 56 km/h against a
rigid barrier is found as a realistic test speed because it gives a balance between
acceptable injuries at lower speeds and minor fatalities at higher speeds (Witteman
1993).
The optimization can also be done for different dummies (e.g. a 5 percentile female
and a 95 percentile male) and seat positions, but for these kind of variables an
intelligent airbag and safety belt system that adjusts itself to the occupants is more
favorable. Now only an optimal start value of the parameters for an average situation
will be found.
Because the influence of one parameter to the injury criteria is dependent of the
current values of the other parameters, the optimum will not be found at the
combination of all separate optimizations. As a result, the influence of some
parameters will not be unambiguously determinable. The optimization is also
performed for an undeformable vehicle interior. This means that flesh and
penetration wounds will not occur and therefore are not taken into account. The belt
pretensioner is originally developed to prevent the occupant from hitting the
dashboard or other parts that move into the interior during a crash. Because the belt
pretensioner has a negative influence on the occupant chest compression and
deceleration, and interior deformation is not taken into account, optimal tuning will
probably be without the pretensioner. For the full overlap collision, for which the
optimization is performed, the safety belt does probably not have to be extremely
tightened. This is confirmed by research done by Volvo (Fokker 1995), which shows
that for the average collision a belt strain of 10 per cent is better than a belt strain of
4 per cent. However, for a partially overlap collision the restraint system, including the
seat belt, the seat itself, and the door, should keep the position of the occupant
predictable relative to the occupant compartment. In particular the restraint system
should not allow the head and torso to lag far behind as the vehicle begins to rotate
at the end of the crash (O’Neill 1996). To prevent ambiguity, an intelligent restraint
system would be of great use, because then conflicting design demands can be
avoided.
A parameter optimization performed by means of an optimization algorithm (de Jager
1993), would be ideal. This method not only provides insight in the sensitivity of
Numerical Optimization of Crash Pulses
13
injuries to certain parameters, but it also provid es the absolute optimum because the
changes are not limited to the assumed 25 per cent. Since these optimizations can
not be performed using only the dummy FEM program PAM-SAFE, and the
implementation of such an optimization process leads to a large calcul ation effort
(about 2 calculations a day on a workstation), the optimization is restricted to 54
variations on already mentioned restraint parameters. In this research the intention is
to show that an optimal pulse deviates much from a traditional vehicle deceleration
pulse and therefore new structural measurements are necessary to obtain such a
new pulse and that for different crash velocities different pulses are required. In
future further optimization with an optimization algorithm and with the aid of f aster
computers can be done for fine tuning of the structure, but now such a detailed study
is not required.
First, a total of 32 different combinations are calculated, where Combination 1 is the
reference start situation. The combinations are given in Ta ble 3 as numbers 1 up to
32 and the results of the simulations are given in Table 4, with the separate values of
the different injury types and the calculated OSI value.
Numerical Optimization of Crash Pulses
14
Table 3. Restraint parameters combinations in 3 phases.
Number
Belt feed
Yes/No
1 reference Yes
2
No
3
Yes
4
No
5
Yes
6
No
7
Yes
8
No
9
Yes
10
No
11
Yes
12
No
13
Yes
14
No
15
Yes
16
No
17
Yes
18
No
19
Yes
20
No
21
Yes
22
No
23
Yes
24
No
25
Yes
26
No
27
Yes
28
No
29
Yes
30
No
31
Yes
32
No
33
Yes
34
Yes
35
Yes
36
Yes
37
Yes
38
Yes
39
Yes
40
Yes
41
Yes
42
Yes
43
Yes
44
Yes
45
Yes
46
Yes
47
Yes
48 Yes
49 Yes
50 Yes
51 Yes
52 Yes
53 Yes
54 Yes
Belt pretensioner
Yes/No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
No
No
Belt strain
Spool
Airbag
Airbag
Airbag
effect
delay time stream out mass flow
%
%
ms
surface %
%
100
100
15
100
100
100
100
15
100
100
125
100
15
100
100
125
100
15
100
100
100
100
5
100
100
100
100
5
100
100
125
100
5
100
100
125
100
5
100
100
100
100
15
125
100
100
100
15
125
100
125
100
15
125
100
125
100
15
125
100
100
100
5
125
100
100
100
5
125
100
125
100
5
125
100
125
100
5
125
100
100
100
15
100
125
100
100
15
100
125
125
100
15
100
125
125
100
15
100
125
100
100
5
100
125
100
100
5
100
125
125
100
5
100
125
125
100
5
100
125
100
100
15
125
125
100
100
15
125
125
125
100
15
125
125
125
100
15
125
125
100
100
5
125
125
100
100
5
125
125
125
100
5
125
125
125
100
5
125
125
100
125
15
100
125
100
100
25
100
125
100
125
25
100
125
75
100
15
100
125
75
125
15
100
125
75
100
25
100
125
75
125
25
100
125
100
100
15
75
125
100
125
15
75
125
100
100
25 75
125
100
125
25 75
125
75
100
15
75
125
75
125
15
75
125
75
100
25 75
125
75
125
25 75
125
100
100
10
75
125
100
100
20 75
125
75
75
15
75
125
75
75
15
75
125
100
100
15
75
125
100
75
15
75
125
125
125
15
75
125
15
Numerical Optimization of Crash Pulses
Table 4. Simulation results for the injury types and the minimized OSI value.
Number
1 reference
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
HIC
297.5
304.1
289.9
309.8
300.2
337.1
343.3
327.0
287.1
318.0
285.2
305.0
308.6
307.5
317.9
323.6
237.5
233.0
258.4
239.4
241.6
250.1
245.6
239.1
236.6
234.3
240.8
230.4
271.9
244.5
266.6
256.6
238.9
285.0
284.4
242.8
251.0
289.9
278.8
276.9
309.7
325.9
346.8
280.2
275.6
313.5
315.4
270.9
289.7
271.2
286.4
282.1
292.6
278.5
CHEST-G
38.1
38.0
36.1
36.2
35.4
34.5
43.2
34.6
36.0
35.0
37.4
36.5
34.0
35.0
41.4
34.4
38.4
40.9
36.8
40.0
35.3
41.5
40.5
41.7
36.5
39.7
35.7
40.9
35.0
40.0
34.9
35.1
38.4
44.2
47.4
36.3
39.9
39.6
39.0
40.8
43.0
42.4
45.1
40.1
42.8
37.7
42.6
37.9
39.7
40.1
38.4
35.4
44.2
36.2
CHEST-D
40.5
37.0
38.4
34.4
36.3
36.0
37.1
35.6
40.1
34.1
35.0
35.8
41.1
39.2
36.6
35.3
30.4
33.7
30.7
31.3
34.3
33.9
33.2
34.1
31.8
34.0
31.3
35.5
35.2
34.4
34.0
33.7
31.5
33.0
31.7
35.8
32.6
32.7
32.9
30.1
30.8
31.8
30.6
31.3
32.4
31.6
31.7
31.4
31.7
30.7
38.2
37.3
30.9
28.8
FEMUR-F
3045.2
2980.6
3019.4
2872.3
3027.4
2974.2
2979.9
3072.2
3062.6
2896.7
2966.4
2976.7
3039.6
2869.4
2964.4
2930.1
3459.0
3275.5
3235.5
3250.4
3220.7
3274.9
3339.8
3333.8
3302.6
3129.1
3116.7
3378.1
3161.0
3107.6
3101.8
3272.6
3249.7
3286.8
3190.6
3218.7
3453.0
3270.2
3478.7
3427.0
3223.5
3348.5
3297.8
3412.0
3530.5
3339.5
3298.7
3428.6
3274.1
3889.8
5357.3
5317.1
3624.1
3440.9
NECK-M
97.1
106.1
85.9
119.5
116.4
100.5
103.5
102.7
116.0
131.1
110.8
126.1
123.6
113.6
126.7
121.4
64.5
72.9
73.0
59.3
89.7
89.9
90.2
82.1
105.7
99.0
96.6
100.2
116.3
111.3
111.5
109.2
68.8
46.4
43.9
55.8
67.1
47.9
40.4
31.6
36.4
28.2
26.7
39.2
37.5
31.2
29.0
46.6
28.1
36.7
83.8
78.3
34.0
30.0
OSI
100.0
96.8
93.8
94.1
95.4
93.3
101.6
92.9
100.5
94.7
93.5
97.0
102.3
98.8
102.1
94.8
79.7
87.0
80.9
81.3
86.7
90.8
88.9
89.7
86.1
90.4
83.7
93.3
92.5
93.1
89.9
89.1
81.8
86.3
86.4
84.4
84.6
83.1
81.6
77.6
82.0
82.3
83.1
79.9
83.0
78.8
82.0
79.3
78.9
78.8
95.9
91.6
82.1
72.5
Numerical Optimization of Crash Pulses
16
From Table 4 the conclusion can be drawn that Combination 17, which is the same
as the Reference Combination 1 but with an enlarged airbag mass flow, gives the
lowest OSI value of the first 32 combinations. The chest and neck injuries participate
heavily in the OSI, so these injury types must be improved, possibly at the expense
of the others. To reduce the Chest-D and the Neck-M, the airbag must be made
larger and stiffer by decreasing the stream out surface and increasing the mass flow.
Assuming this assumption, belt feed must be present without a pretensioner in order
to let the improved airbag do its protective job. New parameter influences are
considered with Combination 17 as reference. The OSI however has still been
calculated on the basis of Combination 1.
Second, a total of 15 different combinations are calculated without using the results
of one calculation in the other. These 15 calculations are the numbers 33 up to 47.
For these combinations the belt strain is considered because this was left out of
consideration in the first series. It is increased with 25 per cent. The spool effect is
decreased with 25 per cent, because an increase had a negative influence with
Combination 17 as reference as shown in the first series. Because an airbag delay of
5 ms increased the OSI compared to Combination 17, a delay of 25 ms was chosen.
Finally, the airbag stream out surface has been decreased, because an increase
magnified the OSI.
From these 15 combinations, Combination 40 is the best, which is the same as the
second Reference Situation 17 but with a smaller airbag stream out surface. New
parameter influences will be considered with Combination 40 as reference.
Third, a total of 7 different combinations are calculated which are the numbers 48 up
to 54. Because airbag delay times of 5 and 25 ms both resulted in an increase of the
OSI as compared to a delay time of 15 ms, new delay times of 10 and 20 ms are
tried. The belt pretensioner is used with and without a decrease of the belt strain and
spool effect.
Still, the airbag delay of 15 ms is optimal, because of the optimal combination of
injury levels. If the delay time is changed some injury levels decrease and some
increase but the combination appears to increase the OSI always.
The belt pretensioner normally lowers the chance of the dummy hitting any
protruding objects. The other injury types increase as a result of the higher
participation of the belt and smaller participation of the airbag. The leg forces
however should decrease because the dummy is better restrained. But the used belt
pretensioner at the sill tensions most off all the chest belt and not so much the lap
17
Numerical Optimization of Crash Pulses
belt, because of the friction with the chest. A pretensioner that pulls the buckle
backwards between the front seats should give better results.
Although parameter Combination 54 reaches a lower OSI as Combination 40, further
research will be performed with Combination 40 as a reference with the second
lowest OSI, because in case of Combination 54 the restraining by the belt in off-set
situations would be far too low. In case of Combination 40 with Combination 1 as
reference, the mass flow of the airbag has been increased and the airbag outflow
surface has been decreased. To prevent numerical instability of the belt/dummy
contact at crashes with 64 km/h (later on described in Section 5.2), small model
adjustments were necessary, which result in little different injury values for the
reference combination in the next section.
4. Optimizing the deceleration pulse at 56 km/h
With the optimal combination of restraint parameters that was obtained, research has
been carried out to find a deceleration pulse (the resulting deceleration -time
signature on the vehicle passenger compartment generated when a collision occurs)
with a minimal injury risk, based on the lowest value of the overall severity index, at a
56 km/h crash during 90 ms. This pulse determines the occupant loading and hence
the injury risk for a passenger in a vehicle involved in an accident. First, the already
mentioned reference pulse (see Figure 4), will be used as input of the numeric
model. The corresponding injury values are plotted in Figure 5. The total injury risk is
expressed with the Overall Severity Index, and is set at a value of 100 as reference
for other pulses to compare.
Reference pulse
Velocity
60
50
40
30
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
40
50
60
70
80
90
time [ms]
Numerical Optimization of Crash Pulses
Figure 4. The reference deceleration pulse and the velocity decrease.
18
19
Numerical Optimization of Crash Pulses
R e fe r e n c e
p u ls e
0 .7
0 .6
mm/ms2
0 .5
0 .4
0 .3
0 .2
0 .1
0
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e l e r a t i o n [ 8 5 0 1 0 1 ] ) H IC
H IC = 2 8 3 . 5
0
))
5 0
1 0 0
T im e
1 5 0
HIC
[m s ]
R e fe r e n c e
p u ls e
0 .5
mm/ms2
0 .4
0 .3
0 .2
0 .1
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [ 8 5 0 3 0 1 ] ) M S
C H E S T G = 3 4 .7 3
0
0
))
5 0
1 0 0
T im e
1 5 0
CHEST-G
[m s ]
R e fe r e n c e
p u ls e
0
B a r A x i a l E lo n g a t i o n [ 5 3 0 0 0 0 t / m
5 3 0 0 0 6 ]
- 1 0
mm
- 2 0
- 3 0
- 4 0
- 5 0
C H E S T D = 2 6 .46
- 6 0
0
5 0
1 0 0
T im e
R e fe r e n c e
7
1 5 0
CHEST-D
[m s ]
p u ls e
S a e 1 8 0 _ 5 ( S P R IN G
L o c a l F o r c e - M a g n itud e
[ 9 1 4 1 0 0 ] = - .
S a e 1 8 0 _ 5 ( S P R IN G
L o c a l F o r c e - M a g n itud e
[ 9 1 8 1 0 0 ] = -
F E M U R F = 3 9 8 3
6
kN
5
4
3
2
1
0
0
5 0
1 0 0
T im e
1 5 0
FEMUR-F
[m s ]
R e fe r e n c e
p u ls e
1 6 0
1 4 0
1 2 0
kN mm
1 0 0
8 0
6 0
4 0
2 0
0
N E C K M = 3 2 .7 7
S a e 1 8 0 _ 5 ( S p h e r e J o i n t M o m e n t a b o u t [9 0 2 1 0 0 ]
0
5 0
1 0 0
T im e
[m s ]
Figure 5. Injury values of the reference pulse.
1 5 0
NECK-M
Numerical Optimization of Crash Pulses
20
The injury values as shown in Figure 5 and also in following pages are plotted with a
SAE180_5 filter with the times in ms on the x-axis and on the y-axis for the HIC the
deceleration in mm/ms2, for the CHEST-G also the deceleration in mm/ms2, for the
CHEST-D the negative elongation (deflection) in mm of 7 bar elements perpendicular
to the chest (where the CHEST-D is calculated as the average of this 7 distances),
for the FEMUR-F the force in kN, and for the NECK-M the flexion moment in kNmm.
In Figure 6 four other crash pulses are viewed. The first pulse V56 -1 (56km/h crash,
first chronological number) is derived from research of Sparke (Sparke 1994,1996).
This pulse is based on the idea that at low crash velocity the deceleration force must
be also low in order to yield minor injuries at low speed crashes. At higher crash
velocities the stiffness increases with increasing deformation, to ensure enough
energy absorption in the end to stop the vehicle before the passenger compartment
deforms. The pulse is a modified shape between a hard and short pulse and a soft
and long pulse. The second pulse V56-2 is based on research of Brantman
(Brantman 1991). This optimized pulse has comparable deceleration regions as the
reference pulse but the change points are earlier in time (10 and 45 ms).
Looking to the shapes of the injury types of the reference pulse in Figure 5, a
possible improvement of the injury values can be reached with a slightly larger
deceleration in the first half of the crash duration and a slightly smaller deceleration
at the end, because the peaks are concentrated in the second half and can be
lowered if the peaks are more spread and start earlier. For this reason pulse V56-3
has a deceleration of 25 g the first 50 ms and for the remaining time a deceleration of
8.4 g. Another noteworthy pulse is pulse V56-4, a constant deceleration of 17 g,
which seems optimal for one crash velocity with maximum structural efficiency and
without a deceleration peak. However, the resulting forces on the dummy are directly
determined by the restraint system and only indirectly determined by the deceleration
forces of the car body, which has no rigid coupling with the dummy.
The results of the simulations with these four pulses as input are shown in Figures 7
and 8 and in Table 5. As can be seen in the table none of the newly tried pulses
show a better OSI value as the reference pulse. In case of pulse V56-3 the
deceleration in the first part of the collision is too high, resulting in a very steep and
high head deceleration, and, therefore, a high HIC value. The maximum chest
deceleration in case of pulses V56-3 and V56-4 lies at 50 ms and is higher than the
value with the reference pulse on interval 70-90 ms. This means that the deceleration
before 50 ms was too high and in case of the reference pulse it was too low before
70 ms. The chest compression peak in pulses V56-3 and V56-4 is moved from 90 to
50 ms with a higher value. Pulse V56-4 leads to one peak force in the legs hitting the
21
Numerical Optimization of Crash Pulses
dashboard. In case of the reference pulse there are two important peaks because
after first contact the legs move backwards relative to the compartment and due to
the increasing deceleration the dummy hits the dashboard for a second time.
However, these peaks are lower than the single peak caused by pulse V56-4. In case
of pulse V56-2 the two peaks of the contact force between the knees and the
dashboard are better balanced as the reference pulse yielding a low er Femur-F
value. The Neck-M value is dramatically increased with all pulses because a larger
deceleration before the head is in contact with the airbag yields the head to move
further forward with its higher relative velocity in relation to the torso.
Pulse V56-1
Pulse V56-2
Velocity
60
60
50
50
40
40
30
30
20
20
10
Velocity
10
deceleration [g],velocity [km/h ]
deceleration [g],velocity [k m/h ]
0
0
0
10
20
30
40
50
60
70
80
90
0
10
20
30
40
50
60
70
time [ms]
Pulse V56-3
Velocity
Pulse V56-4
60
60
50
50
40
40
30
30
20
20
10
80
90
time [ms]
Velocity
10
deceleration [g],velocity [km/h ]
deceleration [g],velocity [km/h ]
0
0
0
10
20
30
40
50
60
70
80
90
0
time [ms]
Figure 6. Four examples of other possible crash pulses.
10
20
30
40
50
60
70
80
90
time [ms]
22
Numerical Optimization of Crash Pulses
Pulse-V56-2
0.7
0.6
0.6
0.5
0.5
mm/ms2
mm/ms2
Pulse-V56-1
0.7
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0
0.1
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
H IC = 6 3 3 . 4
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
H IC = 3 5 0
0
50
Time [ms]
HIC V56-1
0.4
0.4
mm/ms2
mm/ms2
0.5
0.3
0.2
0.1
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=63.56
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=35.99
0
50
Time [ms]
Time [ms]
CHEST-G V56-1
CHEST-G V56-2
Pulse-V56-1
Pulse-V56-2
0
0
Bar Axial Elongation [530000 t/m 530006]
-10
-10
-20
-20
mm
mm
Bar Axial Elongation [530000 t/m 530006]
-30
-40
-30
-40
-50
-50
CHESTD=28.63
CHESTD=32.05
0
50
100
-60
150
0
50
Time [ms]
150
CHEST-D V56-2
Pulse-V56-1
Pulse-V56-2
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
100
Time [ms]
CHEST-D V56-1
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=5724
6
6
5
5
4
4
kN
kN
150
0.3
0.2
0.1
3
3
2
2
1
1
0
100
Pulse-V56-2
0.5
-60
150
HIC V56-2
Pulse-V56-1
0
100
Time [ms]
0
50
100
0
150
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=3132
0
50
Time [ms]
100
150
100
150
Time [ms]
FEMUR-F V56-1
FEMUR-F V56-2
Pulse-V56-1
Pulse-V56-2
150
90
80
70
100
kN mm
kN mm
60
50
50
40
30
20
10
0
0
NECKM=154.4
Sae180_5(SphereJoint Moment about [902100]
0
50
100
-10
150
NECKM=99.48
Sae180_5(SphereJoint Moment about [902100]
0
Time [ms]
NECK-M V56-1
50
Time [ms]
NECK-M V56-2
Figure 7. Injury values of pulses V56-1 and V56-2.
23
Numerical Optimization of Crash Pulses
Pulse-V56-4
0.7
0.6
0.6
0.5
0.5
mm/ms2
mm/ms2
Pulse-V56-3
0.7
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0
0.1
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
H IC = 4 5 4 . 7
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
H IC = 3 3 6 . 7
0
50
Time [ms]
HIC V56-3
0.4
0.4
mm/ms2
mm/ms2
0.5
0.3
0.2
0.1
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=62.77
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=45.34
0
50
Time [ms]
Time [ms]
CHEST-G V56-3
CHEST-G V56-4
Pulse-V56-3
Pulse-V56-4
0
0
Bar Axial Elongation [530000 t/m 530006]
-10
-10
-20
-20
mm
mm
Bar Axial Elongation [530000 t/m 530006]
-30
-40
-30
-40
-50
-50
CHESTD=33.88
CHESTD=27.11
0
50
100
-60
150
0
50
Time [ms]
150
CHEST-D V56-4
Pulse-V56-3
Pulse-V56-4
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
100
Time [ms]
CHEST-D V56-3
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=5467
6
6
5
5
4
4
kN
kN
150
0.3
0.2
0.1
3
3
2
2
1
1
0
100
Pulse-V56-4
0.5
-60
150
HIC V56-4
Pulse-V56-3
0
100
Time [ms]
0
50
100
0
150
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=5018
0
50
Time [ms]
100
150
100
150
Time [ms]
FEMUR-F V56-3
FEMUR-F V56-4
Pulse-V56-3
Pulse-V56-4
90
90
80
80
70
70
60
60
kN mm
kN mm
50
40
50
40
30
30
20
20
10
10
0
-10
NECKM=76.91
Sae180_5(SphereJoint Moment about [902100]
0
50
0
100
-10
150
NECKM=96.82
Sae180_5(SphereJoint Moment about [902100]
0
Time [ms]
NECK-M V56-3
50
Time [ms]
NECK-M V56-4
Figure 8. Injury values of pulses V56-3 and V56-4.
24
Numerical Optimization of Crash Pulses
Table 5.
Simulation results for the injury types to find the optimized pulse.
Pulse
Reference
V56-1
V56-2
V56-3
V56-4
HIC
283.5
633.5
350.0
454.7
336.7
CHEST-G
34.7
63.6
36.0
62.8
45.3
CHEST-D
26.5
28.6
32.0
33.9
27.1
FEMUR-F
3983.5
5724.3
3132.5
5466.7
5017.7
NECK-M
32.8
154.4
99.5
76.9
96.8
OSI
100.0
143.3
108.6
136.9
113.4
The lowest OSI value of the reference pulse justifies the further optimization of the
reference pulse by changing the deceleration levels and the ch ange points in time to
see the influence on the OSI value. First, some changes are made to improve
several injury types, based on the experience with the already simulated pulses. Of
course an improvement in one injury type can lead to a deterioration in other injury
types, but the OSI gives a weighted final result to compare carefully.
To improve the HIC value, the head deceleration at the end of the collision must be
lowered, however, not as low as in pulse V56-3. Lowering the deceleration level in
the third time region needs to increase the level in the first region. This gives pulse
V56-5, shown in Figure 9. Another possibility to lower the HIC value is extending the
second time interval which results in a lower contact velocity between the head and
the airbag. The head deceleration is lower over a larger time interval. This pulse is
V56-6. In Table 6 the results of the simulations are given. The HIC of pulse V56 -5
and mostly V56-6 is indeed lower as the reference pulse. In case of pulse V56-5 the
OSI is also improved.
To improve the Chest-G, the third deceleration level can be lowered a bit and the first
two levels can be increased (see pulse V56-7). In pulse V56-8 the deceleration
adjustments are smaller than in pulse V56-7 but the duration of the third deceleration
level is shortened by increasing the time duration of the first one. From Table 6 it is
clear that V56-7 improves the Chest-G and the OSI, while pulse V56-8 deteriorates
both. For pulse V56-8 the deceleration in the first part of the collis ion is too high. It is
difficult to improve the Chest-G any more because the curve path of the Chest-G of
the reference pulse is almost balanced perfectly, see Figure 5.
The Chest-D can be improved by lowering the deceleration level before the dummy is
restrained, and increasing the deceleration after the dummy is restrained. Pulse V56 9 lacks the first high deceleration level. In Table 6 an improvement of the Chest -D
can be seen but the OSI is higher, as expected, because the adjustments are
opposite to those improving the HIC and Chest-G.
Numerical Optimization of Crash Pulses
25
The Femur-F can be improved as mentioned earlier to balance the two leg force
peaks. This can be obtained by decreasing the difference between deceleration level
two and three of the reference pulse, see pulse V56-10. From Table 6 it is clear that
Femur-F and the OSI value improve a little.
From Table 5 it is clear that it is difficult to improve the Neck -M. However, pulses
V56-5, V56-7 and V56-8 improved the Neck-M, see Table 6. These pulses have a
higher deceleration level before 50 ms and a lower deceleration after 50 ms. This
means that the chest is moved further forward, which leads to minor displacement
between head and torso resulting in a lower neck moment. The lower deceleration in
the third interval yields a dec rease of the further head displacement.
26
Numerical Optimization of Crash Pulses
R e ference
Pulse V56-5
Velocity
R e ference
60
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
Pulse V56-6
Velocity
10
deceleration [g],velocity [km/h ]
40
50
60
70
80
90
0
0
10
20
30
40
50
60
70
time [ms]
R e ference
Pulse V56-7
Velocity
R e ference
60
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
Pulse V56-8
10
deceleration [g],velocity [km/h ]
40
50
60
70
80
90
0
0
10
20
30
40
50
60
70
Pulse V56-9
Velocity
R e ference
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [km/h ]
0
10
20
30
80
90
time [ms]
60
0
90
Velocity
time [ms]
R e ference
80
time [ms]
Pulse V56-10
Velocity
40
70
10
deceleration [g],velocity [km/h ]
40
50
60
70
80
90
0
0
10
20
30
50
60
time [ms]
Figure 9. Deceleration pulses V56-5 – V56-10 as variation on the reference pulse.
80
90
time [ms]
27
Numerical Optimization of Crash Pulses
Table 6.
Simulation results for the injury types to find a better pulse as the reference.
Pulse
Reference
V56-5
V56-6
V56-7
V56-8
V56-9
V56-10
HIC
283.5
276.0
225.3
278.6
292.6
331.1
284.8
CHEST-G
34.7
33.5
41.5
32.5
40.3
49.0
34.4
CHEST-D
26.5
19.6
28.7
20.7
23.6
21.2
25.0
FEMUR-F
3983.5
4407.6
6228.2
4385.9
4706.1
4577.8
3964.6
NECK-M
32.8
28.4
43.6
26.8
27.4
57.1
33.3
OSI
100.0
93.8
106.9
94.0
103.1
110.1
98.6
From the simulation results in Table 6 it can be concluded that the reference pulse is
improved by pulse V56-5 yielding an OSI value of 93.8. In the next step this new
pulse will be used as reference for further improvements. In order to find the optimal
pulse with the lowest OSI value, it is necessary to know which injury criteria have
much influence and which are not so important. For every injury criterion the
importance can be calculated by using the reference injury value of the reference
pulse and the limit of that injury criterion as mentioned in Table 1. Using the OSI
formula with these values and a supposed modified value of 10 per cent
improvement, this leads to the following factors in the summation for the OSI
calculation, presented in Table 7. These factors give an OSI value of 91.7.
Table 7.
Relative importance factors of the injury types at the reference pulse in the calculated
OSI value.
Injury Criterion
Summation factor
Relative importance
HIC
-0.0082
13 %
CHEST-G
-0.0304
49 %
CHEST-D
-0.0211
34 %
FEMUR-F
-0.0019
3%
NECK-M
-0.0005
1%
Chest-G and Chest-D have a large influence on the OSI and must therefore be
improved likely to the detriment of the other injury criterions to lower the OSI va lue.
Since more deceleration in the first and less in the third time interval improves the
Chest-G while for an improvement of the Chest-D the opposite adjustments are
desired, pulse V56-5 will only adjusted a little in both directions to see if the optimu m
is already obtained. Pulse V56-11 has a lower deceleration in the first and a little
higher deceleration in the third time interval, while pulse V56 -12 has a higher
deceleration in the first and a little lower deceleration in the third time interval with
respect to pulse V56-5, see Figure 10.
28
Numerical Optimization of Crash Pulses
In Table 8 the results are presented, in case of pulse V56-11 Chest-G improves but
Chest-D deteriorates. For pulse V56-12 Chest-G improves less and Chest-D
deteriorates less as for pulse V56-11. However, only the OSI of pulse V56-11
improves. To see if pulse V56-11 can give a better new optimum, the tendency is
persisted in pulse V56-13. However, this pulse deteriorates the OSI value as can be
seen in Table 8. At this moment pulse V56-11 can be the new reference pulse.
R e f. V56-5
Pulse V56-11
Velocity
40
70
60
50
40
30
20
10
deceleration [g],velocity [k m/h ]
0
0
10
20
30
50
60
80
90
time [ms]
R e f. V56-5
Pulse V56-12
Velocity
R e f. V56-5
60
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [k m/h ]
0
0
10
20
30
Pulse V56-13
Velocity
40
70
10
deceleration [g],velocity [k m/h ]
40
50
60
70
80
90
0
0
10
20
30
50
60
time [ms]
Figure 10. Deceleration pulses V56-11 –13 as variation on reference pulse V56-5.
80
90
time [ms]
29
Numerical Optimization of Crash Pulses
Table 8.
Simulation results for the injury types to find a better pulse as pulse V56-5.
Pulse
V56-5
V56-11
V56-12
V56-13
HIC
276.0
268.9
270.1
268.9
CHEST-G
33.5
30.0
31.8
33.3
CHEST-D
19.6
22.8
22.3
22.8
FEMUR-F
4407.6
4546.8
4203.3
4494.6
NECK-M
28.4
25.0
36.0
31.9
OSI
93.8
93.5
94.4
96.0
With pulse V56-14, see Figure 11, research is done on the influence of the lowest
deceleration level of the middle time interval. This lowest deceleration level (9 g for
reference pulse V56-11) determines the maximum stiffness of the longitudinals
during that time interval. For structural reasons it is preferable if this lowest level is
higher, also because of the large differences in deceleration resulting in abrupt
stiffness changes during the deformation. For this test the second level is increa sed
to 12 g and the first and third level are lowered to compensate for the same total
deceleration. From the results in Table 9 it is clear that the HIC and the Chest -G
increase a lot, which is the result of the higher velocity at which the dummy hits the
airbag, resulting in a higher deceleration.
The next step is to vary reference pulse V56-11 in several directions to gain insight in
the OSI sensitivity of the pulse for specific variations, see Figure 11. In pulses V56 15 and V56-16 the middle time interval is respectively increased and decreased with
constant deceleration level and with the necessary deceleration level adjustments of
the first and third time interval. In pulses V56 -17 and V56-18 the middle time interval
is respectively decreased and increased with constant deceleration level of the first
and third time interval and with the necessary deceleration level adjustment of the
second time interval. In pulses V56-19 and V56-20 the second time interval is moved
respectively later and earlier in t ime with all deceleration levels kept constant.
In Table 9 it can be seen that at most three injury types improve a little by the
represented adjustments and that only for pulse V56-15 the OSI value improves. To
see if pulse V56-15 is an optimum, the tendency of increasing the middle time
interval and increasing the first and third deceleration level could be persisted with a
further adjusted pulse. However, this new pulse will be far unrealistic. The first
deceleration level will be too high for a real ve hicle and the third higher deceleration
level starts too late after 60 ms, which is also unrealistic. Normally the higher
deceleration level in the third time interval could exist by pressing motor aggregates
together against the firewall. Pulse V56-15 can be seen as the found optimal pulse
for the lowest OSI value. In Figure 12 the specific injury time plots of this new optimal
30
Numerical Optimization of Crash Pulses
pulse are given and can be compared with the old reference pulse injuries in Figure
5.
R e f. V56-11
Pulse V56-14
Velocity
40
70
60
50
40
30
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
50
60
80
90
time [ms]
R e f. V56-11
Pulse V56-15
Velocity
R e f. V56-11
60
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [k m/h ]
0
0
10
20
30
Pulse V56-16
Velocity
40
70
10
deceleration [g],velocity [k m/h ]
40
50
60
70
80
90
0
0
10
20
30
50
60
time [ms]
R e f. V56-11
Pulse V56-17
Velocity
R e f. V56-11
60
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
80
90
time [ms]
Pulse V56-18
Velocity
40
70
10
deceleration [g],velocity [km/h ]
40
50
60
70
80
90
time [ms]
0
0
10
20
30
50
60
80
90
time [ms]
31
Numerical Optimization of Crash Pulses
R e f. V56-11
Pulse V56-19
Velocity
R e f. V56-11
60
60
50
50
40
40
30
30
20
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
Pulse V56-20
Velocity
40
70
10
deceleration [g],velocity [km/h ]
40
50
60
70
80
90
0
0
10
20
30
50
60
time [ms]
80
Figure 11. Deceleration pulses V56-14 – 20 as variation on ref. pulse V56-11.
Table 9.
Simulation results for the injury types to find the optimized pulse.
Pulse
V56-11
V56-14
V56-15
V56-16
V56-17
V56-18
V56-19
V56-20
V56-21
HIC
268.9
298.6
251.2
299.8
260.4
273.0
272.3
275.6
315.1
CHEST-G
30.0
39.2
31.0
36.3
30.6
37.6
33.7
32.9
39.1
CHEST-D
22.8
21.2
20.9
20.6
22.8
20.4
20.3
24.2
21.8
90
time [ms]
FEMUR-F
4546.8
4183.5
5066.0
4174.1
4551.5
4329.3
4561.9
3962.7
4169.6
NECK-M
25.0
29.6
28.9
25.8
26.6
23.5
25.8
30.4
33.9
OSI
93.5
100.1
92.6
97.4
93.6
97.3
94.6
96.5
101.4
32
Numerical Optimization of Crash Pulses
P u l s e - V 5 6 -1 5
0 .7
0 .6
mm/ms2
0 .5
0 .4
0 .3
0 .2
0 .1
0
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e l e r a t i o n [ 8 5 0 1 0 1 ] ) H IC
H IC = 2 5 1 .2
0
))
5 0
1 0 0
1 5 0
HIC
Tim e [m s ]
P u ls e - V 5 6 - 1 5
0 .5
mm/ms2
0 .4
0 .3
0 .2
0 .1
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [ 8 5 0 3 0 1 ] ) M S
C H E S T G = 3 1 .0 2
0
0
))
5 0
1 0 0
T im e
1 5 0
CHEST-G
[m s ]
P u ls e - V 5 6 - 1 5
0
B a r A x i a l E lo n g a t i o n [ 5 3 0 0 0 0 t / m
5 3 0 0 0 6 ]
- 1 0
mm
- 2 0
- 3 0
- 4 0
- 5 0
C H E S T D = 2 0 .85
- 6 0
0
5 0
1 0 0
T im e
1 5 0
CHEST-D
[m s ]
P u ls e - V 5 6 - 1 5
7
S a e 1 8 0 _ 5 ( S P R IN G
L o c a l F o r c e - M a g n itud e
[ 9 1 4 1 0 0 ] = - .
S a e 1 8 0 _ 5 ( S P R IN G
L o c a l F o r c e - M a g n itud e
[ 9 1 8 1 0 0 ] = -
F E M U R F = 5 0 6 6
6
kN
5
4
3
2
1
0
0
5 0
1 0 0
T im e
1 5 0
FEMUR-F
[m s ]
P u ls e - V 5 6 - 1 5
9 0
8 0
7 0
6 0
kN mm
5 0
4 0
3 0
2 0
1 0
0
- 1 0
N E C K M = 2 8 .9 2
S a e 1 8 0 _ 5 ( S p h e r e J o i n t M o m e n t a b o u t [9 0 2 1 0 0 ]
0
5 0
1 0 0
T im e
[m s ]
Figure 12. Injury values of the optimal pulse V56-15.
1 5 0
NECK-M
33
Numerical Optimization of Crash Pulses
The unusual higher deceleration level of the optimal pulse during the first 12.5 ms
with respect to traditional pulses can be made more plausible with next example. The
interception velocity, the velocity difference between the dummy and the vehicle at
approximately t=50 ms which determines the contact velocity with the airbag, is very
important to the injury severity. This relative velocity is almost completely determined
by the surface beneath the vehicle deceleration curve during the first 50 ms. At first
sight, averaging the first two intervals of pulse V56-15 into one interval, see pulse
V56-21 in Figure 13, seems to have almost no influence to the OSI value. However
this is not the case, the interception velocity increased. The reason for this is that if
the first two intervals are averaged, the deceleration at first is lower and the velocity
does not decrease as fast. This implies that the vehicle has traveled a larger distance
at the time that the dummy would be intercepted in the reference case. Since the
dummy velocity did not change, the interception occurs at a later point of time. At that
moment the vehicle has decelerated more yielding a higher interception velocity. In
Table 9 it can be seen that the OSI value of pulse V56-21 deteriorates a lot. So a
deceleration pulse with the same average deceleration but more deceleration in the
first time of the collision, not only shortens the final deformation length but also
lowers the interception velocity between dummy and vehicle interior and therefore
probably the OSI value.
R e f. V56-15
Pulse V56-21
Velocity
40
70
60
50
40
30
20
10
deceleration [g],velocity [km/h ]
0
0
10
20
30
50
60
80
90
time [ms]
Figure 13. Deceleration pulse V56-21 as variation on reference pulse V56-15.
5. Optimizing the deceleration pulse for other velocities
In the preceding section an optimal crash pulse has been found for only one collision
situation and therefore one certain combination of crash parameters. Parameters as
dummy positioning and vehicle interior are different for all types of occupants and
Numerical Optimization of Crash Pulses
34
vehicles. The influence of belt and airbag parameters has already been examined in
Section 3. Since more than 90 per cent of all frontal collisions occurs at a velocity
lower than the prescribed crash velocity of 56 km/h (Witteman 1993), also an optimal
pulse for lower collision velocities is necessary to minimize the occupant injury level
at that lower velocity. This is analyzed in Section 5.1. Although only 2 to 10 per cent
(dependent of the overlap percentage) of all crashes takes place at a velocity higher
than 56 km/h, also a pulse optimized at such a velocity is interesting because of the
higher energy level yielding larger vehicle deformations and higher injury levels. The
higher velocities are examined in Section 5.2. If the initial crash velocity is decreased
to 32 km/h, this results in a decrease of crash energy of 67 per cent with respect to a
crash speed of 56 km/h, so the vehicle might just be too stiff. An increase of the initial
velocity to 64 km/h results in an increase of energy of 31 per cent, so the structure
might be too supple.
The pulse shapes for the higher and lower velocities are based on the found optimal
pulse V56-15 at 56 km/h. This pulse is simply shortened in case of lower speeds and
extended in case of higher speeds. However, there are two different conversion
criteria to render pulse V56-15 into the needed pulses. The time-based criterion or
the distance-based criterion can be used. In case of the time-based criterion, the
deceleration intervals start at the same time and have the same time duration as
pulse V56-15 except for the last interval. In case of a lower velocity the last interval is
shortened and in case of a higher velocity the interval is extended. The distance based criterion is a little more complicated. Each deceleration time interval in case of
pulse V56-15 results in an increase of the deformation length, and at these distances
of the frontal structure the reinforcements or weakenings could be inserted in a
certain manner to generate the intended crash pulse. If this structure is used for other
velocities, the deceleration intervals have another duration and a deviating start and
end time than in case of pulse V56-15. The mentioned frontal structure distances
must be used to calculate for each other velocity the new time intervals.
The pulse V56-15 is set as new reference, so its OSI value is set to 100. In Figure 14
reference pulse V56-15 is given with the corresponding velocity and deformation
length curve against time.
35
Numerical Optimization of Crash Pulses
Decel.V 56-15
Deformation
V elocity
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
deceleration [g], deformation [cm], velocit y [km/h ]
time [ms]
Figure 14. Optimal deceleration pulse V56-15 and the velocity and deformation curve.
5.1. Optimizing the deceleration pulse at 32 km/h
In Figure 15 the time-based pulse V32-1 and the distance-based pulse V32-2 are
shown.
R e f. V56-15
Pulse V32-1
R e f. V56-15
Velocity
35
35
30
30
25
25
20
20
15
15
10
10
5
deceleration
[g],velocity [km/h ]
0
0
10
20
30
Pulse V32-2
Velocity
5
deceleration
[g],velocity [k m/h ]
40
50
60
70
80
90
0
0
10
20
30
40
50
60
time [ms]
Figure 15. Deceleration pulses V32-1 - 2 as variation on reference pulse V56-15.
70
80
90
time [ms]
Numerical Optimization of Crash Pulses
36
If the time-based pulse is considered, a certain phenomenon can be spotted.
Because the deceleration curve is the same as pulse V56-15 up to 60 ms, all injury
curves (see Figure 16) show the same curve for this first part of the collision. So the
injury curves resulting from pulse V32-1, have the same peak value as pulse V56-15
but are nevertheless much smaller. The smaller surface beneath the head
acceleration curve establishes a lower HIC value but the Chest -G and Chest-D are
so called ‘maximum value’criteria and therefore differ very little from pulse V56 -15.
The Femur-F curve lacks a large second peak, because the deceleration after the
first large peak is almost zero. The absence of this second peak lowers the Femur-F
value, as this peak was the larger of the two. An exception is the Neck-M value,
which has increased dramatically given the 29 Nm for the pulse at 56 km/h. If the
collision velocity is lowered, the dummy torso displacement is less and the dummy
torso moves backwards sooner. The head almost does not touch the airbag and
when the torso moves backwards it drags the still forward moving head also
backwards. This results of course in a large Neck -M value, which does not emerge in
case of much larger collision velocities. In case of pulse V56 -15 the torso moves
much further forwards and moves much later backwards again. This gives the airbag
the time and opportunity to restrain the head and lower the forward velocity. If the
torso then moves backwards, it can drag the head along without a large neck
moment. The head is even thrown backwards by the airbag just as the torso.
In case of the distance-based pulse V32-2, the vehicle deceleration before the
dummy has been restraint is much larger than in case of the time-based pulse. The
interception also occurs earlier, because of the small displacement of the vehicle as
result of this large deceleration. The dummy hits the belt with such an impact, that
most injuries are worsened compared to a velocity of 56 km/h. The OSI reaches a
value of 119.5. See Table 10.
As could be expected, the time-based pulse is preferable to the distance-based
pulse. Due to the much lower kinetic energy of the vehicle at this velocity, the
deformation length is only 253 mm in case of the time-based pulse and even 127 mm
for the distance-based criterion.
With the used deformation length it seems hard to improve the OSI for this velocity.
For pulse V32-3 the same pulse V32-1 is used but the airbag has been disengaged
and for pulse V32-4 the airbag has been made less stiff. This could be done with an
intelligent airbag system. Also bottoming out of the airbag occurs less fast at this
velocity.
37
Numerical Optimization of Crash Pulses
Pulse-V32-6
0.7
0.6
0.6
0.5
0.5
mm/ms2
mm/ms2
Pulse-V32-1
0.7
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0
0.1
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
HIC=159
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
HIC=68.27
0
50
Time [ms]
0.4
0.4
mm/ms2
mm/ms2
0.5
0.3
0.2
0.1
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=29.02
0
50
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=21.54
100
0
150
0
50
Time [ms]
Time [ms]
CHEST-G V32-1
CHEST-G V32-5
Pulse-V32-1
Pulse-V32-6
0
0
Bar Axial Elongation [530000 t/m 530006]
-10
-10
-20
-20
mm
mm
Bar Axial Elongation [530000 t/m 530006]
-30
-40
-30
-40
-50
-50
CHESTD=17.66
0
CHESTD=14.35
50
100
-60
150
0
50
Time [ms]
Pulse-V32-1
Pulse-V32-6
6
6
5
5
4
4
kN
kN
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=3519
3
3
2
2
1
1
0
50
100
0
150
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=2057
0
50
Time [ms]
100
150
100
150
Time [ms]
FEMUR-F V32-1
FEMUR-F V32-5
Pulse-V32-1
Pulse-V32-6
90
90
80
80
70
70
60
60
50
50
kN mm
kN mm
150
CHEST-D V32-5
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
100
Time [ms]
CHEST-D V32-1
40
40
30
30
20
20
10
10
0
-10
150
0.3
0.2
0.1
0
100
Pulse-V32-6
0.5
-60
150
HIC V32-5
Pulse-V32-1
0
100
Time [ms]
HIC V32-1
NECKM=52
Sae180_5(SphereJoint Moment about [902100]
0
50
0
100
150
-10
NECKM=69.08
Sae180_5(SphereJoint Moment about [902100]
0
Time [ms]
NECK-M V32-1
50
Time [ms]
NECK-M V32-5
Figure 16. Injury values of pulses V32-1 and V32-5.
38
Numerical Optimization of Crash Pulses
Because the head and chest in both cases are decelerated less fast, these injuries
are considerably lower than with the stiff airbag. The Chest -D stays almost the same
as the head has not a large influence on the chest compression and the chest
already was completely restraint by the belt. Logically also the Femur -F stays almost
the same, but the main dismay is caused by the Neck-M. The effect mentioned
earlier of the torso dragging the head backwards is intensified by both measu res. The
combination of the injuries in case of the less stiff airbag (pulse V32 -4) results in a
lower OSI value as from pulse V32-1.
The only way to decrease the OSI further at this point is to change the belt
parameters with an intelligent restraint sys tem or by increasing the vehicle
deformation length. The belt changes are not investigated further but a deformation
length increase is discussed below.
To lower the injury level by increasing the deformation length, pulse V32 -5 is created
with the same deceleration level as the minimum level used in all previous pulses,
see Figure 17. This pulse results in a deformation length of 448 mm instead of 253
mm of pulse V32-1.
R e f. V32-1
Pulse V32-5
Velocity
35
30
25
20
15
10
5
deceleration
[g],velocity [km/h ]
0
0
10
20
30
40
50
60
70
80
90
100
time [ms]
Figure 17. Deceleration pulse V32-5 as variation on pulse V32-1.
This new pulse shape indeed lowers the OSI significantly, that is from 92.3 to 80.2 ,
see Table 10. Most injuries decrease, but the Neck-M prevents a lower OSI, see
Figure 16. As the interception velocity and therefore the dummy deceleration at
interception are lower, all injuries except for the Neck -M are lower. With this pulse the
deceleration continues after interception, the torso is thrown backwards by the
restraint system while the head is moving forwards by the deceleration. This results
in a larger Neck-M, which could be lowered by increasing the deceleration in the first
39
Numerical Optimization of Crash Pulses
part. This will lower the deformation length and Neck-M, but will increase all other
injury types. A deceleration of 9 g over a distance of 448 mm seems unrealistic, but
pulse V56-15 needs 586 mm deformation length for the first two time intervals, see
Figure 14. So if the first deceleration level of 31.8 g of pulse V56 -15 could be
changed to the level of 9 g of the second interval by means of a structural solution,
both crash velocities yield a minimal OSI.
Table 10.
Simulation results for the injury types to find the optimized pulse.
Pulse
V56-15
V32-1
V32-2
V32-3
V32-4
V32-5
HIC
251.2
159.0
349.0
122.5
93.9
68.3
CHEST-G
31.0
29.0
49.1
27.3
26.8
21.5
CHEST-D
20.9
17.7
28.1
18.3
18.5
14.3
FEMUR-F
5066.0
3518.6
4858.1
3442.2
3473.2
2057.2
NECK-M
28.9
52.0
49.5
248.4
92.4
69.1
OSI
100.0
92.3
119.5
92.8
89.4
80.2
Since the OSI in case of a crash velocity of 32 km/h only dropped with 7.7 per cent
for pulse V32-1 with respect to pulse V56-15, also lower velocities based on this
reference pulse are examined. To prevent unnecessary injuries resulting from this
pulse, which was optimized for higher velocities, a relationshi p has been searched
between the collision velocity and the OSI for this pulse. All pulses for the different
velocities are generated the same way as pulse V32-1. Pulse V56-15 is used and the
last part has been cut off until the end velocity reaches zero. N ormally the airbag is
disengaged for velocities below some 30 km/h, but to investigate its use, all
calculations have been performed with and without an airbag. The examined
velocities are 28, 24, 16 and 8 km/h and the results are shown in Table 11. In Fig ure
18 the OSI is displayed as function of the crash velocity, for the velocities of 28 km/h
and lower also without an airbag.
40
Numerical Optimization of Crash Pulses
Table 11.
Simulation results for the injury types to find the optimized pulse.
Pulse
V56-15
V8t_airbag
V8t_no bag
V16t_airbag
V16t_no bag
V24t_airbag
V24t_no bag
V28t_airbag
V28t_no bag
HIC
251.2
5.6
3.8
42.8
39.6
112.1
100.1
150.1
138.8
CHEST-G
31.0
9.5
7.7
19.4
16.9
28.5
26.5
32.3
29.1
CHEST-D
20.9
5.3
2.1
12.1
9.4
18.4
17.1
20.0
19.4
Without airbag up to 28
FEMUR-F
5066.0
1050.1
1050.1
1812.3
1811.4
3847.1
3889.4
4245.3
4243.0
NECK-M
28.9
49.6
52.8
74.8
175.4
73.3
251.1
63.9
276.0
OSI
100.0
62.3
59.5
75.3
74.0
90.6
91.5
95.8
96.9
With airbag
110
100
90
80
70
60
OSI 50
40
30
20
10
0
0
4
8 12 16 20 24 28 32 36 40 44 48 52 56 60 64
velocity [km/h]
Figure 18. The OSI for other velocities.
From Figure 18 it is clear that the injuries do not lower that much as would be
expected. For a parking collision velocity o f 8 km/h the injury level is too high, but for
higher velocities from 24 until 64 km/h the increase of the injury level is low and even
at high velocities not life -threatening (see Section 5.2). At low velocities the airbag
causes even little higher injuri es due to the kinetic energy of the airbag. However,
these results do not completely support the decision of car manufacturers of
deactivating the airbag for crash velocities below 30 km/h, resulting in repair cost
saving. Most injuries are slightly increa sed by an airbag for low velocities, but the
Neck-M is decreased a lot for velocities higher than 8 km/h. This decrease amounts
57 per cent at 16 km/h up to 77 per cent at 28 km/h. See Figure 19. Without an
airbag the Neck-M exceeds the limit value of 189 Nm above a velocity of 16 km/h.
41
Numerical Optimization of Crash Pulses
The head acceleration lowers for decreasing velocities just as the HIC value. If no
airbag is used, the summit of the head acceleration logically moves further in time
until the head hits the steering wheel. In case of an airbag used at low velocities
where the steering wheel is not reached by the head, the HIC is higher as result of
the high unnecessary head deceleration.
The Chest-G deteriorates a little if the belt and the airbag intercept the torso. The
airbag increases the torso deceleration because it generates an extra force on the
torso. At high velocities the airbag takes over a part of the dummy interception,
resulting in lower chest deformations. However, if the crash velocity is low, then the
belt does not carve the chest anymore and the airbag generates a larger force on the
chest as necessary, resulting in a larger average chest deformation. This
phenomenon increases with decreasing velocity. The lower interception velocity also
decreases the chest deformation. Leg forces of course decrease as the velocity
decreases.
Without airbag up to 28
With airbag
280
260
240
220
200
180
160
140
120
Neck-M
100 [Nm ]
80
60
40
20
0
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64
velocity [km/h]
Figure 19. The Neck-M for other velocities.
The Neck-M has no logical relation to the crash velocity as shown in Figure 19. The
airbag plays for all velocities an important role in restricting the head movement and
therefore limiting the neck moment. Also in case of a velocity of only 8 km/h the head
still touches the airbag, which limits the movement. As the crash velocity drops, the
Neck-M decreases because the kinetic energy of the head lowers. However, one
aspect lowers the Neck-M for increasing velocity. If the velocity increases, the airbag
42
Numerical Optimization of Crash Pulses
has a larger part of the collision time to decelerate the head. In this way the head is
not dragged backwards by the back bouncing torso while still moving forward relative
to the interior. These two influences generate a Neck-M maximum at about 20 km/h.
At 56 km/h there is a perfect balance of the head and torso deceleration. If the
velocity is decreased, the torso moves back sooner dragging the head along. This
increases the Neck-M at lower velocities and the lower kinetic energy of the head
decreases the Neck-M. A possibility to prevent the Neck-M from increasing at lower
velocities is bringing the airbag further into the interior or increase the dummy
movement by decreasing the belt restrainment.
5.2. Optimizing the deceleration pulse at 64 km/h
In Figure 20 the time-based pulse V64-1 and the distance-based pulse V64-2 are
shown on basis of the optimal reference pulse V56-15.
R e f. V56-15
Pulse V64-1
Velocity
R e f. V56-15
70
70
60
60
50
50
40
40
30
30
20
20
10
deceleration
[g],velocity [km/h ]
0
0
10
20
30
40
Pulse V64-2
Velocity
10
deceleration
[g],velocity [km/h ]
50
60
70
80
90
100
0
0
10
20
30
40
50
60
70
time [ms]
80
90
100
time [ms]
Figure 20. Deceleration pulses V64-1 - 2 as variation on reference pulse V56-15.
When the time-based pulse V64-1 is considered, the head deceleration curve has
almost the same shape as the graph of 56 km/h but logically lasts a little longer
resulting in a higher HIC value, see Figure 21. The Chest deceleration curve is only
extended a bit, but a different local maximum determines the Chest-G. In case of the
pulse at 56 km/h both summits of the Chest-D curve were perfectly balanced. The
second summit reached its maximum when the pulse ended but now the pulse lasts
longer so the summits are not balanced anymore yielding a larger Chest-D value. For
the Femur-F there are not much changes because the pulse is the same, up to the
Numerical Optimization of Crash Pulses
43
summit of the second peak. The still lasting deceleration can not prevent the legs
from bouncing back resulting in a similar peak value. The lower volume of the airbag
in the last part of the collision yields a further forward movement of the head, so the
Neck-M has increased in the last part compared to pulse V56-15, see Table 12.
44
Numerical Optimization of Crash Pulses
Pulse-V64-2
0.7
0.6
0.6
0.5
0.5
mm/ms2
mm/ms2
Pulse-V64-1
0.7
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0
0.1
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
HIC=286
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
HIC=367.5
0
50
Time [ms]
0.4
0.4
mm/ms2
mm/ms2
0.5
0.3
0.2
0.1
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=29.63
0
50
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=38.92
100
0
150
0
50
Time [ms]
Time [ms]
CHEST-G V64-1
CHEST-G V64-2
Pulse-V64-1
Pulse-V64-2
0
0
Bar Axial Elongation [530000 t/m 530006]
-10
-10
-20
-20
mm
mm
Bar Axial Elongation [530000 t/m 530006]
-30
-40
-30
-40
-50
-50
CHESTD=24.71
0
CHESTD=30.14
50
100
-60
150
0
50
Time [ms]
Pulse-V64-1
Pulse-V64-2
6
6
5
5
4
4
kN
kN
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=4921
3
3
2
2
1
1
0
50
100
0
150
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=4617
0
50
Time [ms]
100
150
100
150
Time [ms]
FEMUR-F V64-1
FEMUR-F V64-2
Pulse-V64-1
Pulse-V64-2
90
90
80
80
70
70
60
60
50
50
kN mm
kN mm
150
CHEST-D V64-2
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
100
Time [ms]
CHEST-D V64-1
40
40
30
30
20
20
10
10
0
-10
150
0.3
0.2
0.1
0
100
Pulse-V64-2
0.5
-60
150
HIC V64-2
Pulse-V64-1
0
100
Time [ms]
HIC V64-1
NECKM=41.3
Sae180_5(SphereJoint Moment about [902100]
0
50
0
100
150
-10
NECKM=25.01
Sae180_5(SphereJoint Moment about [902100]
0
Time [ms]
NECK-M V64-1
50
Time [ms]
NECK-M V64-2
Figure 21. Injury values of pulses V64-1 and V64-2.
45
Numerical Optimization of Crash Pulses
In case of the distance-based pulse V64-2 the head deceleration starts a little later
than in case of pulse V56-15, as can be seen in Figure 21, because of the shorter
deceleration before airbag interception. However, the duration and level of the head
deceleration is larger due to the long lasting third part of the crash. This results of
course in a higher HIC value. The same effect is true for the Chest-G and Chest-D,
which also start later but reach a higher value. The Femur -F is lower than in case of
56 km/h because the vehicle is decelerated extra just when the legs hit the
dashboard for the first time, so they bounce less back. Also the relative velocity
between the legs and the vehicle is lower. The large velocity has no negative
influence on the Neck moment because the opposite as mentioned earlier happens.
The interception velocity is lower than it would be with pulse V56 -15, because the
deceleration before interception is lower. When the head and torso are now
intercepted they are both decelerated immediately on a high level and just when the
airbag has its largest volume. This results in a more simultaneous deceleration of the
head and the torso, yielding a lower Neck-M.
Also for larger velocities than 56 km/h, the time-based pulse can be given preference
to. As can be seen in Table 12, the OSI of 103.5 is much lower than the OSI value of
114.3 of the distance-based pulse. This because the pulse was already optimized up
to 90 ms and an extension may cause some minor injury enlargements, but a large
shift of the third interval obviously has la rger influences. Unfortunately the time-based
pulse has a deformation length of 934 mm and the distance-based pulse a
deformation length of 887 mm. The reference pulse V56-15 needs only 724 mm.
Since the lowest OSI for 64 km/h is only 103.5 where the OSI of the reference for 56
km/h was 100, the search for a lower OSI will be ceased. On account of the
unacceptable large deformation length, a pulse will be searched for to keep the
deformation length beneath 800 mm. For the previous pulses the surface beneath
the pulse needed to be constant, as it represents the decrease in velocity, but the
pulses could be varied and the deformation length was gained afterwards. For the
next pulses, the total deceleration had still to be the same, but the deformation length
must be maximal 800 mm. If two deceleration intervals are known, then the level and
the duration of the third interval will follow from two equations. The surface beneath
the first two intervals has been increased compared to pulse V56-15 to meet the
higher initial velocity. This already lowers the deformation length, and the level and
duration of the third interval follows from the next basic equations:
v = v0 − a ⋅t
(4).
46
Numerical Optimization of Crash Pulses
s = s0 + v0 ⋅t − 1 2 ⋅a ⋅t 2
(5).
To find a low OSI value for 64 km/h and meet the maximum 800 mm deformation
length requirement, pulses V64-3, -4 and -5 are analyzed as variation on pulse V641, see Figure 22.
R e f. V64-1
Pulse V64-3
Velocity
R e f. V64-1
70
70
60
60
50
50
40
40
30
30
20
20
0
0
10
20
30
40
Velocity
10
deceleration
[g],velocity [km/h ]
10
deceleration
[g],velocity [km/h ]
0
Pulse V64-4
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
Pulse V64-5
Velocity
Ref. V64-1
70
70
60
60
50
50
40
40
30
30
20
20
P ulse V 64-6
Velocity
10
deceleration
[g], velocit y [km/h ]
10
deceleration
[g],velocity [km/h ]
0
100
time [ms]
time [ms]
R e f. V64-1
90
0
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
time [ms ]
80
90
100
time [ms]
Figure 22. Deceleration pulses V64-3 - 6 as variation on reference pulse V64-1.
Pulse V64-3 only lasts 77 ms, and the high deceleration levels have a negative effect
on almost all injury levels. Although the torso and head are decelerated almost
simultaneously as explained for pulse V64-2, the deceleration is much higher so the
head moves far into the stiff airbag, yielding a not much lower Neck-M. A lowering of
the third deceleration level as shown by pulse V64-5 lowers the Neck-M, but there is
an optimum. At lower decelerations as already explained, the torso will move
backwards while the head is still being decelerated by the less stiff airbag. In pulse
V64-4 the third deceleration level is further decreased. The influence on the Neck -M
47
Numerical Optimization of Crash Pulses
graph of both pulses V64-4 and V64-5 is shown in Figure 23. The interception
velocity for pulse V64-5 is smaller than for pulse V64-4, but the deceleration during
interception (12 g) and afterwards is higher, resulting in slightly higher injuries, except
as mentioned for the Neck-M. Pulse V64-4 has a deformation length of 800 mm and
the lowest OSI, namely 110.3. An earlier activation of the airbag could bring some
improvements. The airbag is now fully inflated at t=50 ms, an earlier activation time
will cause the airbag to take over some forces from the belt.
Pulse V64-6 in Figure 22 is a small variation of the almost optimal pulse V64 -4. The
deceleration levels of the first 50 ms are the same, but the third level is a little
increased to 23 g, more like the optimal level of pulse V64-5 and the same level as
reference pulse V56-15. Although the injury levels for this pulse are not calculated, it
could be expected that the OSI will be a little better as for pulse V64-4 because of the
already mentioned lower Neck-M.
Table 12.
Simulation results for the injury types to find the optimized pulse.
Pulse
V56-15
V64-1
V64-2
V64-3
V64-4
V64-5
HIC
251.2
286.1
367.5
461.6
352.7
386.1
CHEST-G
31.0
29.6
38.9
45.5
39.3
40.5
CHEST-D
20.9
24.7
30.1
32.0
23.8
24.2
FEMUR-F
5066.0
4921.2
4617.4
6002.4
4297.2
4627.6
NECK-M
28.9
41.3
25.0
40.2
39.0
25.0
OSI
100.0
103.5
114.3
125.5
110.3
112.8
48
Numerical Optimization of Crash Pulses
Pulse-V64-8
0.7
0.6
0.6
0.5
0.5
mm/ms2
mm/ms2
Pulse-V64-7
0.7
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0
0.1
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
HIC=352.7
0
50
100
0
150
Magnitude(Sae180_5(Acceleration[850101]) HIC ))
HIC=386.1
0
50
Time [ms]
0.4
0.4
mm/ms2
mm/ms2
0.5
0.3
0.2
0.1
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=39.34
0
50
Magnitude(Sae180_5(Acceleration[850301]) MS ))
CHESTG=40.48
100
0
150
0
50
Time [ms]
Time [ms]
CHEST-G V64-4
CHEST-G V64-5
Pulse-V64-7
Pulse-V64-8
0
0
Bar Axial Elongation [530000 t/m 530006]
-10
-10
-20
-20
mm
mm
Bar Axial Elongation [530000 t/m 530006]
-30
-40
-30
-40
-50
-50
CHESTD=23.82
0
CHESTD=24.2
50
100
-60
150
0
50
Time [ms]
Pulse-V64-7
Pulse-V64-8
6
6
5
5
4
4
kN
kN
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=4297
3
3
2
2
1
1
0
50
100
0
150
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = FEMURF=4628
0
50
Time [ms]
100
150
100
150
Time [ms]
FEMUR-F V64-4
FEMUR-F V64-5
Pulse-V64-7
Pulse-V64-8
90
90
80
80
70
70
60
60
50
50
kN mm
kN mm
150
CHEST-D V64-5
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
100
Time [ms]
CHEST-D V64-4
40
40
30
30
20
20
10
10
0
-10
150
0.3
0.2
0.1
0
100
Pulse-V64-8
0.5
-60
150
HIC V64-5
Pulse-V64-7
0
100
Time [ms]
HIC V64-4
NECKM=39.04
Sae180_5(SphereJoint Moment about [902100]
0
50
0
100
150
-10
NECKM=25.03
Sae180_5(SphereJoint Moment about [902100]
0
Time [ms]
NECK-M V64-4
50
Time [ms]
NECK-M V64-5
Figure 23. Injury values of pulses V64-4 and V64-5.
49
Numerical Optimization of Crash Pulses
6. Structural design specifications for different crash velocities
The optimal pulse for a speed of 64 km/h (V64-6) (total deformation length of 762
mm) is plotted in Figure 24 together with the optimal pulse for 56 km/h (V56-15) (total
deformation length of 724 mm) and with the optimal pulse for a collision with 32 km/h
(V32-5) (total deformation length of 448 mm).
56 km/h
64 km/h
32 km/h
64
60
45g
56
52
48
44
32g
40
36
9g
9g
32
23g
28
2 4 y [km/h ]
velocit
9g
20
16
23g
12
9g
8
4
0
0
4
8
12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80
deformation length [cm]
Figure 24. Three optimal decelerations curves in three phases.
Before designing structural solutions to realize the desired deceleration pulses
(Witteman 1999), first the specifications for this design will be mentioned. For these
specifications the optimal curves that were obtained for three different crash
velocities as shown in Figure 24 will be used. In this figure the velocity decrease is
plotted against deformation length instead of time, because it is more interesting to
know on which length position in the car structural measures are necessary to realize
a change in stiffness corresponding with the desired change in deceleration level. In
Table 13 the time duration and deformation length of each deceleration interval are
50
Numerical Optimization of Crash Pulses
presented. As can be seen the difference in deformation length in the first interval of
the 56km/h and the 64 km/h collision is small. So for simplification the lengths of 170
mm and 188 mm could be joined together on 179 mm. At the end of the second
interval the deformation length is already identical for both velocities, vz. 586 mm.
These interval borders are visualized in Figure 24 as two vertical lines. For the 32
km/h collision there is no difference in deceleration between the first two intervals.
Table 13
Deceleration parameters of 3 crash velocities.
Crash velocity
Phase 1
Deceleration
Deformation length
Time duration
Phase 2
Deceleration
Deformation length
Time duration
Phase 3
Deceleration
Deformation length
Time duration
32 km/h
56 km/h
64 km/h
9g
32 g
170 mm
12.5 ms
45 g
188 mm
12.5 ms
9g
9g
9g
total 448 mm 416 mm, total 586 mm 398 mm, total 586 mm
100.7 ms
42.5 ms, total 55 ms
37.5 ms, total 50 ms
23 g
23 g
138 mm, total 724 mm 176 mm, total 762 mm
35 ms, total 90 ms
39.4 ms, total 89.4 ms
7. Conclusions
An optimal crash pulse for 56 km/h is obtained, which has an OSI value of 92.6 with
respect to the initially used reference pulse with an OSI value of 100. To adjust this
crash pulse in order to obtain a pulse for different velocities, the time -based and the
distance-based criterions are analyzed. The time-based criterion proves to be the
best, as the dummy uses a part or an extended version of the optimal pulse.
Unfortunately, also the time-based pulse gives unsatisfactory results for high as well
as low velocities. For high velocities the injuries are relatively low but the deformation
length is too large. For low velocities the deformation length is short but the injuries
are too high.
The optimal pulse obtained for higher velocities (with a smaller deformation length
than 800 mm), has a higher deceleration level in the first interval and the levels of the
Numerical Optimization of Crash Pulses
51
middle and the third interval must remain unchanged in comparison with the
optimized reference pulse for 56 km/h. The OSI for this higher velocity crash with 30
per cent more kinetic energy increases by no more than 10 per cent.
The optimal pulse for a collision with 32 km/h has a constant deceleration level of 9
g, the same level as the higher velocity pulses have during their middle interval. The
total deformation length is 448 mm. For this velocity the OSI is decreased with 20 per
cent in comparison with the 56 km/h collision. The injuries decrease further with
decreasing velocity except for the Neck-M. To prevent a much higher Neck-M at
lower velocities, an airbag is already useful for velocities as low as 8 km/h, instead of
the 30 km/h and higher as assumed by the car manufacturers. In order to lower
injuries for all velocities even more, an intelligent restraint system is required.
From these observations (see also Table 13) it can be concluded with the considered
numerical model, that for minimal injury for crash velocities starting at 32 km/h the
vehicle structure needs a constant stiffness to decelerate 9 g during the first 586 mm.
For higher velocities as 32 km/h the stiffness of the first 179 mm must be directly
increased to decelerate up to 45 g for the highest velocity of 64 km/h. After 586 mm
deformation has been reached the stiffness must be increased to decelerate to 23 g
for all relevant velocities.
8. References
Auto Motor und Sport, spezial, Überleben in der Golf-Klasse, 1992.
Bakker, S.O., Optimising Deceleration Pulses - Lowering injury levels for different collision
velocities, Master’s thesis, Internal report WOC/VT/R/97.04, Eindhoven University of
Technology, Laboratory for Automotive Engineering, Eindhoven, The Netherlands, 1997.
Bosch, Automotive Handbook 3rd edition, pp. 649, 1993.
Brantman, R., Achievable Optimum Crash Pulses for Compartment Sensing and
Airbag Performance, Thirteenth International Technical Conference on Experimental
Safety Vehicles (ESV), Paper S9-O-22, pp. 1134-1138, Paris, France, 1991.
ESI Group Software Product company, Occupant Database, PAM-SAFE™ Version 1996,
PAM System International, 1996.
Numerical Optimization of Crash Pulses
52
Fokker, P., Minder rekkende gordels blijken niet veiliger te zijn (Less stretching belts are not
safer), in Dutch, from: Utrechts Nieuwsblad 6 april 1995.
Hoffman, R., Pickett, A.K., Ulrich, D., Haug, E., Lasry, D., Clinkemaillie, J., A Finite
Element Approach to Occupant Simulation: The PAM-CRASH Airbag Model, SAE paper no.
890754, 1989.
Hoffmann, R., Ulrich, D., Protard, J.-B., Wester, H., Jaehn, N., Scharnhorst, T., Finite
Element Analysis of Occupant Restraint System Interaction with PAM-CRASH, 34th Stapp
Car Crash Conference, Orlando, Florida, 1990.
Jager, J.A. de, Optimisation of Restraint System Design using mid-range approximation
concepts, Master’s thesis, code WfW/93.183, Eindhoven University of Technology,
Eindhoven, The Netherlands, 1993.
Kebschull, S.A., Zellner, J.W., Auken, M. van, Rogers, N.M., Injury Risk/Benefit Analysis of
Motorcyclist Protective Devices Using Computer Simulation and ISO 13232, Proceedings of
the Sixteenth International Technical Conference on the Enhanced Safety of Vehicles (ESV),
Paper 98-S10-W-26, pp. 2357-2374, Windsor, Canada, 1998.
Khalil, T.B., Sheh, M.Y., Chalons, P., Dubois, P.A., Integrated Vehicle-Dummy-Airbag
Model for Frontal Crash Simulation by FE Analysis, AMD-Vol. 210/BED-Vol. 30,
Crashworthiness and Occupant Protection in Transportation Systems, ASME, pp. 355-382, 1995.
Landheer jr., D., Witteman, W.J., Kriens, R.F.C., Crashworthiness in the Concept Stage: A
Parametric Body Design Method, Proceedings of the 26th FISITA Congress, Paper B16.54,
16p, Prague, Czech Republic, 1996.
Lasry, D., Hoffmann, R., Protard, J.-B., Numerical Simulation of Fully Folded Airbags and
Their Interaction with Occupants with Pam-Safe, SAE Paper no. 910150, 1991.
Levine, R.S., Head and Neck Injury, Chapter 5, USA, SAE P-276, ISBN 1560914998, 1994.
Mertz, H., Patrick, L., Strength and response of the human neck, pp. 821-846, From:
Biomechanics of impact injury and injury tolerances of the head-neck complex, USA, SAE
PT-43, ISBN 1560913630, 1993.
Ni, X., Lasry, D., Haug, E., Hoffmann, R., Advances in Problem-Adaptive Occupant Modelling
with PAM-SAFE, Proceedings of the Thirteenth International Technical Conference on
Experimental Safety Vehicles (ESV), Paper 91-S9-O-20, Paris, France, 1991.
Numerical Optimization of Crash Pulses
53
O’Neill, B., Lund, A.K., Zuby, D.S., Estep, C.R., New Vehicle Crashworthiness Evaluations
by the Insurance Institute for Highway Safety, Proceedings of the Fifteenth International
Technical Conference on the Enhanced Safety of Vehicles (ESV), pp. 2030-2039, Melbourne,
Australia, 1996.
Relou, J.J.M.G., Integrated Crash Simulation of a Frontal Crash Structure and a Dummy,
Master’s thesis, Internal report WOC/VT/R/95.66, Eindhoven University of Technology,
Laboratory for Automotive Engineering, Eindhoven, The Netherlands, 1995.
Schlosser, J., Ullrich, P., Industrial Applications of PAM-SAFE in Occupant Simulation at
Audi, PAM’95 Fifth European Workshop on Advanced Finite Element Simulation
Techniques, Bad Soden, Germany, 1995.
Seiffert, U., Scharnhorst T., Die Bedeutung von Berechnungen und Simulationen für den
Automobilbau - Teil 1, Automobiltechnische Zeitschrift 91, pp. 241-246, 1989.
Seiffert, U., Fahrzeugsicherheit, VDI-Verlag, ISBN 3-18-401264-6, Düsseldorf, 1992.
Seiffert, U., Möglichkeiten und Grenzen der neuen Frontal- und Seitenaufprall-Gesetzgebung,
ATZ 9, pp. 494-504, 1997.
Sparke, L.J., Tomas, J.A., Crash Pulse Optimisation for Minimum Injury Risk to Car
Occupants, SAE paper no. 945162, XXV FISITA congress Beijing, China, ISBN 780003-311-2/U.7, 1994.
Sparke, L.J., Optimisation of Crash Pulse Through Frontal Structure Design, Proceedings of
the Fifteenth International Technical Conference on the Enhanced Safety of Vehicles (ESV),
Paper 96-S4-W-21, pp. 720-725, Melbourne, Australia, 1996.
Sparke, L.J., Restraint System Optimisation for Minimum Societal Harm, Proceedings of the
Sixteenth International Technical Conference on the Enhanced Safety of Vehicles (ESV), Paper
98-S1-O-12, pp. 255-258, Windsor, Canada, 1998.
Stolinski, R., The Development of Result Presentation in Australia’s New Car Assessment
Program, Proceedings of the Fifteenth International Technical Conference on the Enhanced
Safety of Vehicles (ESV), Paper 96-S1-W-27, pp. 261-268, Melbourne, Australia, 1996.
Viano, D.C., Arepally, S., Assessing the Safety Performance of Occupant Restraint Systems,
SAE paper no. 902328, 1990.
Numerical Optimization of Crash Pulses
54
Wijntuin, A.L.M., Evaluation of vehicle interior dimensions and numerical modeling of
interior foam padding, Practical assignment, Internal report WOC/VT/R/95.56, Eindhoven
University of Technology, Laboratory for Automotive Engineering, Eindhoven, The
Netherlands, 1995.
Wismans, J., Bruijs, W., Janssen, E.G., Beusenberg, M., Injury Biomechanics, Lecture notes
college 4J610, Eindhoven University of Technology, Eindhoven, The Netherlands, 1992.
Witteman, W.J., Insufficiency of a Single Frontal Impact test for Vehicle Crashworthiness
Assessment, Proceedings of the 26th ISATA Conference, Paper 93SF069, pp. 307-314, Aachen,
Germany, 1993.
Witteman, W.J., Kriens, R.F.C., Modeling of an Innovative Frontal Car Structure: Similar
Deceleration Curves at Full Overlap, 40 per cent Offset and 30 Degrees Collisions, Proceedings
of the Sixteenth International Technical Conference on the Enhanced Safety of Vehicles (ESV),
Paper 98-S1-O-04, pp. 194-212, Windsor, Canada, 1998.
Witteman, W.J., Improved Vehicle Crashworthiness Design by Control of the Energy
Absorption for Different Collision Situations, Ph.D. thesis, Eindhoven University of
Technology, Laboratory for Automotive Engineering, Eindhoven, The Netherlands, ISBN 90386-0880-2, 1999.
Yang, J., Haland, Y., Modeling of Adaptive Passenger Airbag Systems in Car Frontal Crashes,
Proceedings of the Fifteenth International Technical Conference on the Enhanced Safety of
Vehicles (ESV), Paper 96-S3-W-10, pp. 486-501, Melbourne, Australia, 1996.
automotive engineering & product design
dr.ir. W.J.Witteman
prof.dr.ir. R.F.C.Kriens
EUROPAM 7-8 October 1999
automotive engineering & product design
Which crash pulse gives the lowest
injury levels with an already optimized
restraint system ?
(instead of optimizing the occupant
restraint system at a given crash pulse)
After optimal pulses are found for
several crash velocities, a structure
must be built which realizes such an
optimal pulse.
automotive engineering & product design
automotive engineering & product design
The following restraint parameters with current values must be
optimized for minimal OSI value:
Belt feed (pull out distance of belt before locking), 6% (200 mm).
yes / no
Belt pretensioner (pyrotechnic tightening of the belt), pull in
distance vs. time curve (150 mm pull in at 12 ms). yes / no
Spool effect, belt force vs. outlet curve (12 kN = 120 mm).
+ / - 25 %
Belt strain, belt force vs. strain curve (30 kN = 18.3 %
elongation).
+ / - 25 %
Airbag delay time (activation time after collision start), 15 ms.
+ / - 10 ms
Airbag stream out surface, 3848.5 mm2.
+ / - 25 %
Airbag mass flow, curve of mass flow dependent on inflation
time.
+ / - 25 %
automotive engineering & product design
The demands for a realistic crash index are formulated as
follows:
1. Some injury types are more important than others.
2. Improvements in high valued injury criteria count more than
improvements in already low valued criteria.
3. If one injury type has a much higher value with respect to
its limited value than the other criteria, then an improvement
on this higher value must be count additionally.
4. The improvement in one injury type counts less than the
improvement of several injuries types.
5. A large improvement in an injury type counts more than a
small one.
automotive engineering & product design
Relevant injury criteria with their by legislation limited values.
Injury Criterion
Limit value
Weight factor
HIC
1000
8
CHEST-G
60 g
2
CHEST-D
50 mm
2
FEMUR-F
10000 N
1
NECK-M
189 Nm
2
The Overall Severity Index (OSI) is given by the following formulae:
  (HICmod − HICref) HICref
HICref

OSI =  
*
*
 *8+

HIClim
HIClim HIClim * TOTREF 
CHEST.Gref
 (CHEST.Gmod − CHEST.Gref) CHEST.Gref

*
*

 *2+

CHEST.Glim
CHEST.Glim CHEST.Glim * TOTREF 
CHEST.Dref
 (CHEST.Dmod − CHEST.Dref) CHEST.Dref

*
*

 *2+

CHEST.Dlim
CHEST.Dlim CHEST.Dlim * TOTREF 
FEMUR.Fref
 (FEMUR.Fmod − FEMUR.Fref) FEMUR.Fref

*
*

 * 1+

FEMUR.Flim
FEMUR.Flim FEMUR.Flim * TOTREF 
NECK.Mref
 (NECK.Mmod − NECK.Mref) NECK.Mref

*
*

 *2

NECK.Mlim
NECK.Mlim NECK.Mlim * TOTREF 
 100
+ 100
*
 0.75
where
TOTREF =
HICref CHEST.Gref CHEST.Dref FEMUR.Fref NECK.Mref
+
+
+
+
HIClim CHEST.Glim CHEST.Dlim FEMUR.Flim NECK.Mlim
automotive engineering & product design
Reference pulse
Velocity
60
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
OSI = 100
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
time [ms]
automotive engineering & product design
Pulse V56-1
Pulse V56-2
Velocity
60
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
Velocity
OSI=143.3
50
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
40
30
20
10
0
OSI=108.6
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
0
10
20
30
40
50
60
70
time [ms]
Pulse V56-3
90
time [ms]
Velocity
Pulse V56-4
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
80
Velocity
60
OSI=136.9
50
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
40
30
20
10
0
OSI=113.4
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
0
time [ms]
Four examples of other possible crash pulses.
10
20
30
40
50
60
70
80
90
time [ms]
p u ls e
0 .6
HIC = 284
mm/ms2
0 .5
0 .4
0 .3
0 .2
(limit 1000)
0 .1
0
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [ 8 5 0 1 0 1 ] ) H IC
H IC = 2 8 3 . 5
0
))
5 0
1 0 0
T im e
1 5 0
HIC
[m s ]
R e fe r e n c e
p u ls e
0 .5
CHEST-G = 35 g
mm/ms2
0 .4
0 .3
0 .2
(limit 60 g)
0 .1
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [ 8 5 0 3 0 1 ] ) M S
C H E S T G = 3 4 .7 3
0
0
))
5 0
1 0 0
T im e
1 5 0
CHEST-G
[m s ]
R e fe r e n c e
p u ls e
0
B a r A x i a l E lo n g a t i o n [5 3 0 0 0 0
t/m
5 3 0 0 0 6 ]
-1 0
CHEST-D = 27 mm
mm
-2 0
-3 0
-4 0
(limit 50 mm)
-5 0
C H E S T D = 2 6 .4 6
-6 0
0
5 0
1 0 0
T im e
7
1 5 0
[m s ]
R e fe r e n c e
CHEST-D
p u ls e
S a e 1 8 0 _ 5 ( S P R IN G
L o c a l F o r c e - M a g n itu d e
[9 1 4 1 0 0 ] =
-.
S a e 1 8 0 _ 5 ( S P R IN G
F E M U R F = 3 9 8 3
L o c a l F o r c e - M a g n itu d e
[9 1 8 1 0 0 ] =
-
FEMUR-F = 3984 N
6
kN
5
4
3
(limit 10.000 N)
2
1
0
0
5 0
1 0 0
T im e
1 5 0
[m s ]
R e fe r e n c e
FEMUR-F
p u ls e
1 6 0
1 4 0
NECK-M = 33 Nm
1 2 0
1 0 0
kN mm
automotive engineering & product design
R e fe r e n c e
0 .7
8 0
6 0
(limit 189 Nm)
4 0
2 0
0
N E C K M = 3 2 .7 7
S a e 1 8 0 _ 5 ( S p h e r e J o i n t M o m e n t a b o u t [9 0 2 1 0 0 ]
0
5 0
1 0 0
T im e
[m s ]
1 5 0
NECK-M
automotive engineering & product design
R efer e nc e
Puls e V56- 5
Velo c ity
R efer e nc e
60
OSI=93.8
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
Ve loc ity
60
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
Pu ls e V5 6- 6
0
10
20
30
40
50
60
70
80
50
40
20
10
0
90
OSI=106.9
30
0
10
20
30
40
50
60
70
time [ms ]
R efer e nc e
Puls e V56- 7
Velo c ity
R efer e nc e
60
OSI=94.0
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
Pu ls e V5 6- 8
Ve loc ity
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
40
30
20
10
0
10
20
30
40
50
60
70
80
50
40
20
10
0
90
OSI=103.1
30
0
10
20
30
40
50
60
70
time [ms ]
R efer e nc e
Puls e V56- 9
OSI=110.1
Velo c ity
90
R efer e nc e
Puls e V56 - 10
Ve loc ity
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
80
time [ms ]
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
90
60
50
0
80
time [ms ]
0
10
20
30
40
50
60
70
80
90
time [ms ]
50
40
OSI=98.6
30
20
10
0
0
10
20
30
40
50
60
70
80
90
time [ms ]
automotive engineering & product design
R ef. V56-5
OSI=93.5
Puls e V56-11
Veloc ity
40
70
60
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
0
10
20
30
50
60
80
OSI=94.4
OSI=96.0
time [ms]
R ef. V56-5
Puls e V56-12
Veloc ity
R ef. V56-5
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
90
Puls e V56-13
Veloc ity
40
70
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
time [ms]
0
10
20
30
50
60
80
90
time [ms]
automotive engineering & product design
R ef. V56 -11
OSI=100.1
Puls e V56 -1 4
Velo c ity
40
70
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
0
10
20
30
50
60
80
time [ms]
OSI=92.6
R ef. V56 -1 1
Puls e V56 -15
Veloc ity
R ef. V56 -1 1
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
OSI=97.4
Puls e V56 -16
Veloc ity
40
70
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
90
0
10
20
30
40
50
60
70
80
90
time [ms]
50
40
30
20
10
0
0
10
20
30
50
60
80
90
time [ms]
automotive engineering & product design
R ef. V56-11
Puls e V56 -1 7
Veloc ity
R ef. V56-11
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
Veloc ity
60
OSI=93.6
50
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
40
30
20
10
0
Puls e V56 -1 8
0
10
20
30
40
50
60
70
80
40
30
20
10
0
90
OSI=97.3
50
0
10
20
30
40
50
60
70
time [ms]
R ef. V56-11
Puls e V56-19
90
time [ms]
Veloc ity
R ef. V56-11
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
80
Puls e V56-20
Veloc ity
60
]
h/
m
k[
yt
i
c
ol
e
v,]
g[
n
oti
ar
el
e
c
e
d
OSI=94.6
50
40
30
20
10
0
OSI=96.5
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
time [ms]
0
10
20
30
40
50
60
70
80
90
time [ms]
Optimized
OSI=100
OSI=92.6
Velocity
60
deceleration [g], velocity [km/h]
automotive engineering & product design
Start reference
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
time [ms]
284
P u ls e - V 5 6 - 1 5
0 .7
0 .6
0 .6
0 .5
0 .5
mm/ms2
mm/ms2
HIC
0 .4
0 .3
0 .2
0 .2
35
251
0 .1
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [ 8 5 0 1 0 1 ] ) H IC ) )
H IC = 2 8 3 . 5
0
5 0
1 00
0
15 0
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [8 5 0 1 0 1 ]) H IC ) )
H IC = 2 5 1 . 2
0
P u ls e - V 5 6 - 1 5
0 .5
0 .4
0 .4
0 .3
0 .2
CHEST-G
0 .2
0 .1
31
0 .1
0
5 0
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [8 5 0 3 0 1 ]) M S ) )
C H E S T G = 3 1 .0 2
1 00
0
15 0
0
50
T i m e [m s ]
T i m e [m s ]
P u ls e - V 5 6 - 1 5
-2 0
-2 0
mm
mm
B a r A x i a l E lo n g a t i o n [5 3 0 0 0 0 t / m 5 3 0 0 0 6 ]
-1 0
-3 0
-4 0
CHEST-D
-3 0
-4 0
-5 0
21
-5 0
C H E S T D = 2 6 .4 6
0
C H E S T D = 2 0 .8 5
5 0
1 00
-6 0
15 0
0
6
5
5
4
4
3
3
2
2
1
1
0
5 0
1 00
1 5 0
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 4 1 0 0 ] = - .
7
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [9 1 8 1 0 0 ] = F E M UR F =39 8 3
6
0
10 0
T i m e [m s ]
P u ls e - V 5 6 - 1 5
kN
kN
3984
50
T i m e [m s ]
R e fe r e n c e p u ls e
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [9 1 4 1 0 0 ] = - .
7
FEMUR-F
1 5 0
0
B a r A x i a l E lo n g a t i o n [5 3 0 0 0 0 t / m 5 3 0 0 0 6 ]
-1 0
-6 0
10 0
R e fe r e n c e p u ls e
0
27
1 50
0 .3
M a g n i t u d e ( S a e 1 8 0 _ 5 ( A c c e le r a t i o n [ 8 5 0 3 0 1 ] ) M S ) )
C H E S T G = 3 4 .7 3
CHEST-D
10 0
T im e [m s ]
R e fe r e n c e p u ls e
0 .5
0
50
T i m e [m s ]
mm/ms2
CHEST-G
mm/ms2
0
HIC
0 .4
0 .3
0 .1
0
15 0
S a e 1 8 0 _ 5 ( S P R IN G L o c a l F o r c e - M a g n i t u d e [ 9 1 8 1 0 0 ] = F E M UR F =5 06 6
FEMUR-F
5066
0
50
T i m e [m s ]
10 0
1 5 0
T i m e [m s ]
R e fe r e n c e p u ls e
P u ls e - V 5 6 - 1 5
1 6 0
9 0
1 4 0
8 0
7 0
NECK-M
1 2 0
33
6 0
5 0
kN mm
NECK-M
1 0 0
kN mm
automotive engineering & product design
R e fe r e n c e p u ls e
0 .7
8 0
4 0
6 0
3 0
4 0
2 0
29
1 0
2 0
0
0
N E C K M = 3 2 .7 7
S a e 1 8 0 _ 5 ( S p h e r e J o i n t M o m e n t a b o u t [9 0 2 1 0 0 ]
0
5 0
1 00
T i m e [m s ]
15 0
-1 0
N E C K M = 2 8 .9 2
S a e 1 8 0 _ 5 ( S p h e r e J o i n t M o m e n t a b o u t [9 0 2 1 0 0 ]
0
50
10 0
T i m e [m s ]
1 5 0
automotive engineering & product design
R e f. V5 6 - 1 5
Puls e V56 -2 1
Velo c ity
40
70
60
]
h/
m
k[
yt
i
c
ol
e
v],
g[
n
oti
ar
el
e
c
e
d
50
40
30
20
10
0
0
10
20
30
50
60
80
90
time [ms]
HIC=251, CHEST-G=31, CHEST-D=21, FEMUR-F=5066, NECK-M=29
OSI=92.6
HIC=315, CHEST-G=39, CHEST-D=22, FEMUR-F=4170, NECK-M=34
OSI=101.4
automotive engineering & product design
W ith ou t airbag u p to 28
]
m
N[
M
k
c
e
N
W ith air ba g
280
260
240
220
200
180
160
140
120
100
80
60
40
20
0
0
4
8 12 16 20 24 28 32 36 40 44 48 52 56 60 64
velocity [km/h]
The Neck-M for other velocities.
automotive engineering & product design
automotive engineering & product design
56 km/h
64
OSI=100
60
64 km/h
OSI=110
32 km/h
OSI=80
45g
56
52
48
]
h/
m
k[
yt
i
c
ol
e
v
44
32g
40
36
9g
9g
32
23g
28
24
9g
20
16
23g
12
9g
8
4
0
0
4
8
12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80
deformation length [cm]
Three optimal decelerations curves in three phases.
automotive engineering & product design