Evaluation of Near-Side Oblique Frontal Impacts Using THOR With SD3 Shoulder

Abstract

Objective: Within the EC Seventh Framework project THORAX, the Mod-Kit THOR was upgraded with a new thorax and shoulder. The aim of this study was to investigate whether the THOR ATD met a set of prerequisites to a greater extent than Hybrid III and by that measure whether the dummy could serve as a potential tool for future evaluation of serious head and chest injuries in near-side oblique frontal impacts.

Method: A small-overlap/oblique sled system was used to reflect occupant forces observed in oblique frontal crashes. The head and thoracic response from THOR was evaluated for 3 combinations: belt only with no deformation of the driver's side door (configuration A), belt only in combination with a predeformed door (configuration B), and prepretensioning belt and driver airbag (PPT+DAB) in combination with a predeformed door (configuration C). To evaluate head injury risk, the head injury criterion (HIC) and brain injury criteria (BrIC) were used. For evaluation of the thoracic injury risk, 3 injury criteria proposed by the THORAX project were evaluated: Dmax, DcTHOR, and strain (dummy rib fractures).

Results: Unlike Hybrid III, the THOR with SD3 shoulder interacted with the side structure in a near-side oblique frontal impact. HIC values for the 3 test configurations corresponded to a 90% (A) and 100% (B and C) risk of Abbreviated Injury Scale (AIS) 2+ head injury, and BrIC values resulted in a 100% risk of AIS 2+ head injury in configurations A and B. In C the risk was reduced to 75%. The AIS 2+ thoracic injury risks based on Dmax were similar (14–18%) for all tests. Based on DcTHOR, AIS 2+ injury risk increased from 29 to 53% as the predeformed door side was introduced (A to B), and the risk increased, to 64%, as a PPT+DAB was added (C). Considering the AIS 2+ injury risk based on strain, tests in A resulted in an average of 3 dummy rib fractures (17%). Introducing the predeformed door (B) increased the average of dummy fractures to 5 (39%), but in C the average number of dummy rib fractures decreased to 4 (28%).

Conclusions: THOR with an SD3 shoulder should be the preferred ATD rather than the Hybrid III for evaluating head and thorax injuries in oblique frontal impacts. Thoracic interaction with the predeformed door was not well captured by the 3D IR-Traccs; hence, use of deflection as an injury predictor in oblique loading is insufficient for evaluating injury risk in this load case. However, injury risk evaluation may be performed using the strain measurements, which characterize loading from seat belt and airbag as well as the lateral contribution of the structural impact in the loading condition used in this study.

Keywords:

• near-side
• oblique impact
• THOR
• head injury
• thorax injury

Introduction

Oblique and small-overlap crashes are the second most common frontal crash type leading to fatal outcome after crashes with a large change in velocity (high delta V; Bean et al. 2009). The oblique or small-overlap crashes result in forward-lateral kinematics of the occupant and, if seated on the near-side adjacent to the impacted side of the vehicle, this oblique motion can result in serious injuries to the head and chest if contact occurs with the A-pillar, instrument panel, intruding door, or steering wheel or by seat belt loading (Hallman et al. 2011; Mueller et al. 2011; Rudd et al. 2011). For near-side frontal oblique crashes chest injuries can be associated with 2 mechanisms: seat belt and airbag loading and structural loading from the lateral side (Iraeus et al. 2013).

To assess occupant safety in the oblique and small-overlap type of crashes it is necessary to find a representative laboratory test capable of replicating typical vehicle damage and corresponding injuries found in real-world data (NHTSA, Insurance Institute for Highway Safety [IIHS]). To replicate structural vehicle behavior, 2 proposed test scenarios are currently being evaluated; the IIHS small-overlap test (IIHS 2012) and the NHTSA oblique moving deformable barrier test procedure (Saunders and Parent 2013; Saunders et al. 2013). In the IIHS small-overlap test procedure (IIHS SO) the Hybrid III 50th percentile male anthropomorphic test device (ATD) is currently used. The Hybrid III was specified whereby a comparative study found no difference in injury risk or occupant kinematics compared to the THOR-NT (Mueller et al. 2011). The THOR Mod-Kit dummy is currently being evaluated for the NHTSA oblique moving deformable barrier procedure (Saunders and Parent 2013).

The THOR Mod-Kit, including an updated shoulder (SD3), was evaluated within the European Union project THORAX. THORAX ATD status includes the Mod-Kit build level (Ridella and Parent 2013) with the addition of the SD3 shoulder and strain gauges attached to ribs 2–7 to detect rib bending complementary to the IR-Tracc measurements (Lemmen et al. 2012, 2013).

According to the NHTSA biofidelity ranking system (Bio Rank), the score was improved for the THOR Mod-Kit with SD3 shoulder (Mod-Kit_SD3) compared to the Hybrid III (Parent et al. 2013). Comparing the THOR Mod-Kit+SD3 to the Hybrid III dummy in diverse configurations, THOR showed a larger chest and head excursion and moved more outboard compared to the Hybrid III. The increased T1 movement and head excursion compared to the Hybrid III more closely represents the motion of PMHS tested in the same setup (Shaw et al. 2000; Yoganandan et al. 2011). The differing kinematic behavior and improved measurement abilities for thorax loading could provide deeper insights of occupant loading compared to previous ATDs, and these improvements should be a benefit of using the THOR Mod-Kit+SD3 in existing and new loading conditions (Sunnevång and Hynd 2014).

To assess the most frequent serious and fatal injuries observed in real-life oblique frontal crashes there is a need for an evaluation tool, such as a crash test dummy, that fulfills the following 3 criteria: (1) That the dummy kinematics allow for interaction of head and thorax with the vehicle interior as is expected in real-life crashes (and is suspected in injury-causing contacts). (2) The dummy should be capable of recording loading to the head and thorax from the seat belt, airbag, and vehicle interior. (3) Injury criteria with corresponding risk functions should be available to evaluate the measured loading.

The aim of this study was to investigate whether the THOR ATD met the above criteria to a greater extent than Hybrid III and by that measure whether the dummy could serve as a potential tool for future evaluation of serious head and chest injuries in near-side oblique frontal impacts.

Method

The Autoliv small-overlap test method described by Boström and Kruse in 2011 was used for evaluating the THOR and Hybrid III responses to near-side oblique frontal impact. The method consists of a sled system based on linkage arms to reflect the intrusion and occupant forces observed in small-overlap crashes. The sled pulse, generated by means of a pneumatic brake and iron bar bending, was tuned to mimic an average crash pulse of the IIHS small-overlap load case (Mueller et al. 2011), and the “vehicle interior” replicated on the sled was based on a 2001 Ford Taurus. Sled acceleration pulse and velocity for all tests are shown in Figure A1 (see online supplement). For this study the sled was positioned 26° counterclockwise to full frontal. To reduce impact complexity, the instrument panel was set to an initial position, and the test was performed without dynamic intrusion. The seat was a reinforced standard seat from a Volvo V70 and driver airbag and belt from a Volvo S40 and S60, respectively. Further details regarding the test setup and components are shown in the Appendix (Figure A3 and Table A-II, see online supplement). For reference, the side structure on the left (near-side) was initially nondeformed (subsequently denoted as test configuration A). In the following 2 test configurations (B and C) the side structure—for example, door panel—was predeformed to represent an oblique or small-overlap impact where the driver's door intruded. For images of the test setup and the predeformed door see Figure A2 (online supplement).

For the evaluation of occupant injury prediction the Test Device for Human Occupant Restraints (THOR) 50th percentile male ATD was used. The THOR used in these tests was one of the demonstrator dummies developed within THORAX (Lemmen et al. 2012, 2013). The THORAX THOR is similar to the THOR ModKit+SD3, further denoted as THOR. One test was also performed using the Hybrid III 50th percentile male dummy as a reference to the ATD currently used in regulation and rating procedures.

The response from THOR was evaluated for the driver position with combinations of restraints and door deformation; belt only (configuration A and B) and with a prepretensioning belt and driver (steering wheel) airbag (PPT+DAB; configuration C), as shown in Table 1. The test performed with the Hybrid III was carried out to investigate whether this dummy could interact with the predeformed door in a similar manner to THOR when a standard restraint system, PPT+DAB+Inflatable Curtain (IC), was used. The Hybrid III test was therefore only performed in configuration C.

Table 1. Test matrix

Test			Door	PPT
configuration	Test no.	Dummy	deformation	belt	Airbag
A	72	THOR	No	No	No
	42	THOR	No	No	No
	43	THOR	No	No	No
B	74	THOR	Yes	No	No
	44	THOR	Yes	No	No
	46	THOR	Yes	No	No
C	76	THOR	Yes	Yes	DAB
	47	THOR	Yes	Yes	DAB
	48	THOR	Yes	Yes	DAB
C	80	Hybrid III	Yes	Yes	DAB IC

In the THOR tests, head center of gravity acceleration as well as head angular rate sensors (according to Takhounts et al. 2013) in 3 directions were measured. In the thorax, 4 3D IR-Traccs (3D Infra-Red Telescoping Rod for the Assessment of Chest Compression) measured deflection in 4 quadrants (upper left and upper right, lower left and lower right). As a complement to the deflection measurement, 72 strain gauges (12 on each rib level, excluding the first rib level) recorded strain variations in the ribs. Accelerations in T1, T12, and the pelvis were recorded, as well as femoral forces. The dummy was painted before each test, with different colors for the head, arm, and thorax. The anterior and posterior part of the left side of the thorax were painted in different colors to record which body part contacted the door side. Belt forces (shoulder and lap) and sled acceleration were measured and each test was filmed with 4 high-speed cameras.

Injury Risk Calculation/Strain Gauge Processing

For head injuries the head injury criterion (HIC) and brain injury criteria (BrIC) with associated Abbreviated Injury Scale (AIS) 2+ injury risk curves were used (NHTSA 1995; Takhounts et al. 2013). For chest injuries the 3 injury criteria proposed by the THORAX project were evaluated (Davidsson et al. 2014a,b). The chest injury risk predictions were based on the number of fractured ribs sustained in the cadaveric sample of Davidsson et al. and therefore represent skeletal AIS more closely than overall (skeletal and soft tissue) thoracic AIS. The injury risk functions supporting the criteria used in this study are expected to relate to a risk of injury where at least 2 fractured ribs are sustained by a living human casualty and it is therefore, approximately, at least AIS 2 in severity according to AIS 1998 or 2005. For these chest injury criteria, Dmax is based on peak deflection and the X-deflection component from any of the 4 3D IR-Traccs. The DcTHOR, combined deflection, is based on the X-deflection component from all 4 IR-Traccs and calculated according to where Dm is the mean deflection of the ribcage, based on the 4 maximum deflections measured by the IR-Traccs in the X-axis as shown in Eq. (1)(1) . ULx (upper left X), URx (upper right X), LLx (lower left X) and LRx (lower right X) are the IR-Tracc X-component time histories with respect to the local coordinate system. (1) dD_up and dD_lw describe upper and lower thoracic twisting and are calculated using Eqs. (2) and (3). (2) (3)

THOR ribs 2–7 were equipped with strain gauges, 12 gauges per rib, 6 on the left-hand side and 6 on the right-hand side of the sternum and spine. During the loading event the rib was exposed to tension and compression over time. Signals were postprocessed to identify each peak value and determine peak strain for each rib. The strain-based criteria takes into account the additive effect of loading several ribs by counting how many ribs with gauges that exceed a value of 1.6 mstrain. This value was determined by computing regressions relating number of PMHS rib fractures to the number of peak strains exceeding a threshold on the dummy. Regressions were computed for various threshold values and the 1.6 mstrain value was found to lead to the higher R² value (Davidsson et al. 2014a). The number of dummy ribs exceeding this threshold (a.k.a. NFRdum) is used as risk predictor to relate the PMHS outcome to the dummy measurement in paired test by the mean of fitting a survival function with an assumed Weibull distribution. The number of dummy rib fractures is then used as the predictor variable (with a corresponding injury risk function) for the number of fractured ribs in a human and the relation to the AIS 2+ severity level.

Results

The Results section follows the 3 issues identified as necessary for a test tool in assessing injury risk from oblique frontal crashes: The first issue was to see whether the THOR, unlike the Hybrid III, could interact with the door side in this generic small-overlap test as presented in the Kinematic Comparison section; the second was to investigate whether the THOR dummy could measure loading produced by restraints and vehicle interiors to the head and chest, as presented in the Head and Chest Injury Measurements section; the third issue was whether injury risk functions could be applied to measurements to evaluate loading from restraints and vehicle interiors as presented in the Injury Risk Prediction section.

Kinematic Comparison

In test configuration A, the THOR dummy head and thorax traveled forward and outboard toward the door side. The head impacted the window line at 130 ms. No mark from the painted thorax was observed on the door. As the predeformed door was introduced, in test configuration B, the THOR head impacted the steering wheel and window line and the arm/thorax impacted the door side. Paint from the THOR indicated impact location. The thorax contacted the predeformed door at approximately 100–120 ms (time for peak left-hand strain) and the head contacted the window line at 125 ms. In configuration C, where THOR was restrained by a PPT+DAB, the thorax and head still impacted the window line and door side. The thorax impacted the door at approximately 100–120 ms and the head slipped off the driver airbag and impacted the window line at 135 ms. In the test using the Hybrid III, the PPT of the seat belt was activated as well as both the DAB and IC. The Hybrid III only interacted with the seat belt and was not close to contact with the door side.

Head Injury Measurements

The BrIC and HIC results for each THOR test are shown in Figure 1. Configuration A head impacts resulted in HIC₁₅ values of 584, 672, and 675 and BrIC values of 0.93, 1.29, and 1.45. For test configuration B the predeformed door was impacted, resulting in the higher HIC₁₅ values of 1316, 1995, and 1919. The BrIC values were calculated to 1.23, 0.90, and 1.00. In test configuration C, where the driver airbag was added, the head slipped off the airbag and struck the door, resulting in HIC₁₅ values of 1,178, 1,457, and 1,345. The BrICe values were calculated to 0.85, 0.61, and 0.63 (Figure 1). The HIC₁₅ for the Hybrid III was calculated as 166 (noncontact).

Fig. 1. THOR HIC and BrIC for the test configurations.

Fig. 2. THOR peak chest deflections in the x-direction (top), peak resultant deflections (mid), and DcTHOR (low).

Fig. 3. Strain peak value distribution for THOR in test configurations A (left), B (mid), and C (right). The line to the left represent the door side and the 3 bars on each rib represent peak strain from each test within the test configuration.

Chest Injury Measurements

In test configuration A, the THOR thorax did not interact with the door. Peak deflection in the x-direction was 30, 30, and 38 mm with the highest peak values measured at the upper right IR-Tracc. Taking y- and z-deflections into consideration, the peak resultant deflection was measured as 44, 47, and 50 mm. The combined deflection was calculated as 28, 28, and 38 mm. Strain peak values were associated with 2 dummy rib fractures in the first and third tests and 4 fractures in the second. In test configuration B, using the predeformed door side, peak chest deflection in the x-direction was measured to 27, 32, and 36 mm, and the upper right IR-Tracc and peak resultant deflection was 37, 38, and 45 mm. Combined deflection was calculated to 32, 35, and 57 mm. Strain peak values were associated with 6, 6, and 3 dummy rib fractures for the 3 repeated tests. Introducing PPT+DAB, peak chest deflections in the x-direction were 31, 32, and 39 mm at the upper right IR-Tracc, and the peak resultant deflection was 35, 43, and 44 mm (the peak resultant deflection was greatest at the lower right measurement position in this setup). Combined deflection was calculated to 38, 44, and 54 mm. Strain peak values resulted in 5 dummy rib fractures in tests 1 and 3 and 2 fractures in the second repetition. Peak deflections are shown in Figure 2, peak strain value distributions in Figure 3, and all THOR peak values in Table A-I (see online supplement). For the Hybrid III (configuration C) the peak chest deflection was 18 mm.

Injury Risk Prediction

In test configuration A the HIC values correspond to 87 and 93% (2 tests) of AIS 2+ head injury. For configurations B and C HIC values for all tests correspond to approximately 100% risk of head injury (NHTSA 1995). The BrIC values resulted in an approximately 100% risk of AIS 2+ injury to the head in test configurations A and B, and although the pretensioned belt and driver airbag was deployed in configuration C one test resulted in 90% risk and the other 2 showed an approximately 65% risk of AIS 2+ head injury. Head measurements and associated AIS 2+ and AIS 3+ injury risks are listed in Table A-I.

Fig. 4. AIS 2+ thoracic injury risk based on THOR measurements.

In Figure 4, the thoracic injury risk (for a 45-year-old occupant, based on the number of fractured ribs and representing the AIS 2+ severity level) for all tests per test configuration are presented. The average AIS 2+ thoracic injury risk based on Dmax was 16% (min 10%, max 27%), 14% (min 7%, max 21%), and 18% (min 10%, max 30%) for configurations A, B, and C, regardless of predeformation of the door and restraint system. Based on resultant deflection the average injury risk was 49% (min 34%, max 57%), 29% (min 17%, max 52%), 25% (min 12%, max 34%) for configurations A, B, and C, respectively. The AIS 2+ thoracic injury risk based on the DcTHOR showed an increased average risk, from 29% (min 19%, max 49%) to 53% (min 31%, max 87%), as the side deformation was introduced (configuration A to B) and the risk increased further, to 64% (min 12%, max 34%), using PPT+DAB (configuration B to C). Considering the thoracic AIS 2+ injury risk based on dummy rib fractures, configuration A resulted in the lowest number of dummy rib fractures (3) corresponding to an average of 17% (min 7%, max 28%) risk of injury. Introducing the predeformed door increased the number of dummy rib fractures (5), resulting in 39% (min 17%, max 48%) risk of AIS 2+ injury. For configuration C, with PPT+DAB, the number of dummy rib fractures was slightly lower (4), resulting in an average of 28% (min 7%, max 39%) risk of AIS 2+ injury. Chest measurements and corresponding AIS 2+ injury risk for each test are listed in Table A-I.

Discussion

Unlike the Hybrid III, THOR interacts with the side structure in a near-side oblique frontal impact as replicated in this study. Introducing deformation to the door side affected the injury measurements of the THOR for both the head and chest, whereas the Hybrid III head does not even contact the driver's (steering wheel) airbag. Based on these kinematic differences, THOR should be the preferred ATD to evaluate injury risk in oblique frontal impacts.

With the improved kinematic behavior, THOR has previously been shown useful in evaluating head injuries (Boström and Kruse 2011). In this study, HIC increased as the deformation of the door was introduced, due to earlier contact with the door side. When PPT+DAB were added, the head still hit the door side after sliding off the driver airbag. Considering the injury risk based on HIC, all tests resulted in a very high (above 85%) risk of AIS 2+ injury. For the BrIC measurements, the highest values were found in test configurations A and B. Using PPT+DAB the measurement decreased, and because the head is caught by the DAB AIS 2+ injury risk based on BrIC was reduced from 100% to approximately 70%.

The high injury risks to the head could be further reduced with a protection system, such as the Oblique Inflatable Curtain (OIC) designed to prevent contact with the vehicle interior in this type of load case (Boström and Kruse 2013). However, until now, evaluation of thoracic injuries has been impossible. In combination with the head protective airbag, most vehicles are equipped with side airbags protecting the thorax in a lateral collision. In the ideal situation (for safety and cost), the existing frontal and side airbags would offer thoracic protection in oblique impacts without need for further developments. However, as used in most vehicles, it is questionable whether the side airbag would have mitigated the thoracic interaction to the door side seen in this testing, due to the forward motion of the thorax.

Studies of real-life data have shown that in combination with head injuries, chest injuries are common in oblique crashes (Brumbelow and Zuby 2009; Lindqvist et al. 2006). The present study aimed to evaluate whether the THOR ATD was also suitable to evaluate thoracic injury risk in near-side oblique frontal impacts. The peak x-deflection values measured by the 3D IR-Traccs, Dmax, were similar for all tests independent of thorax-to-door contact, and hence the thoracic AIS 2+ injury risk predictions based on Dmax(x) were also similar. The peak value was always found in the upper right IR-Tracc due to the belt routing. The similar deflections and location for peak value were expected because the chest deflections in the x-direction corresponded to the belt load only. In the 3 configurations the belt characteristics were the same except that PPT was activated in configuration C.

Injury risk based on Dmax (x-direction) takes only the belt-induced risk into consideration, showing no increased risk due to interaction with the deformed door side. If measurements incorporating the axes other than the x-axis could offer a better discrimination of injury risk than the x-axis alone, then it could be important to consider the peak resultant deflections in future oblique and small-overlap testing applications. In these tests, the peak resultant deflection showed slightly more variation, with the highest deflection in configuration A followed by B and C. Beyond the x-axis, the contribution to the resultant deflection was mainly due to deflection in the z-direction, which was greater in the tests with most forward excursion. Contact with the door side in configuration B reduced excursion, as did PPT+DAB and door contact in configuration C. Therefore, injury risk based on resultant deflection still did not capture the thorax-to-door interaction in this test series.

When considering the combined deformation using all 4 IR-Traccs, the DcTHOR was greater as the predeformed door side was introduced. This was also the case for configuration C, even with the pretensioned belt and driver airbag. The AIS 2+ injury risk increased in configurations B and C compared to configuration A. Based on DcTHOR it seems as if PPT+DAB in configuration C generates larger differential loading on the thorax than the less advanced belt in configuration B, which may require further investigation.

The only measurements that clearly show the difference between the 3 test configurations are the strain gauge results. Comparing the peak strain value distributions in Figure 3 for the different test configurations (A–C) shows that for configuration A (belt only) the maximum loading occurs at the top right ribs, due to belt loading as the THOR moves outboard and forward without interaction with the door side. For configuration B, where the thorax interacts with the predeformed side structure, there is a substantial increase in the peak strain on the lower left side of the thorax while the right side rib peak strains remain similar (and in one test are even higher) to those in configuration A. For configuration C, PPT changes the kinematics of the head and thorax compared to A and B. Despite the earlier coupling to the seat and the driver (steering wheel airbag), the THOR still interacted with the door side, and high peak strain values were measured on the left side (ribs 4–6). Injury risk based on strain and number of dummy fractured ribs is a noncontinuous function and variations in dummy rib fractures can therefore result in large step changes in the estimated risk. This was the case for all test configurations in this study, which explains the large variation in injury risk based on dummy rib fractures.

Although the sled pulse and test setup were very similar in each test, there was a notable spread in the THOR measurements within each test configuration. The largest overall variations were found for the tests in configuration B and the least variation for tests in configuration A. In test configuration A the dummy is only loaded by the seat belt and can move rather freely forward and outboard. When the interaction with the predeformed door occurs, in test configuration B, the loading response of the dummy becomes more complex. In addition to spreading of the IR-Tracc measurements, the peak strain profiles in Figure 3 show a variation in peak strain on the right side of the thorax. This difference causes the variation in dummy rib fractures within each test configuration. Having the strain gauge information provides insight into the complexity of this loading condition. The difference in results is most likely a result of small variations in thorax impact location on the door, arm position (especially the left arm but also the right), and belt slippage during forward excursion.

• The THOR Mod-Kit with SD3 shoulder should be the preferred ATD before the Hybrid III for evaluating head and chest injuries in oblique frontal impacts. The limited kinematic behavior of the Hybrid III dummy prevents the expected interaction between thorax and door side even if the side is predeformed.

• In this test series, the THOR impact with the intruding structure occurred at approximately 100–120 ms, which is beyond the time of peak deflection in the x-direction, and the impact was not visible on the resultant deflection time history curve. Hence the use of deflection as an injury predictor in oblique loading will not be adequate for detecting injury-producing mechanisms from the restraint system and vehicle structure contacts in near-side oblique frontal impact conditions as studied here.

• Injury risk evaluation in near-side oblique frontal impacts can be performed using the strain measurements, which reflect loading from seat belt and airbag as well as the lateral contribution from the structural impact (e.g., door side).

Supplemental Materials

Supplemental data for this article can be accessed on the publisher's website.

Additional information

Funding

Parts of this study were conducted and financed through the THORAX project. The authors would like to thank all participating partners.