EUROPEAN COMMISSION DG RTD

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public EUROPEAN COMMISSION DG RTD SEVENTH FRAMEWORK PROGRAMME THEME 7 TRANSPORT - SST SST.2007.4.1.2: Human physical and behavioural components GA No. 218516 THORAX Thoracic injury assessment for improved vehicle safety Deliverable No. THORAX D1.1 Deliverable Title A comparison between crash test results and real-world accident outcomes in terms of injury mechanisms and occupant characteristics. Dissemination level Public Written By Carroll J A and Smith S (TRL), Adolph T and Eggers A (BASt) 11 November 2009 Checked by Hynd D (TRL) 18 December 2009 Approved by Lemmen P (FTSS) Date Issue date Date Page 1

Executive summary The attainment of an ever-higher safety level on European roads is a long-term and ongoing process, which has seen significant progress already. Improvements in occupant protection (passive safety) have contributed to this progress. For example, the combination of EU legislation for crash test standards and improved consumer information through the Euro NCAP has substantially raised the survivability for vehicle occupants in a crash. However, around 41,600 people were killed and more than 1.7 million injured in European road accidents in 2005. Therefore, whilst the number of road fatalities has declined by more than 17 percent since 2001, greater efforts will have to be made if the European Commission s target of halving the number of deaths on the roads by 2010 is to be met. Although efforts are needed on all levels of road safety, the COVER (Coordination of Vehicle and Road Safety Initiatives) project has been set-up to develop a harmonised and consistent direction of research and to accelerate the implementation of research findings of four complimentary initiatives in the field of crash biomechanics (including the Thorax and THOMO projects). From a review of in-depth accident data, from around Europe (the UK, Germany, and France), the COVER Project Task 1.2 has already provided an overview of the current situation with regard to thoracic injuries resulting from frontal impact car accidents. Of the body regions in the accidents analysed, the thorax was the most frequently injured region for all killed and seriously injured occupants in frontal impact accidents. With this knowledge it is reasonable to investigate why thorax injury hasn t been reduced as much as may have been expected. Motivated by findings of previous projects (including EC Framework projects) the Thorax and THOMO projects were set-up to study thoracic injuries for a wide variety of car occupants and transfer results into test and design tools. Overall differences between accidents and crash test results have been studied in great detail by the SARAC II EC Project (e.g. Newstead et al., 2006). It highlighted the need for further investigation of the relationship between the front impact component of the Euro NCAP test score and real world injury outcomes in front impact crashes (Delaney et al., 2006). This task of the Thorax Project was devised to follow on from both the initial accident analysis of COVER Task 1.2 and the statistical analyses carried out with SARAC II. Individual frontal impact accident cases from the CCIS and GIDAS in-depth accident studies, as used in the COVER work, were selected for further in-depth analysis. These cases were chosen such that the impact conditions were close to those used in the Euro NCAP frontal impact test and where Euro NCAP has tested the vehicle being investigated. For each case, a comparison was made between the thoracic injury outcome for the occupants predicted from the Euro NCAP crash test of that vehicle and the real-world accident. Thirty-four cases where the impact conditions were similar to the Euro NCAP frontal impact crash test were reviewed. Other impact types known to cause torso injuries (for instance, with single vehicle impacts with narrow objects, such as trees) were not included in this study. Page 2

As in the COVER accident analysis which formed the basis to this investigation, typical thorax injuries were: rib fractures, sternum fractures, lung contusions and clavicle fractures. This study reinforces these injuries as being priorities for investigation with the Thorax and THOMO projects. An improved dummy thorax should be able to predict injuries such as these. In almost all cases in which a front passenger was present, the front passenger suffered the more severe injuries, despite being on the non-struck side of the vehicle and therefore likely to be subjected to a less severe acceleration pulse. Although mostly females were sitting on the passenger seat, it could not be identified if gender was the main factor or the restraint system was better optimised for the driver than for the front passenger. As a result, injury risk curves for women should be developed and compared with those for men. Additionally, further investigations are needed to identify: o If the protection level afforded by the front passenger restraint system is equivalent or lower than the driver restraint system. o Why women may be more severely injured than male occupants? It was very interesting to note that young occupants tended to receive only slight injuries in some quite severe accidents. This illustrates the effect of age on the ability of the human body to tolerate loads applied through modern restraint systems in frontal crashes. It also suggests that protection in offset frontal crashes, for younger occupants, is good generally. It was concluded that: Injury risk curves for elderly occupants should be developed. It should be investigated to see if the dummy design (i.e. biofidelity; rib stiffness, etc.) as well as the associated injury risk functions need to be specific to an elderly occupant. This will be important for Work Package 3 of the Thorax project. The mechanism of lung injuries with no or relatively minor hard thorax injuries (e.g. rib, sternum, or clavicle fractures) is not clear and deserves further investigation. o The mechanism of the non-penetrative lung injuries (e.g. compression, velocity of compression), both with and without skeletal injury, is not clear and it is recommended that this is investigated in Thorax WP2 and in THOMO. Consideration should be given to whether the current thorax instrumentation and injury risk functions provide an adequate prediction of the risk of this type of injury. Also, spinal column injury mechanisms need to be analysed further. When clavicle and rib fractures occurred together, there was a high risk of mortality for the injured occupant. It is recommended that the correlation of this injury combination with other serious, or fatal, injuries and the mechanism responsible is investigated within the rest of the Thorax and THOMO projects. This may give an insight as to how these injuries could be prevented. General recommendations regarding the frontal impact crash testing of cars, arising from this work, are: A female dummy should be used on the front passenger seat or at least specific risk curves should be used for front seat passenger injury risk evaluation. Page 3

An additional lower speed frontal impact test may also be of benefit in assessing thoracic injury risk. If more than one frontal impact test is conducted, consideration should be given to testing with a closer seating position for the driver. This is expected to represent a sub-optimal seating position. Page 4

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public Contents 1 Introduction... 7 1.1 Background... 8 1.2 Objective... 8 2 Methods and results... 9 2.1 Euro NCAP... 9 2.1.1 Frontal impact test... 9 2.1.2 Impact speed equivalence... 9 2.1.3 Overlap... 9 2.1.4 Crash Pulse... 9 2.1.5 Injury risk assessment...10 2.2 Real-world accident cases...11 3 Summary of findings from the case reviews...13 3.1 CCIS case overview...13 3.2 GIDAS case overview...16 3.3 Case summary...18 3.3.1 Occupant factors...19 3.3.2 Dummy factors...19 3.3.3 Crash factors...19 3.3.4 Other observations...20 3.3.5 Potential intervention...20 4 Risk Register...21 5 Discussion...23 6 Conclusions...27 7 Recommendations...29 Acknowledgements...30 References...31 Appendix A. Data reporting template...33 Appendix B. CCIS Case Overview...35 Appendix C. UK CCIS Cases...39 Appendix D. GIDAS Cases...41 Appendix E. GIDAS Case comments and explanations...43 Appendix F. GIDAS Case Overview...45 Page 5

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 1 Introduction The attainment of an ever-higher safety level on European roads is a long-term and ongoing process, which has seen significant progress already. The last 30 years have seen a tripling of traffic on European roads while the number of casualties has halved during the same period (European Commission, 2006a). Improvements in occupant protection (passive safety) have contributed significantly to this progress. For example, the combination of EU legislation for crash test standards and improved consumer information through the Euro NCAP has substantially raised the survivability for vehicle occupants in a crash. In the 2001 White Paper on Transport Policy, the European Commission set the ambitious objective of halving the number of fatalities on European roads by 2010. However, based on a review of the 2004 data, the EC published that progress did not appear to be sufficient in order to reach the Community s 2010 target (European Commission, 2006a). Around 41,600 people were killed and more than 1.7 million injured in European road accidents in 2005 (European Commission, 2006b). Although the number of road fatalities has declined by more than 17 percent since 2001, greater efforts will have to be made if the European Commission s target of halving the number of deaths on the roads by 2010 is to be met. Although efforts are needed on all levels of road safety, the COVER (Coordination of Vehicle and Road Safety Initiatives) project has been set-up to develop a harmonised and consistent direction of research and to accelerate the implementation of research findings of four complimentary research initiatives in the field of crash biomechanics (including the Thorax and THOMO projects. Motivated by findings of previous projects (including EC Framework projects) the Thorax and THOMO projects were set-up to study thoracic injuries for a wide variety of car occupants and transfer results into test and design tools. In order to maximise the safety benefits gained from new vehicle and restraint technology for various genders, ages and sizes of occupants, these tools will have to be much more sensitive to the in-vehicle occupant environment than is the case with existing test tools. From a review of in-depth accident data from around Europe (the UK, Germany, and France), the COVER Project Task 1.2 provided an overview of the current situation with regard to thoracic injuries resulting from frontal impact car accidents (Carroll et al., 2009). Of the body regions injured in the accidents analysed, the thorax was the most frequently injured body region for all killed and seriously injured occupants in frontal impact accidents. With this knowledge it is reasonable to investigate why thorax injury hasn t been reduced as much as may have been expected. Such expectations are based on improvements in protection observed for occupants generally and for other specific body regions, such as the head. Assuming thoracic safety improvements are lagging behind improvements for other body regions, then research effort is needed to understand why recent advances in crash safety have not been as effective for the thorax as for other body regions and what more can be done to protect the thorax during impact events. Page 7

1.1 Background Overall differences between accidents and crash test results have been studied in great detail by the SARAC II EC Project (e.g. Newstead et al., 2006). This project examined the real-world performance of Euro NCAP tested vehicles in frontal impacts, compared with the overall star rating for the vehicle and with a weighted index of the body-region scores. It highlighted the need for further investigation of the relationship between the front impact component of the Euro NCAP test score and real world injury outcomes in front impact crashes (Delaney et al., 2006). It is considered that SARAC II exhausted the statistical approach to this problem. 1.2 Objective This task of the Thorax Project was devised to follow on from both the initial accident analysis of COVER Task 1.2 and the statistical analyses carried out with SARAC II. Individual cases from the dataset of frontal impact accidents used in the COVER work have been selected for further in-depth analysis. These cases have been chosen such that the impact conditions are close to those used in the Euro NCAP frontal impact test and where Euro NCAP has tested the vehicle being investigated. The information collated within this task is intended to show whether occupant factors (such as age, size, and gender which will affect seating position, distance to the steering wheel, and so forth) and dummy factors (such as the inability to discriminate between different loading conditions) contribute to any discrepancy between real-world thorax injury risk and crash test results. Ideally, this will be used to determine whether for key injury groups, test procedure or dummy issues dominate. For instance, if occupant characteristics are dominant then the test procedure is not representing those occupants very well. It may be possible to adjust the injury risk function that is used with the dummy to account for such differences (injury risk functions for use with the Thorax Project dummy hardware are considered within Work Package 2), though relevant biofidelity may also be required. Alternatively, if the occupant characteristics tend to be similar to those of the dummy, this would suggest that improvements in the dummy are necessary. As such, the detailed case-by-case study reported here will provide some focus for biomechanics work within Work Package 2 (WP2) of the Thorax Project. Recommendations for WP2 and WP3 (the test tool demonstrator development and validation work package) will be made from this analysis. In particular, it will be identified whether improvement of the existing test tools or better targeting of the occupant characteristics through the development of improved injury risk functions would be expected to give the greatest benefit in addressing the injuries investigated. Page 8

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 2 Methods and results 2.1 Euro NCAP 2.1.1 Frontal impact test The Euro NCAP frontal impact test involves crashing a vehicle into a deformable barrier element, fixed to a rigid immovable block, at the target speed of 64 km.h -1 (Euro NCAP, 2009a). The car is aligned with the barrier so that the initial level of overlap is limited to 40 percent of the car s width. The barrier stiffness is approximately equal to that of the front of another passenger car. The test impact conditions are used to represent a relatively severe car-to-car offset crash. For this test 50 th percentile male Hybrid III dummies are used in both of the front seating positions. The assessment of the adult occupant injury risk is based on the outputs from these dummies together with other modifiers for undesirable aspects of the vehicle s behaviour during the crash. 2.1.2 Impact speed equivalence The energy absorbed by the test barrier during a frontal impact Euro NCAP test was assessed by Lenard et al. (1998), based on 26 tests carried out between 1996 and 1998. According to Lenard et al., the barrier absorbed between 37 and 56 kj, with a mean value of 45 kj. In addition Lenard et al. also predicted the change in velocity, v, of the crashing vehicle, based on the vehicle deformation; as would be the case for accident investigations. They found that the calculated v was 57 km.h -1 on average, or 7 km.h -1 less than that measured during the test (around the impact speed of 64 km.h -1 ). The scatter associated with this assessment was ± 10 km.h -1. It should be noted that the Lenard et al. work was based on cars that would have been released during the middle of the 1990s. These are likely to be lighter and less stiff than more modern vehicle designs. This is particularly the case as initially Euro NCAP was biased towards super-mini and small family car models. Therefore, we can expect more energy to be absorbed by the barrier during an impact with modern vehicles than with older vehicles. However, it is not certain how the extra energy absorption will affect the v calculation. It is expected that the mean v, as calculated based on vehicle deformation, and expected with Euro NCAP tests of modern vehicles, will be lower than 57 km.h -1. However, this remains to be proved empirically, and as such the extent to which the expected value is less than 57 km.h -1 is unknown. 2.1.3 Overlap The 40% overlap used in the Euro NCAP test is designed to load the vehicle structures in a similar manner to a 50% overlap car-to-car crash (Lowne, 1994). 2.1.4 Crash Pulse It should be noted that a car-to-car crash with an overlap of 50% and impact speed of 55 km.hr -1 may not give exactly the same deceleration pulse to the occupant compartment, or occupant loading of the restraint system, as a 40% offset test with a target speed of 64 km.hr -1 (e.g. Wykes et al., 1998). Page 9

2.1.5 Injury risk assessment The rating scheme used in the Euro NCAP adult occupant protection assessment is defined by their protocol (Euro NCAP, 2009b). In the Euro NCAP frontal impact test, two criteria are used to determine the thoracic injury risk. These are the chest compression measurement and the viscous criterion (V*C, the velocity of compression multiplied by the compression value itself). Both of these criteria are based on measurements taken from the dummy s single-point sternal deflection measurement transducer. Points are awarded depending on the measured output in relation to two threshold values, a higher and a lower performance limit. A value lower than or equal to the higher performance limit will result in maximum points being awarded. A value higher than or equal to the lower performance limit will result in no points being awarded. A sliding scale system is used to determine the points to award if the value falls between the two thresholds, and this is calculated by linear interpolation between the two limits. The threshold values for the chest compression and V*C, used in the Euro NCAP frontal impact test are shown in the following table (Table 2-1). Table 2-1 Euro NCAP frontal impact adult occupant performance limits for the chest (Euro NCAP, 2009b) Higher performance limits Lower performance limits Criterion Value Risk of injury Compression 22 mm 5 % risk of injury AIS 3 Viscous Criterion 0.5 m.s -1 5 % risk of injury AIS 4 Compression 50 mm 50 % risk of injury AIS 3 Viscous Criterion 1.0 m.s -1 25 % risk of injury AIS 4 The chest result modifiers are: Displacement of the A pillar o The score is reduced for excessive rearward displacement of the driver s front door pillar, at a height of 100 mm below the lowest level of the side window aperture. Up to 100 mm displacement there is no penalty. Above 200 mm there is a penalty of two points. Between these limits, the penalty is generated by linear interpolation. Integrity of the passenger compartment o Where the structural integrity of the passenger compartment is deemed to have been compromised, a penalty of one point is applied. The loss of structural integrity may be indicated by characteristics such as: door latch or hinge failure, unless the door is adequately retained by the door frame. buckling or other failure of the door resulting in severe loss of fore/aft compressive strength. Separation or near separation of the cross fascia rail to A pillar joint. Severe loss of strength of the door aperture. Page 10

The protection provided for adults for each body region is shown in the Euro NCAP results using coloured segments within body outlines. The colour used is based on the points awarded for that body region, as follows: Green 4.000 points Yellow 2.670 to 3.999 points Orange 1.330 to 2.669 points Brown 0.001 to 1.329 points Red 0.000 points Results are shown separately for the driver and for the passenger. 2.2 Real-world accident cases From the frontal impact case samples used in COVER Task 1.2, accidents were identified that broadly matched the Euro NCAP impact conditions. The selection criteria for the COVER accident cases were as follows: Cars (or car-derivatives) Registered in 2000 or later Had one significant frontal impact o CCIS cases were selected to give frontal accidents according to the Collision Damage Classification (CDC) o GIDAS cases were selected with Vehicle Directions of Impact (VDI) with the main damage to the front of the vehicle and be either 11, 12, or 1 o clock No roll-over at any point Occupants were seat-belted Occupants were known to be 12 years old or over Occupants had a known overall MAIS Cases from the CCIS were judged to be potentially like the Euro NCAP crash test if: The change in velocity for the case vehicle ( v) was 65 km.h -1 or less (this eliminated cases where the accident severity was overwhelmingly high). The off-side longitudinal was loaded directly, and the near side longitudinal was not loaded directly. Also, due to the availability of cases, CCIS accidents were only selected if at least one occupant sustained a serious or fatal injury, to any body region. Injury causing crashes were deemed to be a priority for investigation. Cases from the GIDAS were judged to be potentially like the Euro NCAP crash test if: The main loading was on driver s side. Also unbelted cases were taken into account (two unbelted occupants were included in the sample). These cases were included to understand the injury mechanisms and to compare it with similar accidents where the occupants were restrained. A lot of cases were excluded from the GIDAS sample because the car collided with a tree which induced severe damage and injuries. Usually these cases had a small overlap and because of this a high intrusion. These accident configurations were not comparable to the Euro NCAP crash test conditions, so they are not included in this Page 11

investigation. However, one impact with a large tree was included in the CCIS analysis. Priority was given to cases where the case vehicle had the driver on the same side as the Euro NCAP crash test vehicle (i.e. left- or right-hand drive as appropriate). This is to avoid any concerns that minor structural differences from one side of the vehicle to the other might affect the crash performance for that vehicle. The thorax injuries sustained by the vehicle occupant were compared with the predictions from the Euro NCAP test. So that injuries could be compared, cases where one occupant was seriously injured were selected. This seems a reasonable stipulation as the Euro NCAP test is intended to assess the vehicle in severe loading. The cases were examined for any additional factors that may have influenced the level of thorax injury (such as an unbelted occupant in the seat behind the driver). For the cases considered below, no additional factors were identified. The difference between the crash test results and the real-world outcome was investigated in terms of the injury mechanisms and occupant characteristics. For example, the COVER accident analysis had shown that older occupants were at a greater risk of sustaining a rib fracture than younger occupants. The total number of cases reviewed was 34. This included 20 cases from the CCIS and 14 from the GIDAS. To make the reporting of case details standard for both CCIS and GIDAS accidents, a data template was drawn-up. This is shown in Appendix A, Table A-1. The cases from the UK CCIS database that were reviewed are described according to the data template in Appendix C. The equivalent cases from the GIDAS database are shown in Appendix D. An appendix showing comments and explanations for some of the abbreviations and codes used in the GIDAS cases is provided in Appendix E. Page 12

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 3 Summary of findings from the case reviews A general summary of the findings from the CCIS and GIDAS cases are presented in the following two sub sections (Sections 3.1 and 3.2) respectively. 3.1 CCIS case overview For this task, 20 cases from the UK CCIS were selected and reviewed. As noted earlier, these cases are described in Appendix C. In general the selection criterion of the damage profile involving one longitudinal beam only gave accidents with similar overlap to the 40 percent specified for the Euro NCAP test, and the engine was also directly loaded in most cases. However, there were a few cases with wider overlap and a few cases with lower levels of overlap than registered in the test. For almost all cases, the change in velocity ( v) of the case vehicle was lower than the 64 km.h -1 impact speed of the test. However, as noted in Section 2.1.2, the change in velocity in the Euro NCAP test would be equivalent to a v lower than 64 km.h -1 (more like 57 km.h -1 ) based on prediction from the vehicle damage. Based on comparison of the vehicle damage, it seems as though a number of the collision severities were much closer to the severity of the Euro NCAP crash test than may have been expected based on v alone. To show how closely the selected cases represent the total frontal impact population, certain key aspects of the cases were compared against the whole frontal impact sample used in the COVER Task 1.2 (2,148 occupants). Also, to provide a relationship with national frontal impact accident data, the injury severity rating was also compared with STATS19 data. These British data have been published previously by Richards and Cuerden (2009). The distribution of the injury severity for the occupants in the cases selected for this task, the CCIS frontal impact dataset and the national statistics are shown in Table 3-1. This table shows how the selection criteria used in the CCIS bias that sample towards the more severely injured occupants than would be seen from a random sample of the national accidents. Furthermore, the CCIS cases selected for comparison with Euro NCAP tests have an even higher proportion of fatally and seriously injured occupants. This can be explained as the cases were chosen only if one of the occupants in the vehicle received a serious injury. Table 3-1 Distribution of occupant injury severity in case files, all COVER Task 1.2 frontal impact dataset, and national Police-reported frontal impacts Injury Case examples (%) Frontal impact dataset National data (%) severity (%) Fatal 6 3 0.5 Serious 62 28 7 Slight 26 55 92 Non-injury 6 14 The MAIS for the occupants from the selected CCIS case examples and from the whole frontal impact dataset are shown in Table 3-2. This shows that there are more uninjured or slightly injured occupants in the frontal impact CCIS sample used in COVER than in the case examples used for this task. Instead the case examples have more occupants with MAIS 2 or MAIS 3 injuries. Again this reflects our intention to look at occupants with a thorax injury, and hence the use of the selection criteria where the case vehicle included a seriously or Page 13

fatally injured occupant. It is also apparent that some occupants who were recorded by the Police as having slight injuries had a MAIS of 2 or greater. Table 3-2 Occupant MAIS for the case examples and the COVER Task 1.2 frontal impact dataset MAIS Case examples (%) Frontal impact dataset (%) 0 6 14 1 22 60 2 38 15 3+ 34 11 The ages of the occupants form the selected cases were in the range from 17 years to 84 years. Generally, the occupants from these cases were slightly older than those from the whole frontal impact dataset. This is shown by Table 3-3, which provides the average and well as the quartile ages for the selected cases and the frontal impact dataset. This difference was not statistically significant at the 95 percent confidence level. Table 3-3 Distribution of occupant age for the case examples and the COVER Task 1.2 frontal impact dataset Age Case examples Frontal impact dataset Average 51.3 41.2 25 th percentile 30.5 25 50 th percentile 54 39 75 th percentile 71.5 55 The percentages of the occupants from the selected cases and from the whole frontal impact dataset who were either male or female are shown in Table 3-4. The slight difference shown (for there to be more male occupants in the selected cases) was not statistically significant. Table 3-4 Sex of the occupants in the selected cases and the COVER Task 1.2 frontal impact dataset Sex Case examples (%) Frontal impact dataset (%) Male 59 56 Female 41 44 The percentages of the occupants from the selected cases and from the whole frontal impact dataset who were either drivers, front seat passengers, or rear seat passengers are shown in Table 3-5. The greater proportion of front seat passengers in the case examples was statistically significant. Table 3-5 Distribution of crash severity, v, for the selected cases and the COVER Task 1.2 frontal impact dataset Seating Position Case examples (%) Frontal impact dataset (%) Driver 59 75 Front seat passenger 38 21 Rear seat passenger 3 5 Page 14

Also, in Table 3-6, the severity of the crashes was compared between the selected cases and the frontal impact dataset via the v measure. Again any difference was not statistically significant at the 95 percent level. Table 3-6 Distribution of crash severity, v, for the selected cases and the COVER Task 1.2 frontal impact dataset v Case examples (km.h -1 ) Frontal impact dataset (km.h -1 ) Average 31.4 31.2 25 th percentile 28.5 22 50 th percentile 31 29 75 th percentile 35.5 39 In summary, when comparing the case examples to the COVER Task 1.2 frontal impact dataset, the distributions of seating position, injury severity, and MAIS were found to be different at the 95% level. No other significant differences were found. Within the accident analysis conducted as part of the COVER Project, it was identified that most torso injuries occurred in cars that had advanced modern restraint systems, which was a feature of selecting only modern vehicles for the analysis. Typically these restraint systems would consist of seat belt, airbag, pretensioners, and a load limiter. On this basis, it was not surprising to see that almost all of the cases reviewed here involved restraint systems incorporating a seat belt, airbag, and pretensioner. However, several vehicles did not seem to have a load limiter. This is interesting as it may indicate that occupants without a load limiter are at a greater risk of sustaining a torso injury than those with a load limiter, in crashes with impact conditions similar to the Euro NCAP test. However, it may just be a feature of the statistically small number of cases analysed in this detail. In one case (Case 13411) the airbags and pretensioners didn t activate. The reason for this is uncertain. In this case the 72 year-old female front seat passenger sustained AIS 5 thorax injuries, despite the low v. This occupant sustained AIS 5 bilateral tension pneumothorax and bilateral AIS 4 lung contusions, despite sustaining only two rib fractures (on the right hand side of the thorax). It is possible that the low-speed deployment and performance of airbags could be checked in a low-speed test. In several cases (Cases 13411 and 86158) the seat-belt pretensioners did not activate. In the first case, the v was quite low (27 km.h -1 ), but in the other case it was towards the high end of the data set (38 km.h -1 ). Most of the vehicles did not have a load limiter, and in one case (86158) the load limiter did not activate. Many of the cases had an unexpectedly high level of intrusion into the occupant compartment, despite the v of the impact being relatively low compared with the Euro NCAP test. Occupants in these cases who were young to middle age generally had no or minor thorax injury despite the intrusion (e.g. Cases 58117, 70660). One younger occupant (Case 86082) had serious thorax injuries at a v of 42 km.hr -1 with severe intrusion. All of the elderly occupant in vehicles with moderate (5-10 cm) or severe (>10 cm) intrusion sustained moderate to severe thorax injuries (Cases 35395, 75680 and 76042). Page 15

Overall, some cases had unexpectedly high intrusion and/or unexpectedly high injury levels compared with the Euro NCAP tests, some had serious injuries as predicted by Euro NCAP despite markedly lower v in the accident, whilst other cases had very minor or no thorax as may be expected for the lower v s even if the Euro NCAP frontal impact score was not very good. There were no confounding factors, such as an unbelted rear seat occupant impacting the rear of the front seat occupant and increasing the restraint loads. Except for the one case mentioned above where the airbags did not deploy, where injuries occurred, this appeared to be because the restraint system had deployed but not provided sufficient protection. This was despite the maximum delta-v being 42 km.hr -1. A number of occupants had clavicle or sternum fractures without any rib fractures or other significant injury (Cases 17211, 17258, 33679, 34041 and 76042); two of these occupants were young and three were elderly. Several occupant had serious lung injuries with minor rib, sternum of clavicle fractures (Cases 13411, 15420, 35395 and 86158), and all of these occupants were middle age to elderly. One occupant had serious lung injuries without skeletal torso injury (Case 86082), and this occupant was young. These findings are in agreement with the Cover accident analysis (Carroll et al., 2009). There were four cases with a male and a female occupant, and the female occupant was more seriously injured in (Cases 13411, 34041, 35127, 86168), but there were as many cases where the female occupant was less seriously or equally injured compared with the male occupant In almost every case, the torso injuries were reported as having been caused through contact with the seat belt webbing. There were a few exceptions to this where the injuries were caused by another object in the vehicle or were judged to have been non-contact injuries. Very few injuries were attributed to either the airbag or steering wheel rim. Apparently the airbag caused a bruise to the upper left anterior chest of the driver in Case 35127. Whereas, despite activation of the airbag and pretensioner, the steering wheel rim contact was responsible allegedly for the thoracic injuries to the driver in Case 35395 (even the spinal and the right clavicle fractures). No other sources of contact (such as the adjacent door, or fascia, or other occupant) were noted in the thoracic injury causation codes. Equally, in no case were the torso injuries attributable to, or was the level of thorax injury influenced substantially by additional factors (or peculiarities) of the accident (such as an unbelted occupant in the seat behind the driver). Several clavicle fractures were noted in the accident cases. Some measure of the force applied to the clavicle of the dummy in the crash test may be helpful in designing against this injury. Currently, the clavicle is not well represented in the Hybrid III, and there is no measurement of the force applied to the shoulder. 3.2 GIDAS case overview The GIDAS cases were a good representation of the total vehicle fleet. They included typical crash configurations and had a high diversity of occupants regarding age, sex, height, and weight. A total listing for all parameters on vehicle and personal level for the GIDAS cases can be found in Appendix F GIDAS Case. Typical values were calculated and are listed below. Page 16

The distribution of MAIS for the selected GIDAS cases shows Table 3-2. The numbers are similar to the MAIS distribution from the CCIS cases. Obviously, there were mostly very severe injuries, because this accident configuration is very severe. However, 10 % of the occupants were uninjured. Table 3-7 Occupant MAIS for the case examples MAIS Case examples (%) 0 10 1 20 2 35 3+ 35 The distribution the occupant ages for the GIDAS cases are listed in next Table 3-3. Age of occupants is similar to the age distribution in CCIS case analyses. The average age is 50 years. The range is from 19 to 82 years old. Generally, most persons were quiet old in this sample, but the range of average occupants was covered. Table 3-8 Distribution of occupant age for the case examples Age Case examples Average 50,4 25 th percentile 28,8 50 th percentile 56,0 75 th percentile 71.3 In this analyses were 20 persons involved, with an equal distribution of gender. Thus, 10 males and 10 females were included. In the COVER analyses there were more males included comparing to females, which is congruent with the analyses from CCIS. Table 3-9 Sex of the occupants in the selected cases Sex Case examples (%) Male 50 Female 50 In Table 3-10 the distribution of crash severity in Delta-v is shown for the GIDAS cases. An average value of 52 kph is calculated. This is below the simulated test velocity of Euro NCAP of 55 kph. CCIS data even has an average delta v of 31 kph, which is lower than the GIDAS value. That can be explained by the use of different calculation methods. Furthermore CCIS skipped all accidents with a higher delta v than 64 kph. Table 3-10 Distribution of crash severity, v, for the selected cases v Case examples (km.h -1 ) Average 52,3 25 th percentile 40,0 50 th percentile 48,0 75 th percentile 56,0 Page 17

It seemed that occupant and crash configurations are such dominant factors that the injury prediction of Euro NCAP is not always a reliable guide to outcome. Even when the real world crash is broadly similar to the conditions of the test, just minor changes in the conditions leads to differences in the real-world outcome. At least in high velocity accidents, conformity between Euro NCAP injury prediction and real-world outcome could be observed. Many thorax injuries, in particular contusions or sternum and rib fractures, were caused due to thoracic loading combined from the seatbelt and airbag.usually the scene investigator would assign the seat belt as first injury causing part and airbag as second injury causing part. In a few cases injuries at spinal column were recorded, in particular the lumbar spine. Further investigation is needed to identify the mechanism responsible for these injuries. The majority of the cases occurred at a minor impact speed than the test speed used in Euro NCAP. As shown in the statistical analyses from COVER project Task 1.2, it could be seen that the age of the persons is a dominant factor. Older persons can suffer severe injuries, without substantial occupant compartment intrusion. On the other hand, in some cases it could be observed that younger persons sustained only minor injuries in a high velocity accident even though there was compartment intrusion. If there was no deformation of the occupant compartment, younger persons usually sustain only minor injuries. When accidents involving trees are excluded (as they have been for this study), the majority of the investigated cases had an overlap more than 40 %. For severe accidents, the Euro NCAP prediction for torso injuries was limited by the lack of assessment of the risk of severe abdominal injuries like spleen and liver ruptures. 3.3 Case summary The following section brings together findings from both the UK and German case reviews. To drive greater protection for the torso region in the future it is necessary to have a test procedure that relates measured outputs to real-world levels of safety. Where differences are observed between the real-world accident outcomes and what would be expected based on crash test results, then the implication is that such a relation is not very strong for those cases. It should be noted that there may be specific impact types or a group of occupants for which the existing relationship between crash test and real world accident outcome is already effective. To improve the connection between the crash test metrics and real world accident outcome, a number of aspects of the test could be investigated (as examples: whether the test conditions reflect an important crash type with respect to thoracic injury epidemiology, whether the dummy size appropriately represents the occupants being injured, or if more focus should be put on the physical tolerance to loading for older occupants than is currently the case). In order to offer advice as to the aspects of frontal impact testing that are in greatest need of being updated, on the basis of improving the link to real-world thoracic injury Page 18

risk, factors related to the occupant, dummy, and crash were investigated. The influence of the factors falling under these categories are described in the following sections. 3.3.1 Occupant factors The CCIS cases showed a useful spread of occupant ages and sizes (at the extremes of the ranges there were two seventeen year old ladies, and one 84 year old lady). In the GIDAS cases the age distribution was similar with a spread of occupant ages from 19 years old to 82 years old. Age seemed to be a major factor. It was very interesting to note that young occupants tended to receive only slight injuries in some quite severe accidents. These cases showed that older persons can sustain severe injuries in accidents without compartment intrusion and at a relatively low crash severity. This illustrates the affect of age on the ability of the human body to tolerate loads applied through modern restraint systems in frontal crashes. Small as well as tall and slim occupants seemed to be more vulnerable to thoracic injury. Big and heavy occupants may have a higher injury tolerance level. Generally, women are more vulnerable to torso injury than men. This may be due to the following factors: Lower injury tolerance level of females Effectiveness of the passenger airbag Position of the front seat passenger immediately before the accident (OOP), for instance sitting too far from the airbag for it to offer effective restraint. From the GIDAS data, it appears that having a larger distance between the occupant and the steering wheel seemed to be beneficial. No conclusion could be made regarding seat back angle. 3.3.2 Dummy factors In the investigation quite a wide range of (well-known) thorax injuries with different severities was found; for instance, thorax bruise or contusion, sternum fracture, rib fracture, series of rib fractures, and lung injuries. The Hybrid III dummy measures thoracic deflection at one point in one dimension, and it is this chest compression or the viscous criterion (V*C), based on the compression measurement, which is used to predict the probability of an AIS 3 (or 4) thorax injury occurring. It may be that the criteria based on this single point chest compression measurement are not able to assess the appropriate injury mechanisms for the full diversity of thoracic injuries observed. This is important because the different injury types could have vastly different implications regarding the prognosis for an occupant. No abdomen injuries can be predicted with the Hybrid III as used in Euro NCAP. Sometimes abdominal injuries do occur in accidents with a high severity, as was shown in these cases. 3.3.3 Crash factors In the analyzed GIDAS cases occupant factors seemed to dominate over crash factors. Sometimes even in the same crash the front seat passenger sustained higher injuries than the driver, which was in some cases contrary to the injury risk predicted by Euro NCAP. Page 19

The cases in which the occupants sustained thorax injuries were not exact matches with the Euro NCAP collision conditions. In general the impact velocity was lower, sometimes much lower, than the test speed. The overlap in the cases reviewed was also more varied than in the crash test, although - like the crash tests - the off-side longitudinal and engine were usually loaded in the accident cases. The deformation pattern of the front structure and compartment resulting from the accident were similar to the deformation pattern of vehicles tested in Euro NCAP. In a number of the CCIS cases, the level of intrusion in the accident vehicles was as great as, or greater than, that observed in the Euro NCAP test, despite a similar overlap and markedly lower delta-v. The reason for this was not clear, but it did not appear to be due to over- or under-run. It seemed that by colliding with another car the loading paths in the front structures were mostly used effectively. In particular the load path of a deforming front wheel which is supported by the rocker could be noticed. In a few cases it could be observed, even though the compartment was stable, the occupants sustained severe injuries, in particular with elderly and female occupants. Compatibility issues could be observed in cases with old against new vehicles. Under-run occurred in one collision with the rear end of another car and in one collision with a motor home. 3.3.4 Other observations Based on the CCIS and GIDAS sample of accident cases, no general conclusion could be made on the efficacy of load limiters. In several cases injuries of the spinal column were recorded, in particular the lower cervical or upper lumbar spine. Further investigation is needed to identify the mechanism responsible for these injuries. 3.3.5 Potential intervention The groups of occupants currently at greatest risk of receiving a thorax injury in accident configurations similar to the Euro NCAP frontal impact test are elderly people and women. Neither of these groups is represented specifically by the dummy used in European regulatory frontal impact crash tests or the Euro NCAP frontal impact test. Page 20

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 4 Risk Register Risk What is the risk No. 0 No risks identified as being specifically related to this task Level Solutions to overcome the risk of risk 1 0 No solution necessary 1 Risk level: 1 = high risk, 2 = medium risk, 3 = Low risk Page 21

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 5 Discussion The general principle used within this study was to use crash performance as determined through a crash test result to predict real world crash behaviour. This prediction was then compared with the outcome from a real world accident. Whilst we have selected cases where the accident was close to the conditions of the test, they are never exactly the same. For instance, the impact severity might be somewhat lower, or the level of overlap somewhat different. As the accident cases never exactly matched the test configuration, there was, in each case, an element of extrapolation from the crash test to the real world accident prediction. This extrapolation is subject to a number of assumptions. For instance, it is assumed generally that at lower impact speeds less energy is transferred to the case vehicle and therefore the occupant should be subjected to lower acceleration levels and hence have a lower injury risk. However, there may be instances where the restraint system gives a suboptimal performance at lower speeds. Under these circumstances the counter-intuitive result could be a greater injury risk at a lower impact severity. This effect was demonstrated by Smith and Cooper (2006). They found that the risk of sustaining a serious thorax injury, as measured through the Hybrid III dummy s chest deflection, could be higher in tests at either 40 or 56 km.h -1 than in a test at 64 km.h -1. In many cases several different factors came together to make the outcome different from that which might have been expected. For instance, in the second CCIS case (number 15369) the driver sustained relatively minor injuries (MAIS 2) being in a car that was assessed as offering only weak protection for the thorax. However, the impact severity appears to have been lower than the test, the occupant was 32 (so younger than age of occupant for which the dummy injury risk functions are usually developed), and approximately 50 th percentile dummy height and weight (so likely to have interacted with the restraint system in the intended way). It is very difficult to judge the degree to which each of these factors would have influenced the outcome, but it is possible to conclude that there is no evidence that a new test procedure, dummy hardware, or injury risk functions would have delivered much safety benefit in this case. Another example where confounding factors make interpretation of the results difficult is where the passenger was more severely injured than the driver. When the driver was seated on the struck side in an offset frontal impact, the driver would be expected to have been subjected to a greater deceleration pulse and therefore be more severely injured than the passenger. Correspondingly, Euro NCAP tends to predict a higher injury risk for the driver than for the front seat passenger. In certain cases from the study the opposite outcome was seen in the real-world accident. Differences in the occupant characteristics between the driver and passenger go some way to explaining these results. However, optimisation of the restraint system may also contribute along with the potential for there to be complex vehicle structure interactions. As such, it is not always simple to prioritise the factors influencing the outcome. In most cases the torso injuries recorded were attributed to loading from the seat belt. Through discussion with the Crash Analysis Team at TRL, it was indicated that the CCIS accident investigators would most likely assign this as the injury causing contact unless there was some definitive other evidence. For example, seat belt webbing would be cited unless the steering wheel was deformed and that could be justified as causing the injuries instead. Whilst the seat belt was worn in every CCIS case and would have contributed to the kinematics of the occupant, consideration should be given as to whether potential other injury causing mechanisms were given appropriate thought. It may be that having an almost default contact code for thoracic injuries may mask investigation of other less obvious contacts. Page 23

Within the COVER project overview of thoracic injuries deliverable (D5: Carroll, 2009) a number of trends were identified regarding the combinations of thoracic injuries being sustained by occupants of different ages. For instance, the vulnerability of women to AIS 1 torso injuries (typically bruises) was observed. Also older occupants had a greater propensity to receive torso injuries than younger occupants. Typical thorax injuries in the COVER accident analyses were lung injuries, a series of rib fractures, sternum fractures, bruises and contusions. These are the thorax injuries seen most frequently in the accident cases reviewed for this task as well. It was noted within the COVER work that at the AIS 3 level, young occupants tended to receive lung injuries with no or relatively minor (AIS 1 or 2) skeletal injuries. With the CCIS cases, there are examples that support the COVER trend. The relatively old drivers from cases 35395, 77041, and 86158 sustained AIS 3 rib fractures or rib and lung injuries in combination, whereas the 48 year old driver from Case 15420 had bilateral lung contusions (AIS 4) with just a sternal fracture (AIS 2). This sternum fracture would not have been shown as occurring in association with the lung contusions by the AIS 3 COVER analysis of torso injuries. The 30 year old driver from case 86082 sustained a contusion to the lower lobe of his right lung (AIS 3) without any skeletal thorax injuries. Therefore, the pattern of injuries and ages of the occupants is in accordance with expectations based on the COVER analysis. The older occupants had severe rib injuries whereas the younger driver from Case 86082 had a lung injury alone. The injuries to the 48 year old driver from Case 15420 bridges the gap between young and old occupants neatly, receiving a fracture of the rib cage (AIS 2 sternum fracture) that was less severe than those sustained by the older occupants, whilst also sustaining bilateral lung contusions (AIS 4). The mechanism of these lung injuries (e.g. compression, velocity of compression) is not clear and it is recommended that this is investigated in Thorax WP2 and in THOMO, with consideration of whether the current thorax instrumentation and injury risk functions provide an adequate prediction of the risk of this type of injury. Another pattern of the torso injuries from the COVER analysis was that occupants tended to sustain rib fractures or clavicle fractures, but rarely both. In the CCIS cases reviewed for this study this was again the general trend. However for four occupants from 72 years to 84 years of age, rib fractures and a clavicle fracture occurred together (sometimes with other thorax injuries). It is known that two of these occupants subsequently died from their injuries. This indicates how rib and clavicle fractures can occur together, but it tends to happen for only the most frail occupants. When such a combination of injuries does occur it seems to mark a poor prognosis for that occupant. Therefore, it is recommended that the mechanism of injury causing rib and clavicle fractures together is investigated in the Thorax and THOMO projects. Investigating why this injury combination is not seen so frequently for younger, stronger occupants may also be useful for trying to prevent such loading in the future. However, it seems quite likely that this is a simple manifestation of decreasing skeletal loading tolerance with aging. Cases were specifically chosen to be similar in configuration to the Euro NCAP test. This meant that impacts with narrow objects were excluded. The GIDAS COVER analysis had shown that for single vehicle accidents, AIS 3 torso injuries were more likely to occur in impacts with narrow objects (with a diameter less than 40 cm) than in collisions with other types of objects. The serious torso injuries occurring in impacts with narrow objects are then one specific tranche of accidents that has not been included in this analysis. Appropriate testing to prevent these injuries should be considered alongside the Euro NCAP and regulatory frontal impact testing regimes. Page 24

As mentioned above, if the accident impact conditions were to match those of the Euro NCAP crash test exactly then it is possible to relate the crash test injury outcome prediction to the observed real-world outcome directly. None of the accident impact conditions in the CCIS and GIDAS database matched a Euro NCAP case exactly. Therefore, it was invariably necessary to extrapolate findings from the test to the accident. The closer the accident conditions are to the crash test, the greater certainty can be afforded in the extrapolation of the predictions, and when the collision factors are similar to Euro NCAP than occupant factors may be expected to prevail. Therefore, this study is fundamentally limited by the way in which the accidents investigated match the impact conditions of the crash test. Generally, the accident cases had a lower crash severity (as estimated by the v or EES) than would have been expected in a Euro NCAP crash test. Therefore the analysis often included consideration of the how much better off the occupants should have been. Judging this is one of the key aspects and limitations of the work reported. It is interesting to discuss the expectation of better protection at lower speeds because in many cases this was not obvious from the accident cases. Cars were often deformed substantially in lower crash severity impacts and occupants still received thoracic injuries in those crashes. This raises the question as to whether an additional lower speed frontal impact test would be of benefit. Such a test could be used to assess sub-optimisation of vehicle structures and restraint systems at a lower impact speed, one at which many torso injuries can still be sustained. Age seemed to be an important factor. It was interesting to note that young occupants tended to receive only slight injuries in some quite severe accidents, including some of the accidents with unexpectedly high intrusion. Older persons can sustain severe injuries in accidents without compartment intrusion and at relatively low crash severity. This illustrates the effect of age on the ability of the human body to tolerate loads applied through modern restraint systems in frontal crashes. This also suggests that protection in offset frontal crashes is generally good, but that there is scope for further improvement, particularly for elderly occupants even in impacts with a markedly lower v than the Euro NCAP test. Page 25

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 6 Conclusions Cases from the CCIS and GIDAS in-depth accident studies were reviewed. For each case, a comparison was made between the thoracic injury outcome for the occupants predicted from the Euro NCAP crash test of that vehicle and the real-world accident. Thirty-four cases were identified where the impact conditions were similar to the Euro NCAP frontal impact crash test. Accidents with similar overlap and accident speed were included; therefore, other impact types known to cause torso injuries (for instance, with single vehicle impacts with narrow objects, such as trees) were not included in this study. Some interesting accident events were reviewed leading to some observations regarding the potential for torso-related safety interventions to improve the expected outcome in similar future accidents. In many cases several different factors came together to make the accident outcome different from that which might have been expected based on the crash test result. In some cases it was difficult to judge the degree to which each of the contributory factors influenced that outcome. As in the COVER accident analysis which formed the basis to this investigation, typical thorax injuries were: rib fractures, sternum fractures, lung contusions and clavicle fractures. This study reinforces these injuries as being priorities for investigation with the Thorax and THOMO projects. In most cases, the torso injuries were attributed to loading from the seat belt. In almost all cases in which a front passenger was present, the front passenger suffered the more severe injuries, despite being on the non-struck side of the vehicle and therefore likely to be subjected to a less severe acceleration pulse. Although mostly females were sitting on the passenger seat, it could not be identified if gender is the main factor or the restraint system is better optimised for the driver than for the front passenger. Thus, the restraint and protection system for the front passenger has potential for improvements. An updated dummy with enhanced risk curves could help to develop better protection for the front passenger. Injury risk curves for women should be developed. It was very interesting to note that young occupants tended to receive only slight injuries in some quite severe accidents. This illustrates the effect of age on the ability of the human body to tolerate loads applied through modern restraint systems in frontal crashes. It also suggests that protection in offset frontal crashes, for younger occupants, is good generally. Injury risk curves for elderly persons should be developed. The mechanism of lung injuries with no or relatively minor injuries is not clear and deserves further investigation. Page 27

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public 7 Recommendations When clavicle and rib fractures occur together, there was a high risk of mortality for the injured occupant. It is recommended that the correlation of this injury combination with other serious, or fatal, injuries and the mechanism responsible is investigated within the rest of the Thorax and THOMO projects. This may give an insight as to how this injury combination could be prevented. The mechanism of the non-penetrative lung injuries (e.g. compression, velocity of compression), both with and without skeletal injury, is not clear and it is recommended that this is investigated in Thorax WP2 and in THOMO. Consideration should be give to whether the current thorax instrumentation and injury risk functions provide an adequate prediction of the risk of this type of injury. WP2 WP3 Injury risk curves for women and elderly persons should be developed. It should be investigated to see if the dummy design (as well as injury risk functions) needs to be specific to an elderly occupant. This will be important for WP 3. An improved dummy thorax should predict injuries like rib fractures, sternum fractures, lung contusions and clavicle fractures. General recommendations: A female dummy should be used on the front passenger seat or at least specific risk curves should be used for front seat passenger injury risk evaluation. An additional lower speed frontal impact test may also be of benefit in assessing thoracic injury risk. Further investigations are needed to identify: If the protection level afforded by the front passenger restraint system is equivalent or lower than the driver restraint system. Why women are more severely injured than male occupants? Cases were specifically chosen to be similar in configuration to the Euro NCAP test. This meant that impacts with narrow objects were excluded. Further investigations should include car to obstacle collisions. Also, spinal column injury mechanisms need to be analysed further. From the GIDAS analysis, it was suggested that having a larger distance between the occupant and the steering wheel seemed to be beneficial. Therefore, to test under sub-optimal conditions, consideration should be given to testing with a closer seating position for the driver. Page 29

Acknowledgements A fundamental part of the analysis reported here was the review of both CCIS and GIDAS in-depth accident case reports. The authors are grateful to the bodies responsible for these databases for making the data available for analysis. The authors are also grateful to David Hynd and Paul Lemmen who carried out the reviews of this report. Page 30

References Carroll J A, (2009). COVER D5 Main summary report: Matrix of serious thorax injuries by occupant characteristics, impact conditions and restraint type and identification of the important injury mechanisms to be considered in THORAX and THOMO; consisting of one main summary report and three annexes. To be made available from the COVER internet site (www.biomechanics-coordination.eu). Delaney A, Newstead S, and Cameron M, (2006). Alternative weighting of NCAP series to improve the relationship to real-world crashes. SARAC II (Quality criteria for the safety assessment of cars based on real-world crashes) Report of sub-task 2.4. Available from the EC internet site: http://ec.europa.eu/transport/roadsafety_library/publications/sarac2_2.4.pdf. Euro NCAP (European New Car Assessment Programme, 2009a). Frontal impact testing protocol. Version 4.3, February 2009. Available from the Euro NCAP internet site: http://www.euroncap.com/files/euro-ncap-frontal-protocol-version-4.3---b8b12883-bd24-4b6e-b663-7898cda98e56.pdf. Euro NCAP (European New Car Assessment Programme, 2009b). Assessment protocol adult occupant protection. Version 5.0, May 2009. Available from the Euro NCAP internet site: http://www.euroncap.com/files/euro-ncap-assessment-protocol---aop---v5.0---0- fa324b84-8073-4e67-b739-3aa33329092b.pdf. European Commission (2006a). CARS 21 - A Competitive Automotive Regulatory System for the 21st century, Final Report. Brussels, Belgium: European Commission, Enterprise and Industry Directorate-General, Automotive Industry unit. European Commission, (2006b). Keep Europe moving. Sustainable mobility for our continent. Mid-term review of the European Commission s 2001 transport White Paper. Luxembourg: Office for Official Publications of the European Communities. Lenard J, Hurley B, and Thomas P, (1998). The accuracy of CRASH3 for calculating collision severity in modern European cars. Proceedings of the 16 th international technical conference on the Enhanced Safety of Vehicles (ESV), 31 May to 4 June 1998, Windsor, Ontario, Canada. Washington, D.C., U.S.A.: US Department of Transportation, National Highway Traffic Safety Administration (NHTSA; available from the NHTSA internet site: http://wwwnrd.nhtsa.dot.gov/departments/esv/16th/). Lowne, R. (1994). EEVC Working Group 11 report on the development of a frontal impact test procedure. Proceedings of the 14th international technical conference on Enhanced Safety of Vehicles (ESV), Munich, Germany, 23-26 May 1994. Washington D.C., U.S.A: US Department of Transportation, National Highway Traffic Safety Administration (NHTSA). Newstead S, Delaney A and Cameron M, (2006). Use of in-depth data in comparing Euro NCAP and real-world crash results. SARAC II (Quality criteria for the safety assessment of cars based on real-world crashes) Report of sub-task 2.3. Available from the EC internet site: http://ec.europa.eu/transport/road/publications/doc/sarac2_2.3.pdf. Richards D and Cuerden R, (2009). Road safety web publication 9: The relationship between speed and car driver injury severity. London: UK Department for Transport. Smith T L and Couper G, (2006). Assessment of advanced restraint systems: Final report (PPR 102), Published Project Report. Wokingham, Berkshire: Transport Research Laboratory (TRL). Page 31

Wykes N, Edwards MJ and Hobbs CA (1998). Compatibility requirements for cars in frontal and side impact. 16th International Technical Conference on the Enhanced Safety of Vehicles, Windsor, Ontario, Canada, 31 May to 4 June, 1998. US Department of Transportation, National Highway Traffic Safety Administration. Paper number 98-S3-O-04. Page 32

Appendix A. Data reporting template Table A-1 Data reporting template for recording details about each of the accident cases reviewed Collision / Accident Data Make: Model: Year: ETS (kph): dv (kph): EES (kph): Object hit: Overlap: Compartment Intrusion: Loading: Description: Photos of crashed vehicles Tested model EuroNCap Test: Picture of test damage Hand of drive Body type Year of publication Frontal Impact Test: Frontal impact driver Frontal impact passenger Rating: Score: Front: Side: Page 33

Driver (right side) Gender: Height (m): Personal data Age: Mass (kg): Pretensioner: Airbags: Seat Position: Seating / Restraint data Load limiter: Seat back angle: Injury Injury data Body Region AIS Injury mechanism Influenced by intrusion? Passenger (left side) Gender: Height (m): Personal data Age: Weight (kg): Pretensioner: Airbags: Seat Position: Seating / Restraint data Load limiter: Seat back angle: Injury Body Region Injury data AIS Injury mechanism Influenced by intrusion? Summary: Page 34

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public Appendix B. CCIS Case Overview Table B-1 Vehicle level data for CCIS cases: Case Make Model EES (kph) DV (kph) Year Overlap (%) Compartment Intrusion 1 Euro NCAP Score 2 Stars Euro NCAP Score 2 Total Euro NCAP Score 2 Front 13411 Vauxhall Astra - 27 2000 77 None 4 25 13 15369 Rover 25 33 33 2000 63 None 3 17 3 15420 Vauxhall Zafira 29 34 2002 47 Slight 3 22 6 15564 Vauxhall Corsa - 37 2000 74 None 3 18 7 17211 Audi A3 39 31 2004 37 None 4 29 12 17258 Ford Mondeo 32 28 2005 41 Slight 4 28 10 33679 Vauxhall Astra - 37 2002 45 Slight 4 25 13 34041 Vauxhall Astra 19 18 2001 50 None 4 25 13 35127 Honda Jazz 23 23 2005 47 None 4 25 11 35395 Rover 25 25 27 2004 30 Moderate 3 17 3 58117 Ford Mondeo 33 34 2005 44 Severe 4 28 10 58127 Ford Fiesta 47 41 2002 30 Slight 4 25 11 70660 Land Rover Freelander - 30 2000 40 Severe 3 20 6 70832 Nissan Micra 36 30 2000 53 None 2 15 4 75680 Vauxhall Vectra - 31 2000 50 Moderate 2.5 18 2 76042 BMW 525 42 33 2003 60 Severe 4 29 11 77041 Honda Jazz 26 30 2005 52 None 4 25 11 86082 Ford Mondeo 45 42 2001 60 Severe 4 28 10 86158 Vauxhall Zafira 34 38 2004 26 Severe 3 22 6 86168 Land Rover Freelander 33 36 2001 52 None 3 20 6 1 Slight 5 cm; Moderate 5 to 10 cm; Severe > 10 cm 2 Adult occupant protection Page 35

Table B-2 Personal level data for CCIS cases: Case Seat Sex Age Height (m) Weight (kg) Seatbelt load limiter Driver M 78 1.78 78 None Seat pretensioner Fitted, not activated Seating position length MAIS MAIS Torso Unknown 1 0 13411 Bilateral tension pnemothorax, Fitted, not FSP F 72 Unknown Unknown None Unknown 5 5T/A2 bilateral lung contusions, 2 R activated rib # (AIS2), seat-belt webbing 15369 Driver M 32 1.77 82 None Activated Mid 2 T1/A2 Burst # L1, non-contact 15420 Driver M 48 1.85 89 None Activated Mid 4 T4/A1 15564 17211 17258 33679 34041 35127 Driver F 53 Unknown Unknown None Activated Mid Unknown Unknown FSP F 17 Unknown Unknown None Activated Far back 5 (fatal) 0 Driver M 25 Unknown Unknown Activated Activated Mid Unknown Unknown FSP F 26 Unknown Unknown Fitted, not activated Activated Far back 2 T2/A0 Driver M 67 Unknown Unknown None Activated Mid 2 T2/A0 FSP M 69 Unknown Unknown None Activated Mid 0 0 Driver F 33 1.8 103 Activated Activated Unknown 1 T1/A1 FSP F 17 1.57 54 Activated Activated Unknown 2 T2/A1 Driver M 75 Unknown Unknown None Activated Far back 2 T2/A0 FSP F 73 Unknown Unknown None Activated Far back 3 0 Driver F 55 1.45 73 None Unknown activation Far back 2 T1/A1 FSP M 27 1.75 64 None Unknown activation Mid 1 T1/A1 35395 Driver F 75 Unknown Unknown Unknown activation Activated Far forward 4 T4/A0 Notes Bilateral lung contusions, upper 1/3rd sternum #, seatbelt webbing Object in vehicle caused head injuries AIS 4, 5, 5 Displaced L clavicle #, no rib #, seat-belt webbing # L clavicle, no rib #, seat-belt webbing Sternum #, no rib #, seat-belt webbing Displaced sternum #, no rib #, seat-belt webbing Bilateral pneumothorax, L8-9 rib #, T8-9 process #, R clavicle # Page 36

Case Seat Sex Age Height (m) Weight (kg) Seatbelt load limiter Seat pretensioner Seating position length Unknown Driver M 39 1.8 86 None Far back 2 T1/A0 activation 58117 FSP M 43 1.73 82 None Activated Far back 1 T1/A0 RSP (L) M 46 1.75 98 None None N/a 1 T1/A0 58127 Driver M 21 1.73 64 None Activated Mid 3 0 70660 Driver M 25 1.8 84 None Activated Unknown 2 T1/A0 Seat-belt webbing FSP F 24 1.7 57 None Activated Unknown 1 0 70832 Driver F 70 1.8 105 None Activated Far forward 2 T1/A1 Seat-belt webbing Driver M 45 Unknown Unknown None Activated Unknown 0 0 75680 Activation FSP M 69 1.83 98 None unknown Unknown 3 (fatal) T3/A2 76042 Driver M 73 Unknown Unknown None Activated Far back 2 T2/A0 FSP F 72 Unknown Unknown None Activated Far back 2 T1/A2 77041 Activation 3 (fatal Driver M 81 1.73 57 Activated Mid T3/A2 unknown 35 days) Activation FSP F 84 Unknown Unknown Activated Mid 2 T2/A0 unknown 86082 Driver M 30 Unknown Unknown None Activated Far back 3 T3/A1 86158 Driver M 57 Unknown Unknown None Not activated Mid 3 T3/A1 Driver M 62 Unknown Unknown None Activated Far back 1 0 86168 Activation FSP F 60 Unknown Unknown None unknown Far back 3 0 MAIS MAIS Torso Notes Page 37

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public Appendix C. UK CCIS Cases Page 39

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public Appendix D. GIDAS Cases Page 41

THORAX D1.1: Comparison between crash tests and real-world accident outcomes Public Appendix E. Body regions: GIDAS Case comments and explanations Table E-1 Body region abbreviations H Head F Face N Neck T Thorax A Abdomen C Column (spine) UE Upper extremity LE Lower extremity Nfs Unknown (Not further specified) BASt used Injury causing part instead of injury mechanism Table E-2 Seat back angle coding RLEHNE Seat back angle 3 Upright, not specified 4 Upright, very steep 5 Upright, midlle position 6 Upright, flat 7 Flat, backwards 8 Other 9 Unknown 10 Folded forwards Table E-3 Seat position height SITZHO Seat position height 1 Existing, not further specified 2 Non-existent 3 Upper position 4 Middle position 5 Lower position 8 Other 9 Unknown Table E-4 Seat position length SITZST Seat position length 2 Not adjustable 3 Front 4 Middle 5 Back 8 Other 9 Unknown Page 43

RLN = seat back recliner angle MBD = floor intersection backrest / seat Allok = backrest steering wheel SWSITZ = foot pedal seat MBD ALLOK RLN SWSITZ Page 44