EUROPEAN COMMISSION DG RTD

Transcription

1 EUROPEAN COMMISSION DG RTD SEVENTH FRAMEWORK PROGRAMME THEME 7 TRANSPORT - SST SST : Human physical and behavioural components GA No THORAX Thoracic injury assessment for improved vehicle safety Deliverable No. THORAX D3.3 Deliverable Title Dummy validation report Dissemination level ] Written By Jolyon Carroll (TRL), Johan Davidsson 3 April 213 (Chalmers), Luis Martinez (UPM), Antonio García (UPM), Philippe Vezin (IFSTTAR), Andre Eggers (BASt) Checked by Bernard Been (Humanetics), 3 April 213 David Hynd (TRL) Approved by Paul Lemmen (Humanetics) 3 April 213 Issue date July 213

2 Executive summary Although the number of road fatalities in Europe has declined by about 45 percent since 21, further efforts are required to make European roads even safer. Although efforts are needed on all aspects of road safety, the THORAX Project is focused on reduction and prevention of thoracic trauma, this being one of the leading types of severe and fatal injuries to occupants in car crashes. One of the scientific and technological objectives identified for the THORAX Project is to produce a mechanical demonstrator consisting of a new dummy thorax and shoulder design capable of representing identified injury mechanisms under real world loading conditions developed and implemented in a THOR-NT frontal crash test dummy. To confirm that the demonstrators meet their design requirements and the relevant objectives of the THORAX Project, the demonstrators have been evaluated by project partners in a broad series of tests. This report documents the evaluation of three THORAX Project Demonstrator dummies in a series of tests as defined in the biofidelity requirements laid out in an earlier report from the project. The tests were designed to replicate, as closely as possible original tests with human subjects (PMHSs [post-mortem human subjects] or volunteers). By repeating tests with the demonstrators the dummy performance was compared with the previous PMHS and volunteer test results and the biofidelity requirements. In certain conditions, Hybrid III tests were carried out to help validate the set-up. The Hybrid III tests generated data which was compared with Hybrid III tests performed by the original authors with the original equipment. Baseline THOR tests were also carried out for two principal reasons. The first being to provide a baseline against which design alterations within the THORAX Project could be evaluated and secondly, to provide initial guidance to the demonstrator design team on the performance of the existing THOR-NT, the starting point for their efforts. The results indicate that the THORAX Demonstrator has improved regional stiffness distribution. This represented a positive result for the Demonstrator design and supported further dummy evaluation. Although in some individual tests differences between dummy response and corridors / requirements from PMHS tests were observed, the THORAX demonstrator generally shows reasonable to good correlation over a broad range of conditions. The upper rib - clavicle complex seems stiffer though than those of the PMHSs in various conditions. Most dummy components seem to have been robust. A single failure of a strain gauge was reported during the biofidelity evaluation. However, the IR-Traccs have been damaged in THORAX Project testing. The mechanism by which this damage can occur, and the strength of the iliac crests may need further consideration. Page 2

3 Contents 1 Introduction Testing performed within THORAX Test matrix Results Sled tests Forman et al. [26] and Bolton et al. [26] Törnvall et al. [28] Rouhana et al. [23] Shaw et al. [29a] Petitjean et al. [22] Table-top tests Cavanaugh et al. [1988] Cesari and Bouquet [199] and other authors Kent et al. [24] Shaw et al. [27] Impactor tests Lebarbé [21] Yoganandan et al. [1997] Out of position / Inertial loading / deploying restraint tests Davidsson [213] Discussion Dummy biofidelity Durability Belt retention Other issues Belt cutting Instrumentation Seating procedure Risk Register Summary Conclusions...53 Reference...54

4 1 Introduction Around 34,8 people were killed and almost 1.2 million injured in European road accidents in 29. Although the number of road fatalities has declined by about 45 percent since 21, further efforts are required to make European roads even safer. This may be particularly challenging taking into account the growing transportation needs of the elderly and the reduced tolerance of older people to injury. Although efforts are needed on all aspects of road safety, the THORAX Project is focused on reduction and prevention of thoracic trauma, this being one of the leading types of severe and fatal injuries to occupants in car crashes. The general objectives of THORAX are to develop the required understanding of thoracic injury mechanisms and to implement this into numerical and experimental tools that will enable the design and evaluation of advanced vehicle restraint systems that offer optimal protection for a wide variety of car occupants. In order to maximise the safety benefits gained from new vehicle and restraint technology, these tools will have to be much more sensitive to the in-vehicle occupant environment than the Hybrid III dummy, as used in regulatory and consumer information frontal impact testing today. One of the scientific and technological objectives identified for the THORAX Project is to produce a mechanical demonstrator consisting of a new dummy thorax and shoulder design capable of representing identified injury mechanisms under real world loading conditions developed and implemented in a THOR NT frontal crash test dummy. Based on the biomechanical requirements identified in Work Package (WP) 2 a dummy with improved thorax and shoulder designs has now been specified (Task 3.1) and three demonstrator prototypes have been developed (Task 3.2). To confirm that the demonstrators meet their design requirements and the relevant objectives of the THORAX Project, the demonstrators have been evaluated by project partners in a broad series of tests. This report documents that evaluation and the validation of the demonstrators. The evaluation testing involved repetition of tests performed on PMHS (post-mortem human subjects) or volunteers with exactly the same protocol. The demonstrator has been evaluated primarily in terms of biofidelity (targets from WP 2) and injury prediction; whilst also giving some information regarding repeatability and sensitivity to test conditions (though these latter aspects will be evaluated further in WP4). The main output of WP 3 is to produce validated demonstrator dummies with an improved upper torso complex. Once approved, the demonstrator dummies will be delivered to those partners involved in WP 4 for testing with modern restraint systems. Based on the findings of the accident surveys and the outcome of the biomechanical work a demonstrator dummy was designed (Been et al. [212]). Extensive design and prototyping efforts were made to realise three demonstrator prototypes for evaluation testing and restraint sensitivity testing Lemmen et al. [213]. The main focus was on the updates to the SD2 shoulder, rib response tuning to corridors and the implementation of instrumentation to capture information for newly proposed injury criteria. Three demonstrators were built based on THOR donor dummies provided by THORAX partners IFSTTAR, Autoliv and TRL. All three included tuned rib sets, modified jackets and SD3 shoulders. The Autoliv and TRL dummies also included the Pelvis-Femur-Knee ModKit described by Ridella and Parent [211]. Page 4

5 In conjunction with the injury criteria proposed, four 3D IR-Traccs in the thorax were adopted from the THOR ModKit Ridella and Parent [211]. The IR-Traccs provide the required input to be used in calculating the Dc criterion (Davidsson et al. [213]). As input to the NFR criterion (Davidsson et al. [213]) two of the demonstrator dummies were equipped with a total of 72 strain gauges on the ribs (Figure 1-1). The gauges are implemented such that the influence on the chest dynamic response is negligible. From the second rib down, all six lower ribs have six strain gauges on both sides equally spaced in ratio to the length of the ribs. Figure 1-1 Example of rib with gauges Page 5

6 2 Testing performed within THORAX To ensure that a crash test dummy loads the vehicle and restraint system in a similar way to the human, biofidelity requirements are used to specify the dummy performance. In addition, these requirements are used to ensure that the response of the dummy to restraint system loading is relevant to be used in the prediction of injury risk in simulated crashes. Biofidelity requirements may be derived from human volunteer, PMHS (post-mortem human subject), or animal tests and the test conditions should be representative of real-world accidents. In 29, ISO/TC22/SC12 (Technical Committee 22, Sub-Committee 12 Passive safety crash protection systems) Working Group 5 (Anthropomorphic test devices) started work on defining frontal impact biofidelity targets in a world-wide expert group. The ISO Task Force led by ACEA (European Automobile Manufacturers Association) and CEESAR (Centre Européen d Etudes de Sécurité et d Analyse des Risques) contributed a literature review and also defined selection criteria to find thorax biofidelity test configurations. In the end, the work focused on three test configurations and biofidelity targets were proposed by ACEA-TFD (the ACEA Dummy Task Force) and CEESAR and released by Lebarbé and Petit [21], [212]. Their current proposal for biofidelity targets is being reviewed by ISO WG5, with a draft Technical Report under revision. Acknowledging this effort and previous biofidelity requirements documented by EEVC and NHTSA, the THORAX project sought further test conditions under which an advanced dummy thorax could be evaluated. In THORAX a literature survey and analysis of existing datasets was conducted to define biomechanical requirements and to identify gaps in the existing datasets. For the analysis of the existing data the load cases used were compared with the loads in actual collisions to prioritise the available information. From a literature survey a first set of thoracic biofidelity target corridors for a 5 th percentile male frontal impact dummy was presented in Deliverable 2.1 Hynd et al. [211]. The list of biofidelity requirements produced in D2.1 was used to define the test matrix for the THORAX Demonstrator dummy evaluation. As noted in D2.1, to ensure good performance under various loading conditions a much broader set of requirements was proposed when compared with those from the EEVC and NHTSA. 2.1 Test matrix Due to the efforts of the THORAX Project partners involved in Task 3.3 it was possible to recreate many of the test conditions specified in D2.1. The following tables show all of the references identified in the D2.1 requirements and those performed as part of the THORAX Project. As noted in D2.1, a sub-set of the evaluated data sets were considered as suitable for defining biofidelity requirements, and suitable for replication within the THORAX project (or at any well-equipped crash test or biomechanics laboratory). These requirements are summarised in Tables 2-1 and 2-2. Table 2-1 shows the selected data sets that give absolute biofidelity requirements, i.e. those that result in, for example, specific targets for the deflection of the mid-sternum in a particular loading condition. Table 2-2 shows the selected data sets that give relative biofidelity requirements, i.e. those that result in, for example, relative compression targets for different regions of the thorax, or under different load distributions. Page 6

7 Table 2-1 Recommended data sets Biofidelity requirements / absolute requirements Author Description Restraints / Loading Lebarbé [21] Pendulum tests, Response including data from corridors using 11 PMHS. Nahum et al. [197] 22.3 or 23.4 kg at an Kroell et al. [1971], impactor velocity of 4.7 or Kroell et al. [1974], 4.3 m/s Bouquet et al. [1994] Pendulum tests. Oblique. 7 PMHS, of which 5 meet the Yoganandan et al. inclusion criterion, exposed [1997] to a 23.5 kg pendulum impact at 4.3 m/s 3-point belt; Sled tests. Passenger force limited Forman et al. [26] position. 9 PMHS of which 8 3-point belt; (not including Shaw et PMHS meets the inclusion and 3-point al. [2]) criterion at 48 and 29 km/h belt with in Ford Taurus 1997 buck airbag Bolton et al. [26] Sled tests. Passenger Airbag with position. 3 PMHS at 48 and knee bolster 29 km/h in Ford Taurus and lap belt 1997 buck Törnvall et al. [28] Sled tests. 9 tests with 3 3-point belt PMHS in full frontal 45 far-side and 3 near-side collisions at 27 km/h in rigid seat Thorax biofidelity requirements Shoulder biofidelity requirements Injury risk functions Dummy TRL, Hub Hub Autoliv, IFSTTAR TRL, IFSTTAR Autoliv Autoliv Autoliv Page 7

8 Table 2-2 Recommended data sets Engineering guidelines / relative requirements Author / Appendix Description Restraints / Loading Chest Cavanaugh et al. [1988] Table top tests. Static. 2 PMHSs loading plate; 4.5 cm x 1 cm Table top tests. Dynamic. 7 PMHS, of which 3 meet the Diagonal inclusion criterion, at 3 m/s shoulder loading rate using a 22.4 kg belt impactor Table top tests. Dynamic. 7 Cesari and Bouquet PMHS, of which 4 meet the Diagonal [199]/ Riordain et al. inclusion criterion, at 7.3 shoulder [1991]/ Cesari and m/s loading rate using a belt Bouquet [1994] 22.4 kg impactor Table top tests. Dynamic. 7 PMHS, of which 4 meet the Diagonal inclusion criterion, at 2.4 shoulder m/s loading rate using a belt 76.1 kg impactor Hub, 2 point Table top tests. Quasistatic. belt, 4 point Kent et al. [24] 15 PMHS, of which belt, or belt 12 meet the inclusion for criterion, loading rate 1 m/s distributed load Chest Table top tests. Quasi-static loading Shaw et al. [27] or dynamic. 5 PMHS using plate; 6,2 load rate 1 m/s cm x 6,2 or 11,3 cm Thorax biofidelity requirements Shoulder Injury risk Dummy biofidelity functions requirements Autoliv TRL TRL TRL IFSTTAR IFSTTAR Another sub-set of the evaluated data was identified that were assessed as being potentially suitable for defining biofidelity requirements with several constraints: In some cases additional information on the test set-up was required in order to reproduce the test conditions accurately. Page 8

9 The complexity of the test set-up was judged to be high, such that it is probably only practicable for the original authors to reproduce the tests with a dummy. Some data sets were judged to be of good quality, but very small (e.g. only three subjects per test condition). These potential biofidelity requirements are summarised in Tables 2-3 and 2-4 for absolute and relative requirements respectively. Table 2-3 Potential data sets Biofidelity requirements / absolute requirements Author Description Restraints / Loading Thorax biofidelity reqs Shoulder biofidelity reqs Injury risk functions Dummy Different restraints. Yoganandan et al. [1993] Rouhana et al. [23] Shaw et al. [29a] Sled tests. Driver position. Important data set 14 PMHS at 32 or 47 km/h comprise 5 PMHS in Ford Tempo 1986 buck tests using an airbag with lap belt Sled tests. 7 PMHS of Mainly force limited 4 which 7 meet the inclusion point belt system incl. criterion, at 4 km/h in dual lap belt specially made seat pretensioner Sled tests. 8 PMHS, of which 6 meets the 3-point belt (separate inclusion criterion, at 4 lap and shoulder belt) km/h in rigid/cable seat X TRL Autoliv Table 2-4 Potential data sets Engineering guidelines / relative requirements Author Description Restraints / Loading Fayon et al. [1975] Table top tests. Static. 7 Diagonal PMHS and volunteers shoulder belt Table top tests. Quasistatic. 3 PMHS using ramp- Salzar et al. [28] Loaded by a and Lessley et al. diagonal hold or sinusoidal load at [28] shoulder belt rates 5 to 1.2 m/s. Thorax Shoulder Injury risk Dummy biofidelity biofidelity functions reqs reqs X X Page 9

10 Within the project there were only three of these datasets which were not reconstructed. Those three were the tests originally reported by: Yoganandan et al. [1993] sled tests Fayon et al. [1975] table-top tests Salzar et al. [28] and Lessely et al. [28] table-top tests Tables 2-1 to 2-4 also define whether the data set is suitable for defining thorax biofidelity requirements or shoulder biofidelity requirements, and whether it should be considered for thorax injury assessment tests in Task 2.6. Also identified within the table is the dummy used for each test series. Page 1

11 3 Results The results from each of the test series are summarised in the following sections. The information presented is taken from a more detailed Annex which has been prepared by each test laboratory. For further information about any particular test series, please see the relevant annex, as appended to this report. The topics described for each series of tests include: Recreation of the original set-up Dummy and measurements used Biofidelity requirements Comparison of dummy performance versus o Requirements o Previous dummy versions Handling and robustness General dummy evaluation The list of Annexes appended to this report is as shown in Table 3-1. Table 3-1 Annexes with detailed information regarding the THORAX Demonstrator dummy evaluation testing Original test series Lebarbé [21] and Yoganandan et al. [1997] Forman et al. [26] and Bolton et al. [26] Törnvall et al. [28] Cavanaugh et al. [1988] and Shaw et al. [27] Cesari and Bouquet [199] and Cesari and Bouquet [1994] Kent et al. [24] Rouhana et al. [23] and Vezin et al. [22] Shaw et al. [29b] Davidsson [213] Petitjean et al. [22] Annex letter B C D E F G H I J K 3.1 Sled tests Forman et al. [26] and Bolton et al. [26] (Recommended data set absolute requirement) To replicate PMHS sled test configurations reported by Forman et al. [26] and Bolton et al. [26] a sled rig was developed (Figure 3-1). The rig consists of a stiffened Ford Taurus buck equipped with airbags, safety belts, dashboard, seat, etc. Four front passenger sled test configurations were replicated using the Autoliv THORAX demonstrator dummy: Page 11

12 No shoulder belt (2pt belt) + airbag. Only 3-point belt load, no airbag. 5 kn Shoulder belt load. 3-point belt (force limiter) + airbag. 5 kn Shoulder belt load. 3-point belt (no force limiter) + AB. 8 kn Shoulder belt load. High speed filming with motion capture was applied to obtain dummy kinematics data. Dummy responses were compared against global kinematics, thoracic spine accelerations and thoracic deformations obtained from the PMHS tests. Figure 3-1 General view of the sled rig To derive the acceleration corridors, the original raw PMHS data from the tests performed by UVa have been used. PMHS corridors were derived for T1, T8 and T12/L1 resultant accelerations. An attempt was made to apply normalisation to the standard 5th percentile size Eppinger et al. [1984]. However, the different temporal scaling factors between subjects appeared to generate different values resulting in a bi modal mean response, which was not representative of any of the PMHS and cannot reasonably be reproduced by the dummy. Therefore normalisation was not applied for the corridors used in the performance comparisons. The acceleration versus time responses were obtained as the mean response with a corridor set at ± one standard deviation. For T1 acceleration corridors, the positions of the accelerometers in the PMHS and in the dummy are different. To derive the T1 corridors, the PMHS response with three accelerations and three rotational velocities have been used to obtain the acceleration corridors for the position of the dummy T1 accelerometer. Therefore the T1 corridors presented in this report are dummy specific corridors. Furthermore, the kinematics behaviour is analysed (trajectories of different targets on the head, T1 or hip) in order to perform a general comparison of the demonstrator dummy with respect to the PMHS. The results from the two tests with the demonstrator dummy in each test configuration, together with the response corridors, are shown in Figure 3-2 and Figure 3-3. The positive numbers in brackets in Figure 3-3 show the expansion of the chest in that region as a percentage of the original chest depth. Based on these results it is observed that: The general kinematic behaviour of the THOR is good. However, the lower part of the dummy underwent a greater excursion than the mean of the PMHSs. It is possible that the main factor was the knee-to-dashboard distance; which, together with the Page 12

13 forward displacement of the pelvis from condition to condition, was variable in the PMHS. The T1 and head excursions are well reproduced. The resultant spine accelerations (T1, T8 and T12) gave good results. Dummy accelerations are greater than the PMHS, but the morphology and timing was well reproduced. The T8 and T12 dummy accelerations are influenced by the direct contact of the knees against the dashboard. The THOR dummy had less chest compression than the PMHS, especially for the more severe configurations. The dummy thoracic deformation was able to discriminate the configurations in terms of their severity. In the three-point belt loading, the PMHS had large uncoupled deflections (the upper compression was 2 to 4 times greater than the lower compression); however, the prototype dummy could not reproduce this. With the three-point belt, the lower left thoracic location had high deflection and the lower right location had very little deflection (even considering the local chest depth at that level) reproducing the behaviour of the PMHS. Forman et al. [26].Taurus buck. 3P belt pretensor & FL + depowered Airbag. 48 km/h. 5 kn SBFL. Displacement (mm) Head displacement X Time (ms) (s) Head 1 Left Head 2 Left Head 3 Left Head 1 Right Head 2 Right Head 3 Right Head PMHS Displacement (mm) Shoulder displacement X Shoulder Left Shoulder Right Shoulder PMHS Time (ms) (s) Displacement (mm) Hip displacement X Hip 1 Right Hip 2 Right Hip PMHS Time (ms) (s) Upper Spine (T1*) Resultant Acceleration Mid Spine (T8) Resultant Acceleration Lower Spine (T12) Resultant Acceleration Acceleration (g) 4 T1 Corridor 35 12LSP24E LSP24E Time (ms) Displacement (mm) Acceleration (g) Time (ms) Forman et al. [26].Taurus buck. 3P standard belt + depowered Airbag. 48 km/h. 8 kn SBL Head displacement X Time (ms) Head 1 Left Head 2 Left Head 3 Left Head 1 Right Head 2 Right Head 3 Right Head PMHS Displacement (mm) Hip displacement X T8 Corridor 12LSP24E11 12LSP24E12 Hip 1 Right Hip 2 Right Hip PMHS Time (ms) Acceleration (g) Displacement (mm) T12 Corridor 12LSP24E11 12LSP24E Time (ms) Knee displacement X Knee Left Knee Right Knee PMHS Time (ms) Acceleration (g) Upper Spine (T1*) Resultant Acceleration 7 T1 Corridor 6 12LSP24E9 5 12LSP24E Time (ms) Acceleration (g) Mid Spine (T8) Resultant Acceleration 7 T8 Corridor 6 12LSP24E9 5 12LSP24E Time (ms) Acceleration (g) Lower Spine (T12) Resultant Acceleration T12 Corridor 12LSP24E9 12LSP24E Time (ms) Figure 3-2 Head, T1 and hip x-displacements; upper spine, mid spine and lower spine resultant accelerations for tests with 3-point force limiting belt (5 kn) and de-powered airbag (top rows) and tests with 3-point standard belt no-force limitation (maximum shoulder belt load of 8 kn) with de-powered airbag by Forman et al. [26] Page 13

14 2P belt + Full-powered Airbag THOR PMHS THOR Right Sternum Left Upper -12.2% -11.% -11.1% Lower -4.1% (5.9%) % (6.3%) 3P standard belt + NO Airbag THOR PMHS THOR Right Sternum Left Upper -6.7% -16% -14.3% Lower -5.2% (2.4%) -7% -13.% (1.4%) 3P belt pretensor & FL + depowered Airbag THOR 3P standard belt + depowered Airbag THOR Right Right Sternum Left Left -1.4% -25.9% -27.6% -24.8% -17.% -4.4% (4.3%) 6.2% -6.9% -11.5% -16.3% (1.7%) THOR PMHS PMHS THOR Right Right Sternum Left Left -16.7% -33.5% -34.1% -38.% -18.5% -6.6% (3.4%) 7.1% -5.9% -13.4% -2.7% (2.7%) Figure 3-3 Comparison of chest x-deflections (as a percentage of the chest depth at measurement location) tests for all four passenger load conditions considered by Forman et al. [26] and Bolton et al. [26] Törnvall et al. [28] (Recommended data set absolute requirement) Shoulder response and belt interaction of the THORAX demonstrator dummy, with improved geometry and enlarged range-of-motion of the shoulder as compared to the THOR-NT, were compared in full frontal collisions, in 45 far-side collisions (away from the diagonal belt anchor) and in 3 near-side collisions (towards the diagonal belt anchor) with the kinematics of post mortem human subject (PMHS), Hybrid III, THOR NT and THOR SD-1NT data. Several tests with the THORAX demonstrator, were conducted for three collision angles, 45 far-side, 3 near-side and full-frontal, and compared with PMHS, Hybrid III, the THOR NT and THOR SD-1NT data from Thorax report D2.1 and Törnvall et al. [28]. Three-dimensional film markers were attached to the metal structure of the THORAX demonstrator dummy (Figure 3-4). The dummy was restrained with a three-point lapshoulder belt without force limitation. Average peak acceleration was 14 g and the average v was 26.7 km/h for both the dummies and PMHSs. The kinematics of the dummy was captured by four video cameras; frontal view, left and right side views and top view. Page 14

15 THORAX D3.3 Dummy validation report Main summary Z X.33m X Y full-frontal Anchor point for diagonal belt Y Z Anchor point for diagonal belt.37m.634m Pulse direction.145m.7m Anchor point for lap belt H-point.6m.28m Anchor point for lap belt m Top view X Y 45 far-side Side view X Z 3 near-side Z Pulse direction Pulse direction Y Top view 3-D film marker for the head Top view 3-D film marker for the T1 position 3 axial accelerometers for the head 3 axial accelerometer for the T1 position 3 axial accelerometers for the arms 3-D film markers for the shoulders Instrumentation Figure 3-4 Test set-up of full-frontal, 45 far-side and 3 near-side tests, and instrumentation of the test subjects. Adopted from Törnvall et al. [28]. Page 15

16 During the full-frontal impacts, the torso of each subject rotated around the Z-axis. Such rotations were observed with all PMHSs and all dummies with the exception of the Hybrid III. The PMHSs and THORAX demonstrator reached their maximum anterior head centre of gravity displacement at about the same time. The shoulder loaded by the diagonal part of the seat belt of all PMHSs moved posterior to T1, on average 9.1 cm, while the motion of the loaded shoulder for the THORAX demonstrator was negligible in relation to T1. After 1 ms the THORAX demonstrator shoulder moved forward relative T1. This was thought to be due, most probably, to the diagonal belt sliding towards the neck in the THORAX demonstrator test; this was not present in the PMHS tests. The non-loaded shoulder moved anterior (forwards) to T1 on average 1 cm for the PMHSs and 16.6 cm for the THORAX demonstrator. In the 45 far-side impact the THORAX demonstrator did not retain the diagonal belt on the shoulder. When the jacket and the bib were modified the belt was retained for a longer period, even though the belt appeared to disengage with the shoulder and apply loads to the upper arm. This was in contrast to the PMHSs; for two subjects the belt never slipped of the shoulder and for one the belt slipped of the shoulder in the off-loading sequence of the impact. In the 3 near-side impact, the THORAX demonstrator exhibited kinematics close to those of the PMHSs. On conclusion the THORAX demonstrator appeared to have responses closer to the PMHSs than the Hybrid III and the THOR NT. However, the THORAX demonstrator response would be more humanlike if the shoulder was design to reduce the belt slip, in any direction, in full frontal collisions and in far-side oblique collisions. Design changes that could help to retain the diagonal belt in far-side oblique collisions would be favourable Rouhana et al. [23] (Potential data set absolute requirement) As described in D2.1, Rouhana et al. [23] carried out sled tests to study the effect of load distribution, by introducing four-point seat-belts, on the number of rib fractures. In the paper sled tests in a vehicle buck investigating different 4-point-belt concepts with the Hybrid III 5% and THOR NT were reported. Further PMHS sled tests comparing a 3-point-belt and a 4-point-V-belt configuration in a generic setup were reported. In these tests eight PMHS were included of which six were restrained by the V-four-point seat-belt system. In this test setup no tests with HIII 5% and THOR NT were carried out. However, pre-tests with the PMHS sled configurations were done with the Hybrid III 5% small female, which could be used to validate the test rig that was built up with in the THORAX project for reproduction of the PMHS tests. The Rouhana et al. [23] test setup was successfully reproduced. It was validated by test with the Hybrid III small female, which showed similar responses compared to the results reported by Rouhana et al. [23]. Especially the belt loads showed good agreement. IR-Tracc reading of the dummy showed high asymmetric loading for the severe belt path in the 3-point belt configuration. This could be useful additional information for injury risk assessment compared to single point measurement which might not be able to correctly assess the severity of this configuration. Page 16

17 The four-point-configuration showed very low x-axis IR-TRACC deflections. The vertical and resultant deflection were significantly higher. Based on this observation it might be necessary to include the vertical or resultant displacement in injury risk criteria. It could also be possible that the dummy does not load the four-point-restraint system in a human-like way. As stated before the test setup was validated against dummy tests done by Rouhana et al. [23] However, detailed results and videos from the PMHS tests were not available to investigate whether the loading of the dummy was similar to the loading of the PMHS. Especially for the four-point-belt configuration it is recommended to further investigate possible differences in kinematics and loading of the test subject by the restraint system, before using the data for injury risk curves. Figure 3-5 shows the x-deflections of the four IR-Traccs in the Rouhana 3-point-belt configuration. This configuration is very severe as the belt path is very high on the chest and close to the neck. This leads to a high asymmetric loading of the chest. The deflections on the right chest are much higher as the belt mainly loads that part of the chest. In the left part of the chest the x-deflection is very low. However, the upper left IR-Traccs measures a significant vertical displacement. This can be seen in Table 3-2, which shows the average peak deflection of the two tests. It should be noted that the resultant deflection will always be a positive value regardless of the sign of the component deflections due to the square root of the sum of squares function used to calculate it. 1 Rib ddsplacment Dx in [mm] -1-2 Upper RH (LCS) Dx Upper LH (LCS) Dx Lower RH (LCS) Dx -3 Lower LH (LCS) Dx Upper RH (LCS) Dx Upper LH (LCS) Dx Lower RH (LCS) Dx Lower LH (LCS) Dx Time in [s] Figure 3-5 Rouhana 23 3-Point-Belt Configuration: IR-Tracc Measurements Dx Page 17

18 Table 3-2 Rouhana 23 3-Point-Belt Configuration: IR-Tracc Peak deflections Upper LH Upper RH Lower LH Lower RH Dx Dy Dz R Dx Dy Dz R Dx Dy Dz R Dx Dy Dz R Max Min The extreme belt position also resulted in an inward movement of the shoulder belt. Figure 3-6 shows the belt after the test. It moved towards the neck and damaged a cable. Due to this the d-ring position was modified for the second test and moved about 5 mm outward as shown in Figure 3-7. In the second test the belt also moved closely toward the neck, but stayed on the shoulder. The loading of the chest by the shoulder belt was similar for both belt attachment positions. Figure 3-6 Photo after the first test with the Rouhana 23 3-point-belt configurations D-ring attachment point moved 5 mm for second test Figure 3-7 D-ring attachment point position moved outward for second test Page 18

19 Figure 3-8 shows the x-deflections measured by IR-Traccs in the Rouhana et al. [23] fourpoint-belt-configuration. All x-deflections are very low (below 1 mm) for this configuration. The highest x-deflections can be observed at the upper left IR-Tracc. The corresponding IR- Tracc at the upper right measures lower deflections, which even though the 4-point-belt should load the chest symmetrically. However, the total amount of deflection in x-direction is quite low. Table 3-3 shows the average peak values of all deflections calculated from the 3-D-IR-Tracc measurements of the two tests. For the upper IR-Traccs significant deflections in vertical direction can be observed, which also result in high resultant deflection compared to the x- deflections. 1 Rib ddsplacment Dx in [mm] -1 Upper RH (LCS) Dx Upper LH (LCS) Dx Lower RH (LCS) Dx Lower LH (LCS) Dx Upper RH (LCS) Dx Upper LH (LCS) Dx Lower RH (LCS) Dx Lower LH (LCS) Dx Time in [s] Figure 3-8 Rouhana 23 4-Point-Belt Configuration: IR-Tracc Measurements Dx Table 3-3 Rouhana 23 4-Point-Belt Configuration: IR-Tracc Peak deflections Upper LH Upper RH Lower LH Lower RH Dx Dy Dz R Dx Dy Dz R Dx Dy Dz R Dx Dy Dz R Max Min In summary: Two sled test PMHS configurations from a study by Rouhana et al. [23] were reproduced with the THORAX demonstrator for the purpose of injury risk curve generation. Page 19

20 The test rigs were validated by reproducing the Hybrid III tests, which were also done by the original authors of the studies Based on this validation, the original test conditions were judged to have been correctly replicated. However, no comparison of PMHS measurements and kinematics with the THORAX Demonstrator were done, which should be taken into account when using the data for risk curves The three point belt configurations showed a significant asymmetric (difference between right and left chest deflection). This should be good input for an asymmetrybased injury risk criterion, which could be a significant improvement compared to the single point based criteria of the Hybrid III Shaw et al. [29a] (Potential data set absolute requirement) In a sled test series, commonly referred to as the Gold Standard, by Shaw et al. [29b] and Shaw et al. [29a], eight PMHSs were positioned on a rigid planar seat with their torso and head supported by a matrix of wires. The subjects were restrained by a three-point shoulder and lap belt, using separate adjustable-length sections joined near the subject s left hip, tensioned to approximately 5 N and 5 N, respectively, prior to each test. In addition, a rigid knee bolster, adjusted to be in contact with the knees, and a footrest with ankle straps restrained the lower extremities. The sled and the PMHSs were subjected to a peak acceleration of 14 m/s 2 that resulted in a velocity change of 4 km/h. Instrumentation was comprehensive and enabled the extraction of acromion, spine, and head displacements and chest compressions using video analysis (Shaw et al. [29a; Shaw et al. [29b] and Lebarbé and Petit [212]). The new THORAX demonstrator and Hybrid III dummy were subjected to Gold Standard sled tests using a replica of the original test rig Crandall [28]. Dummy rib cage deformations were measured using internal sensors and recalculated to the coordinate system used in the the Gold Standard series; anterior chest displacements in a T8 vertebra coordinate system. T8 location and coordinate system attitudes were transferred to dummy drawings from drawings of seated humans Schneider et al. [1983]. THORAX demonstrator upper chest deformations were larger than those of the THOR NT Crandall [28] but still lower than the response target (Figure 3-9). In the tests with the demonstrator dummy slightly larger belt downward sliding was observed compared to the original PMHS tests Shaw et al. [29a]. This off-loaded the left chest and may, to some degree, explain the smaller upper left chest deformations. For the belted shoulder, the clavicle of the PMHSs moved somewhat rearward and exposed the chest to belt loads. This was not fully reproduced by the demonstrator; the clavicle shielded the upper right chest to a higher degree in the demonstrator. In addition, the demonstrator exhibited slightly larger thorax rotation around its vertical axes and this could explain why the demonstrator exhibited relatively large lower left chest compressions as compared to those observed in the upper chest. Another reason for larger lower left chest deformations were differences in pelvis motions; the demonstrators moved on average 35 mm forward while the PMHSs only moved on average 25 mm. The results indicate that the dummy chest may be slightly stiffer than that of the average PMHS in this loading condition. Hybrid III chest deformations were uniform but also small; upper right and lower left chest deformations were smaller than the biofidelity targets (Figure 3-9). Page 2

21 Upper Right x-displacement (mm) THORAX project requirement THORAX 1 THORAX 2 THORAX 3 Hybrid III Time (ms) Upper Left x-displacement (mm) Time (ms) Figure 3-9 Chest deformations relative T8 for THORAX demonstrator with SD3, Hybrid III compared with biofidelity target provided by Lebarbé and Petit [17] Lower Left x-displacement (mm) ( ) Time (ms) In addition to chest stiffness, two other factors highly influenced the chest response of the Hybrid III; the Hybrid III thorax spine is rigid and as a result the thorax did not flex as did the spine of the demonstrator and the PMHSs (Figure 3-1); the Hybrid III exhibited smaller head forward displacements than those specified in the project requirements (Figure 3-11). These differences may not seem to be important in this sled condition where there is nothing for the head to contact. However, differences in head excursions may be more important in vehicles both with respect to the potential for head contacts and also in restraint system interactions. Neither the THORAX demonstrator nor the Hybrid III exhibited lower right chest bulge out as did the average PMHS. The THORAX demonstrator appears to be repeatable based on these three tests. Figure 3-1 Stills from high speed video of tests in Gold standard conditions with a representative PMHS, THORAX and Hybrid III at 12 ms Page 21

22 Head x-displacement relative sled (mm) Thorax project requirement THORAX 1 Hybrid III Time (ms) Figure 3-11 Head displacement relative sled for THORAX demonstrator and Hybrid III compared with project requirements that were derived from Ash et al. [212] Petitjean et al. [22] Although they were not specified as a biofidelity requirement, the frontal sled tests of Petitjean et al. were recreated at LAB with a THORAX Demonstrator instead of the four PMHS. These tests were predominantly carried out to provide information to support the development of injury risk curves for use with the dummy. Two configurations were tested in this sled test series: 6 kn load-limiting belt without airbag, referred as "6kN" 4 kn load-limiting belt with airbag, referred as "4kN+AB" Each restraint configuration was tested two times. PMHS external responses have been described in the THORAX project deliverable D2.1. However due to the low number of surrogates and since no PMHS deflection was recorded, those tests do not provide biofidelity targets. Therefore the purpose of the study was to evaluate dummies ability to discriminate both restraint types. From accidental study, the 6kN configuration is expected to be more severe than the 4kN+AB one. The test configuration is set to be representative of a real-world car crash event: deceleration pulse, type of restraint and environment geometry. The restraint system involves seat, belt with pretensioning device, steering wheel with airbag and knee bolsters. The time to fire for the pretensioner was 15 ms. The airbag time to fire was 2 ms. The airbag s volume was 65 litres. Page 22

23 Figure 3-12 Test set-up general view Seat contact forces were used to provide a comparison of the general loading for the PMHS and dummy. As far as the x-axis force components are concerned, it was observed that the PMHS and dummy exhibited very similar force-time responses. With the z direction, the PMHS response is scattered and the peak force tends to be higher when compared with the dummy. In both configurations, upper shoulder belt load-limiting systems were used, leading to the PMHS and dummy responses matching very well in the case of the shoulder belt force. The Lap belt sensors indicated that the dummy pelvis is loaded less than the PMHS. This observation might be related to the dummy knees loading the knee bolster in a more significant way as compared to the PMHS. nee bolster forces were recorded, and despite variations in the PMHS responses related to anthropometry differences, the x-axis dummy loads were generally higher than for the PMHS. This is true for both the 4 kn and airbag (Figure 3-13) and 6 kn belt only tests (Figure 3-14). Page 23

24 Figure 3-13 Knee bolster contact forces: 4kN+AB configuration Page 24

25 Figure 3-14 Knee bolster contact forces: 6kN configuration In accordance with accidental studies, the 6 kn condition led to higher injury outcomes for the two PMHS tested under this condition than the 4 kn and airbag condition. Although, some overlap can be observed for PMHS SL4_2 and SL6_2. The location of the rib fractures for the PMHS are shown below (Figures 3-15 and 3-16) and a description of the symbols used is given in the following table. Page 25

26 V D ND PI PE Rib length (mm) # V D ND PI PE Rib length (mm) # Left side Right side 2 Left side Right side Rib number 5 6 Rib number CCJ CVJ CCJ CCJ SL4_1 AIS: 2 SL4_2 AIS: 4 CVJ CCJ Figure 3-15 Rib fracture schematic localization, 4kN+AB configuration V D ND PI PE Rib length (mm) # V D ND PI PE Rib length (mm) # Left side Right side 2 Left side Right side Rib number 5 6 Rib number CCJ CVJ CCJ CCJ SL6_1 MAIS: 5 SL6_2 MAIS: 4 CVJ CCJ Figure 3-16 PMHS rib fracture schematic localization: 6kN configuration The peak strain values from the THORAX Demonstrator dummy rib gauges are shown in Figure The LAB authors noted that, as far as raw dummy strains are concerned, predicted injury outcome will be higher in the case of the 4kN+AB configuration which is in contrast to the expected result. Figure 3-17 Dummy peak rib strain: 4kN+AB (left) and 6kN (right) configurations Page 26

27 Indeed, the previous figure demonstrates that whatever the strain failure threshold, the number of fractured ribs indicated by the dummy will always be higher in the 4kN+AB configuration. In summary, due to the limited number of surrogates and the lack of PMHS deflections, the original study does not intend to provide biofidelity targets. However, the highly instrumented setup allows for a better understanding of the loading force balance in between the various restraint components. Among others it emphasis the issue of mimicking PMHS responses when they are highly related to initial positioning and anthropometry such as the knee bolster interaction. It also showed that in its current form, the NFR criteria based on raw rib strains fails to rightly rank those two configurations and that further investigation are required in that field to mitigate this latter observation. 3.2 Table-top tests Cavanaugh et al. [1988] (Recommended data set relative requirement) The testing of the THORAX Demonstrator under conditions matching those reported by Cavanaugh et al. [1988] was carried out by UPM-INSIA and is reported in full in Annex E. UPM-INSIA has reproduced 18 tests according the Cavanaugh test set up with different configurations: 9 quasi-static tests and 9 dynamic tests 2 Different spine supports (only spine or spine and ribs supported) Cavanaugh et al. [1988] used 2 different PMHS to perform the tests. The subjects were mounted in a supine position and were loaded at each of six load locations with a 5 x 1 mm indenter surface. There are six different locations for applying load as shown with the dummy replications as identified in Table 3-4. Page 27

28 Table 3-4 Replicated test from Cavanaugh et al. [1988] Position Indentor Stroke Vel (mm/s) Support CSU 5x1 25 CSM 5x1 25 CSL 5x1 25 CRU 5x1 25 CRM 5x1 25 CRL 5x POS CRU POS CRM POS CRL Spine support Support 1 Spine and rib support Support 2 POS CSU POS CSM POS CSL The Table 3-5 summarises the calculated chest stiffness for the Cavanaugh configuration, the stiffness values of the PMHS have been obtained directly from the paper (Cavanaugh et al. [1988]) The THOR stiffness is greater than the PMHS stiffness (in quasi-static tests from 2.3 up to 3.1 greater in the ribs and from 1.3 to 2.8 in the sternum). The dynamic effects can be confirmed in the THOR Demonstrator with an increment from 1.24 to 1.43 with respect the quasi-static stiffness. In the PMHS, practically no viscous effects are observed (in several cases the dynamic stiffness is lower than the quasi-static). In the PMHS, the upper and middle parts of the chest have very similar stiffness whilst the THORAX Demonstrator has greater stiffness in the upper part of the chest than the mid part and less stiffness in the lower part. Table 3-5 THOR Demonstrator and PMHS stiffness Cavanaugh configuration Cavanaugh configuration - stiffness (N/mm) Quasistatic Rib Sternum Upper Middle Lower Upper Middle Lower Dynamic PMHS THOR PMHS THOR PMHS THOR Spine + rib Spine Spine + rib Page 28

29 The Table 3-6 shows the coupling effects of the THORAX Demonstrator with the Cavanaugh loading configuration. As it can be seen, when the load is applied in the sternum area it is not affected by the type of support of the dummy (only spine or spine and rib supported). When the load is applied in the sternum, a symmetric coupling is seen. There is a strong coupling in the upper sternum area with the upper ribs. The main difference between the CSM and CSL coupling effect is due to the CSL compress the abdominal area whereas the CSM is compressed practically all the sternum area and therefore more coupling effects (in the Error! Reference source not found. is shown a graphic representation with the indenter size and location). Finally, when the load is applied at the ribs, it can be observed different behaviour depending if the load is applied at the upper chest or at the lower chest. The upper rib location have strong coupling with the ribs on the same side but low with the ribs on the other side (the sternum has half deflection than the applied load). The CRM and CRL configurations show that the lower ribs (5 th, 6 th and 7 th ) of the THOR Demonstrator are fully decoupled from the rest of the chest. This fact is verified as well in the Shaw configuration. The low coupling in the lower ribs of the dummy is caused by the dummy design. The load patter from the lower ribs is performed with the urethane outer bib. In this area the thickness of the bib is bigger than the upper part bat there is no load path through the sternum area and due to this fact the coupling in this area is very low. Table 3-6 Cavanaugh configuration coupling Cavanaugh configuration Rib + spine support Spine support Rib Sternum Sternum 12LSP24E4 - Coupling 12LSP24E3 - Coupling 12LSP24E42 - Coupling Upper L.7 Str.1 R Up Mid.9.9 Low L.. Str..2 R. Up Mid.9.9 Low L.. Str.2. R Up Mid.9.9 Low LSP24E38 - Coupling 12LSP24E32 - Coupling 12LSP24E44 - Coupling Middle 1..9 L 1..8 Str R Up Mid.6.7 Low L Str R Up Mid.8.9 Low L Str R Up Mid.9.9 Low Page 29

30 Cavanaugh configuration Rib + spine support Spine support Rib Sternum Sternum 12LSP24E36 - Coupling 12LSP24E34 - Coupling 12LSP24E46 - Coupling Lower L 1..9 Str R Up Mid.. Low L Str R.5 Up Mid.8.9 Low L Str R.6 Up Mid.8 Low Cesari and Bouquet [199] and other authors (Recommended data set relative requirement) Continuing the work of L'Abbé et al. [1982], Cesari and Bouquet [199] reported the results from a series of tests with PMHS. The tests provided dynamic seatbelt loading to examine thoracic deflection characteristics. The PMHS were laid supine on a rigid table with the legs in a sitting position. They were loaded by a diagonal seatbelt passing from the left clavicle down to the lower right ribs. The deflection of the thorax was measured at eleven locations including a mid-clavicle point (see Figure 3-18). In these tests the impactor which loaded the belt had a mass of either 22.4 or 76.1 kg and the impact velocities ranged from 3 to 9 m/s. Figure 3-18 Belt position and deflection measurement points (L'Abbé et al. [1982]) In the TRL reproduction, the belt movement was generated by a bungee-powered sled, this had seat-belt webbing linking it to the spreader bar. Figure 3-19 shows the TRL set-up with the THOR-NT dummy. Within the THORAX Project, tests were carried out with the Hybrid III dummy to relate the TRL results to previous results from the literature. The purpose of this was to check that the set-up had been re-created in an appropriate manner. Page 3

31 Figure 3-19 Image of set-up from TRL test with THOR-NT The combined belt force from the TRL tests at three impact energy conditions was compared with the spread of tests performed at the Biokinetics facility (L'Abbé et al. [1982]). The input force was about the correct level for the lowest energy group of tests. For the higher severity tests, the TRL combined seat belt force was slightly below that expected with the Biokinetics set-up. This indicates that the dynamic stiffness of the applied belt loading may have been slightly lower for the TRL tests compared with the original tests. However, this difference was small and was not expected to compromise the findings at each energy level regarding the distribution of loading over the thorax or the relative deflection at different locations. A graph of the internal chest deflection measurement taken from the Hybrid III dummy and the input energy is shown in Figure 3-2. From this figure it seems as though the TRL set-up is a closer replication of the thoracic loading for a Hybrid III dummy than was expected based on the belt forces alone. This gives additional confidence that the loading of the thorax by the belt is similar when considering the historic set-up compared with the TRL reproduction of conditions. Hybrid III chest potentiometer (mm) Impact energy (J) TRL tests Biokinetics Figure 3-2 Internal Hybrid III chest deflection from the TRL tests and from previous results of tests from Biokinetics (as described by (L'Abbé et al. [1982]) Page 31

32 For each dummy the internal chest deflection was measured. External deflection measurements were also required to compare directly with the original studies. Direct measurements were made through the use of LVDTs (linear variable differential transformers). Ten or eleven LVDTs were mounted above the supine dummy with vertical attachments to the anterior aspect of the torso. The test matrix consisted of three tests carried out at each of three conditions: Low mass, low speed High mass, low speed Low mass, high speed The biofidelity requirements for these table-top tests are defined in the THORAX Project deliverable D2.1. The requirements relate to the low velocity tests, at either mass. They define relative displacement of each measurement point with regard to the mid-sternum measurement. The normalised chest compression biofidelity requirements are reproduced in Table 3-7. Table 3-7 Normalised thorax compression biofidelity requirements for Cesari and Bouquet table-top test condition Right Side Sternum Left Side Clavicle.3.46 Upper.5.83 Clavicle Rib Mid 1. Rib Rib Lower Rib Tests with the THORAX Demonstrator dummy were then performed to investigate whether the modifications to the THOR shoulder complex and torso had improved the regional stiffness of the thorax compared with the THOR-NT and Hybrid III. Full peak value results from the tests with the THORAX Demonstrator are shown in Annex F. The results from the Hybrid III plotted with the THOR-NT and THORAX Demonstrator results as well as the PMHS response limits are shown in Figure The Hybrid III showed thoracic stiffness which met the THORAX biofidelity requirements for most measurement points but at a few points gave too little chest deflection. The THOR-NT was shown also to be close to meeting the THORAX requirements but with a few measurement points exceeding the limits. This was interpreted to be an improvement over the Hybrid III behaviour. Unlike the Hybrid III, the THOR-NT allowed too much deflection at two measurement points. Page 32

33 Figure 3-21 Relative deflections for HIII, THOR NT and THORAX demonstrator with SD3 in comparison to corridor from PMHS data. As with the THOR-NT, the THORAX Demonstrator is close to meeting the biofidelity requirements. However, it still deviates from those requirements at a couple of the measurement points. In particular, too much deflection was recorded at Point 5 and too little at Point 7. However, these results are closer to the required targets than was the case with the THOR-NT. Therefore, it can be said that the THORAX modifications to the chest have not adversely affected the regional stiffness of the THOR torso, instead there seems to have been slight improvements (particularly over the Hybrid III) when evaluated under the Cesari and Bouquet table-top test condition. A limitation of the test set-up is in the accuracy with which the external measurement points were positioned around the thorax of the subject. The generic position description from the original authors was converted into exact locations by considering the standard anthropometry of the UMTRI mid-sized male. These locations then gave target positions for measurement point attachments with each dummy. However, in practice the points were manipulated in order to align them with available hard points throughout the thorax. There are several steps in this process where alignment deviations could have been introduced. Whilst efforts were taken to minimise those variations, it is considered that small variations in the positions of the external deflection measurement points could have an influence on the biofidelity results. In summary, the THORAX results showed similar behaviour to the THOR-NT demonstrating that there had been no negative effect of the modifications to the chest and shoulder complex on the regional stiffness distribution, as evaluated in this test condition. The THORAX Demonstrator dummy was even closer to meeting the THORAX biofidelity requirements than both the Hybrid III and THOR-NT (marginally). In this way, of the dummies tested, the THORAX Demonstrator seems to have the most humanlike thorax stiffness distribution and coupling when evaluated in this table-top test condition. Page 33

34 3.2.3 Kent et al. [24] (Recommended data set relative requirement) In addition to the Cesari and Bouquet tests the THORAX demonstrator was subjected to table-top conditions described in (Cesari and Bouquet [199]). In their research, Kent et al. studied the effect of four different loading conditions on the biomechanical response of the human thorax using PMHS. For the purpose of the evaluation of the demonstrator, three of four loading conditions applied by Kent et al. were reproduced. These loadings conditions were: hub, single diagonal belt and double diagonal belt. The hub had the same geometry as the one used in Kent et al. The belt for the single and double diagonal belt loading conditions was positioned on the dummy chest similarly to the cadaver tests. The breadth of the belt used was a little smaller than the original belt (4.6 cm instead of 5 cm). The test rig was not reproduced identical to the original set-up but the positions and orientations of the loading devices as well as the loading conditions were identical to the PMHS tests. A universal tensile test machine was used to generate chest deflection at a rate similar to the rate applied on the PMHS chest. This chest deflection is supposed to mimic the deflection of a restrained PMHS in 48 km/h frontal sled tests. This corresponds to a linear displacement with a constant velocity of 1 m/s. In order to replicate the PMHS protocol accurately, the pre-test 1 cycle preconditioning regime described in the Kent paper was also applied to precondition the thorax of the demonstrator. The dummy was positioned in the same way as the PMHS with its back lying on the table. Applied and reaction force were recorded with the same sampling frequency as in the PMHS tests. In addition to the dummy instrumentation, the chest deflection at mid-sternum was also measured with a linear transducer (LVDT) to facilitate the comparison with the PMHS corridor established by Kent et al. [24]. Kent et al. performed non-injurious tests with the different loading conditions on each cadaver and a final injurious test with one of the four loading conditions for each PMHS. For the THORAX demonstrator, successive tests were performed by increasing the chest deflection from 1% up to 3%. The 3% of chest deflection corresponds roughly to the injurious tests performed on the PMHS. This allowed to check the repeatability of the demonstrator response and to avoid damage to the dummy by checking the rib conditions after each test and before increasing the deflection. The responses of the demonstrator were compared with the PMHS corridors (see Figure 3-22). The demonstrator response is within the corridor for the hub loading condition. Nevertheless, for low chest deflection (<1%) the reaction force was low compared to cadavers. The reaction force increased rapidly between 1 and 15% and then the demonstrator followed the cadaver response with a value slightly higher than the average characteristics of the PMHS. For the single diagonal belt loading, the demonstrator behaviour fitted very well with the corridor up to 1% of chest deflection. Then, the stiffness of the dummy chest increased and became higher than the upper corridor above 15% of deflection. The THORAX demonstrator appears too stiff compared with the PMHS corridor with the double diagonal belt loading condition. It should be noted that permanent deformation of the lower ribs of the demonstrator were observed for this series, which is to be expected considering the peak load of 18 kn applied. Page 34

35 Posterior Reaction Force (N) Scaled to 45 year-old, 5th male Hub Loading Condition % 5% 1% 15% 2% 25% 3% Chest Deflection (% ) Figure 3-22 Force deflection data for the THORAX demonstrator compare with the characteristic average and corridors of the PMHS for single diagonal belt (left), hub (middle) and double diagonal belt (right). PMHS corridors are scaled to a 45 year-old, 5th percentile male Shaw et al. [27] (Recommended data set relative requirement) In this table top configuration (Shaw et al., 27), five PMHS were used for evaluating the response of the torso to quasi-static and dynamic anterior loading. The loading was implemented via rigid rectangular indenters designed to approximate a section of a shoulder belt and registered the deflection and the indenter load in three dimensions. At least the three-dimensional deflection from eight to nine points was recorded using Vicon Cameras equipment and the indenter loads were recorded by a load cell. The scheme of the size and location of the indenter load as well as the target markers for tracking technology are shown in Figure Figure 3-23 Indentor and Vicon marker sites for Shaw table top configuration (from Shaw et al. [27]) For supporting the dummy only in the spine support, UPM-INSIA considered that the best way to implement this configuration is using the two screws that attach each rib eye to the spine (these screws are located at the back of the dummy). With this fixture, the anterior of the chest of the dummy remains horizontal. Page 35

36 The stiffness results for the Shaw configuration are shown in Table 3-8. The PMHS stiffness data represented is the average of five subjects. As it can be seen the THOR Demonstrator has a very similar stiffness with respect to the PMHS: from 1 to 1.45 greater in the THOR dummy, except for a single case (marked in blue) the 2x4 upper rib location that have 3.45 times the PMHS stiffness. The stiffness in the THOR dummy for the upper rib with the 2x4 indenter is very high and does not follow the tendency of the other regions. In this case, the viscous effect can be seen both in the PMHS and THOR data. The viscous effect doubles the regional stiffness. Except this particular case, the THOR Demonstrator has the same tendency of the stiffness than the PMHS data, the greater stiffness are located in the sternum and the lower stiffness in the lower part of the chest. Table 3-8 THOR Demonstrator and PMHS stiffness Shaw configuration Shaw configuration - stiffness (N/mm) Quasistatic 2x2 Dynamic 2x2 Dynamic 2x4 Upper rib Middle Sternum Lower rib PMHS THOR PMHS THOR PMHS THOR For the Shaw PMHS data, the comparisons are shown in the Table 3 4, Table 3 5 and Table 3 6. The corridors were developed for the THORAX Deliverable D2.1 and are superposed as well the THORAX Demonstrator response (red colour). In the quasi-static configuration (Table 3 4), the PMHS stiffness is linear (at least in the measurement range). In the THORAX Demonstrator, the middle and lower locations presents a linear tendency, whilst the upper location has a non-linear stiffness behaviour. The stiffness value calculated by Shaw et al. [27] for comparison is the force at 15 mm divided by its deflection. The THOR force value at 15 mm is closed to the PMHS but the slope of the force-deflection curve is quite different. Page 36

37 Table 3-9 Force versus deflection for the THORAX Demonstrator and PMHS (Shaw et al. [27] quasistatic data) Shaw configuration - Quasistatic Upper Force (N) S2 S3 S4 S5 Mean THOR Force (N) S2-S3-S4-S5 S2-S3-S5 S2-S5 S2 Mean THOR Displacement (mm) Displacement (mm) Middle Force (N) S2 S3 S4 S5 Mean THOR Displacement (mm) Force (N) S2-S3-S4-S5 S2-S3-S5 S2-S5 S2 Mean THOR Displacement (mm) Lower Force (N) S2 S3 S4 S5 Mean THOR Force (N) S2-S3-S4-S5 S2-S4-S5 S2-S5 S2 Mean THOR Displacement (mm) Displacement (mm) Page 37

38 Table 3-1 Force vs deflection THOR Demonstrator and PMHS Shaw dynamic data Shaw configuration - Dynamic Upper Force (N) S1 S2 S3 S4 S5 Mean THOR Force (N) S1-S2-S3-S4-S5 S1-S2-S3-S5 S1-S2-S5 S1-S2 S2 Mean THOR Displacement (mm) Displacement (mm) Middle Force (N) S1 S2 S3 S4 S5 Mean THOR Force (N) S1-S2-S3-S4-S5 S1-S2-S3-S5 S1-S2-S5 S2-S5 S5 Mean THOR Displacement (mm) Displacement (mm) Lower Force (N) S1 S2 S3 S4 S5 Mean THOR Force (N) S1-S2-S3-S4-S5 S1-S2-S3-S5 S1-S2-S5 S1-S2 S2 Mean THOR Displacement (mm) Displacement (mm) In the Shaw et al. loading condition there are three loading locations with two different deflections (25 and 5 mm). In these tests the conclusions obtained with the Cavanaugh configuration are confirmed. The lower rib of the dummy has very low coupling effects, the sternum area has a strong coupling and upper rib area has greater coupling with the ribs of the same side but very low with the ribs of the other side. The coupling effects with the largest stroke are greater than with the standard stroke. With the 5 mm stroke, the conclusions are reaffirmed. When the load is applied at the upper ribs, the sternum area has approximately half of the deformation, while the other side of the ribs reaches 2-3% of the maximum deformation. The lower ribs are virtually not deformed. At the middle sternum location, there is a symmetric pattern and the upper rib (1st) and the lowest ribs (6th and 7th) have very low deflections. Finally, for the lower rib load location, the lower sternum is 5% coupled and for this fact, the 4th dummy rib is coupled (2% of the deformation). With the upper location (at the rib), the coupling effect between the PMHS and the THOR Demonstrator is very similar. Only in the lower ribs the dummy has less coupling effect than the PMHS. When the load is applied at the mid sternum, the THOR Demonstrator is less coupled in practically all the regions (the upper sternum is 3 times less, and the lower ribs are 2 up to 4 times lower). Finally for the lower rib load location, the THOR dummy has less coupling effect than the PMHS (the other rib side is coupled 1 times less than the PMHS). Page 38

39 3.3 Impactor tests Pendulum impactor tests replicating the Kroell frontal sternum and Yoganandan oblique lower rib tests, were evaluated against corridors defined by Lebarbé [Lebarbé and Petit], 212] and, for the oblique impacts, by the THORAX project itself based on Yoganandan pendulum tests [Hynd et al.], 212] Lebarbé [21] (Recommended data set absolute requirement) The requirements for the frontal sternum impacts were developed as part of the ACEA-TFD support of ISO Working Group 5. They define an external (surface of the chest) deflection measurement. As this cannot be measured with the IR-Traccs inside the thorax the pendulum penetration was recorded with High Speed Video analysis. A standard 23.4 kg and 152 mm diameter pendulum was used at an impact speed of 4.3 m/s. The dummy was positioned according the PMHS test positions (see Figure 3-24). Figure 3-24 Dummy in frontal and oblique pendulum impact test set-up (jacket removed for photo) Results for the upper thorax impact tests are given in Figure 9. The peak penetration in these tests was between 67 and 72 mm. The response is in fairly good correlation with corridors defined by Lebarbé and Petit [212]. The peak penetration corresponds well with the average found in PMHS tests. The peak pendulum force exceeds the target by around half the corridor width. The unloading phase is entirely within the corridor. Page 39

40 Figure 3-25 Pendulum force versus pendulum penetration. Corridors from Lebarbé and Petit [212] Yoganandan et al. [1997] (Recommended data set absolute requirement) For the Yoganandan tests the same 23.4 kg pendulum was used, but the pendulum was lined with 19 mm Rubatex foam, as specified. In THORAX [Hynd et al.], 211] the original Yoganandan [1997] data was reanalysed and Moorhouse normalisation applied, which was found to reduce scatter more than other methods. An artefact of this method is that the corridors do not represent the variation of peak deflection seen in the original tests, and corridors are narrow at peak deflection; in contrast to the Lebarbé corridors, which are wide at peak deflection. Figure 3-26 shows resulting corridors and the demonstrator dummy responses. Due to inaccuracy of the time zero of pendulum contact with the dummy, the penetration derived from high speed video analysis was possibly overestimated by up to 17mm. The peak forces of the dummy exceed the PMHS responses. Considering that dummy peak penetrations were almost certainly overestimated, the penetration response was close to the requirement. Page 4

41 Figure 3-26 Pendulum force penetration for oblique impact tests on lower thorax. Normalised PMHS responses grey shaded [Hynd et al.], 211] 3.4 Out of position / Inertial loading / deploying restraint tests Davidsson [213] In the Thorax project a test rig for evaluation of the human shoulder stiffness was developed (Figure 3-27). Belted volunteers were seated and their shoulder bone motions measured when loaded forward ( ), forward-upward (45 ), upward (9 ) and rearward (18 ). The forward and upward loads to the shoulders were applied through the arms by means of arm brackets fastened to the elbows. Rearward loads were applied by means of a padded strap wrapped around the shoulder complex. To block torso movement the volunteers were restrained by two shoulder belts, routed close to the neck, that were pre-tensioned to 1 N each. The arms or shoulders were statically loaded with 5 N increments to a maximum of 2 N/side. Each volunteer was exposed to three tests for each loading direction. The position of the shoulder complex was recorded by three digital cameras. The left acromion process relative to T1 displacements were used to calculate shoulder motion in 3D. Belt loads and seat back loads were recorded to facilitate a comparison between dummy interactions during testing as compared to those of the volunteers. Page 41

42 Figure 3-27 Test rig used for shoulder stiffness tests Tests were done with six volunteers and reproduced using the THORAX demonstrator dummies as well as a standard Hybrid III. Evaluation of the dummy performance in this loading condition is regarded as complimentary to PMHS sled tests like, for instance, the Gold Standard sled tests Shaw et al. [29a]. Average T1 change in position for each loading condition was below 32 mm forward-rearward and 14 mm upward-downward when maximum load was applied (both for volunteers and dummies). This means that the shoulder motion was more successfully isolated from other motions in this study than in similar previous studies Törnvall et al. [28] and therefore more suitable for evaluation of crash test dummy shoulders. The THORAX demonstrator produced similar shoulder motions as the volunteers did when the loads were applied forward, oblique and upward (Figure 3-28). For series the shoulder relative to T1 forward motion was 54 mm for the average volunteer, 45 mm for the THORAX demonstrator and 2 mm for the Hybrid III when maximum load was applied. For series 45 the resultant maximum shoulder relative to T1 motion was 68 mm for the average volunteer, 64 mm for the THORAX demonstrator and 2 mm for the Hybrid III. Both the THORAX demonstrator and the Hybrid III exhibited less than half the rearward shoulder motion of the volunteers in the 18 tests (Figure 3-28:). For this loading condition the THORAX demonstrator exhibited 22 mm rearward motion whereas the average volunteer exhibited 47 mm. Page 42

43 8 Series - Volunteers Series - HYBRID III Series - THORAX Series 45 - Volunteers Series 45 - HYBRID III Series 45 - THORAX Series 8 - Volunteers Series 8 - HYBRID III Series 8 - THORAX Series 18 - Volunteers Series 18 - HYBRID III Series 18 - THORAX Acromion rel. T1 z-displacement (mm) Z X Series Z X Series 8 Z Z X Series 45 X Series HYBRID III Acromion rel. T1 x-displacement (mm) Figure 3-28: THORAX project shoulder response corridors, Hybrid III and THORAX displacements in four loading directions. Page 43

44 4 Discussion In the THORAX project a new shoulder-thorax complex to be fitted on the THOR-NT was developed and three demonstrator dummies were updated with this design. The dummies were then evaluated against a broad list of biomechanical requirements for the thorax and shoulder. Those requirements included, but were not limited to those published previously by NHTSA and Lebarbé. 4.1 Dummy biofidelity In summary the performance of the demonstrator dummies in each test condition was as follows: 1. In pendulum impactor tests generally a reasonable to good correlation between demonstrator dummy responses and corridors was obtained. Peak forces for frontal pendulum tests (Lebarbé and Petit [212]) and oblique tests in the lower thorax region (Yoganandan et al. [1997]) tended to be higher than the corridors. Pendulum penetration could not be determined accurately; however, it appeared to be a good match in both conditions. 2. In table-top tests the regional coupling, the influence of loads in one part of the thorax on deflections in another part was investigated. Comparison of the demonstrator dummy with corridors identified from PMHS table top tests by Cesari and Bouquet [199] revealed that the dummy performs reasonably well compared with other dummies like the Hybrid III. However, slight deviations from the PMHS responses were still observed. Also, in sled tests, it appears that the balance of the upper to the lower measurement point deflections may not be the same as for PMHS. As injury risk development work now considers a combined deflection criterion incorporating the differential deflections from the four measurement points (left-right, top-bottom), it seems important that those efforts consider the biofidelity of the dummy. Implications of the performance of the dummy in terms of regional coupling and stiffness need to be viewed in relation to the injury criterion used. 3. In table-top tests considering the influence of loading condition (Kent et al. [24]), the demonstrator dummy also performed reasonably well. The stiffness of the dummy increased from the hub to the single diagonal belt to the double diagonal belt conditions as for the PMHS. Results were within the force-compression corridors for hub and initially for the diagonal belt loads while being too stiff for the double diagonal belt. This may indicate that the SD3 shoulder or the top of the thorax is too stiff compared to humans under the belt conditions tested. The consequences of this for the behaviour in frontal impacts are not clear from this test environment though. 4. Very good correlation was obtained between dummy tests and corridors for the shoulder range of motion tests, matched with a new set of volunteer tests. The human shoulder motions in the forward and/or upward directions are represented well by the dummy including sufficient range. Rearward shoulder motions however, were found to be limited but still showed a substantial improvement compared with the Hybrid III. The results for rearward shoulder loading with limited motion compared to volunteers might help explain the belt condition performance in the Kent table top tests, but this is to be investigated. Page 44

45 5. In sled tests (Forman et al. [26], Bolton et al. [26]) representing various restraint conditions the demonstrator dummy generally shows good behaviour, although, some differences with respect to the PMHS corridors were observed. It is to be noted that reproducing these test conditions is difficult in terms of finding representative test components like airbags, belts and seats. Furthermore for each restraint type only a very limited number of subjects were used to define the corridors. Therefore, the requirements could be biased to the specific subject characteristics and responses, for instance in their interaction with the knee bolster. 6. The gold standard tests were specified to address some of the limitations noted above by selecting subjects with anthropometry close to the average male. Also, the set-up was intended to be easily reproduced in other laboratories. In reproductions of these tests the demonstrator dummy showed improved overall kinematics in terms of head forward displacement and spine rotation and bending in comparison with the Hybrid III. Also the chest deformations were improved when compared to published THOR-NT results. However, the maximum upper right chest deformation was somewhat lower than the project requirements while the upper and lower left maximum deformations were within the required range. The reason for low upper left chest deformation may be due to higher stiffness of the upper ribcage quadrant but may also be due to the first rib-clavicle complex. This was also indicated in the comparisons with volunteer shoulder data, Kent table top test data, table top tests by Cesari and Bouquet and the Forman and Bolton sled test series. Although for some of the individual test conditions the dummy response deviates from the project requirements, reasonable correlation is obtained over the broad range of conditions considered. Detailed comparison of chest deflections in various conditions indicates that the upper rib - clavicle complex appears stiffer than those of the PMHSs. No specific biomechanical data for this region was available and it is recommended to collect such data in future studies. The demonstrator exhibited differential displacements in upper and lower, as well as left to right thorax regions in line with PMHS data. The Hybrid III dummy also showed differential deflections but to a much lesser extent. These relative deflection components are input to injury criteria proposed in THORAX and regarded as important when distinguishing between different loading conditions. In defining such criteria the relatively stiff response in the upper thorax region should be kept in mind as it reduces dummy displacement readings in this part. In some cases the lack of correlation could clearly be attributed to specifics of the set-up. As an example, differences in kinematics in sled tests could be attributed to knee contact with the bolster which is highly dependent on subject size and seating position. For certain test conditions which, particularly sled tests, data for only a limited number of subjects were available. However, if broad evaluation in a variety of test conditions is deemed a priority, then also conditions of small subject count must be included. In this respect it should also be noted that at this stage of the research the comparison of the dummy responses with the biofidelity targets has been a subjective one. It is encouraged that future efforts will incorporate an objective validation of the dummy performance. This will require an approved set of requirements and assessment method, which are not formally available at the time of writing. Page 45

46 During the biofidelity evaluations it became evident that the THORAX demonstrator dummies showed sensitivity of the chest deflection measurements to the different restraint systems or loading devices used. This suggests that the THORAX demonstrator can be valuable as a tool to investigate performance differences between various types of restraint. However, to offer equivalent or comparable risk of injury predictions under localised or distributed loading a criterion which is independent of the restraint system should be used. The potential use of the THORAX demonstrators and injury criterion candidates is being investigated in sled tests using modern restraint systems within the final work package of the THORAX Project. 4.2 Durability It was noted that it would be possible for the neck load cell cable to interact with the seat belt during a test in the way that could lead to robustness issues. This behaviour is not a result of any changes made within the THORAX Project. There was a problem with the initial position of the connector for the data umbilical with the TRL dummy. Due to the cables from the dummy being short before an extension was connected, that connector could come under substantial strain during test events. This should be subject to a relatively simple remedy to move, or even remove, the connector. The plastic end stop in the upper arm to maintain the orientation of the arm bone in the arm flesh was cracked in some of the early sled tests. This part could be made to be more durable. In the first few sled tests with the dummy, it was noted that contact between the pelvis flesh and the top of the iliac crest could result in the pelvis flesh being cut. This may be due to the design of the iliac crest or the absence of smooth iliac load cell replacements in the TRL dummy. Three iliac wing failures have been reported during testing with the THORAX Demonstrators within the THORAX Project, two of which occurred during the biofidelity evaluation testing as described within this report (a third failure occurred during the subsequent WP4 evaluation). The lap belt forces at the time of failure were around 5 kn, which is relatively low for restraint systems in a modern vehicle. 4.3 Belt retention During the THORAX Demonstrator evaluation test programme, several project partners mentioned that the shoulder portion of the seat belt had slipped over the shoulder of the dummy. In the case of the Rouhana et al. [23] testing, the belt slipped inwards towards the neck of the dummy (Figure 4-1). This behaviour was unlikely to have occurred to the same extent in the original PMHS tests. Therefore BASt adjusted the test set-up to correct for this. Page 46

47 Figure 4-1: Belt and neck interaction after BASt sled test 4.4 Other issues Belt cutting During the tests performed with the demonstrator in the Vezin et al. [22] configuration there were no further limitations with measurements and no robustness issues. However, in both tests with the THORAX demonstrator in this configuration damage to the lap belt was observed after the test. Figure 4-2 shows the damage to the belt, which was most likely caused by a sharp edge of the iliac wing (without a load cell or structural replacement) of the pelvis. It is expected that this behaviour would not be an issue if the iliac crest load cells were fitted. Figure 4-2 Damages to the lap belt caused by the THORAX demonstrator THOR dummy in the Vezin et al. [22] 5 km/h belt/airbag configuration Page 47