- Email: [email protected]
Restrained occupants on the nonstruck side in lateral collisions
Accid. Anal. & Prev. Vol. 25, No. 2, pp. 147-152, Printed in the U.S.A.
$6.00 + .oo 0001-4575/93 0 1993 Pergamon Press Ltd.
1993
RESTRAINED OCCUPANTS ON THE NONSTRUCK SIDE IN LATERAL COLLISIONS G. M. MACKAY, J. HILL, S. PARKIN, Accident Research Centre, The University of Birmingham,
and
J. A. R. MUNNS
Birmingham
Bl5 2TT, United Kingdom
(Receiwd 30 Sepfember 1991;in revise~.~$~ 25 January 1992)
Abstract-Ofinju~-pr~ucing collisions with high seat belt use, some 25% to 30% are lateral collisions. This paper describes some of the characteristics of those collisions as they relate to the front-seat occupant sitting on the side opposite to the impact. The data came from a stratified sample of in-depth crash investigations conducted in the Birmingham region in the period 1983 to 1989 involving current model cars. Crash severity was assessed using the Vehicle Deformation Index (VDI) of the Collision Deformation Classification (CDC) ratings and velocity change. Injury severity was assessed using AIS 85 for each body region. 193 cases of restrained occupants in nonstruck side collisions were examined. Of those occupants with head injuries of AIS I 2, 35% came out of the shoulder section of the seat belt. Of abdominal injuries of AIS 5 2,72% came from the seat belt itself. Interaction between front seat occupants was not a frequent cause of injury to the nonstruck side occupant. Some aspects of seat belt geometry might be changed so that the trajectory and loading of the nonstruck side occupant are improved.
INTRODUCTION
(Harms et al. 1987), and those studies emphasize the relative importance of head injuries for nonstruckside, restrained occupants. Experimental work by Herbert et al. ( 1976) and Horsch ( 1980) explored the limits of motion of the chest and hence the head in angled impacts, using dummies in sled tests. That work showed that the limit of retention of the torso occurred when the direction of the crash force was at 45 degrees from straight ahead. At greater angles, head excursion markedly increased, although Horsch noted that even at angles of around 60 degrees, significant energy was removed from the upper part of the body by the seat belt before the torso slipped out of the belt. This present study examined the injury causation aspects of nonstruck-side, restrained occupants in an effort to assess how seat belts function in a representative sample of real-world collisions. Lateral collisions are seldom like sled tests in the sense that the direction of principal force is often changing during the crash phase because of rotation of the occupant’s car. We wished therefore to assess how seat belts perform for a condition for which they are obviously not designed. It was our intention to describe the distribution of vehicle impact damage and the causes of injury, and to identify any seat belt problems. Cause of injury was to be described in terms of vehicle component struck, interaction with other car occupants, and loads applied through the seat belt system. Head contact evidence was to be used to es-
Lap/shoulder seat belts are designed to protect occupants in frontal collisions, primarily. The geometry of the anchorages and the placement of the belt across the chest, the iliac spines of the pelvis, and the lower abdomen, when combined with appropriate seat characteristics, limit forward excursion of the head, chest, and knees and apply loads through the seat belt webbing that are tolerable for most of the exposed population in all but the most extreme collisions. Side impacts, however, constitute a singificant part of the collision population at around 20°h to 30%, increasing to 36% for fatalities with higher levels of seat belt use (Mackay et al. 1990). In side impacts most attention has been focused on the struck-side occupant, and the recent rulemaking through Federal Motor Vehicle Safety Standard 2 14 addresses that condition. For that type of loading, the seat belt has relatively little influence on outcome, because the main loads come from the occupant’s striking the side header and the door, contacts normally not influenced by a seat belt. Head excursion out through the side window can be reduced by a seat belt, and of course ejection risk is diminished. For an occupant on the nonstruck side of a lateral collision, however, the seat belt is of greater consequence. Crash severities and injury patterns have been described in other projects in Canada (Dalmotas 1983), Germany (Otte et al. 1984) and Britain 147
G. M. MACXAYet al.
148
tablish if occupants remained fully restrained throughout the impact. Furthermore, seat belts were examined in anticipation of component malfunction under lateral impact conditions. MATERIAL
AND
METHODS
The Accident Research Centre at The University of Birmingham investigate 300 car accidents in the West Midlands of England every year. The Centre employ medical and enginee~ng specialists who work closely with local hospitals, coroners, the police, garages, and scrap yards. A major project has been in progress since 1983, contributing results to a national study with the object of evaluating vehicle crashworthiness and occupant injury. A stratified sample ofcar occupant collisions has been examined in the Birmingham region. The sampling plan includes all fatalities, 50% of the police-reported “serious” cases, i.e. requiring hospital admission for fractures and severe lacerations, and one-third of lesser levels of injury in current-model cars less than six years old. The cars are inspected at garages soon after their collisions, and data from that inspection are later coordinated with medical information from hospital records and questionnaire information from the people involved. The methodology is described elsewhere (Mackay et al. 1985). This paper presents an analysis of accidents from the West Midlands. Thus, 507 side impacts-27% of all collision types-were available for analysis. A precise definition of what a side impact is cannot be clear-cut. For example, occupants can move laterally to strike side structures when the front of the car is struck obliquely. We chose only those cases where the impact was on the side of the vehicle and where the direction of principal force on the case vehicle was between 1 to 5 o’clock on the right side and 7 to 1 1 o’clock on the left. Thus sideswipe collisions were not included. Crush was measured in each case, and a Collision Deformation Classification (CDC) (Society of Automotive Engineers 1985) was assigned to each vehicle. This is a seven-digit code commonly used to describe crash damage that has been caused by directly contacting another object. Only the part of CDC that indicates crush extent will be used here-the Vehicle Deformation Index (VDI). VDI takes an integer value between I and 9 to indicate the maximum extent of crush perpendicular to the vehicle’s struck side. Where possible, velocity change at impact (delta V) was calculated from crush measurements and the known weights of the vehicles involved, but for side
Tahlc I Selection ofnonstruck
side lateral impacts
.-
Total side impacts
Cases used
‘IhIof cases used ..-
Driver side impacts Passenger side impacts
281 220
62 131
22 60
Total
507
193
-
impacts the appropriate stiffness coefficients to be used are not well established for many of the cars in the sample. Injury severity codes according to the Abbreviated Injury Scale (AIS) (American Association for Automotive Medicine 1985) were assigned by body region. Occupant height, weight, sex, and age were known in most cases. Contacts within the vehicle were assessed for the main body regions. RESULTS The numbers of candidate side impacts are shown in Table 1. 193 cases where there was a restrained occupant on the nonstruck side were identified. In Britain, driving on the left side of the road in right-hand driver cars produces a preponderance of drivers alone in left-side, i.e. passenger-side, collisions. Principal
dirtwion
offorce
The direction of principal force in the collision was defined in each case. The cases analysed exclude side-swipe collisions. Table 2 shows that directions 2 and 3 o’clock, and 9 and 10 o’clock predominate, at 71%.
The amount of crush was assessed using the VDI protocol, and the frequency is illustrated in Table 3. Level 1 is the door skin width, level 2 is the width of the door window horizontal to the cant rail level (side header). Levels 3, 4, and 5 are equal zones to the centreline of the car, and then levels 6 to 8 repeat the process to the other side of the vehicle. Level 9 is crush all the way to the opposite side of the car, essentially breaking the car in half. with crush extending beyond the opposite cant rail. Table 2. Direction ofprincipal force on the clock direction ___~
No. of cases ‘%,
l/i 1
Z/IO
3/9
4/x
5/7 ---
NK
37 I9
68 35
69 36
I0
4 2
5 3
5
Restrained occupants in lateral coltisions
Table 3. Amount ofcrush according to VDI numbers code
-
No.of cases 0x1 -I-
I
2
3
4
5
6
11 5
43 22
70 36
32 16
13 7
6 5
Of the cases with known crush therefore, I.55 out of f80, or 87’%, had crush rhat was limited to VIX 4 or Iess, i.e. cases where the crush was less than twothirds of the way across the sitting zone on the opposite side from that of the occupant being considered. This is relevant to assessing the effect of partial restraint of the nonstruck-side occupant regardless of collision angle.
Velocity change values were calculated for 69% of the relevant eases. Within the limits of the known cases it was found that VDI was hnearIy related to delta V up to a value ofVDI6, at 54 km/h. Results are shown in Table 4. Within this genera1 condusion, however, there is a range of variation influenced mainly by the direction of force. With that reservation, the above data can be summarised by saying that the great majority of the cases were in the 2/3 or 9/ 10 o’clock direction with velocity change values less than 40 km/h, with crush less than two-thirds of the width of the opposite passenger-compartment sitting space (87%).
For 180 eases injury details and positive belt use were established. These are now considered. Dividing the cases between passenger- and driver-side impacts suggests, although the numbers are small, that there is benefit in having an occupant between you and the impact. In driver-side impacts, there was a struck-side occupant (the driver) present in every case. In passenger-side impacts there was a passenger present in only 32% of the cases. Table 5 ilhtstrates the injury outcome according to the Maximum AXS(MAIS),
For the nonstruck-side occupant, an identified contact in the vehicle for the head is the prime indicator of the occupant’s trajectory. Such contacts are given in Table 6.
I
~ 8
9
NK
Total
2 1
2 I
13 7
193 100%
--
Thirty-nine percent ofthe nonstruck-side occupants received an AIS 5 2 head injury, mainly from contacts with the opposite header, the roof, side-window glass, opposite door, or the other occupant. Each case was evaluated individually to assess if it was necessary for the occupant to come out of the seat belt in order to make the contact. Table 7 shows that that occurred in 35% of cases. There were 13 additional head injury cases of AIS 2 3. In 9 of the 13 cases the crush was at least to the centreline ofthe car (VDI 5+), and VDI 4 in the remaining 4 cases. With such amounts of crush it was not possible to assess the function ofthe seat belt, and it became irrelevant if the intrusion extends up to or beyond the centreline of the ear.
Of the 180 occupants, 42% received an injury to the chest. Only 5 received a chest injury of AIS z 3, (3%) mainly multiple rib fractures with haemo/ pneumothorax. The causes of the chest injuries are illustrated in Table 8. These data show that chest injuries are caused mainly by the seat belt. The incidence of chest contacts with far-side structures in driver-side impacts is Iow, because in those collisions the driver was present between the case occupant and the zone of impact. Few injuries to the case occupant were caused by the other (near-side) person, and thus the presence of that other occupant is beneficial in diminishing contacts with far-side structures and the resulting injuries,
Forty-one of the I80 occupants on the nonstruck side received an injury to the abdomen (23%). The Causes of those injuries are shown in Table 9; passenger- and driver-side cases are combined because no differences were apparent. The great majority (7 1%) of abdominal injuries came from the seat belt.
Table 4. Delta v and crush level I_-I--
-II -
149
VD! no.
Delta Y Km/hr
1
2
3
4
5
6
~$8
(26
<32
<40
<48
(56
^x1-.
7+ <62
I 50
G.
Table 5. Overall injury severity ofnonstruck-sideoccupant of car struck
by side
MAIS i
2 (‘W)
MAIS 2 3 (“LI)
52 (96)
MACKAY
M.
et al
Table 7. lncidenceofcomingout
h
Driver-side impacts Passenger-side impacts
I I3 (90)
2 (4) 13(10)
54 126
Total
165 (9’)
15 (8)
180
Remained in belt Driver-side impacts Passcngcr-side impacts Both sides
Seut hrlr prrjiwmanc’e
Table 6. Contacts
Impact
forAlS
i 2 head
out of belt
‘%Iout of belt
8 IO I8
47 29 35
9 24 33
DISCUSSION
In-depth investigations provide useful feedback to the designer of the performance of various subsystems in the car in circumstances that cannot be predicted in every respect in laboratory testing. Table 10 details some factors relating to seat belt performance. The incidence of damage to buckle casings illustrates that in lateral collisions the buckle is exposed to specific high-energy blows from the tunnel, the seat, or the central armrest structures. In none of those cases did the belt release, but in some instances there were difficulties in undoing the damaged buckle after the collision. Eight percent of occupants had put their belts on with a twist. That conceivably may result in a less favourable loading across the abdomen in particular. The roping (and jamming) of the webbing at the D ring is probably a consequence of the lateral direction in which the seat belt is being loaded, but it indicates the need for careful consideration of the dynamics of the D ring in collisions that are not frontal. The ultimate consequence may be to encourage the webbing to break prematurely. and there were two cases of that. Given the principle of a swivelling D ring, it is difficult to see how that can be avoided at the extremes of crash severity and lateral loading. In 3% of cases there was evidence that the belt was loose when loaded. This was probably due to poor adjustment of the belt combined with a retraction spring that was beginning to fatigue. No belts with comfort features were present in this sample. In no instances was the performance of the locking mechanism of the retractor judged to be unsatisfactory.
Far door
ofshoulderbelt injury cases
Side window glass
causing
This study has examined some of the characteristics ofthe restrained occupant on the nonstruck side in lateral collisions. The sample is biased towards the more serious injury cases, due to the overall sampling plan, so that it gives insights primarily into the limits of performance of the seat belt for this crash configuration. The typical collision investigated is a side impact from the 2 and 3 o’clock direction (or 9 and 10 o’clock) with a velocity change of the struck car of less than 40 km/h (25 mph) and with crush limited to less than two-thirds of the sitting space on the struck side. The pattern and depth of crush are major factors considered in determining an amount of energy required to crush a car, and hence in calculating delta V. Naturally. therefore, delta V and crush may be expected to have a positive relationship. It was found that VDI was linearly related to delta V, although there was a range of variation influenced mainly by the direction of force. For a given velocity change there is more crush if the impacts come from 10/l 1 o’clock or l/2 o’clock than if the collisions are more nearly perpendicular at 9 or 3 o’clock. This is usually because the angled collision is with the front corner of a striking car rather than being distributed across the entire front. Delta V is also dependent on the stiffness and mass of the case vehicle and the striking object. The case vehicle will experience a higher speed change after impact with an object of increased stiffness or mass, all other factors being equal. For these collisions there is obvious asymmetry in that a driver is present in every case. but a front-
head injury AIS i
Z for non struck-side
Facia
Seat
A pillar
Steering wheel
occupants
Other occupant
Header roof. etc.
‘2
Driver side Passenger side
0 5
? 1;
0 2
, ;
0 2
? ;
4 3
9 20
I9 51
Both (“/I)*
5 (3)
I9 (IO)
Z(l)
3 (2)
J(l)
3 (1)
7 (4)
29(16)
70 (39)
*Percentages
indicate
the proportion
of all I80 occupants
so injured
Restrained
occupants
in lateral collisions
151
Table 8. Chest injury causes Other occupant
Far-side structure
Impact Driver side Passenger side Both sides (%)* *Percentages
0 II
Seat belt
2 Z(3)
I6 28 44 (24)
11(6) indicate
the proportion
of all 180 occupants
Other components 2 4 6 (3)
Seat belt
Seat
29 16%
2 I%
Note: Percentages so injured.
5 3% indicate
Induced
2 I% the proportion
3 2% ofall
27 48 75 (42)
IO. Seat belt function n
to the abdomen
Other occupant
7
cupant between the impact and the nonstruck side occupant appears to be beneficial in limiting farside contacts. Chest and abdominal injury occurred in 42% and 23% of cases, respectively. Because the seat belt is providing the main load paths for the crash forces into the occupant on the nonstruck side, it is inevitable that the chest and abdomen will be injured from the seat belt. The great majority (7 1%) of abdominal injuries came from the seat belt. With lateral flexion of the upper part of the body occurring, in 35% of cases the torso was coming out of the shoulder belt, it is highly likely that the orientation of the pelvis is changing. That results in the inboard section of the seat belt loading the lateral aspect of the abdomen directly. This is reflected in the severe abdominal injury cases in which there were ruptures of the spleen or tears to the liver. One case occurred with a lateral flexion fracture of the lumbar spine, with neurological involvement. There were, however, only five cases of abdominal injuries of AIS L 3. All of the cars in this study had two bucket seats in the front, there were no bench seat designs. Thus most of the seat belt buckles were located relatively low, near to the hip joint, often on rigid brackets attached to the seat. Even with such geometry, the relatively high incidence of abdominal injury (23%) suggests that the belt is moving from the pelvis into the abdomen, although the shoulder section of the seat belt is also loading at least the upper quadrant of the abdomen as well as the chest. In a number of cases, we were able to verify this from the position and nature of bruises. In some cars, the buckle is relatively highly positioned and thus the
Table
Intruding structures
N
so injured
seat passenger is present in only 32% of the cases. Thus, for U.K. data, left-side impacts predominate; the opposite occurs in the United States. Table 5 shows that 8% of nonstruck-side occupants scored MAIS L 3. Thomas and Bradford ( 1989) used data from the same national study from which the present sample has been taken and found that 243 out of 1,6 18 (15%) struck-side occupants scored MAIS >_ 3. Their analysis included unrestrained occupants, so we might expect a higher proportion of serious injury. Nevertheless, with nearly half the number of MAIS L 3 cases observed in the present nonstruck-side sample, this comparison does suggest that nonstruck-side occupants are less likely to be seriously injured. In one-third of cases, the occupant came out of the shoulder section of the seat belt. This was established from the positions of the head contacts within the car. As well as the direction of the impact a number of other factors have a bearing on this event-the position of the upper anchorage, the size of the occupant, the seat position, the adjustment of the upper anchorage, and the looseness of the seat belt. In the marginal condition, the geometry of the upper anchorage will influence the retention of the shoulder, but for pure lateral impacts the torso will translate out of the belt anyway. Further work will examine differences between two- and four-door cars and adjustable versus fixed upper anchorages, but that requires substantially more cases. Experimental work might attempt to alleviate the problem with improved seat belt geometry and webbing pretensioning. This study illustrates, however, the exposure of the head particularly, as a result of translating across the passenger compartment. The presence of an oc-
Table 9. Causes of injuries
NK
N 41 23%
180 occupants
%
Satisfactory Buckle casmg fractured I80 degrees of twist in webbing Webbing roped at D ring Belt loose when loaded Webbing broken
140 16 I5 II 6 2
74 8 8 6 3 I
No. of belts examined
190
100
152
G.
M. MACKAY
buckle itself is loading the abdomen. The properties of the seat base should not be overlooked; in some cases, they may enable submarining to occur. Experimental work could be done to examine the relative risks of lateral abdominal loading from various seat belt geometries and with modified seat base characteristics. This would be of interest now that a potential means of assessing abdominal loads is available with the foam insert for Hybrid III developed by Rouhana et al. ( 1989). Not mentioned in this analysis are injuries to other body regions. This is because injuries to the upper and lower limbs were infrequent and almost always minor. There was one fatal neck injury to a 7% year-old with advanced arthritis of the cervical vertebrae. Otherwise, the injuries of main consequence were confined to the head, chest, and abdomen. This field study also points to some aspects of seat belt function in lateral collisions where the restraint is loaded somewhat differently than in the main design condition of the frontal impact. Buckle performance and D ring operation particularly are of concern in some instances. In conclusion, the typical nonstruck-side occupant is injured in conditions that are broadly similar to side-impact crash tests that are currently proposed. However, 13 cases of serious head injury occurred with considerable passenger-compartment intrusion. In these cases, the seat belt performance was considered irrelevant. The kinematics of the nonstuck-side occupant are very different from those of the struckside occupant. While the nonstick-side occupant is at lower risk, serious injuries did occur, and these were mitigated when a struck-side occupant was present. Future work on seats and seat belt design for lateral loading conditions might well reduce the likelihood of torso egress from the belt, limit abdominal injury, and enhance the functional integrity of seat belt components. ncknowk~~~~~~rnc~~ts_Theauthors wish to thank the sponsors of this research, the Transport and Road Research Laboratory of the U. K. Department of Transport, the Ford Motor Company, the Rover Group, and Nissan (U. K.) Ltd. We are also grateful to all
et al
members of the Cooperative Crash Injury Project who have contributed to the data collection process, and the many police officers, hospital staff, garage owners, and others who form an invisible laboratory for our research.
REFERENCES American Association for Automotive Medicine (A.A.A.M.). The Abbreviated Injury Scale. 1985 Revision. Arlington Heights, IL: A.A.A.M.; 1985. Dalmotas, D. J. Injury mechanisms to occupants restrained by three-point belts in side impacts. S.A.E. Paper 830462. Warrendale, PA: Society of Automotive Engineers (S.A.E.); 1983. Harms, P. L.; Renouf, M.; Thomas, P. D.; Bradford, M. Injuries to restrained car occupants, what are the outstanding problems? Proceedings of the 1lth E.S.V. ConferWashington, ence. DC: U.S. Department of Transportation, National Highway Traffic Safety Administration; 1987: 183-201. Herbert, D. C.; Stott, J. D.; Corben, C. W.; Cutting, D.; Gillies, N. Occupant head space in passenger cars. Sydney, N.S.W., Australia: Traffic Accident Research Unit; 1976. Horsch. J. D. Occupant dynamics as a function of impact angle and belt restraint. S.A.E. p-88. Proceedings of the 24th Stapp Car Crash Conference. Warrendale, PA: S.A.E.; 1980: 4 17-438. Mackay, G. M.; Ashton, S. J.; Galer, M. D.; Thomas. P. D. The methodology of in-depth car crashes in Britain. S.A.E., P-159. Warrendale, PA: S.A.E.; 1985: 365-390. Mackay, G. M.; Cheng, L.; Smith, M.; Parkin, S. Restrained front seat occupant fatalities. Proceedings ofthe 34th A.A.A.M. Conference, Scottsdale, AZ, Oct. 1-3, 1990; 139-162. Otte, D.; Suren, E. G.; Appel, H.; Nehmzow, J. Vehicle parts causing injuries to front seat car occupants in lateral impacts. S.A.E., P-152. Proceedings of the 28th Stapp Car Crash Conference Warrendale, PA: S.A.E.; 1984: 13-24. Rouhana, S. W.; Viano, D. C.; Jedrzejczak, E. A.: McCleary, J. D. Assessing submarining and abdominal injury risk in the Hybrid III family ofdummies. S.A.E., P227. Proceedings of the 33rd Stapp Car Crash Conference. Warrendale, PA: S.A.E. 1989: 257-280. Society of Automotive Engineers. Collision Deformation Classification. Handbook Volume 4. Warrendale. PA: S.A.E.; 1985: 34.131-34.141. Thomas, P.: Bradford, M. Side impact regulations-how do they relate to real world accidents? Proc. 12th E.S.V. Conf., 9 19-929, 1989.
$6.00 + .oo 0001-4575/93 0 1993 Pergamon Press Ltd.
1993
RESTRAINED OCCUPANTS ON THE NONSTRUCK SIDE IN LATERAL COLLISIONS G. M. MACKAY, J. HILL, S. PARKIN, Accident Research Centre, The University of Birmingham,
and
J. A. R. MUNNS
Birmingham
Bl5 2TT, United Kingdom
(Receiwd 30 Sepfember 1991;in revise~.~$~ 25 January 1992)
Abstract-Ofinju~-pr~ucing collisions with high seat belt use, some 25% to 30% are lateral collisions. This paper describes some of the characteristics of those collisions as they relate to the front-seat occupant sitting on the side opposite to the impact. The data came from a stratified sample of in-depth crash investigations conducted in the Birmingham region in the period 1983 to 1989 involving current model cars. Crash severity was assessed using the Vehicle Deformation Index (VDI) of the Collision Deformation Classification (CDC) ratings and velocity change. Injury severity was assessed using AIS 85 for each body region. 193 cases of restrained occupants in nonstruck side collisions were examined. Of those occupants with head injuries of AIS I 2, 35% came out of the shoulder section of the seat belt. Of abdominal injuries of AIS 5 2,72% came from the seat belt itself. Interaction between front seat occupants was not a frequent cause of injury to the nonstruck side occupant. Some aspects of seat belt geometry might be changed so that the trajectory and loading of the nonstruck side occupant are improved.
INTRODUCTION
(Harms et al. 1987), and those studies emphasize the relative importance of head injuries for nonstruckside, restrained occupants. Experimental work by Herbert et al. ( 1976) and Horsch ( 1980) explored the limits of motion of the chest and hence the head in angled impacts, using dummies in sled tests. That work showed that the limit of retention of the torso occurred when the direction of the crash force was at 45 degrees from straight ahead. At greater angles, head excursion markedly increased, although Horsch noted that even at angles of around 60 degrees, significant energy was removed from the upper part of the body by the seat belt before the torso slipped out of the belt. This present study examined the injury causation aspects of nonstruck-side, restrained occupants in an effort to assess how seat belts function in a representative sample of real-world collisions. Lateral collisions are seldom like sled tests in the sense that the direction of principal force is often changing during the crash phase because of rotation of the occupant’s car. We wished therefore to assess how seat belts perform for a condition for which they are obviously not designed. It was our intention to describe the distribution of vehicle impact damage and the causes of injury, and to identify any seat belt problems. Cause of injury was to be described in terms of vehicle component struck, interaction with other car occupants, and loads applied through the seat belt system. Head contact evidence was to be used to es-
Lap/shoulder seat belts are designed to protect occupants in frontal collisions, primarily. The geometry of the anchorages and the placement of the belt across the chest, the iliac spines of the pelvis, and the lower abdomen, when combined with appropriate seat characteristics, limit forward excursion of the head, chest, and knees and apply loads through the seat belt webbing that are tolerable for most of the exposed population in all but the most extreme collisions. Side impacts, however, constitute a singificant part of the collision population at around 20°h to 30%, increasing to 36% for fatalities with higher levels of seat belt use (Mackay et al. 1990). In side impacts most attention has been focused on the struck-side occupant, and the recent rulemaking through Federal Motor Vehicle Safety Standard 2 14 addresses that condition. For that type of loading, the seat belt has relatively little influence on outcome, because the main loads come from the occupant’s striking the side header and the door, contacts normally not influenced by a seat belt. Head excursion out through the side window can be reduced by a seat belt, and of course ejection risk is diminished. For an occupant on the nonstruck side of a lateral collision, however, the seat belt is of greater consequence. Crash severities and injury patterns have been described in other projects in Canada (Dalmotas 1983), Germany (Otte et al. 1984) and Britain 147
G. M. MACXAYet al.
148
tablish if occupants remained fully restrained throughout the impact. Furthermore, seat belts were examined in anticipation of component malfunction under lateral impact conditions. MATERIAL
AND
METHODS
The Accident Research Centre at The University of Birmingham investigate 300 car accidents in the West Midlands of England every year. The Centre employ medical and enginee~ng specialists who work closely with local hospitals, coroners, the police, garages, and scrap yards. A major project has been in progress since 1983, contributing results to a national study with the object of evaluating vehicle crashworthiness and occupant injury. A stratified sample ofcar occupant collisions has been examined in the Birmingham region. The sampling plan includes all fatalities, 50% of the police-reported “serious” cases, i.e. requiring hospital admission for fractures and severe lacerations, and one-third of lesser levels of injury in current-model cars less than six years old. The cars are inspected at garages soon after their collisions, and data from that inspection are later coordinated with medical information from hospital records and questionnaire information from the people involved. The methodology is described elsewhere (Mackay et al. 1985). This paper presents an analysis of accidents from the West Midlands. Thus, 507 side impacts-27% of all collision types-were available for analysis. A precise definition of what a side impact is cannot be clear-cut. For example, occupants can move laterally to strike side structures when the front of the car is struck obliquely. We chose only those cases where the impact was on the side of the vehicle and where the direction of principal force on the case vehicle was between 1 to 5 o’clock on the right side and 7 to 1 1 o’clock on the left. Thus sideswipe collisions were not included. Crush was measured in each case, and a Collision Deformation Classification (CDC) (Society of Automotive Engineers 1985) was assigned to each vehicle. This is a seven-digit code commonly used to describe crash damage that has been caused by directly contacting another object. Only the part of CDC that indicates crush extent will be used here-the Vehicle Deformation Index (VDI). VDI takes an integer value between I and 9 to indicate the maximum extent of crush perpendicular to the vehicle’s struck side. Where possible, velocity change at impact (delta V) was calculated from crush measurements and the known weights of the vehicles involved, but for side
Tahlc I Selection ofnonstruck
side lateral impacts
.-
Total side impacts
Cases used
‘IhIof cases used ..-
Driver side impacts Passenger side impacts
281 220
62 131
22 60
Total
507
193
-
impacts the appropriate stiffness coefficients to be used are not well established for many of the cars in the sample. Injury severity codes according to the Abbreviated Injury Scale (AIS) (American Association for Automotive Medicine 1985) were assigned by body region. Occupant height, weight, sex, and age were known in most cases. Contacts within the vehicle were assessed for the main body regions. RESULTS The numbers of candidate side impacts are shown in Table 1. 193 cases where there was a restrained occupant on the nonstruck side were identified. In Britain, driving on the left side of the road in right-hand driver cars produces a preponderance of drivers alone in left-side, i.e. passenger-side, collisions. Principal
dirtwion
offorce
The direction of principal force in the collision was defined in each case. The cases analysed exclude side-swipe collisions. Table 2 shows that directions 2 and 3 o’clock, and 9 and 10 o’clock predominate, at 71%.
The amount of crush was assessed using the VDI protocol, and the frequency is illustrated in Table 3. Level 1 is the door skin width, level 2 is the width of the door window horizontal to the cant rail level (side header). Levels 3, 4, and 5 are equal zones to the centreline of the car, and then levels 6 to 8 repeat the process to the other side of the vehicle. Level 9 is crush all the way to the opposite side of the car, essentially breaking the car in half. with crush extending beyond the opposite cant rail. Table 2. Direction ofprincipal force on the clock direction ___~
No. of cases ‘%,
l/i 1
Z/IO
3/9
4/x
5/7 ---
NK
37 I9
68 35
69 36
I0
4 2
5 3
5
Restrained occupants in lateral coltisions
Table 3. Amount ofcrush according to VDI numbers code
-
No.of cases 0x1 -I-
I
2
3
4
5
6
11 5
43 22
70 36
32 16
13 7
6 5
Of the cases with known crush therefore, I.55 out of f80, or 87’%, had crush rhat was limited to VIX 4 or Iess, i.e. cases where the crush was less than twothirds of the way across the sitting zone on the opposite side from that of the occupant being considered. This is relevant to assessing the effect of partial restraint of the nonstruck-side occupant regardless of collision angle.
Velocity change values were calculated for 69% of the relevant eases. Within the limits of the known cases it was found that VDI was hnearIy related to delta V up to a value ofVDI6, at 54 km/h. Results are shown in Table 4. Within this genera1 condusion, however, there is a range of variation influenced mainly by the direction of force. With that reservation, the above data can be summarised by saying that the great majority of the cases were in the 2/3 or 9/ 10 o’clock direction with velocity change values less than 40 km/h, with crush less than two-thirds of the width of the opposite passenger-compartment sitting space (87%).
For 180 eases injury details and positive belt use were established. These are now considered. Dividing the cases between passenger- and driver-side impacts suggests, although the numbers are small, that there is benefit in having an occupant between you and the impact. In driver-side impacts, there was a struck-side occupant (the driver) present in every case. In passenger-side impacts there was a passenger present in only 32% of the cases. Table 5 ilhtstrates the injury outcome according to the Maximum AXS(MAIS),
For the nonstruck-side occupant, an identified contact in the vehicle for the head is the prime indicator of the occupant’s trajectory. Such contacts are given in Table 6.
I
~ 8
9
NK
Total
2 1
2 I
13 7
193 100%
--
Thirty-nine percent ofthe nonstruck-side occupants received an AIS 5 2 head injury, mainly from contacts with the opposite header, the roof, side-window glass, opposite door, or the other occupant. Each case was evaluated individually to assess if it was necessary for the occupant to come out of the seat belt in order to make the contact. Table 7 shows that that occurred in 35% of cases. There were 13 additional head injury cases of AIS 2 3. In 9 of the 13 cases the crush was at least to the centreline ofthe car (VDI 5+), and VDI 4 in the remaining 4 cases. With such amounts of crush it was not possible to assess the function ofthe seat belt, and it became irrelevant if the intrusion extends up to or beyond the centreline of the ear.
Of the 180 occupants, 42% received an injury to the chest. Only 5 received a chest injury of AIS z 3, (3%) mainly multiple rib fractures with haemo/ pneumothorax. The causes of the chest injuries are illustrated in Table 8. These data show that chest injuries are caused mainly by the seat belt. The incidence of chest contacts with far-side structures in driver-side impacts is Iow, because in those collisions the driver was present between the case occupant and the zone of impact. Few injuries to the case occupant were caused by the other (near-side) person, and thus the presence of that other occupant is beneficial in diminishing contacts with far-side structures and the resulting injuries,
Forty-one of the I80 occupants on the nonstruck side received an injury to the abdomen (23%). The Causes of those injuries are shown in Table 9; passenger- and driver-side cases are combined because no differences were apparent. The great majority (7 1%) of abdominal injuries came from the seat belt.
Table 4. Delta v and crush level I_-I--
-II -
149
VD! no.
Delta Y Km/hr
1
2
3
4
5
6
~$8
(26
<32
<40
<48
(56
^x1-.
7+ <62
I 50
G.
Table 5. Overall injury severity ofnonstruck-sideoccupant of car struck
by side
MAIS i
2 (‘W)
MAIS 2 3 (“LI)
52 (96)
MACKAY
M.
et al
Table 7. lncidenceofcomingout
h
Driver-side impacts Passenger-side impacts
I I3 (90)
2 (4) 13(10)
54 126
Total
165 (9’)
15 (8)
180
Remained in belt Driver-side impacts Passcngcr-side impacts Both sides
Seut hrlr prrjiwmanc’e
Table 6. Contacts
Impact
forAlS
i 2 head
out of belt
‘%Iout of belt
8 IO I8
47 29 35
9 24 33
DISCUSSION
In-depth investigations provide useful feedback to the designer of the performance of various subsystems in the car in circumstances that cannot be predicted in every respect in laboratory testing. Table 10 details some factors relating to seat belt performance. The incidence of damage to buckle casings illustrates that in lateral collisions the buckle is exposed to specific high-energy blows from the tunnel, the seat, or the central armrest structures. In none of those cases did the belt release, but in some instances there were difficulties in undoing the damaged buckle after the collision. Eight percent of occupants had put their belts on with a twist. That conceivably may result in a less favourable loading across the abdomen in particular. The roping (and jamming) of the webbing at the D ring is probably a consequence of the lateral direction in which the seat belt is being loaded, but it indicates the need for careful consideration of the dynamics of the D ring in collisions that are not frontal. The ultimate consequence may be to encourage the webbing to break prematurely. and there were two cases of that. Given the principle of a swivelling D ring, it is difficult to see how that can be avoided at the extremes of crash severity and lateral loading. In 3% of cases there was evidence that the belt was loose when loaded. This was probably due to poor adjustment of the belt combined with a retraction spring that was beginning to fatigue. No belts with comfort features were present in this sample. In no instances was the performance of the locking mechanism of the retractor judged to be unsatisfactory.
Far door
ofshoulderbelt injury cases
Side window glass
causing
This study has examined some of the characteristics ofthe restrained occupant on the nonstruck side in lateral collisions. The sample is biased towards the more serious injury cases, due to the overall sampling plan, so that it gives insights primarily into the limits of performance of the seat belt for this crash configuration. The typical collision investigated is a side impact from the 2 and 3 o’clock direction (or 9 and 10 o’clock) with a velocity change of the struck car of less than 40 km/h (25 mph) and with crush limited to less than two-thirds of the sitting space on the struck side. The pattern and depth of crush are major factors considered in determining an amount of energy required to crush a car, and hence in calculating delta V. Naturally. therefore, delta V and crush may be expected to have a positive relationship. It was found that VDI was linearly related to delta V, although there was a range of variation influenced mainly by the direction of force. For a given velocity change there is more crush if the impacts come from 10/l 1 o’clock or l/2 o’clock than if the collisions are more nearly perpendicular at 9 or 3 o’clock. This is usually because the angled collision is with the front corner of a striking car rather than being distributed across the entire front. Delta V is also dependent on the stiffness and mass of the case vehicle and the striking object. The case vehicle will experience a higher speed change after impact with an object of increased stiffness or mass, all other factors being equal. For these collisions there is obvious asymmetry in that a driver is present in every case. but a front-
head injury AIS i
Z for non struck-side
Facia
Seat
A pillar
Steering wheel
occupants
Other occupant
Header roof. etc.
‘2
Driver side Passenger side
0 5
? 1;
0 2
, ;
0 2
? ;
4 3
9 20
I9 51
Both (“/I)*
5 (3)
I9 (IO)
Z(l)
3 (2)
J(l)
3 (1)
7 (4)
29(16)
70 (39)
*Percentages
indicate
the proportion
of all I80 occupants
so injured
Restrained
occupants
in lateral collisions
151
Table 8. Chest injury causes Other occupant
Far-side structure
Impact Driver side Passenger side Both sides (%)* *Percentages
0 II
Seat belt
2 Z(3)
I6 28 44 (24)
11(6) indicate
the proportion
of all 180 occupants
Other components 2 4 6 (3)
Seat belt
Seat
29 16%
2 I%
Note: Percentages so injured.
5 3% indicate
Induced
2 I% the proportion
3 2% ofall
27 48 75 (42)
IO. Seat belt function n
to the abdomen
Other occupant
7
cupant between the impact and the nonstruck side occupant appears to be beneficial in limiting farside contacts. Chest and abdominal injury occurred in 42% and 23% of cases, respectively. Because the seat belt is providing the main load paths for the crash forces into the occupant on the nonstruck side, it is inevitable that the chest and abdomen will be injured from the seat belt. The great majority (7 1%) of abdominal injuries came from the seat belt. With lateral flexion of the upper part of the body occurring, in 35% of cases the torso was coming out of the shoulder belt, it is highly likely that the orientation of the pelvis is changing. That results in the inboard section of the seat belt loading the lateral aspect of the abdomen directly. This is reflected in the severe abdominal injury cases in which there were ruptures of the spleen or tears to the liver. One case occurred with a lateral flexion fracture of the lumbar spine, with neurological involvement. There were, however, only five cases of abdominal injuries of AIS L 3. All of the cars in this study had two bucket seats in the front, there were no bench seat designs. Thus most of the seat belt buckles were located relatively low, near to the hip joint, often on rigid brackets attached to the seat. Even with such geometry, the relatively high incidence of abdominal injury (23%) suggests that the belt is moving from the pelvis into the abdomen, although the shoulder section of the seat belt is also loading at least the upper quadrant of the abdomen as well as the chest. In a number of cases, we were able to verify this from the position and nature of bruises. In some cars, the buckle is relatively highly positioned and thus the
Table
Intruding structures
N
so injured
seat passenger is present in only 32% of the cases. Thus, for U.K. data, left-side impacts predominate; the opposite occurs in the United States. Table 5 shows that 8% of nonstruck-side occupants scored MAIS L 3. Thomas and Bradford ( 1989) used data from the same national study from which the present sample has been taken and found that 243 out of 1,6 18 (15%) struck-side occupants scored MAIS >_ 3. Their analysis included unrestrained occupants, so we might expect a higher proportion of serious injury. Nevertheless, with nearly half the number of MAIS L 3 cases observed in the present nonstruck-side sample, this comparison does suggest that nonstruck-side occupants are less likely to be seriously injured. In one-third of cases, the occupant came out of the shoulder section of the seat belt. This was established from the positions of the head contacts within the car. As well as the direction of the impact a number of other factors have a bearing on this event-the position of the upper anchorage, the size of the occupant, the seat position, the adjustment of the upper anchorage, and the looseness of the seat belt. In the marginal condition, the geometry of the upper anchorage will influence the retention of the shoulder, but for pure lateral impacts the torso will translate out of the belt anyway. Further work will examine differences between two- and four-door cars and adjustable versus fixed upper anchorages, but that requires substantially more cases. Experimental work might attempt to alleviate the problem with improved seat belt geometry and webbing pretensioning. This study illustrates, however, the exposure of the head particularly, as a result of translating across the passenger compartment. The presence of an oc-
Table 9. Causes of injuries
NK
N 41 23%
180 occupants
%
Satisfactory Buckle casmg fractured I80 degrees of twist in webbing Webbing roped at D ring Belt loose when loaded Webbing broken
140 16 I5 II 6 2
74 8 8 6 3 I
No. of belts examined
190
100
152
G.
M. MACKAY
buckle itself is loading the abdomen. The properties of the seat base should not be overlooked; in some cases, they may enable submarining to occur. Experimental work could be done to examine the relative risks of lateral abdominal loading from various seat belt geometries and with modified seat base characteristics. This would be of interest now that a potential means of assessing abdominal loads is available with the foam insert for Hybrid III developed by Rouhana et al. ( 1989). Not mentioned in this analysis are injuries to other body regions. This is because injuries to the upper and lower limbs were infrequent and almost always minor. There was one fatal neck injury to a 7% year-old with advanced arthritis of the cervical vertebrae. Otherwise, the injuries of main consequence were confined to the head, chest, and abdomen. This field study also points to some aspects of seat belt function in lateral collisions where the restraint is loaded somewhat differently than in the main design condition of the frontal impact. Buckle performance and D ring operation particularly are of concern in some instances. In conclusion, the typical nonstruck-side occupant is injured in conditions that are broadly similar to side-impact crash tests that are currently proposed. However, 13 cases of serious head injury occurred with considerable passenger-compartment intrusion. In these cases, the seat belt performance was considered irrelevant. The kinematics of the nonstuck-side occupant are very different from those of the struckside occupant. While the nonstick-side occupant is at lower risk, serious injuries did occur, and these were mitigated when a struck-side occupant was present. Future work on seats and seat belt design for lateral loading conditions might well reduce the likelihood of torso egress from the belt, limit abdominal injury, and enhance the functional integrity of seat belt components. ncknowk~~~~~~rnc~~ts_Theauthors wish to thank the sponsors of this research, the Transport and Road Research Laboratory of the U. K. Department of Transport, the Ford Motor Company, the Rover Group, and Nissan (U. K.) Ltd. We are also grateful to all
et al
members of the Cooperative Crash Injury Project who have contributed to the data collection process, and the many police officers, hospital staff, garage owners, and others who form an invisible laboratory for our research.
REFERENCES American Association for Automotive Medicine (A.A.A.M.). The Abbreviated Injury Scale. 1985 Revision. Arlington Heights, IL: A.A.A.M.; 1985. Dalmotas, D. J. Injury mechanisms to occupants restrained by three-point belts in side impacts. S.A.E. Paper 830462. Warrendale, PA: Society of Automotive Engineers (S.A.E.); 1983. Harms, P. L.; Renouf, M.; Thomas, P. D.; Bradford, M. Injuries to restrained car occupants, what are the outstanding problems? Proceedings of the 1lth E.S.V. ConferWashington, ence. DC: U.S. Department of Transportation, National Highway Traffic Safety Administration; 1987: 183-201. Herbert, D. C.; Stott, J. D.; Corben, C. W.; Cutting, D.; Gillies, N. Occupant head space in passenger cars. Sydney, N.S.W., Australia: Traffic Accident Research Unit; 1976. Horsch. J. D. Occupant dynamics as a function of impact angle and belt restraint. S.A.E. p-88. Proceedings of the 24th Stapp Car Crash Conference. Warrendale, PA: S.A.E.; 1980: 4 17-438. Mackay, G. M.; Ashton, S. J.; Galer, M. D.; Thomas. P. D. The methodology of in-depth car crashes in Britain. S.A.E., P-159. Warrendale, PA: S.A.E.; 1985: 365-390. Mackay, G. M.; Cheng, L.; Smith, M.; Parkin, S. Restrained front seat occupant fatalities. Proceedings ofthe 34th A.A.A.M. Conference, Scottsdale, AZ, Oct. 1-3, 1990; 139-162. Otte, D.; Suren, E. G.; Appel, H.; Nehmzow, J. Vehicle parts causing injuries to front seat car occupants in lateral impacts. S.A.E., P-152. Proceedings of the 28th Stapp Car Crash Conference Warrendale, PA: S.A.E.; 1984: 13-24. Rouhana, S. W.; Viano, D. C.; Jedrzejczak, E. A.: McCleary, J. D. Assessing submarining and abdominal injury risk in the Hybrid III family ofdummies. S.A.E., P227. Proceedings of the 33rd Stapp Car Crash Conference. Warrendale, PA: S.A.E. 1989: 257-280. Society of Automotive Engineers. Collision Deformation Classification. Handbook Volume 4. Warrendale. PA: S.A.E.; 1985: 34.131-34.141. Thomas, P.: Bradford, M. Side impact regulations-how do they relate to real world accidents? Proc. 12th E.S.V. Conf., 9 19-929, 1989.