SAFETY ENHANCED INNOVATIONS FOR OLDER ROAD USERS. EUROPEAN COMMISSION EIGHTH FRAMEWORK PROGRAMME HORIZON 2020 GA No

Transcription

1 SAFETY ENHANCED INNOVATIONS FOR OLDER ROAD USERS EUROPEAN COMMISSION EIGHTH FRAMEWORK PROGRAMME HORIZON 2020 GA No Deliverable No. 4.3 Deliverable Title Dissemination level Benefit Analysis Public Written by Thomas, Alan Hynd, David Kent, Jonathan Appleby, Joshua Zander, Oliver TRL TRL TRL TRL BASt Zander, Oliver BASt 30/05/2018 Checked by Burleigh, Mark Humanetics 30/05/2018 Fornells, Alba IDIADA 30/05/2018 Approved by Wisch, Marcus BASt 31/05/2018 Issue date 31/05/2018 The research leading to the results of this work has received funding from the European Community's Eighth Framework Program (Horizon2020) under grant agreement n

2 EXECUTIVE SUMMARY Elderly occupant safety has been the priority for this project: literature reviews and data gathering from RAIDS, GIDAS, STRADA, TraumaRegister DGU and TARN revealed that elderly occupants were sustaining serious injuries to the thorax in moderate-severity vehicle collisions. As such, current restraint systems were assessed because thorax loading led to the most significant amount of AIS 3+ injuries within this demographic. Front seat restraint systems have improved over the last few decades and as such the loading applied to the occupant s thorax has been greatly reduced, reducing the injury risk for younger and mid-aged occupants. Older occupants, however, have a lower biomechanical tolerance and the collision data shows that they can sustain serious and life-threatening thorax injuries (especially rib fractures) despite the advances in restraint system and vehicle design. To reduce the thorax loading for senior occupants, novel restraint system concepts were tested, with the aim of reducing the risk of serious or life-threatening chest injuries. The benefits of applying these new technologies to the entire fleet were also modelled within this report. The model focused on two restraint system designs in particular the Split Buckle and Criss-Cross seat-belts and how they would reduce European wide casualties, fatalities and the associated costs. The model calculated the benefit for car occupants of regulating each design in 2020 with mandatory fitting in Note that it is not expected that such a regulation would be implemented in this timescale; rather, this study uses this scenario as a way to explore the potential casualty savings and the societal cost reductions that could be delivered by these systems. The analysis showed that, in this scenario, the EU has the potential to prevent between 800 and 1,200 car occupant fatalities among the 65+ age group by implementing one of the seat-belt designs. There is also the potential to prevent between 6,500 and 10,500 serious occupant injuries and have an economic benefit in the range of billion, over the period SENIORS wanted to assess how changes to head testing tools and methods, the Thorax Injury Prediction Tool (TIPT), and the FlexPLI with Upper Body Mass (FlexPLI-UBM) test tool would affect the Euro NCAP test parameters. Would these changes have a benefit for pedestrians and cyclists? If there were changes, what costs or benefits, would they have on consumers and the manufacturing industry? Head impact tests, literature reviews and data assessment indicated that the current Euro NCAP head impact test area was not capturing cyclist head impacts or some taller pedestrians. Extending the upper boundary WAD (Wrap Around Distance) to 2500 mm, would address an additional proportion of pedestrians, and a higher proportion of cyclists, even without changing the head half-diameter exemption zones at the edges, or the head performance limits. However, for OEMs there would be little incentive for additional benefit in protecting VRUs from impacts with the A-pillars. The completely new test tool TIPT (Thorax Injury Prediction Tool) was found to offer the possibility of more informative and biofidelic ways of testing vehicles with higher Page 2 out of 50

3 BLE (Bonnet Leading Edges) especially concerning the growing SUV and MPV market share. Finally benefits of the FlexPLI-UBM were mainly qualitative such as improved biofidelity. Euro NCAP s existing lower leg tests would become more relevant to the real world as they are more biofidelic, particularly for femur and knee injury risk assessment. Page 3 out of 50

4 CONTENTS Executive summary Occupants Benefit Analysis Introduction Novel Restraint System Technologies Split Buckle Criss-Cross Summary of Results Adaptive Restraint Technologies Benefit Analysis Method Baseline Fatalities Fleet penetration of the technology Predicting KSI Benefits Predicting Economic Benefit Applying Discounting Break-even costs Limitations Summary Pedestrians and Cyclists Benefit Analysis Introduction Target Groups Head Impact Headform Neck Impactor (HNI) Windscreen Area Future Changes and Euro NCAP Proposed Combined VRU Testing Protocol Thorax Impactor (TIPT) FlexPLI with Upper Body Mass (FlexPLI-UBM) Tibia and Knee Benefits Femur Benefits Summary of qualitative benefits Head Impact Thorax Impactor (TIPT) FlexPLI with Upper Body Mass (UBM) Euro NCAP Box 3 Assessment Bibliography Acknowledgments Disclaimer Page 4 out of 50

5 1 OCCUPANTS BENEFIT ANALYSIS 1.1 INTRODUCTION Elderly occupant safety has been the priority for this project and as such current restraint systems were assessed. From the outset literature reviews and initial data gathering, see Deliverable 1.2 (Wisch, et al., 2017) and the related paper (Wisch, et al., 2017), revealed that many older occupants were sustaining serious or lifethreatening thorax injuries even in moderate severity collisions. According to STRADA, TraumaRegister DGU and TARN, thorax loading led to the most significant amount of AIS 3+ injuries within this demographic. The thorax was also generally the most severely injured or joint most severely injured body region. During a collision or under heavy braking, the thorax is restrained by the seat-belt to prevent impact with the steering wheel, instrument panel or the seat in front. Over the last few decades, front seat restraint systems have improved considerably and the loads applied to the thorax have been much reduced, which has greatly reduced the risk and incidence of serious thorax injury for younger occupants. However, for seniors this loading can still exceed their biomechanical tolerance, leading to large numbers of fractured ribs. Passenger vehicle seat-belts are currently designed with three anchorage points: above the left or right shoulder and two lap anchor points either side of the seat. As part of the SENIORS project, novel new restraint system concepts were tested and documented in Deliverable 4.2a (Wisch, et al., 2018) and shown to greatly reduce the risk of serious and or life-threatening chest injury (based on reducing the predicted number of fractured ribs). 1.2 NOVEL RESTRAINT SYSTEM TECHNOLOGIES Conventional passenger seat-belts are known as three-point restraint systems. They consist of a lap belt that restrains the pelvis and a diagonal/shoulder belt that lies over the rib cage and clavicle and which restrains the thorax. The lap belt often has a pretensioner, which removes the slack from the belt, which helps to ensure that the lap belt engages with the upper front part of the pelvis, minimising forward motion of the pelvis and thigh in a collision. The diagonal belt typically has a pretensioner and a load limiter. The latter sets the maximum force that the diagonal belt can apply to the upper thorax and shoulder. Improvements in pretensioners, airbags, collapsible steering columns etc. mean that the load limit has been reduced over the last few decades (typically to kn), which has helped to reduce the risk of injury to many occupants in a frontal collision. Nevertheless, this load level may be too high for some older occupants because rib cage strength reduces with increasing age. However, the load limit cannot be reduced indefinitely because a lower load limit allows greater forward motion of the head, which increases the risk of an injurious head contact with the interior of the vehicle (e.g. by bottoming-out the airbag and hitting the steering wheel). Several new restraint system designs were tested within the SENIORS project in order to understand how best to reduce the loading experienced by older occupants. Page 5 out of 50

6 Two of these systems were evaluated in detail: these are described in the following sections, which summarise the work described in D2.5a (Eggers, et al., 2018) Split Buckle During testing and simulation activities, it was observed that the most significant chest loading was often due to the lower part of the diagonal belt. This may in part be due to the buckle anchorage being pulled downwards and rearwards by the pretensioner, which helps the lap belt engage with and restrain the pelvis which is an important part of safely managing the collision forces on an occupant. In order to reduce the thorax compression due to the lower part of the diagonal belt, Autoliv developed a concept seat-belt that separates the buckle anchorage into two separate belt systems upon impact, then moving the diagonal belt lower anchorage forward to reduce thorax loading. In simulations and physical testing within the SENIORS project, the split buckle belt concept was implemented by using a separate lap belt and diagonal belt, i.e. with the lower diagonal belt anchorage already in its deployed position. A split buckle design that can deploy dynamically in a collision has been developed. In simulations with THUMS and simulation and testing with THOR it was observed that the greatest chest deflection for the driver was in the lower right part of the thorax. Therefore to reduce the load on the lower part of the THORAX and increase the load on the upper part, (the upper ribs, clavicle) the lower part of the chest and lap portions of the belt were separated and the load limiter force was increased. The lower attachment point of the diagonal belt was moved forward in the vehicle Criss-Cross The second design became known as a criss-cross belt system. This consists of a standard three-point lap and diagonal belt system plus a secondary (separate) diagonal belt across the in-board shoulder. The second belt has its own pretensioner, load limiter and lower anchorage, and the upper anchorage would likely be seat mounted (similar to the sort of outboard seat-mounted upper anchorage often used in convertible cars). The second diagonal belt meant that the load limiter on both belts could be reduced considerably, to 1 kn in the testing performed in SENIORS. Several considerations for the criss-cross belt should be noted: 1. The load limiter on the standard lap and diagonal belt would have to be switchable, e.g. a standard 3.5 kn load limit when only the standard belt is worn, switching to 1 kn if the secondary belt is engaged. 2. An interlock would be necessary to ensure that the vehicle could not be driven if only the secondary diagonal belt was worn (i.e. as a minimum the standard lap and diagonal is necessary for safety). 3. Regulation 16 requires seat-belts to be removable with a single press of a buckle latch; changes to the regulation may be required to allow this design of belt. Page 6 out of 50

7 As well as reducing thorax loading (see below), the criss-cross belt reduced head excursion. With a standard lap and diagonal belt the torso partially rotates about the upper diagonal belt, giving greater excursion of the unrestrained shoulder than the restrained shoulder which in turn increases head excursion. The second diagonal belt balances the shoulder restraints, thereby reducing head forward excursion and reducing the risk of hard head contacts, and therefore head injury, for a given restraint force. Indeed, it may be possible to reduce the load limits even further, below the 1 kn used in the SENIORS project. The criss-cross belt may also have benefits for occupants in far-side collisions. In this collision scenario, the car sustains a side impact on the side opposite the occupant (e.g. on the front passenger side when considering the driver). This configuration has a high risk of serious or fatal injury in part because the far-side occupant tends to slip out of the diagonal belt, leading to contact with the near-side occupant (if any) or the intruding struck-side structures of the car. In the near future, Euro NCAP intends to introduce an assessment of far-side occupant safety; as a result OEMs and Suppliers are developing centre air bags to contain the far-side occupant and prevent these injurious contacts. However, a second, in-board, seat-belt would likely be even more effective at controlling occupant motion in far-side impacts, especially for oblique farside impacts although this hasn t been assessed within the SENIORS project Summary of Results Assessments were conducted with THOR dummies to understand the differences in chest deflections at the two speeds of interest: a typical crash test speed of 56 km/h (e.g. UN Regulation 94) and the speed at which most 65yo+ AIS3+ injuries occur (35 km/h). Figure 1 depicts the peak resultant deflections from repeated tests at 35km/h and 56km/h (note: at the point of writing this report, not all tests had been performed and analysed yet; therefore, a reduced dataset was used) with the following restraint system configurations (also shown in Figure 2): Standard three-point seat-belt (3-p baseline); Three-point seat-belt with dual-retractor (3-p 2-ret); Criss-cross belt (3+2 2-ret Cr-Cr); Split-buckle system (Split Buckle). The criss-cross seat-belt system produced the greatest difference for the resultant maximum deflection (Rmax): at 35 km/h Rmax was reduced from 42 mm to 19 mm, while at 56 km/h the Rmax reduced from 54 mm to 33 mm (see Figure 1). In comparison, the split buckle system only reduced the deflection from 42 mm to 30 mm at 35 km/h, and from 54 mm to 38 mm at 56 km/h. These differences are due to the way in which the criss-cross belt system evenly distributed deflections at the upper measurement points and the very low load limiter force that could be used with this design of restraint system. Page 7 out of 50

8 Figure 1: Resultant deflections for each belt system and IR-TRACC position Figure 2: Restraint system configurations 3-p 2-ret (left); ret Cr-Cr (centre); and split buckle (right) Figure 3 depicts the injury risk percentages for both 35 km/h and 56 km/h, for the four restraint system options. The injury risks were calculated for the maximum resultant chest defection at any of the four THOR measurement locations (Rmax) and Principle Component (PC) Score, labelled as Principle Component Analysis (PCA), which is a new injury criterion developed by the University of Virginia in the US. Figure 3 shows the Rmax and PC Score injury risk predictions for AIS 3+ thorax injury, calculated using the UVa injury risk functions. Figure 3 shows the injury risk predictions for AIS 2+ injury, based on the SENIORS Rmax and PC Score injury risk functions, see D2.5b (Zander, et al., 2018). The injury risk results for Rmax are shown in Table 1 and for PCA in Table 2. It was possible to calculate the percentage of thorax injury risk for Rmax and PCA from the deflection tests. For 35 km/h: The criss-cross seat-belt reduced AIS3+ Rmax injury risk from 62% to 2% for the over 65s and comparably the Split Buckle system reduced the injury risk by 17%. At 56 km/h the AIS 3+ thorax injury risk dropped from 96% to 23% when the Criss-Cross seat-belt was in use and 43% for the split buckle system. Similar reductions were also observed for the PCA at 35 km/h and 56 km/h. Page 8 out of 50

9 Figure 3: UVa Rmax 65yo and PC Score 65yo Injury Risk AIS3+ (%) for the baseline and novel passive seat-belt technologies Table 1: Rmax AIS 2+ (calculated from SENIORS injury risk functions) and AIS 3+ (calculated from UVa injury risk functions) injury risks for baseline, split buckle and criss-cross restraint systems 35 km/h 56 km/h Rmax (mm) Prob. AIS 2+ Prob. AIS 3+ Rmax (mm) Prob. AIS 2+ Prob. AIS 3+ Baseline 42 74% 61% 54 99% 96% Split 30 21% 17% 38 55% 43% Criss-cross 19 2% 2% 33 32% 23% Table 2: PC Score AIS 3+ (calculated from UVa injury risk functions) injury risks for baseline, split buckle and criss-cross restraint systems 35 km/h 56 km/h PC Score Prob. AIS 3+ PC Score Prob. AIS 3+ Baseline 5,86 67% 7,59 97% Split 4,15 21% 4,05 19% Criss-cross 2,88 5% 3,78 15% 1.3 ADAPTIVE RESTRAINT TECHNOLOGIES Adaptive technologies, in terms of occupant safety, are technologies that are activated pre-emptively before a collision or can actively increase the safety of the occupant dependent upon the type of collision the vehicle is going to experience. These technologies may also relax the constraints, such as seat-belts, if the occupant is going to experience a lower severity crash. Having an adaptive system for activating the upper most limits for seat-belt load limiters only when necessary has been attempted before by the automotive industry. More recently, attention has been turned towards implementing variable/adaptive systems which change the load limits of the belts depending on whether or not the Automatic Emergency Braking (AEB) system believes the vehicle will have a crash and if so what the speed and likely mass of the collision partner is (e.g. small city Page 9 out of 50

10 car or large SUV). Therefore, the adaptive belt system would rely on the AEB s system of cameras and radar. If the system cannot ascertain what type of crash the vehicle is going to have, the load limits will remain unchanged (i.e. at the level of a standard, current restraint system). If the system can ascertain the type of crash the vehicle and occupant are going to have, it will vary the load limits of the seat-belts accordingly. If the current review of the General Safety Regulation (GSR) mandates the implementation of AEB, most of the system s cost will already be implemented in the vehicle. However, these restraint systems are not expected to be as effective as the novel restraint technologies proposed by Autoliv, because there will be a proportion of collisions where the system is not sufficiently confident in predicting the collision characteristics and therefore does not take the opportunity to tune the restraint system. 1.4 BENEFIT ANALYSIS Method The method applied here considers the implications on car occupant Killed or Seriously Injured (KSI) of implementing the split buckle and criss-cross designs into the vehicle fleet. Specifically, the model calculates separately the benefit for car occupants of regulating each design in In practice, this means that the design would be fitted in all new models from 2020 and all new cars from 2022, to give manufacturers sufficient time to respond to the requirements. The main steps for the model are outlined in Figure 4. Note that there are no current plans to implement regulation that would require or encourage these restraint systems and a reward in Euro NCAP may be a faster route to encouraging implementation in at least part of the fleet, see (Wisch, et al., 2018), but this analysis shows what the effect would be if regulation were implemented Estimate the number of car occupant KSIs between 2020 and 2030 in the 65+ age group, assuming current trends continue. Estimate in turn the potential reduction in car occupant KSIs achieved by implementation of a regulation for a split buckle or criss-cross seat belt system in 2020 Estimate in turn the casualty economic benefit from if either of the regulations are mandated and implemented Figure 4: Steps for the model Page 10 out of 50

11 1.4.2 Baseline Fatalities Prior to estimating how many lives could be saved through implementing the seatbelt designs, it is important to understand how many car occupant KSIs there will likely be in the 65+ age group in the EU28 between now and 2030 if vehicle safety standards develop as planned. These assume that the designs will not be implemented by any manufacturers if there is no regulation enforced. To make these predictions, the number of 65+ car occupant fatalities and serious injuries have been extrapolated forwards assuming an exponential trend 1. Figure 5 presents the estimated number of 65+ car occupant KSIs in the EU28 from , as well as the breakdown into fatalities and serious injuries. One limitation of the numbers presented here is that data is not available for every country across the whole of this period. Any missing data has been assumed to be equal to data from neighbouring years (e.g. the annual number of KSIs in Cyprus from is assumed to be equal to the number in 2005). In addition, three members of the EU28 (Estonia, Finland up to 2016, and Italy) do not distinguish between serious injuries and slight injuries, but instead present an overall number of injured car occupants, which does not include fatalities. In these cases, the number of serious injuries has been estimated by calculating the proportion of serious injuries relative to slight injuries across the rest of the EU28 and then applying this proportion to the data from these three countries. The figure shows that the trend in KSIs has been slightly upward over the past few years, following a general downward trend from The majority of this increase has come in the form of more serious injuries, although there has also been a slight increase in fatalities. Figure 5: Number of 65+ car occupant KSIs in the EU28, (Source: CARE) 1 An exponential trend was chosen since this was shown to fit the data well and ensures that the fatality rate does not reach zero (which would be unrealistic) for any country during the timescale of interest for this analysis ( ). Page 11 out of 50

12 Figure 6 presents the predicted number of KSIs among the 65+ age group up to 2030, along with the breakdown into fatalities and serious injuries. Whilst in general, an exponential trend has been applied to the data up to 2016, there are a couple of cases where this was not suitable. Firstly, in Lithuania, the limited amount of data that was available ( ) meant that applying an exponential trend to the fatality numbers gave predictions that were not sensible. Therefore, a constant trend was assumed instead, meaning that the number of fatalities each year from was predicted to remain at the 2016 value. Secondly, in Finland, there is a jump in the estimated number of serious injuries from 2015 to This is because the data up to 2015 has been estimated by applying a ratio to the number of slight injuries, as explained above, whereas actual data is available for Again, this means that when an exponential trend was applied to all the data, the predictions that were generated were not realistic. Therefore, a similar constant trend was assumed on the number of serious injuries from Figure 6: Predicted number of car occupant KSIs in the EU28, The figure shows that the number of fatalities is expected to decrease slightly between 2017 and 2030, whilst the number of serious injuries is expected to be fairly stable, with a slight increase towards the end of the period. By 2030, it is anticipated that there will be around 16,000 KSI car occupant casualties in the 65+ age group across the EU28. Of these, just under 2,000 will be fatalities, and just over 14,000 will be serious injuries. It is worth noting that there is some uncertainty in the predicted trend of serious injuries, due to the recent upward trend since 2013, which is in contrast to the more gradual downward trend seen prior to that. If the more recent trend were to continue, then this would lead to a greater number of serious injuries than predicted here. Page 12 out of 50

13 1.4.3 Fleet penetration of the technology Before the potential reduction in KSIs from implementing either of the new seat-belt designs could be calculated, it was necessary to predict how quickly the designs would penetrate the vehicle fleet. In order to do this, the first step involves predicting the total number of registered cars in the fleet, as well as the number of new car registrations, up to Figure 7 presents the total car registrations in the EU28 from in the top panel, and the number of new car registrations from in the bottom panel. The data has come from a mixture of two sources, ACEA (European Automobile Manufacturers Association) and OICA (Organisation des Internationale des Constructeurs d Automobiles). Data from 2005 onwards comes from OICA, whereas data from before then comes from ACEA. The data on new car registrations from has been adjusted slightly upwards, to reflect the fact that for the years where data is available from both sources, the numbers from OICA are marginally higher than those from ACEA. Figure 7: Total number of registered cars (top panel) (OICA, 2017) and number of new car registrations (bottom panel) (OICA, 2017) in the EU28, (where data available) Page 13 out of 50

14 The figure shows that the total number of registered cars in EU28 has been gradually increasing over recent years, with a 12% increase between 2005 and 2015 (from 229 million to 256 million). The trend in new car registrations has been less stable, with a steady increase from , followed a decline from (likely to have been caused mainly by the global financial crisis in 2008) and then a recovery from The number of new registrations in 2017 was just under 15.2 million, around the same as the number in The data in Figure 7 has been extrapolated forward to 2030, assuming a linear trend. The two sets of predictions are presented in Figure 8. The solid lines represent actual data, and the dotted lines represent the predictions. Figure 8: Predicted number of total registered cars and new car registrations in the EU28, Page 14 out of 50

15 The figure shows that the total number of registered cars is expected to continue to increase gradually, reaching 291 million by The number of new car registrations is expected to remain relatively stable overall, although it is likely that the trend will be less stable than the dotted line suggests, depending on economic conditions across Europe between now and The extrapolations in Figure 8 were used to predict the rate at which either of the seat-belt designs would penetrate the fleet. This involved making the following assumptions, under the basis that the regulation would be implemented for all new models from 2020, and all new cars from 2022: Up to and including 2020, no new cars would be fitted with the design In 2021, 50% of all new cars would be fitted with the design From 2022 onwards, all new cars would be fitted with the design Up until 2030, there would be no cars leaving the fleet with the design fitted The last assumption is based on the fact that the average age of cars in Europe is more than 10 years (ACEA fact sheet 2 ). Figure 9 presents the estimated percentage of cars that would be expected to have the new design fitted up to Figure 9: Predicted percentage of all cars in EU28 countries fitted with the new seat-belt design, if the regulation comes into effect in 2020 The figure shows that under the above assumptions, an estimated 48% of all cars in the EU28 would have the new seat-belt design fitted by 2030, if a regulation were to be implemented in Page 15 out of 50

16 1.4.4 Predicting KSI Benefits To estimate how many of the predicted KSIs in Figure 6 could be saved from by implementing one of the seat-belt designs, the following pieces of information are required: An estimate of the target population (i.e. the KSIs which could be targeted if the seat-belt design was implemented) An estimate of how effective each design is (i.e. in the target population, the proportion of KSIs which would be prevented) As presented in Section 1.2.3, the effectiveness of the designs has been evaluated at both 35 km/h and 56 km/h. Therefore, two separate target populations have been created, to which the two sets of effectiveness estimates can be applied. The first step was to estimate the proportion of fatalities and serious injuries to which the estimates could be applied. This was done using data from RAIDS (the Road Accident In Depth Studies collision investigation study in the UK), in particular the information on the total change in velocity (Delta-V) in the accident. The estimates were applied as follows: The 35 km/h estimate was applied to KSIs where the total delta-v was between km/h The 56 km/h estimate was applied to KSIs where the total delta-v was between km/h It was assumed that the seat-belt designs would have no effect on KSIs where the total delta-v was either less than 20 km/h or more than 56 km/h. The relevant proportions are presented in Table 3. Table 3: Proportion of 65+ KSIs in the target Delta-V ranges km/h delta-v km/h delta-v Fatalities 39% 29% Serious Injuries 47% 32% The two seat-belt designs being considered in this study are both aimed at preventing injury to the thorax area of the body. Therefore, the second step in obtaining the target population was to estimate the proportion of KSI injuries among the 65+ age group which affect the thorax. This was done using data from GIDAS (German In-Depth Accident Study), iglad (initiative for harmonisation of Global indepth traffic Accident Data), RAIDS and STRADA (Swedish Traffic Accident Data Acquisition) (see D1.2). Out of all injuries of severity AIS2 or greater (the equivalent classification of KSI) suffered by the 65+ age group, 41% were injuries to the thorax. This percentage was applied to the proportions in Table 3 to give the target populations to which the effectiveness estimates should be applied. These are presented in Table 4. Page 16 out of 50

17 Table 4: Proportion of 65+ KSIs to apply effectiveness estimates to 35 km/h 56 km/h Fatalities 16% 12% Serious Injuries 20% 13% Once the target populations had been estimated, the next step was to estimate the effectiveness of each seat-belt design in preventing these KSIs. Section presents two sets of effectiveness estimates, one for AIS2+ injuries (calculated from SENIORS injury risk functions) and AIS 3+ injuries (calculated from UVa injury risk functions). There are separate estimates for the split buckle and criss-cross restraint systems. These estimates have been used to calculate two predictions of the number of car occupant KSIs that could be prevented in the EU28 from by implementing either seat-belt design. Figure 10 presents the predicted number of KSIs that could be prevented by implementing the split buckle design, with results presented each year from inclusive and broken down into fatalities and serious injuries. Page 17 out of 50

18 Figure 10: Predicted number of 65+ car occupant KSIs saved in the EU28 from by implemeting the split buckle regulation in 2020, applying the AIS2+ (top panel) and AIS3+ (bottom panel) effectiveness estimates The figure shows that the saving of lives/injuries is expected to increase at a linear rate. This matches the linear increase that is also expected to be seen in the fleet fitment rate of the design, as presented in Figure 9. By 2030, the design is predicted to prevent around KSIs a year among the 65+ age group in the EU28, including around 150 fatalities. Figure 11 presents the equivalent set of results for the criss-cross design. Page 18 out of 50

19 Figure 11: : Predicted number of 65+ car occupant KSIs saved in the EU28 by implemeting the criss-cross regulation in 2020, applying the AIS2+ (top panel) and AIS3+ (bottom panel) effectiveness estimates, The figure shows that a similar upward linear trend in the saving of lives/injuries is predicted if this design were implemented in By 2030, around KSIs are expected to be prevented, including around 200 fatalities. Table 5 presents the total number of KSIs among the 65+ age group that are expected to be saved from in the EU28 by implementing the split buckle regulation. Again, results are presented separately for each set of effectiveness estimates, and broken down into fatalities and serious injuries. Table 6 presents the equivalent set of results for the criss-cross design. Page 19 out of 50

20 Table 5: Predicted number of 65+ car occupant KSIs saved in the EU28 from by implementing the split buckle regulation in 2020 Applying AIS2+ effectiveness Applying AIS3+ effectiveness Fatalities Serious Injuries 6,572 7,738 KSIs 7,343 8,645 Table 6: Predicted number of 65+ car occupant KSIs saved in the EU28 from by implementing the criss-cross regulation in 2020 Applying AIS2+ effectiveness Applying AIS3+ effectiveness Fatalities 1,089 1,224 Serious Injuries 9,259 10,432 KSIs 10,348 11,656 Comparing the two tables, as well as the above figures, shows that the potential benefit from implementing the criss-cross design is expected to be greater than what could be achieved by the split buckle design, regardless of which effectiveness estimate is used Predicting Economic Benefit In line with the method used by Wallbank et al. (2016), the valuation of a statistical life (VSL) method has been used to quantify the economic benefit of the casualty reductions predicted due to the introduction of the seat-belt regulations. VSL methods are based on a willingness to pay to avoid injury and are related to GDP per capita. These figures can be compared cross-nationally and are readily computable from health burden data. However, since the evidence on willingness-topay is varied, a range of estimates are produced for each country. Bhalla et al. (2013) reviewed a number of relevant VSL studies and showed that the economic loss of death due to a traffic collision has been estimated to be between 70 and units of GDP per capita. In addition, they estimate the economic loss of a serious injury to be 17 units of GDP per capita. Whilst their study is focused on the Latin America region, the same methodology can be applied to other regions. Table 7 applies these estimates to forecasts of GDP per capita in 2018, to produce estimates of the cost of a death and serious injury in each of the EU28. The GDP forecasts have been calculated by extrapolating forward data from Eurostat, which runs from , assuming a linear trend. For the purposes of this analysis, the mid estimate of the two values presented for the economic loss of one death in each country was used to estimate the benefit from the reduction in fatalities. As outlined in Wallbank et al.(2016), there is a lack of agreement in what should be included in these VSL estimates, in particular in relation to essentially unquantifiable measures such as the value of pain, grief and suffering. Therefore, they should be interpreted with some care. Page 20 out of 50

21 Table 7: Economic loss of fatalities and serious injuries Forecast GDP per capita (2018) (2018 ) Source: Eurostat Economic loss of one death due to traffic collision (2018, thousands) Economic loss of one serious injury due to traffic collision (2018, thousands) Austria 42,295 2,961 5, Belgium 38,579 2,701 5, Bulgaria 7, , Croatia 11, , Cyprus 21,183 1,483 2, Czech Republic 17,486 1,224 2, Denmark 49,941 3,496 6, Estonia 17,288 1,210 2, Finland 40,297 2,821 5, France 34,138 2,390 4, Germany 39,803 2,786 5, Greece 14,991 1,049 2, Hungary 11, , Ireland 55,356 3,875 7, Italy 27,573 1,930 3, Latvia 13, , Lithuania 14,791 1,035 2, Luxembourg 95,382 6,677 13, Malta 23,659 1,656 3, Netherlands 42,064 2,944 5, Poland 12, , Portugal 17,911 1,254 2, Romania 9, , Slovakia 16,323 1,143 2, Slovenia 19,980 1,399 2, Spain 23,512 1,646 3, Sweden 49,176 3,442 6, United Kingdom 35,905 2,513 4, Page 21 out of 50

22 1.4.6 Applying Discounting In this analysis a discount rate is applied to the benefits (presented in today s terms). The application of a discount rate reflects that benefits further into future are valued less highly than present benefits. For private sector project evaluation it is usual to use the return that an investor could get in the open market as a discount rate (the Opportunity Cost of Capital). Projects are then evaluated more worthwhile if they can offer a better return than would be gained from investing elsewhere. For projects with a social benefit a lower rate is normally used to reflect the difference in people s expectations of returns on social projects versus private ones. A social discount rate was chosen with reference to Seidl et al. (2018), who presented social discount rates for European countries. Lower social discount rates are used for projects with longer time-horizons. Because we are evaluating the benefits from 2020 to 2030 a social discount rate of 4.25%, for projects with a 10-year time horizon, has been used. While these rates were calculated a number of years ago, in other parts of the world social discount rates have stayed relatively stable over the same period. For example the social discount rate used by the UK treasury in 2003 was 3.5%, and the same rate has been used in subsequent updates up to the most recent (HM Treasury, 2003; HM Treasury, 2013). Figure 12 presents the estimated economic benefit that could be achieved from among 65+ car occupants by implementing the split buckle. This has been calculated by first applying the VSL estimates from Table 7 to the KSI savings in Table 5, and then accounting for discounting. The overall benefit has also broken down into the benefit attributable to fatality and serious injury savings respectively. Page 22 out of 50

23 Figure 12: Estimated economic benefit for 65+ car occupants due to implementation of split buckle regulation in 2020, applying the AIS2+ (top panel) and AIS3+ (bottom panel) effectiveness estimates, The figure shows that the level of economic benefit is expected to increase at a fairly linear rate, with a slight tailing off towards 2030, by which point the overall level of benefit would reach somewhere between million. Approximately one third of the benefit is attributed to fatality savings, and around two thirds to serious injury savings. Figure 13 presents the equivalent results for the criss-cross design. Page 23 out of 50

24 Figure 13: Estimated economic benefit for 65+ car occupants due to implementation of criss-cross regulation in 2020, applying the AIS2+ (top panel) and AIS3+ (bottom panel) effectiveness estimates, The figure shows a similar upward trend in benefit, with a slight tailing off towards 2030, by which point it is estimated that the overall level of benefit would reach somewhere between billion. Again, approximately one third of the benefit is attributed to fatality savings, and around two thirds to serious injury savings. Table 8 presents the total level of economic benefit that would be expected to be seen from as a result of implementing the split buckle regulation in Again, the results are broken down into the benefit attributable to fatality and serious injury savings respectively, and is presented separately for the two sets of effectiveness estimates. Table 9 presents the equivalent set of results for the crisscross design. Page 24 out of 50

25 Table 8: Estimated economic benefit for 65+ car occupants due to implementation of split buckle regulation in 2020, Applying AIS2+ effectiveness Applying AIS3+ effectiveness Fatalities prevented Serious injuries prevented 6,572 7,738 Discounted economic benefit due 1, , to fatalities, (2018, millions) Discounted economic benefit due 2, , to serious injuries, (2018, millions) Total discounted economic benefit, (2018, millions) 4, , Table 9: Estimated economic benefit for 65+ car occupants due to implementation of criss-cross regulation in 2020, Applying AIS2+ effectiveness Applying AIS3+ effectiveness Fatalities prevented 1,089 1,224 Serious injuries prevented 9,259 10,432 Discounted economic benefit 2, , due to fatalities, (2018, millions) Discounted economic benefit 3, , due to serious injuries, (2018, millions) Discounted economic benefit due to fatalities, (2018, millions) 6, , Comparing the two tables, as well as the above figures, shows that the potential economic benefit from implementing the criss-cross design is expected to be greater than what could be achieved by the split buckle design, regardless of which effectiveness estimate is used. However, the benefit from implementing either regulation is expected to be considerable, with a minimum economic benefit of 4.7 billion from implementing the split buckle design, and a maximum of 8.1 billion from implementing the criss-cross design. Page 25 out of 50

26 1.4.7 Break-even costs This analysis has focused on predicting the benefits to 65+ car occupants of implementing either one of the seat-belt designs. However, it is also important to consider how much it would cost to implement a regulation, so that a decision can be made as to whether it would be economically viable to do so. Whilst an estimate of the cost of fitting the seat-belt designs to a vehicle is not available, what has been calculated is a break-even cost. That is, the fitment cost at which the total benefits would equal the total costs. This has been calculated for both designs being considered, with separate values for the two sets of effectiveness estimates. The calculation consists of dividing the estimated total level of benefit, after accounting for discounting, and then dividing by the total number of cars which would be fitted with the design from , using predictions of the number of new cars (Figure 8, bottom panel) and the fitment rate (Figure 9). The results are presented in Table 10. Table 10: Estimated break-even cost of each seat-belt design, applying either the AIS2+ or AIS3+ effectiveness estimates (2018 ) Applying AIS2+ effectiveness Applying AIS3+ effectiveness Split buckle Criss-cross The table shows that the break-even cost for the criss-cross design is higher than for the split buckle design, which follows from the fact that the criss-cross design is expected to generate a larger benefit (see Table 8 and Table 9) Limitations The modelling presented in this report has used various input values and assumptions to predict the effects of introducing either of two new seat-belt designs across the EU. Predictions of the future are by definition inherently subject to a degree of uncertainty. This study has used input values based on historical trends, which cannot completely account for extreme changes in circumstances. The following important limitations of the simulation model and the input value estimates should be taken into account when interpreting the results. The predictions are based on the assumption that current trends in road traffic fatalities continue as they have done in recent years. For some countries these trends are based on 10+ years of data (e.g. Austria, Belgium, France), but for others the data is limited (e.g. Lithuania, Malta, Slovakia). Furthermore, three of the countries (Estonia, Finland, Italy) do not split their injury data into serious and slight injuries, so the number of serious injuries has had to be estimated, and historically there is a lack of agreement across the EU as to the definition of a serious injury. Therefore, the authors cannot comment on how accurate these figures are. For Page 26 out of 50

27 example, if underreporting levels have changed, or there are inherent biases in the data reported, then these trends may not hold true into the future. In addition, this approach cannot capture any potential disruptions that might occur in the mobility market in the future, such as autonomous driving radically changing the collision landscape, mobility as a service reducing private car ownership, or a severe economic crisis reducing new vehicle uptake. Disruptions are highly uncertain and impossible to predict as to when, if, and to what extent they will happen and hence their impact could not be captured in the models. If any such changes were to occur then the effect of implementing new seat-belt designs in the EU could be vastly different than predicted here. Finally, the effectiveness estimates are subject to a degree of uncertainty. Both seatbelt designs are currently prototypes, rather than a final product. Therefore, if they were to be launched into the market, their design and subsequent effectiveness would be subject to change, either positively or negatively. Also, the estimation of fatalities from AIS 2+ and AIS 3+ risk levels is subject to uncertainty. However, this is balanced to some extent by the pessimistic assumption that there would be no benefit to occupants under 65 years old. The results of the benefit analysis should be interpreted with these limitations in mind and understood as an evidence-based prediction of the potential benefit of the new designs, if historic trends continue within a range of expected uncertainty. Furthermore, with the implementation of any regulation then appropriate measures will be necessary to enforce compliance. This research has not investigated in detail whether substantive changes would be required to legislative vehicle safety enforcement in any of the 28 EU countries to obtain the benefits predicted. However, this need should be recognised alongside the predictions presented Summary This analysis has shown that the EU has the potential to prevent somewhere between 800 and 1,200 car occupant fatalities among the 65+ age group by implementing one of the seat-belt designs discussed in this study. There is also the potential to prevent between 6,500 and 10,500 serious injuries. This would result in an economic benefit somewhere in the range of billion, over the period This would clearly be a substantial economic boost to the EU, and just as importantly would be a positive step towards improving road safety for older drivers. Page 27 out of 50

28 2 PEDESTRIANS AND CYCLISTS BENEFIT ANALYSIS 2.1 INTRODUCTION SENIORS aimed to reduce the number of senior road user fatalities and severe injuries, including pedestrians and cyclists. The partners and contributors to SENIORS wanted to assess how vulnerable road users would benefit from changes to head testing tools and methods, the introduction of an Injury Prediction Tool for the thorax (TIPT), and the modification of the FlexPLI, introducing an Upper Body Mass. A further point of interest was to what extent new or revised test tools would affect vehicle safety design in order to fulfil the updated requirements of legislation and consumer programmes such as Euro NCAP. If there were changes, what costs would they have on consumers and the manufacturing industry? Target Groups The European Commission keep records of road fatalities in the EU-28 and in 2015 pedestrians made up 21% of all road fatalities (5435), while cyclists were 7.8% (2043) of all road fatalities. These figures did not distinguish between the pedestrians and cyclists impacted by the front of the vehicle or those impacted by another section of the vehicle. Since 2013 these figures have also remained relatively constant (plateaued). However, they remain the two largest VRU groups and therefore required the most attention within SENIORS, especially due to the changing vehicle fleet make up (increase in the use of SUVs and 4x4s). As reported in SENIORS Deliverable 4.1b (Zander & Hynd, 2018), in-depth accident studies of the German GIDAS and the Swedish STRADA database identified the head, the thorax and the lower extremities as the most affected body regions in vehicle to pedestrian and cyclist accidents. Furthermore, injuries to these body regions become more severe for the elderly. Thus, it was already clear at an early point in time of the project that measures mainly need to focus on these body regions. It was unknown at the time of the report how an increased average bonnet leading edge height would affect the pedestrian and cyclist injury rate/probability 3. The European Commission have been publishing collision and fatality statistics for pedestrians (European Road Safety Observatory, 2017a) and cyclists (European Road Safety Observatory, 2017b) since They found that from 2005 to 2008, 73,600 patients from nine of the EU-28 member states were taken to hospital due to road traffic collisions (of a non-fatal nature), of which 23,568 suffered injuries severe enough to be admitted. Nearly 40% of all the patients involved in road traffic 3 In an analysis for the European Parliament of the likely impact of higher and heavier cars on casualty outcomes, Cuerden et al. noted that the present review could not identify real world accident data type evidence showing that modern European SUVs or MPVs were more aggressive towards vulnerable road users than smaller passenger cars. This is in stark contrast with international findings which show a higher risk for these vehicles. This might be explained by more pedestrian-friendly vehicle design, based on European regulatory requirements, or perhaps more likely by an insufficient European evidence base and research to be able to identify this trend. However, it should be noted that the kinematics are different for pedestrians struck by SUVs compared to passenger cars because of the higher bonnet leading edge. This causes a more severe impact to the pedestrian s femur and pelvis area with less upper body and head rotation leading to different injury patterns in which some injuries can be more severe (see section below for more details). Page 28 out of 50

29 collisions taken to hospital were cyclists, compared to 10% for pedestrians. Of the patients admitted; cyclists made up 23% (of those taken to hospital) and 45% were pedestrians (of those taken to hospital). Of the accident casualties roughly 2,300 pedestrians and 6,800 cyclists were admitted to hospitals due to their injuries. Over the same three-year period it is possible to assume that the total collisions across Europe would be nearly six times these values, when making the assumption that the population scales across the rest of Europe linearly. Zander et al. (2013), were able to assess the percentage of head impacts as a factor of wrap around distance for both pedestrians and cyclists. Figure 14, from Zander et al. (2013) highlighted that Euro NCAP s test constraint of WAD 2100 mm covered 80% of pedestrian head strikes but only 65% of cyclist head impacts. Increasing the WAD to 2500 mm will increase the target groups, for Germany, by 38% for cyclists and 24% for pedestrians. The HNI did not show the expected benefits given the two dimensional frontends and needs to be further investigated. Improved kinematics alongside the introduction of a new injury criterion (based on rotational acceleration) should be investigated in a follow-up project. However, a modification of the test procedure by mainly longitudinally extending the impact area rearwards up to WAD 2500 mm and slightly modified impact angles will address a huge number of cyclists as second big group of vulnerable road users. Figure 14: Percentage of cyclist and pedestrian front of vehicle head impacts recodrded, plotted against WAD (Wrap Around Distance) (Zander, et al. 2013) 2.2 HEAD IMPACT Headform Neck Impactor (HNI) Research led to GIDAS, UDV (German Insurers Accident Research) and the EC funded project APROSYS (Advanced Protective Systems), whereby collision data was collated for 2015: 78,176 bicyclists injured in road traffic accidents in Germany (of which 383 fatally and 14,230 seriously); pedestrians were involved in 537 fatalities Page 29 out of 50

30 and 7,792 severe injuries. Following work conducted by Fredriksson & Rosén (2012) and Zander et al. (2013), the data set became restricted to only those which suffered an AIS 3+ head impact with the front of the incident vehicle, which itself was travelling no more than 40 km/h - altering pedestrian and cyclist target groups to 1032 and 1699 accidents respectively. The Euro NCAP headform testing procedure (EuroNCAP, 2017) requires a standardised headform. SENIORS investigated the effect of adding a neck element with a mass representing a portion of the upper thorax/shoulder as suggested in Deliverable 3 from the APROSYS project. The objective was to inhibit the rolling of the headform immediately following the first contact with the vehicle due to the friction and/or the offset of the path of the headform s centre of gravity from the contact point of the skin. Although these effects are present in the real world, the headform is free to rotate during contact and the following rebound to a greater extent than the human head because of its connection to neck and torso. This lacks biofidelity as the relative motion of the cranial bones and brain and vascular tissues cannot be inferred, even with rotational accelerometers fitted to the current headform. However, the expected improvements in kinematics were not fully realised in the human body and HNI modelling in SENIORS. This is likely because the simulations used two-dimensional bonnets, with none of the rounded surfaces/curvatures that occur in vehicle tests, which meant that the rotation of the impactors was low and close to the real world impact kinematics for that kind of vehicle frontend already. While it was not possible to fully identify any HNI benefits within the scope of this project, it could be a topic for further research. A benefit in terms of improved impact kinematics would need to be further investigated during simulations and tests on modern vehicle frontends with rounded shapes and curvatures. A second point that should be taken into account in further studies are new injury criteria, based on rotational acceleration, which indeed would also be significantly better reflected when based on a more realistic kinematics of the impactor during subsystem tests Windscreen Area Within SENIORS, extensive work on the types, body segments and severities of injuries to pedestrians has been conducted, particularly the hospital data supplementing the results from analyses of in-depth crash databases [Deliverable 1.2]. The relation between these injury data and the pedestrian contact locations was not part of the SENIORS project, because earlier work by project partners and others (e.g. Otte(1989), Mizuno & Kajzer (2000) and Yao et al. (2008)) had already identified the highest risk contact locations on the vehicle and, importantly, the high proportion of injuries resulting from impact with the ground, typically 17 to 36% (Liers (2009), Liers (2011), Otte & Pohlemann (2001) and Zhang et al. (2008)), since it is possible to receive a skull fracture (AIS 3+) from a simple fall from standing height simply due to loss of balance (Hill, et al., 2003). Gehring and Zander (2013) studied the different performance of laminated safety glass panes and polycarbonate panes as alternative windscreen glazing technology and performed comparative tests with laminated safety glass panes and polycarbonate panes. In this context, a series of drop tests and vehicle tests with the adult head impactor according to Regulation (EC) 631/2009 and drop tests with the Page 30 out of 50

31 phantom head impactor according to UN Regulation No. 43 were carried out. The aim of the test series was to study the injury risk for vulnerable road users, especially pedestrians, in case of being impacted by a motor vehicle in a way described within the European Regulations (EC) 78/2009 and (EC) 631/2009. Furthermore, the applicability of the phantom head drop test described in UN Regulation No. 43 for plastic glazing was investigated. Altogether, 30 drop tests, thereof 18 with the adult head impactor and 12 with the phantom head impactor, and 49 vehicle tests with the adult head impactor were carried out on panes of laminated safety glass (VSG), polycarbonate (PC) and laminated polycarbonate (LPC). The influence of parameters such as the particular material properties, test point locations, fixations, ambient conditions (temperature and impact angle) was investigated in detail. In general, higher values of the Head Injury Criterion (HIC) were observed in tests on polycarbonate glazing. As the HIC is the current criterion for the assessment of head injury risk, polycarbonate glazing has to be seen as more injurious in terms of vulnerable road user protection. In addition, the significantly higher rebound of the head observed in tests with polycarbonate glazing is suspected to lead to higher neck loads and may also cause higher injury risks in secondary impacts of vulnerable road users. However, as in all tests with PC glazing no damage of the panes was observed, the risk of skin cut injuries may be expected to be reduced significantly. The performance of the windscreen area is not relevant for vehicle type approval according to current legislation (UN-R 127, Regulation (EC) No. 78/2009. However, the authors recommended that pedestrian protection being considered for plastic windscreens should have at least the same level of protection as glass windscreens. Since laminated windscreen technology was developed in previous decades mainly to protect the head of an unrestraint occupant when impacting the inside of the windscreen in a collision, an inherent energy-absorbing deformation is offered. This does not translate fully into the same protection for a pedestrian head impact at an equivalent velocity for several reasons: Optimisation of the glazing for occupants has led to asymmetric specifications for the inner and outer layers of glass. For instance, the inner layer can be thinner, and tempered to produce small fragments on rupture to reduce the size and depth of any lacerations of the occupant, and the outer layer can be made to form larger pieces to retain the butyl middle layer more completely, absorbing more energy and reducing the chance of full penetration by the head. Use of seat-belts increased the speeds at which such occupant protection was useful, but the advent of driver and passenger airbags made reliable support for the bag important once more using this type of windscreen. Curvature of the windscreen in two or three dimensions is believed to provide a degree of increasing resistance to deformation from the outside due to the compressive stresses as the windscreen is loaded against the constraints of the frame, reducing the susceptibility to fracture in the earliest phase of impact until local deformations develop into a concave, tensile action. For occupants the glazing is subject to tensile loads throughout the impact phases, leading to more immediate fracture and consequently lower stiffness. Page 31 out of 50

32 Nevertheless, in general, at the majority of the speed range for which vehicle to pedestrian impacts should be survivable in respect of other body regions and head impact is to the glass only, HIC levels are comparatively low although they may still be highly injurious. A proportion of head to windscreen impacts involve partial contact with the scuttle or the A-pillars, or close enough to them to limit the energy absorption capacity of the glass, or meet almost immediate support against to the deforming windscreen from the top of the instrument panel, or the interior trim covering the A-pillar and side-impact curtain airbag. It should be noted that some types of Head- Up Displays now reaching the market may add a further interaction with head impacts. These areas have been the focus of much research and the first production wave of windscreen airbags arrived with the Volvo V40 in 2012, followed by the Land Rover Discovery Sport (2015) and Subaru Impreza (2016). These all provide protection over the lower windscreen and up the A-pillars, leaving the central area upwards clear. This is important because, unlike vehicle-to-vehicle or vehicle-to-infrastructure injury collisions in which damage can make on going control impossible, after impacts with cyclists and pedestrians the vehicle is almost always still travelling at speed and under control, so the driver needs to be able to see sufficiently to steer and brake. Future vehicles may be able to brake to a halt automatically (whilst staying in their lane) after a collision with a pedestrian or cyclist, so this limitation may become less important over time. The central windscreen area is currently untested and may have good performance on some vehicles; for others, improvements would require changes to the windscreen material, shape or construction. This is currently being evaluated as part of proposed changes to the General Safety Regulation. A limitation with the existing external airbags is that a small but noteworthy proportion (around 4.2% of a pedestrian data set restricted to 40 km/h or less (Mizuno & Kajzer, 2000)) of AIS 2+ head injuries arise from impacts with the top of the windscreen and the windscreen header rail, i.e. the roof structure at the top of the windscreen. This is particularly the case with cyclists and against vehicles with shorter fronts. There are therefore suggestions (e.g. Watson (2010), Fredriksson (2011), Zander & Hamacher (2017)) to extend the test area from the existing WAD 2100 mm ((EuroNCAP, 2017), (European Commission, 2009)) to 2300 or 2500 mm (Zander & Hamacher, 2017). Fredriksson et al. (2010) report a study of GIDAS pedestrian cases where 26% of head impacts were between WAD 2100 and WAD 2300 mm, with 14% beyond that. Of the 39 AIS3+ cases involving head impact to windscreen and frame area, and based on vehicle speed as a (poor) surrogate for head impact speed, six fell within the regime of regulatory head impact tests (35 km/h). A further four were between 35 km/h and the 40 km/h of Euro NCAP tests, while eight of these were also at or less than the WAD 2100 mm limit of both types of adult headform tests, and two were unknown. A further 10 higher speed cases also fell at or below WAD 2100 mm, and nine were between 2100 and 2300 mm WAD. Of the area not currently protected by the three production airbag systems, Fredriksson et al. (2010) had three AIS 3+ cases striking the windscreen header, in the outboard quarter, and one on the joint with the passenger-side A-pillar, as shown in Table 11. Page 32 out of 50

33 Table 11: Head impacts to upper windscreen area categorised by Fredriksson et al. (2010) Impact Head Injury type Vehicle Speed km/h Head Impact WAD cm 32 pillar at top Soft Tissue & joint Skeletal 37 windscreen Soft Tissue header 38 windscreen Soft Tissue & header Skeletal 39 windscreen header Soft Tissue It should be noted that WAD in GIDAS is measured following the impact points of the VRU from leg impact to bumper to head impact, while the regulatory and Euro NCAP protocols are measured parallel to the vehicle s vertical central plane. Therefore, any angled GIDAS measurements will be recorded as a greater WAD than the same impact location in those laboratory protocols. In parallel, for cyclists a similar situation exists in which 80% of cyclist head impacts could be covered within a WAD of 2300 mm(zander & Hamacher, 2017) also supported by Watson (2010) for large family cars and MPVs. Fredriksson & Rosén (2012) report a second study of GIDAS data on the same methodology as the pedestrian study mentioned above, published without the case-by-case vehicle speed and WAD details. Here the windscreen header was impacted in 26% of AIS 3+ cases (n=134) compared with 8% of the equivalent pedestrians, and were all in the outboard thirds of the header, with a further case nearer the centreline on the glass but affected by the header structure. Of the 46 cases with measured data, the average cyclist head impact WAD was 2260 mm compared with 1930 mm for the pedestrians. In D4.1b it was recommended to test up to WAD 2500, independently from the contacted vehicle parts. In our simulations we found the following: In particular the higher statures (50 th male and 95 th male) for Limousine, Compact Car and Sports Car often impact the vehicle front with the head beyond WAD 2100 mm, up to WAD 2500 mm and more. This observation is also valid for Van and SUV when focusing on the perpendicular corner impact. Furthermore, the wrap around distances for the corner impact are higher than for the centred perpendicular impact and the turning scenario. A modified test procedure would therefore suggest a rearward extension of the head test area until WAD 2500 mm except for OneBox vehicles where the current limitation of WAD 2100 mm could be sufficient It should also be noted that the centre of the header rail and the adjacent windscreen is the prime location for effective ADAS sensors such as cameras and LIDAR for pedestrian, cyclist and other hazard detection. These systems while very light in themselves normally require a stable support and therefore provide a stiffer area for external head impacts. So far, this has not figured noticeably in literature from field Page 33 out of 50

34 data as it is a relatively recent development. Although not assessed within SENIORS, it is conceivable that the benefits accruing from the ADAS would justify less stringent head impact requirements in the sensor mounting zones. The windscreen header plays a crucial structural role in vehicle integrity in normal use, and of course in occupant protection in front, side, front under-run, and rollover collisions, and therefore cannot itself be made less stiff. In current and proposed Euro NCAP protocols, the tests avoid damage to the impactor by giving the windscreen test area a boundary of half the impactor diameter between the path of the impactor centreline and the A-pillars and (in cases where it would be within the 2100 / 2300 / 2500 mm WAD) the windscreen header too. Apart from impactor damage, addressing the risk in this area would require additional airbag, or airbag coverage, or the addition of an energy-absorbing surface with some 60 mm of deformability. As this is also a critical area for efficient aerodynamics, the effect on energy consumption and emissions would have to be traded off. Creating a thicker A pillar would also have an effect on visibility and the driver may not see a pedestrian or cyclist soon enough. The vehicles most affected will of course be those with a short, low bonnet, or MPVs and flat-fronted van-style vehicles. Disregarding these somewhat significant difficulties, extending the WAD by 200 mm to 2300 mm and including the A-pillars could address an additional 30% of AIS 3+ pedestrian injuries, particularly at impacts up to 40 km/h, depending on where these WAD map onto different vehicle sizes and types (Fredriksson et al., 2010). Some of the corresponding cyclist AIS 3+ incidents would also be mitigated, of which some will be cascaded from fatal and some serious injuries will be replaced by minor ones for both types of road users. However, for commensurate benefit to cyclists, inclusion of the windscreen header area lying at less than WAD 2500 mm may be required. In assessing the costs and benefits, this could be an area where self-protection would have a greater value, and cyclist helmets would protect in other impacts, even with no motor vehicle involved. Fredriksson & Rosén (2012) encountered just four GIDAS cases out of 2327 injury cases where the cyclist was wearing a helmet and none sustained a severe head injury, which is too small in that study to prove effectiveness, but puts down a marker for further research. It is likely that airbags would be the default solution for the A-pillars because this is the only working system currently available. One class of collisions unlikely to be mitigated by airbag systems or deployable bonnets are those scenarios where a vehicle enters a cyclist s path at a high angle such that the rider s trajectory crosses the front of the car from the side and will not be detected by forward sensors. In this collision configuration, the cyclists head risks lateral contact with the A-pillars, or header; most studies of deployable bonnets or airbags have excluded such cases from the populations studied and they therefore cannot be included in this benefit evaluation. In other words, for this collision scenario AEB (Autonomous Emergency Braking) or AES (Autonomous Emergency Steering) won t be effective and deployable protection may not deploy, so head contact areas need to have inherently good protection. Page 34 out of 50

35 2.2.3 Future Changes and Euro NCAP Proposed Combined VRU Testing Protocol The benefit from isolated passive measures (including deployable pedestrian protection systems) is expected to be complemented by AEB systems. Cyclists are not considered sufficiently in legislation or in consumer ratings tests such as Euro NCAP, despite being the second largest VRU (Vulnerable Road User) group. Contributions to SENIORS, also involved investigating field accident data and EU Reg 78/2009 (which specifies the monitoring procedures for testing the pedestrian safety features of vehicles; EU Reg 631/2009 establishes how to conduct the testing). From this regulation an attempt to was made to produce a test procedure to include cyclists and modify the current headform boundary conditions of the Euro NCAP pedestrian tests. The test procedure developed was based on accident analysis for Germany and simulations with vehicle and bicycle models. Cyclists heights were also considered, ranging from a 6 year old child to a 95%-male. Reconstruction of real accidents defined simulation parameters and could therefore be validated in advance. Real world testing was conducted with a Polar II dummy to correlate results with simulations. 2.3 THORAX IMPACTOR (TIPT) It had been found from the accident studies in D1.2 that at the AIS 3+ level for cyclists head, thorax and lower extremity injuries are the priority, and for pedestrians, head and thorax injuries are the priority at AIS 2 level. Consequently, as described in D3.2b (Zander, et al., 2018) and D4.1b (Zander & Hynd, 2018), SENIORS has made a first attempt to assess a thorax impactor that potentially could be used to guide vehicle design to reduce thorax injury risk. The approach used the thorax segment of a mildly modified ES-2 ATD and assessed its suitability as a tool to measure injury values for the thorax. This required constructing pushers for the laboratories to propel the thorax onto the vehicle at the required angles and velocities as determined from Human Body Modelling simulations of pedestrians and cyclists in D2.5b (Zander, et al., 2018) corresponding to different vehicle categories. The modifications included managing the ES-2 arm, which in its intended side impact work is kept away from interfering with the impact of the vehicle s side against the thorax. However, for SENIORS it was deemed necessary to keep the arm in between the thorax and the vehicle surface, and this was achieved by a new jacket with the sleeve attached to the chest material to prevent it rotating away for repeatability. In the course of a series of tests with the TIPT, as reported in SENIORS D4.2b, the arm and shoulder were completely removed in order to limit rotation of the impactor and obtain more representative rib deflections. Besides, rubber layers were placed on top of the ribcage for more homogeneous loadings along the entire impactor. Injury risk curves developed by Lowne et al. (n.d.) were used to derive the injury thresholds for this application, and then create the points scheme proposed for future Euro NCAP assessments (as shown in Table 12). Page 35 out of 50

36 Table 12: TIPT thresholds (maximum rib intrusion) Colour (points) Maximum rib intrusion Covering (human injury risk) Green (1) < 28mm 5% AIS3 (67 YO) Yellow(0.75) 28mm 35mm 20% AIS3+ (45 YO) Orange (0.5) 35mm 40 mm 30% AIS3+ (45 YO) Brown (0.25) 40mm 44mm 40% AIS3+ (45 YO) Red (0) 44mm 50% AIS3+ (45 YO) The complexity of the rib cage in terms of sensitivity to orientation in mechanical behaviour and instrumentation, plus the difficulty of controlling its trajectory, exacerbates the representation of the human pedestrian thorax in a collision. The influence of the head mass, arm dynamics, abdomen, pelvic masses and lower limbs are eliminated for consistency and repeatability in the TIPT, while emulating the resulting thoracic trajectory understood to be most likely to replicate the more serious injury mechanisms in the target population. It is a characteristic of some vehicle frontal collisions with pedestrians struck laterally that the elbow of the leading arm makes contact with the bonnet, cowl or windscreen before the thorax or head, loading the upper arm and the shoulder and so modifying the thorax and/or head trajectory and impact velocity of thorax and/or head. This is also seen in many vehicle/dummy reconstructions and HBM simulations (Thomas, 2017). Therefore, it is expected that eliminating this possibility for the TIPT should represent a realistic worst case condition for rib loading, with the restrained or removed arm providing a more concentrated loading than otherwise with a larger portion of the rib cage against a larger surface area of the vehicle. The findings of the TIPT testing (D4.2b) were, however, that the injury levels were not exceeding the lower performance limits so no design changes and therefore no benefits would be expected if this tool were to be employed within a pedestrian test protocol for legislation. However, several of the results exceeded the proposed upper performance limit and fell within a sliding scale as proposed in Table 12 for Euro NCAP assessment. Thus, for a good overall performance of the vehicles, design modification would be expected. The altogether low rib intrusions that were observed in the majority of tests were because the direction of impact was not parallel to the potentiometers and ES-2 was not optimised for oblique impacts. During impacts with steeper impact angles or perpendicular orientation of the TIPT the ribs were loaded with higher deflections. Alternative sensor systems (e.g. 2D IR-Traccs) would be worth investigating in the future., The testing method in general was practicable and demonstrated the feasibility of this type of testing. In view of the evident need to address the situation defined by (Wisch, et al., 2017), where thorax injuries were established as a priority, there is a requirement for further Page 36 out of 50

37 investigation using HBM of the way in which the thorax might be loaded in the relevant cases by the masses of the abdomen, pelvic region, head, and non-struckside arm, which could result in higher rib loading, and greater deformation of the vehicle and therefore greater possibility of bottoming out on harder underlying components. This strategy would be consistent with, and an extension of, adopting the FlexPLI-UBM in SENIORS. Note that if representing such a load transfer is found to be necessary, it may not be possible to use existing launch systems, due to the increased mass at least for an impactor representing and average sized adult. In this case, FE simulations with Human Body Models may provide a better way forward. An additional purpose of the TIPT that was discussed in D3.1b (Lemmen, et al., 2016) is assessment of the protection of external road users on vehicles with a high Bonnet leading Edge (BLE) such as some SUVs, pickup trucks and HGVs. This requirement is currently not presented in Euro NCAP or legislation for use in the final assessment. It should address some of the injuries noted in the accident research, but a thorough investigation of the stringency of the TIPT injury risk limits for elderly pedestrians, compared with the existing upper legform impactor (ULI) limits on the design of the vehicle, would be required as the TIPT levels could be higher or lower than the current design targets, and therefore might not encourage any changes. The use of the FlexPLI-UBM representing a 50% adult male has limitations in respect of the UBM interaction with very high BLE on some vehicles, and a TIPT test covering the areas likely to impact a 5% adult female or even a 10yo child may be found to be beneficial. This may be a suitable avenue for further research effort, ensuring perpendicular loading to the TIPT. 2.4 FLEXPLI WITH UPPER BODY MASS (FLEXPLI-UBM) The existing FlexPLI is an improvement over the EEVC legform impactor, as has been recognised in regulations (UN R127.01) and consumer information testing (Euro NCAP since 2014). However, it is acknowledged as having limitations in representing the kinematics (and hence injury mechanisms) sufficiently for the femur section, which was a basis for the FlexPLI-UBM development in SENIORS as described in D4.1b Tibia and Knee Benefits D3.2b section compared simulations of a rigid and flexibly mounted UBM on the FlexPLI, the standard FlexPLI and a human body model (HBM), impacting on the SAE buck and on real vehicles, at the centre and at the bumper beam ends where the outer surface is swept back in plan view. D4.2b compared physical tests with the final FlexPLI-UBM and FE simulations with HBM against the generic sedan buck and actual vehicles, confirming significantly improved kinematics and time history correlations due to the introduction of the upper body mass. The differences in measured peak tibia loadings of the FlexPLI-UBM compared with the existing FlexPLI are of limited nature and addressed by transfer functions between the readings. On the basis of tibia bending moments as the sole criteria for regulatory and consumer testing, the major benefit is shown in the improved kinematics. Page 37 out of 50

38 Currently, VRU injuries to the upper leg are only represented during upper legform tests to WAD 775 mm in Euro NCAP; however, the monitoring phase of Regulation (EC) No. 78/2009 has ceased. UN R127 only foresees the usage of the upper legform impactor during high bumper tests, but not to the bonnet leading edge nor to any fixed WAD. Introducing the FlexPLI-UBM offers the possibility of appropriately assessing injuries to the femur, in particular during tests against vehicles with high leading edges and thus hard contact of the femur to injurious structures. This scenario is currently not reflected at all by legislators nor appropriately by Euro NCAP, where at this point in time only a monitoring test with the child headform impactor to the bonnet leading edge is foreseen in very rare cases, while the tests with FlexPLI and with the upper legform impactor do not have the BLE in their focus. This will change with the introduction of the FlexPLI-UBM. Figure 15: Current UL high bumper test procedure in Euro NCAP and Regulation (EC) No. 78/2009 and UN-R 127 Figure 16: Current UL to WAD 775 test procedure in Euro NCAP Figure 17: Current FlexPLI to bumper test procedure in Euro NCAP and UN-R Page 38 out of 50

39 Oliver Zander, BASt Figure 18: Current Headform to BLE test procedure in Euro NCAP Euro NCAP, Regulation (EC) No. 78/2009, (2009) and UN Reg 127, specify the upper legform test to high-bumper vehicles. These specifications determine at which height the upper leg form test (instead of the lower legform test with FlexPLI) must be conducted against the bumper). Currently if the lower bumper height is below 425 mm the lower leg form impact test must be conducted on the bumper. When the lower bumper height exceeds 500 mm the ULI must be impacted against the bumper, allowing for horizontal ULI tests if the design of the vehicle would dictate that the tibia would not impact any part of the vehicle. The test will be conducted in line with the established ULI test but it may be noted that flatter fronted vehicles will have a more difficult time deflecting the force of the impact but may reduce the bending moment experienced by the ULI femur. According to the UN GTR the most common vehicles to exceed this lower bumper height are SUVs and MPVs, due to increased ride heights compared with sedans and sports vehicles. Within the Euro NCAP testing SUVs and MPVs may be scoring well on the pelvis impact tests due to these test changes and the ULI s are impacting bumpers that are designed to absorb energy, rather than a stiffer bonnet leading edge. These regulations and testing conducted by Euro NCAP may already be having an impact on vehicle designs. Looking forward at the growing market of SUVs, Euro NCAP (2014) altered one of its testing conditions to not neglect injury potential caused by bonnet leading edges. WAD 1000 mm to WAD 930 mm is an untested region on all vehicles because tests with the UL to WAD 775 cover an area up to approximately WAD 930 and the headform test zone starts at WAD In case of the BLE being located between WAD 930 and WAD 1000, this is an area expected to have stiff/hard structures capable of seriously injuring a VRU. Headform testing is usually conducted past WAD 1000 mm; however, SUVs and MPVs can have a bonnet leading edge which exceeds WAD 930 mm and places WAD 930 and 1000 mm into a region which could be considered seriously harmful to VRUs. Testing of this region is consequently conducted with a head form and not an ULI. These test changes may already be having an effect on Euro NCAP scores for the vehicle fleet and OEMs may be altering designs to incorporate head impacts of VRUs to the front of their vehicles. Benefits to an improved FlexPLI with an upper body mass also included improved kinematics and biofidelity. The addition of an upper body mass increased the rotational inertia about the z-axis of the FlexPLI. Therefore when interacting with Page 39 out of 50

40 vehicles with chamfered corners the FlexPLI-UBM would wrap around the front of the vehicle in a more biofidelic way, starting its rebound at a more realistic point in time, before bouncing off the front of the vehicle. This was more representative of human impacts with vehicle fronts. SENIORS Deliverable D4.2b compares, amongst other things, tests with the FlexPLI Baseline, the FlexPLI-UBM and HBM simulations with THUMSv4 against actual vehicles. The vehicle centreline and both ends of the bumper beam were chosen as impact locations. In these tests, the FlexPLI-UBM demonstrated its superior behaviour not only at vehicle centreline, but in particular in angled impacts at the end of the beam. This was demonstrated by the improved kinematics, reducing extraordinary high to a more humanlike impactor rotation and was underlined by the produced time histories better correlating to those of the human model Femur Benefits In SENIORS D1.2 it was found in the hospital data that AIS 3 open femur shaft fractures were the fifth most common pedestrian injury of the top five (n=55) recorded in one hospital study for those aged 65+ (see Table 13). Whereas among the age group all five most frequent pedestrian injuries (n=205) are AIS 1 except AIS 2 hand fractures, the older group sustained more AIS 2 injuries. At a detail level this underlines typical findings of vulnerable road user collision studies. Table 13: Top five injuries recorded at hospitals, for both the mid-aged and SENIORS target group Replacement of the current EEVC Upper Legform Impactor (ULI) with the FlexPLI - UBM would accommodate more biofidelic injury representation and limits. It would also address concerns over the possibility of the ULI prompting design solutions that may be unhelpful in some real world collisions, while accepting that the relative reduction in femur injuries in cases involving modern vehicles has much to do with design changes for aerodynamics for fuel efficiency and emissions reduction independently of the ULI test requirements. These design changes are, however, not irreversible or universal, as the greater numbers of SUVs, pickups and HGVs demonstrates. Similarly, a class of vehicles may emerge much more specifically intended for urban use at lower speeds (less aerodynamically stringent) and increasingly interacting with pedestrian-dominated areas. Furthermore, just as safety- Page 40 out of 50

41 related ADAS systems require top-centre windscreen hardware, they also require stable mountings in the bumper and grille area for radar and other sensors; this same area is also one of the most convenient locations for charging sockets on plug-in electric vehicles, and has to resist repeated positive insertion and withdrawal loads. Together these create stiff areas which may increasingly conflict with leg protection. Matsui et al. (1998) analysed 54 cases from in the JARI database involving known contact locations on the BLE that resulted in femoral and/or pelvic and/or lumbar vertebra injuries. In 16 cases there were sufficient data which covered a range of 25 to 50 km/h with AIS 0 to AIS 3, age 12 to 78 years, male and female. The same vehicle types were tested using the ULI test procedure to validate the test conditions and injury criteria of the EEVC upper legform impact test. Only five out of 16 accident cases (31%) were in good agreement. All of the AIS 0 cases were from vehicles producing forces and bending moments above the 4 kn and 220 Nm limits and covered a wider range of test results than all the AIS 1-3 cases. At the very least the Matsui et al. (1998) work demonstrated that the test procedure and tool can be unrepresentative of the very wide variation of human impacts on the BLE, and they called for reconsideration of the injury criteria and test conditions for the EEVC ULI. Over the succeeding developments of the Euro NCAP protocols, the EU and subsequent adoptions into Japan, UN Reg.127 and GTR-9, and Korean KMVSS requirements, the injury criteria were changed while the fundamental test set-up and procedure and ULI remained much the same. However, the UL test to the BLE was never part of UN-R 127 and UN-GTR 9. The existing FlexPLI is an improvement over the EEVC legform impactor, as has been recognised in regulations (UN R127) and consumer information testing, (Euro NCAP since 2014). However, it is acknowledged as having limitations in representing the kinematics (and hence injury mechanisms) sufficiently for the femur section, which was a basis for the FlexPLI-UBM development in SENIORS as described in D4.1. The tendency of the upper section of the FlexPLI to rebound with the lower section if contacting the vehicle near the knee, or after it makes separate contact higher up, indicates that the loading on the femur is different to behaviour in a human, PMHS or dummy impact when the pelvic masses oblige the upper leg to continue rotation in the opposite sense. As a result, accelerations of the human femur will be lower and bending moments will be sustained longer at the knee to rotate and lift the lower leg as the torso continues over the bonnet. The duration and timing when using the FlexPLI-UBM are greatly improved in most cases. While the emphasis has been on vehicles with higher BLEs such as SUVs, these effects are still present on the majority of vehicle shapes. The use to date of the ULI has motivated OEMs to consider the thigh and pelvis in vehicle design with variable outcomes, but the uncertainty over its value in the real world, low contribution to ratings in Euro NCAP, and lack of focus in regulations has made it less of a priority than other requirements. There is therefore a clear qualitative advantage in principle in designing a vehicle for real-world upper leg pedestrian protection with the FlexPLI-UBM over the use of the combination of the FlexPLI and ULI tests. Quantifying this projected benefit is an exercise in relating the potential vehicle design changes to the reduction in transducer outputs such changes Page 41 out of 50

42 could deliver, and thence to the associated injury levels via the risk curves and applying the change of risk to the collision population. D3.2b compared simulations of a rigid and flexibly mounted UBM on the FlexPLI, the standard FlexPLI and HBM, impacting on the SAE buck and on vehicles, at the centre and at the bumper beam ends. In particular the femur sections as well as the knee show better correlating waveform characteristics in terms of shape, timing and maximum loadings. The time histories and maximum values of the FlexPLI-UBM during impacts at the end of the beam do not display the extraordinary impactor rotation about its long axis very often observed at such positions when using the FlexPLI Baseline: the inertia of the upper body mass emulates the resistance to rotation about the Z-axis of the leg in PMHS and HBM, which is not represented in FlexPLI-Baseline. Furthermore, impacting the vehicle at the end of beam at 30 to its axis is meant to further reduce rotation of the impactor but does not contribute to a further improved correlation. With the FlexPLI-UBM there is now a tool which allows to test at the end of the bumper beams with confidence. SENIORS is aware that FlexPLI displays reported high, non-biofidelic ACL and PCL deflections (D3.2, section 2.2 Design Considerations and Implementation, also Annex 1) in some situations and considered modification of the outer medial condyle. 2.5 SUMMARY OF QUALITATIVE BENEFITS Head Impact SENIORS evaluated a head-neck impactor (HNI) and found that it is unlikely to drive vehicle design changes that reduce headform rotation on impact, at least for the current level of development of the HNI. Nevertheless, improvements to head injury protection can still be made with existing tools through modifying the test zone and impact angles to address a bigger proportion of pedestrian and cyclist collisions. Extending the Euro NCAP adult head impact test area to WAD 2500 would address an additional proportion of pedestrians, and importantly cyclists, even without changing the head half-diameter exemption zones at the edges, or the head performance limits. Assessment of the windscreen area would further reduce the risk of head injury. It would also curb the tendency to position ADAS sensors in the upper centre of the windscreen without provision for minimising pedestrian harm Thorax Impactor (TIPT) Thorax injuries to pedestrians are a priority, and particularly for the elderly. SENIORS has established an approach to meet this imperative, but there are several areas of refinement needed to be able to quantify the scale of improvement and the associated benefits. TIPT would not only benefit the pedestrians and cyclists that suffer chest impacts on the bonnet, it offers the possibility of more informative and biofidelic ways of testing the liability of the increasing volume of vehicles with high BLEs (SUVs, pickups) and potentially new emerging types of vehicles that do not need aerodynamics as a priority. Page 42 out of 50

43 In either case there is a need for further research to define the most appropriate tool for a dedicated pedestrian thorax injury risk assessment. For the time being, the TIPT has proven to be an appropriate tool for tests forward of the bonnet leading edge, with an upright position of the impactor, perpendicular to the direction of travel FlexPLI with Upper Body Mass (UBM) The proposed FlexPLI-UBM gives benefits to three main areas of pedestrian lower extremity protection: 1. Testing becomes more realistic, with the kinematics of impact (Figure 19) and corresponding time histories (Figure 20) much closer to those of the human being. In particular, femur and knee injuries can be more appropriately assessed with a more biofidelic, flexible and adaptable test tool with an injury assessment ability superior to that of the rigid upper legform impactor during tests to WAD 775. Additionally as demonstrated in Figure 20, also the tibia area significantly benefits from more appropriate interaction with the vehicle front with the UBM application. Figure 19: Kinematics of THUMSv4, FlexPLI-UBM and FlexPLI Baseline during impact against Van/MPV centreline Page 43 out of 50

44 Figure 20: Time histories of THUMSv4 (black), FlexPLI-UBM (blue) and FlexPLI Baseline (green) for MCL and adjacent femur and tibia sections during impact against Van/MPC centreline 2. Angled vehicle areas at or adjacent to the end of the bumper beam can be more appropriately assessed, eliminating excessive and unrealistic rotation of the FlexPLI Baseline impactor around its Z-axis (see Figure 21). Figure 21: Kinematics of THUMSv4, FlexPLI-UBM and FlexPLI Baseline during impact against Van/MPV end of beam, LHS 3. Vehicles with very high frontend geometries can be tested and assessed with a more humanlike test tool. As reported in SENIORS Deliverable D4.2b, tests with the FlexPLI Baseline against the SUV representative resulted in very low femur bending moments at both vehicle centreline (see Figure 22) and the end of beam. Also MCL of the FlexPLI Baseline was far away from the real world in most cases. This was not the case when the femur was correctly loaded due to the applied upper body mass. The bending moments experienced an increase of and beyond 200%, getting much closer to the femur loadings of the human model. Also for MCL and in most cases tibia the results were significantly improved, compare Figure 22: Page 44 out of 50

45 Figure 22: Time histories of THUMSv4 (black), FlexPLI-UBM (blue) and FlexPLI Baseline (green) for MCL and adjacent femur and tibia sections during impact against SUV centreline The existing lower leg tests would become more relevant to the real world as they are more biofidelic, particularly femur and knee injury measurements. By reducing rotation about the z-axis (through the addition of the UBM), testing around the edge of the bumper beam can be performed with much more confidence of an appropriate assessment of the safety features of the vehicle. Current upper leg tests direct design solutions in isolation, with the outcome that they may have no benefit, or even be counterproductive. FlexPLI-UBM provides a comprehensive and consistent alternative, combining the injury mechanisms of femur, knee and tibia in one single test. Replacing the separate ULI test with the FlexPLI-UBM will likely reduce design, development and testing costs Euro NCAP Box 3 Assessment SENIORS Deliverable D4.2b proposes the replacement of the ULI to WAD 775 mm test by the introduction of the FlexPLI-UBM with femur injury assessment ability and the introduction of a test with thorax injury prediction tool. It also proposes a change in weighting of those points allocated to passive pedestrian protection without changing the balancing between passive and active safety in Box 3. Nonetheless, a change in weighting of the available 36 passive points is proposed, equally allocating 12 points each to the most affected body regions head, thorax and lower extremities. According to the new assessment procedures, the weighting for femur injuries therein is lowered from a current maximum of 6 points (UL to WAD 775 mm test) to a maximum of 3.6 points (equal to a 30% weighting within the FlexPLI-UBM procedure), the weighting for MCL is increased from a current maximum of 3 points (half of the points of current FlexPLI tests) to a maximum of 4.8 points and the weighting for tibia experiences an increase from a current maximum of 3 points (half of the points of current FlexPLI test) to a maximum of 3.6 points. Thus, femur is allocated a decrease of 40% of the points (6 to 3.6), the knee is allocated an increase of 60% (3 to 4.8) and tibia an increase of 20% of the points (3 to 3.6). A comparison of the performance of the impacted test locations for both tests with the UL to WAD 775 mm and tests with the FlexPLI-UBM shows under the currently proposed threshold values an increase in requirements for the lower leg in case of the compact car, not passing the proposed tibia thresholds at vehicle centreline, while passing the tests with the FlexPLI. MCL, as stated before, experienced a Page 45 out of 50