Thoracic Injury Risk as a Function of Crash Severity – Car-to-car Side Impact Tests with WorldSID Compared to Real-life Crashes

Abstract

Side airbags reduce the risk of fatal injury by approximately 30%. Due to limited real-life data the risk reducing effect for serious injury has not yet been established. Since side airbags are mainly designed and validated for crash severities used in available test procedures little is known regarding the protective effect when severity increases.

The objective of this study was to understand for which crash severities AIS3+ thorax occupant protection in car-to-car nearside collisions need to and can be improved. The aim was fulfilled by means of real life data, for older cars without side airbag, and a series of car-to-car tests performed with the WorldSID 50%-ile in modern and older cars at different impact speeds.

The real life data showed that the risk of AIS3+ injury was highest for the thorax followed by the pelvis and head. For both non-senior and senior occupants, most thorax injuries were sustained at lateral delta-v from 20 km/h to 40 km/h. In this severity range, senior occupants were found to have approximately four times higher risk of thoracic injury than non-senior occupants. The crash tests at lateral impact speed 55 km/h (delta-v 32 km/h) confirmed the improved performance at severities represented in current legal and rating tests. The structural integrity of the modern car impacted at 70 km/h showed a potential for improved side impact protection by interior countermeasures.

INTRODUCTION

Since the mid 1990s, legal and rating test procedures in Europe and the United States have encouraged car manufacturers to improve occupant protection in side impacts. Four test procedures evaluate near side occupant protection in a vehicle-to-vehicle like event. In Europe a 950 kg moving deformable barrier impacts the side of the vehicle at 90° at 50 km/h (EuroNCAP and 96/27/EC). The occupant injury risk is evaluated with the antrophomorphic device (ATD) EuroSID-2 (ES-2). In the US, a 1367 kg moving deformable barrier crabbed at an angle of 27° strikes the driver side of the target vehicle in 54 km/h (FMVSS 214) and 62 km/h (USNCAP) respectively. Between 1998 and 2009 the injury risk to the front seat near-side occupant has been evaluated using USSID. From 2010 the US load cases will evaluate occupant protection with the EuroSID-2re (ES-2re). IIHS performs a side impact rating test using a barrier with high bumper in a 90° lateral impact at 50 km/h. Occupant protection is evaluated with a small female dummy, SID2s. With the current legal and rating test procedures side impact protection is evaluated for lateral impact speeds between 45–55 km/h, resulting in a lateral velocity change (delta-v) in the range of 20–35 km/h for the target vehicle. Different barrier masses, stiffness and design result in slightly different crash severities, which needs to be considered in the design of occupant side impact protection.

The introduction of the legal and rating tests described above have resulted in improved occupant side impact protection (Welsh et al. 2007, Kahane 2007). Moreover, according to Braver and Kyrychenko (2004), McCartt and Kyryschenko (2007) and Kahane (2007) side airbags do reduce the risk of fatal injury by approximately 30%. Due to limited real-life data the side airbag risk reducing effect for serious injury (AIS3+) has not yet been established.

The elderly population of today is mobile and has a high risk awareness (Braitman and McCartt 2008). However, degenerating cognitive as well as biomechanical abilities do put senior occupants at a higher risk for crash involvement and injury outcome. Several studies have shown that senior occupants are overrepresented in side car-to-car crashes (Augenstein 2005, Kent 2005). It has also been shown that due to fragility and frailty a less severe crash can have a more serious outcome (Kent 2005, Sunnevang et al. 2009). Because side impact dummies represent 45 year old occupants (a 50%-ile mid-size male, and a 5%- small female), there is a challenge to evaluate senior occupant safety (Yoganandan 2007, Kent 2005).

According to both ISO and NHTSA biofidelity ranking, the WorldSID 50%-ile mid-size male biofidelity is improved compared to the ES-2re. The WorldSID scored an overall 7.1 (out of max 10.0) compared to the ES-2re which scored 4.2 on the ISO biofidelity scale (ISO TR9790). Comparative studies have shown that for thoracic loading the WorldSID dummy is closer to PMHS responses than other side impact dummies (Yoganandan et al. 2005). To evaluate risk of injury with the WorldSID 50 %-ile a set of risk curves have been derived using available biomechanical data from previous tests using post mortem human subjects (PMHS) correlated to dummy tests in similar setups (Petitjean et al. 2009). The set of risk curves include AIS3+ and AIS2+ (where insufficient data for AIS3+ was available to the PMHS) risks to shoulder, thorax, abdomen and pelvis. The risk curves were derived using a number of statistical methods, and scaled to represent a 45-year-old mid-sized male.

In car-to-car near side crashes with fatal and serious injury outcomes, chest injuries are most prevalent, both for vehicles manufactured pre and post regulatory introduction (Håland et al. 1993, Farmer et al. 1997, Welsh et al. 2007, Yoganandan et al. 2007a). Rib cage, lungs and arteries have the highest injury frequency (Thomas and Frampton 1999). In tests where post mortem human subjects (PMHS) were exposed to lateral loads, chest deflection was found to be a good predictor of injury (Kuppa et al. 2003). Chest deflection is correlated to skeletal thorax injury, but the injury measurement was to some extent also correlated to soft tissue injury. For soft tissue the viscous criteria is presented as a good complement as injury predictor (Viano 1989).

Studies have also shown that crashes with serious or fatal injury outcome occur in severities higher than those currently evaluated in legal and rating tests (Thomas and Frampton 1999, Frampton and Thomas 2003, Arbelaez et al. 2005, Sunnevang et al. 2009). Since side airbags are mainly designed and validated for crash severities used in available test procedures little is known regarding the protective effect when severity increases.

The objective of this study was to understand for which crash severities, in terms of lateral impact speed and lateral delta-v, AIS3+ thorax non-senior and senior occupant protection in car-to-car nearside collisions need to be and can be improved. The aim was fulfilled by real life data, for older cars without side airbag, and a series of car-to-car tests performed with the WorldSID 50%-ile in modern and old cars at different impact speeds.

METHODS

Real life data

The National Automotive Sampling System – Crashworthiness Data System (NASS/CDS) contains a selection of crashes from twenty-seven primary sampling units throughout the US. To be included in the database, the crash must be police reported involving at least one vehicle towed away from the scene. The in-depth investigation contains information of the vehicle, occupant, and crash characteristics. Injuries are recorded and registered according to the Abbreviated Injury Scale (AIS; AAAM 1990). Approximately 5000 crashes are investigated annually. With established weight factors, the NASS/CDS data can be made nationally representative for the U.S.

To associate the laboratory scenario for legal and rating tests to real life data, NASS/CDS data from 1994–2008 was queried for crashes including front seat occupants in near-side (primary event) car-to-car crashes. A near-side crash was defined as an impact with the greatest damage to the right or left side (CDC general area of deformation, GAD, = R or L) of the target vehicle, with principal direction of force (PDOF) 2–4 o’clock for the right side passenger and 8–10 o’clock for the left side driver. Crashes involving rollover and occupant ejection were excluded from the sample. Occupants were belted and unbelted. Only cars without side airbags were considered due to limited numbers of cases (<10% of total sample) with side airbag equipped vehicles available. Lateral impact speed was calculated from lateral delta-v (taken directly from NASS/CDS) and the masses of the cars involved, according to equation 1, which excluded crashes where this information was missing. To have a real life dataset with crashes remniscent of the car-to-car tests, inclusion critera such as specific horizontal location in primary collision in the passenger compartment (CDC specific horizontal location = Y, P or D) were added, and that the bullet vehicle had a frontal impact (general area of deformation = F)

$$V_{\text{impact}}=\left(1+m_{\text{target}}/m_{\text{bullet}}\right)\times \left|\Delta v_{\text{target}}\right|$$ (1)

Equation 1 – Calculation of impact speed from NASS/CDS reported delta-v.

The sample was stratified with respect to age into a non-senior group of occupants, 10–59 years old, and a senior group including 60-year-olds and above. The NASS/CDS data was weighted to become nationally representative. The content of the raw file and the weighted data is presented in Table 1. All analyses are based on weighted data only.

Table 1.

Nass/CDS data from 1994–2008

	Number of cases, raw file	Number of cases, weighted data
Non-Senior occupants
Exposed	972	333 403
AIS3+ thorax	247	10 401
Senior occupants
Exposed	250	57 816
AIS3+ thorax	75	7235

The empirical risk of sustaining at least one thoracic AIS3+ injury was determined by dividing the number of injured occupants with the number of exposed occupants. Risk curves for thoracic AIS3+ injury was derived by logistic regression using SAS 9 software. Risk curves were constructed for both non-senior and senior occupants. Note that these risks give the probability of injury given that one of the vehicles involved in the crash is towed away. The weight factors of the NASS database are nearly exponentially distributed, implying that some cases with large weight factors may have a very high influence on the risk curves. To investigate the presence of highly influential cases a jackknife approach was taken. The cases were then removed from the sample one at a time and new risk curves derived. One particular case was then found to have a very high influence on the risk curve for non-seniors (this case had a very high CI displacement C, see Pregibon (1980)). This was an uninjured 20-year-old male exposed to a lateral delta-v of 57 km/h. A closer inspection of this case revealed that the delta-v’s for the target and impacting cars were not consistent with the principle of momentum conservation (delta-v_bullet=30 km/h, delta-v_target=57 km/h, mass_bullet=1370 kg, mass_target=1470 kg). This particular case was therefore removed from the analysis.

Evaluating the real life data in terms of exposure incidence and risk of AIS3+ thoracic injury as a function of lateral impact speed and lateral delta-v, provides a context for the car-to-car tests performed in terms of severity and occupant protection.

Car-to-car tests

Three car-to-car tests, using modern cars from 2008, were performed at three impact speeds; 55 km/h (USNCAP representative), 70 km/h and 80 km/h. In each test, the bullet vehicle (test weight 1735 kg) impacted the left side of the stationary target vehicle (test weight 1863) at 90°. The impact point was set by aligning the bullet vehicle centerline to the mid of target vehicle wheelbase. The vehicle tested was a modern, mid-size passenger car. Bullet and target vehicles were the same car models with similar ride heights. To compare occupant thorax protection in a modern passenger car to the previous situation, with older cars without side airbags as in the real life sample, an additional car-to-car test (older car-to-car) was performed. This test was performed at 55 km/h in the same setup as for the modern car, but with mid-size sedan vehicles from 1992 as bullet and target vehicles. In these tests the vehicle test weight was 1700 kg for both target and bullet vehicle.

In all four crash tests, a WorldSID 50%-ile was seated in the left front seat. The WorldSID was instrumented in head, thorax, abdomen and pelvis. For shoulder, thorax and abdomen deflection measurement, the standard one-dimensional infrared deflection transducers (IR-Tracc) were replaced by 2D versions (2D IR-Tracc), which include the ability to measure possible forward displacement of the shoulder and the five ribs (three thoracic and two abdominal).

In the tests with modern cars, a seat-mounted airbag (SAB) and inflatable curtain (IC) provided protection for head, thorax, abdomen, and pelvis. In all three tests the WorldSID was restrained by a 3p-belt and belt pretensioning was activated at the same time as airbag deployment. For the test with the older vehicle model, the WorldSID was restrained with a 3p-belt without pretensioning.

All target vehicles were equipped with a 3-axial accelerometer placed at vehicle center of gravity. Left front door and left B-pillar were equipped with 4+4 uni-axial accelerometers in order to derive the velocity of the intruding structure. Dummy and vehicle data was filtered according to SAE J211. Dummy deflection measurements were translated to AIS3+ thoracic skeletal injury risk according to proposed risk curves for chest deflection measured by the IR-Tracc (Petitjean et al., 2009). The risk curves used for the translation were derived using logistic regression and the survival method. All five ribs (three thoracic and two abdominal) were treated as thoracic ribs, due to the construction of the injury risk curves. The maximum deflection for each test was translated to AIS3+ thoracic skeletal injury risk. For the risk of soft tissue AIS3+ injuries, the risk curve for abdomen VC was used. (thoracic VC was related to skeletal injuries by Petitjean et al., 2009).

To evaluate senior occupant protection the risk corresponding to the measured dummy injury values would need to be evauated with a different age scaling. This information is not yet available for the WorldSID dummy. To add a senior occupant perspective to the test results, the risk derived from the dummy injury values were scaled using the relative risk between non-senior and senior occupants from the NASS/CDS sample in the severity range (e.g. lateral impact speed and delta-v) in which the car-to-car tests with modern cars were performed.

RESULTS

Real life data

The AIS3+ injury risks in the NASS/CDS data showed that thorax, followed by the pelvis, is the primarily injured body region for both non-senior and senior occupants, see figure 1. Injury risks for the weighted NASS/CDS data are shown in figure 1. For both non-senior and senior occupants, AIS3+ thorax injuries were primarily skeletal (non-seniors 41%, seniors 66%) followed by injuries to the lungs (non-seniors 30%, seniors 26%).

Figure 1.

Risk of AIS3+ injuries from real-life data.

Exposure to crashes and incidence of AIS3+ thorax injuries as functions of lateral impact speed and lateral delta-v, for the two age groups, were found to be similar. For the incidence, lateral median impact speed for non-seniors was 66 km/h and for senior occupants 67 km/h. Median lateral delta-v was found to be 32 km/h and 30 km/h respectively. For both age groups, in cars without side airbag, the incidence of AIS3+ thoracic injury is highest in the impact speed interval between 40–80 km/h and delta-v interval of 20–40 km/h. A histogram over the incidence and the derived risk of thoracic AIS3+ injury with respect to lateral impact speed is presented in figure 2a for non-seniors and figure 3a for seniors. Corresponding graphs with respect to lateral delta-v is presented in Figure 2b and 3b.

Figure 2.

Non-senior occupants: Incidence (histogram) and risk of AIS3+ thoracic injuries as functions of a) lateral impact speed and b) lateral delta-v.

Figure 3.

Senior occupants: Incidence (histogram) and risk of AIS3+ thoracic injuries as functions of (a) lateral impact speed and (b) lateral delta-v.

At a lateral impact speed of 55 km/h the risk of AIS3+ thorax injury for a non-senior occupant is 6% (95% confidence interval: 4–9%). For a senior occupant the corresponding risk is 28% (95% CI: 19–39%). For an impact speed of 70 km/h the thoracic AIS3+ injury risk increases to 19% (95% CI: 12–28%) and 66% (95% CI: 48–81%) for non-senior and senior occupants respectively. In the impact speed interval between 40 km/h to 80 km/h, and lateral delta-v 20 km/h to 40 km/h, the senior occupant AIS3+ thoracic injury risk is, on average, approximately four times higher than for non-senior occupants (R_senior = R_non-senior × 4).

Car-to-car tests at rating severity (55 km/h)

At a lateral impact speed of 55 km/h, the test with the older cars resulted in a delta-v of 32 km/h for the target vehicle. WorldSID peak rib deflection was 28 mm, which corresponds to a 0.9–1.4% AIS3+ skeletal thoracic injury risk. The lower level is the WorldSID AIS3+ thoracic skeletal injury risk derived using logistic regression and the higher when using the risk curve derived using the survival method (Petitjean et al., 2009). Door velocity at time for peak chest deflection was 10 m/s, and maximum residual intrusion (at front door occupant thorax area) was measured after crash to 410 mm. For the same lateral impact speed (55 km/h) using modern cars, the target delta-v was measured to 32 km/h. WorldSID peak deflection was 15 mm, which corresponds to a 0.2% risk (for both risk curve calculations) of AIS3+ skeletal thoracic injury. Door velocity at the time of peak rib deflection was 8 m/s, and maximum residual intrusion was measured after crash to 250 mm.

A comparison of door velocity and peak deflection response is presented in figure 4.

Figure 4.

Modern versus older car: Door velocity, lateral delta-v and chest deflection (for rib with peak deflection), as a function of time, for the tests at 55 km/h.

Car-to-car tests with modern cars at different impact speeds

In the tests with the modern cars, lateral delta-v in the three tests was measured to 32 km/h, 39 km/h, and 43 km/h. At impact speed 55 km/h, door velocities (measured at window line, thorax and pelvis) at time for peak rib deflection ranged from 7–9 m/s. At impact speed 70 km/h the door velocities were increased to 9–13 m/s. At impact speed 80 km/h, all accelerometers placed in the side structure failed 30 ms after impact. Maximum residual intrusion on the impacted side structure was measured after crash to 250 mm, 370 mm, and 430 mm respectively for the three tests.

In all tests the side airbag and inflatable curtain deployed and positioned as intended. Dummy measurement values to head (acceleration), thorax (deflection, VC), abdomen (deflection, VC) and pelvis (force) were increased as the impact speed was increased. For injury measurements other than chest deflection see table 2 in appendix.

WorldSID rib deflection increased with crash severity, resulting in peak thoracic rib deflections of 15 mm, 35 mm, and 43 mm, which correspond to 0.2%, 3.3–5.6%, and 10.1–13.9% risk of AIS3+ thoracic injury. Door velocity and rib deflection (for rib with highest peak) versus time for all three tests is presented in figure 5.

Figure 5.

Modern cars: Door velocity and chest deflection (rib with peak deflection), as a function of time, for the tests at 55, 70 and 80 km/h. The accelerometer measuring door velocity at impact speed 80 km/h failed at 30 ms.

Rib deflections, versus lateral impact speed and delta-v, for the three modern car-to-car tests are presented in figure 6, and the corresponding AIS3+ injury risk in figure 7.

Figure 6.

Modern Cars: Chest deflection (IR-Tracc lateral measurement for all five ribs) versus impact speed (a) and delta-v (b).

Figure 7.

Modern Cars: Peak deflection translated to AIS3+ thoracic injury risk versus impact speed (a) and delta-v (b). From risk curves derived using logistic regression and survival method (Petitjean et al., 2009).

DISCUSSION

From the real life data for older cars without side airbag in old traffic (i.e., 1994–2008), the largest portion of AIS3+ occupant injuries were thoracic injuries, see Figure 1. Furthermore, the incidence of thoracic AIS3+ injuries, for both non-senior and senior occupants, was highest at lateral delta-v ranging from 20–40 km/h. The median lateral delta-v, representing 50% of all AIS3+ injuries, were found to be 32 km/h and 30 km/h for non-senior and senior occupants, respectively. The risk for AIS3+ thorax injuries was approximately four times higher for senior than non-senior occupants in the delta-v interval 20–40 km/h.

The risk of thoracic AIS3+ injury for non-senior occupants in older cars without side airbags was 2% to 6% for impact speeds in the interval of 40 to 55 km/h, which is low. Corresponding risks for delta-v in the range of 20–35 km/h was 1% to 14%. Many crashes do, however, occur in this severity range, and are addressed by current legal and rating procedures. For higher crash severities the AIS3+ thoracic injury risk increases significantly. Higher crash severity was addressed with the modern car-to-car tests performed in this study.

The three car-to-car tests with modern vehicles were performed at impact speeds ranging from those similar to USNCAP to significantly higher. The test at 55 km/h is close to the USNCAP lateral impact speed although the bullet weight in this test was higher than the weight of the barrier in the USNCAP setup. Injury values measured by the dummy were very low resulting in less than 0.5% risk of thoracic skeletal injury. For head, abdomen and pelvis the AIS3+ injury risks were also below 0.5%. This test supports previous research findings that thoracic AIS3+ injuries for non-senior occupants are rare at this severity level (e.g. risk of AIS3+ injury is low at this impact speed and delta-v). For non-senior occupants at the high end of the severity scale, the test at 80 km/h resulted in a 10–15% risk of AIS3+ thoracic injury, and the dummy was exposed to increased loading to the head and pelvis. Taking into consideration that the structural measurements in the vehicle failed at 30 ms, this severity is challenging for the side airbag and vehicle structure. Increasing the impact speed from 55 km/h to 80 km/h is to more than double the impact energy. In between these two tests, the test at 70 km/h resulted in a 3.5–5.5% AIS3+ thorax injury risk as well as increased injury risk to other body regions compared to the test at 55 km/h. B-pillar velocities and residual intrusion was also higher than for the test at 55 km/h.

Comparing the older and modern vehicle impacted at 55 km/h (modern-to-modern and older-to-older car): the improved side structure and side airbag in the test with modern cars reduced the risk of thoracic AIS3+ injury by approximately 80% compared to the test with old cars (as a result of a 50% reduction of maximum chest deflection). The AIS3+ thoracic injury risk was reduced from 1.2% to 0.2% and AIS3+ pelvic risk was reduced from 3.7% to 0.5%. The residual intrusion was also decreased from 410 mm to 250 mm although door velocity at the time of occupant loading was similar. Assuming a car fleet with only modern side airbag equipped vehicles, and applying a similar risk reduction as observed in the two tests performed in 55 km/h, the high incidence of thoracic AIS3+ injuries in figure 2 and 3 (older cars without side airbags) would be significantly reduced in crash severities addressed by current legal and rating procedures.

However, the high incidence of AIS3+ thorax injuries in severities above those represented in current rating procedures still remains, although the tests with modern cars in 70 km/h and 80 km/h indicated a reduced risk compared to the real life risk at similar severity levels, see figure 7 compared to fig 2. Injury measurements to head, thorax (abdomen included) and pelvis increased when impact speed and delta-v increased for the tests performed with the modern, side airbag equipped, vehicles. It can therefore be assumed, that although improvements made to modern cars may also have beneficial effects in higher severities, a high incidence of AIS3+ and fatal injuries at higher delta-v may remain. Assuming that the incidence of AIS3+ thorax injuries in the severity levels represented in current rating tests (lateral impact speed 40–55 km/h, and lateral delta-v 20–35 km/h) are reduced, the median impact speed for both non-senior and senior occupants would be even higher considering a vehicle fleet with side impact protection similar to the modern vehicle tested (Sunnevang et al. 2009).

This study evaluates thoracic AIS3+ injury risk as a function of crash severity in terms of lateral impact speed and delta-v. Both metrics are important measures of crash severity in side impacts (Tingvall et al., 2003). Target and bullet vehicle height, weight, stiffness and impact speed determines the crash severity transferred from the bullet to the target vehicle. Side impact structure and protection countermeasures determine load transfer experienced by the near-side occupant in the target vehicle. In the comparison between the older cars and the modern cars, impact speeds were the same. Measured delta-v in both test were also found to be the same. Door velocity profile and the residual intrusion however, differ considerably. For the modern vehicle the maximum door velocity is significantly reduced, and the side structure and floor pan reaches a common velocity earlier than for the older car. The softer structure of the older car allows a high door velocity and intrusion for a longer period of time. The difference in structural behavior is important for future understanding and analysis of near-side occupant protection. For cars with a softer side structure, impact speed is most critical in terms of crash severity. For modern vehicles with a more stiff side structure, delta-v is becoming more important as a crash severity parameter.

Taking into account that senior occupants have approximately four times higher risk of thorax injury, the fragility and frailty of senior occupants should be evaluated and considered when future safety systems are developed that address the upper range of impact speeds.

For impact speeds above those represented in current legal and rating physical properties, and practical use of the car, limit how far the AIS3+ injuries can be further reduced. It should be noted that improvements in the traffic environment, for example, can contribute to reducing the occurrence of high severity impacts. One example is to reduce the possibility of vehicles approaching each other at high speeds in intersections.

The lateral impact speed for the real life data was calculated using the vehicle masses and lateral delta-v listed in the NASS/CDS database. The delta-v for the crashes was calculated using the residual intrusion on the target vehicle. Equation 1 assumes an inelastic crash which is not always accurate.

CONCLUSION

Considering car-to-car near-side collisions and occupants without side airbag protection, the real life data showed that the risk of AIS3+ injury was highest for the thorax followed by the pelvis and head. For both non-senior and senior occupants, most AIS3+ thorax injuries were sustained at crash severities ranging from 20 km/h to 40 km/h in lateral delta-v. In this interval, senior occupants were found to have approximately four times higher risk of AIS3+ thoracic injury than non-senior occupants.

Current legal and rating tests include impact speeds and lateral delta-v in the range of 45–55 km/h and 20–35 km/h respectively. Within this range, car-to-car tests were performed with modern cars equipped with side airbags and older cars without. The AIS3+ thorax injury risk and also door intrusion speed were found to be considerably lower in the modern target-car.

Car-to-car tests were also performed with modern cars at crash severities higher than expected in current legal and rating tests. The structural integrity of the impacted car showed a potential for improved side impact protection by interior countermeasures. Consideration should then be made to senior occupants while not jeopardizing the current protective performance. The WorldSID used was able to produce what seems to be valid measurement data also at the higher impacts speeds.

APPENDIX

Table 2.

Dummy measurements to all body regions with the translated AIS risk. The pubic force cell was damaged in the high speed tests (70 and 80 km/h) and the listed value is scaled using the force measurement to left side iliac wing.