Advanced methods for improved child safety

A three- year old child (3YOC) was scanned with a medical scanner. The images were first semi-automatically segmented in order to extract the soft tissues and the bones. In the second step, we separate the different bones slice by slice on the model previously reconstructed. The anatomic structures are identified and each vertebra is reconstructed independently with special attention for the particular process. A polymer physical model of the bones is obtained with a prototyping machine and is used as geometrical reference. Our 3YOC finite element neck model is based on the geometry of an existing adult model developed by Golinski (2000) and modified by Meyer (2003). The scaling process is conduced in order to fit the finite element model dimensions as close as possible to the physical model. The anatomical structures modellised are the head, the seven cervical and the first thoracic vertebrae, the inter-vertebral discs and the principle ligaments. These are modelled using non-linear shock-absorbing spring elements. The stiffness values used are issued from Myklebust (1988) and Chazal (1985), adjusted to the model using the stiffness scaling factors found by Yoganandan (2000). Contact between the articular surfaces is represented by interfaces permitting frictionless movement. At least, the model is validated in comparison of the Q3 Madymo neck model.

The reconstruction database is a tool easy to handle to analyse the results of full-scale crash tests, component tests and numerical simulations to support task 3.3. TUB included all information coming from the partner, which performed it. All accepted 37 full- scale tests, some additional sled test, and the old CREST cases are already included in the database and every partner checked the information stored in the database. For several times TUB provided all partners an updated version of the crash test database.

This document is reporting results from an intensive testing programme on the decrease of efficiency of child restraint systems due to misuse. This deliverable is the compilation of reporting test results per misuse situation, of course, the effect of misuse can differ from one seat to another, due to design or material characteristics. An "ad-hoc" group including partners from the CHILD Consortium (PSA, RENAULT, VSRC, TRL, FIAT, Applus+IDIADA, INRETS, Chalmers, FTSS, TNO), and partners from outside the CHILD Consortium decided to work together in order to share experiences and to improve their knowledge about the CRS misuse. The way the ad-hoc has been working is based on exchange, of expertise and points of view. This represents different interests and approaches, to write a synthesis document on the effect of misuse in terms of child protection. A large dissemination of knowledge gained within this ad-hoc group is planned. The use of a car environment was often necessary to be as close as possible to the real life accident conditions. Only single misuse configurations (no combination) were considered for the moment and testing was base on volunteering according to the interests of the different participants. The test matrix, rather large and containing more than 80 misuses configurations was not completed but about 80 tests having been performed, representing 60 different misuse situations and compared to the standard installation. The group has been also working on interaction of CRS's with airbags and advanced safety devices but the test matrix was not yet completed. The group hopes to complete the matrix and make publications on the subject. Results dissemination: The dispatch of knowledge of this group is important for the improvement of the situation of children transported in car. The activity of the group has been presented on different occasions with different public and different approach of the problems. In addition technical groups dedicated to the improvement of child safety (ISO, EEVC) are informed of the work of this ad-hoc group on misuse. Main conclusions are that misuse situation is never safer than the normal use but seems better than no restraint at all. Misuse represents a high risk of additional injury for different body segments. That's the reason why the data coming from this ad-hoc group are available for education and training in the public area. According to the quality of the results obtained in the first phase, it is important to bring continue bringing new knowledge in the field of misuse and the corresponding danger for children. For this, test matrix for frontal impact could be completed by new test series, the work on advanced devices interaction with CRS should be enlarged and lead to a publication on the subject, and a similar work should be initiated on the effect of misuse in side impact. In addition the introduction of new items like psycho-sociological analysis of misuse should bring new ways in the understanding for the public of the importance of misuse. This will be only possible by the enlargement of the number of partners in the ad-hoc group.

In addition to the original deliverables of the CHILD project a further activity was undertaken relating to the misuse of child restraints. From the work carried out under Task 1.2, it was established that the misuse and inappropriate use of child restraints systems is widespread but relatively little is known about the effects of that misuse on the way the child restraint behaves in a crash. This output from the additional activity is a review of the prevalence of misuse and inappropriate use and recommendations for testing to evaluate the effects of misuse on child restraint performance in order to better understand the effects of misuse on the performance of CRS. The objectives of this activity were to: - Review all studies undertaken in Task 1.2 of the CHILD project, which considered misuse/inappropriate use of CRS, in order to establish types and rates of misuse and inappropriate use for each category of CRS. - Review studies of injury risks or causation in relation to misuse/inappropriate use of CRS. - Identify all cases within the CHILD database where there is evidence of misuse/inappropriate use of CRS and identify injury mechanisms in relation to CRS. - Identify cases within the CHILD database where misuse/inappropriate use was not identified but the crash circumstances were such that they could be easily replicated within tests where misuse/inappropriate use could be introduced. - Make recommendations for a testing schedule to investigate injury risks relating to misuse/inappropriate use. This review and the results from the misuse- testing programme begin to provide an understanding of the effects of misuse on the performance of CRS in use. Further research will increase the understanding of the effects of misuse, in order that these effects can be minimised or eliminated by design, education and awareness.

The literature review consists of a world-wide review of literature and information about child restraint system (CRS) use and misuse. In addition the report presents the findings from 2 observational field studies, which were carried out in France and Spain. The report incorporates national reviews of literature from the project member states (UK, Spain, Sweden, France and Germany) along with a review from some non-European countries. Issues addressed in this report, wherever possible, include the extent of use of different types of child restraint system, levels of inappropriate use (e.g. wrong CRS for size of child) and levels of misuse. This information was gathered from government figures, safety organisations, local government offices, parent groups, police forces etc. The objective of this report is to improve understanding of the issues concerning CRS use that may help in the recognition of modes of misuse and inappropriate use to enable more accurate reconstruction of accidents. Also, to provide a better understanding the potential implications of such factors on CRS design and on child safety as a whole. In addition to the lecture review, two surveys of use were undertaken. These field studies were carried out in France and Spain. The results include data about the various types of misuse observed according to parameters such as CRS groups (0+, G1)types and duration of trips, etc..

The first prototype of the instrumented abdomen was finished end of August 2003. Initially the twenty FlexiForce sensors were fixed with tape on the surface of a Q3-abdomen. Every sensor is assigned to a small area on the abdomen's surface. The effective local force can be calculated by using the measured pressure and the area. A series of sled tests with different child restraint systems was carried out to assess the reliability of the abdomen and the measured signals. The first design was very delicate, because of the low durability of the connection of sensor and wires. Nevertheless there was a good correlation of measured abdominal force, total abdominal force and lap belt force, especially for CRS with shield. The prototypes of the instrumented abdomen of the Q3 dummy are available and were already used by some CHILD partners. A third sensor for the Q6 is completed. The instrumented abdomen is the first try to measure abdominal loads directly on the surface of the abdomen with a series of single sensors. Besides the magnitude of the load also position and direction can be identified. The new approach with sensors protected by latex layers is much closer to a future commercial use of the instrumented abdomen than the first prototype.

In the field of protection improvement dedicated to children submitted to the risk of road collisions, the use of biofidelic dummies fitted with reliable sensors and the definition of appropriate protection reference values (PRV) for every body segment exposed to injury risk, is an essential objective. It is the reason why, INRETS is working in order to: enhance the dynamic response to compression of the thorax of the Q dummy family; improve the design and the dynamic response to compression of the abdominal insert; develop a new abdominal sensor and finally derive an injury criterion based on intra-abdominal pressure and rate of pressure change. At the end of the first year of the CHILD project, the concept and the design of the modified abdominal insert and the new abdominal sensor were in the process of completion. In order to decrease the stiffness of the lumbo-sacral joint, anterior abdominal wall and abdominal foam core will be separated. The new abdominal pressure sensor which has to detect lap strap and/or shoulder strap penetration in the frontal impact and lateral load in the lateral collision will be composed of a right and a left lobe. Every part will be composed of a cylindrical bag made of polyurethane rubber, filled with gel, pressurised with air at 15E-3Mpa and fitted with a current industrial pressure cell. The sensor will provide two pressure signals which will be exploited in order to calculate the value of the AIC (Abdominal Injury Criterion) =V (rate of pressure change)*P (pressure magnitude). The validation of the sensor and the derived injury risk curve will be carried out owing to the experimentations (accident reconstructions and complementary parametric sled tests) planned in the CHILD project. The design of INRETS's new abdominal pressure sensor for the Q3 is finished. INRETS's design is based on axonometric photographs and measurements taken directly on the physical dummy torso. The first prototype of the sensor has been manufactured by the company CLUZEL in Saint Bonnet de Mure near LYON. The INRETS's abdominal device can be used in addition with the piezo-electrical multi-point detector of TUB after modification of the abdomen (see drawing in annexe). By finishing the prototype of the sensor and the modification of the abdominal block at the end of February 2004, the instrumented abdomen will be provided to the partners at the beginning of April after evaluation and validation. A series of sled tests were performed in order to evaluate the features and capabilities of the new abdominal instrument in terms of dynamic response to belt penetration, durability, repeatability, sensitivity to deceleration, etc. The sensor provide relevant signals when used on the Q3 dummy as well as the Q6 (same bladders embedded in the respective abdominal inserts). Furthermore, the Abdominal Pressure Twin-Sensor has been currently utilized on the Q3 and Q6 dummies used for accident reconstructions. Correlations between abdominal injury severity sustained by the child and Injury Criteria value P*V seem to be acceptable. The whole set of values acquired on the planned experimentations (full scale tests and parametric sled tests) and the logistic regression law calculated from these values (Injury Risk Curve) will indicate whether the pressure sensor and the proposed criteria are reliable and can be used for predicting injury risk at abdomen.

This document is describing the methodology and main results of the works conducted in the CHILD project on the development of injury criteria for different body regions. These results are essential for the development of new test procedures to evaluate the level of protection offered by child restraint systems. The source of data is the reproduction of real life injuries sustained by restrained children in accidents cars, through accident reconstructions performed in crash test laboratories. These injury criteria and procedure works have to be reflective of the conditions of loads of child dummies in physical accident reconstructions. Results of the previous EC CREST program are the base of the work of this task and they were completed with new tests performed during the CHILD project, which correspond to the needs of data in term of injured body segments per type of impact. Head, neck, chest and abdomen were considered as body segments to be protected in priority. 38 new accident reconstructions of accidents were performed, because of the interest for the biomechanical data they were bringing. As the objective of this task was to identify the physical parameters associated with various child injury mechanisms specific analysis in frontal and side impacts conditions for all body segments were conducted. Injury risk curves were constructed or at least to limit were recommended and thresholds for injury have been defined when sufficient data were available. Different methodologies were used to construct injury risk curves: the certainty method, the CTE (Consistent Threshold Estimate) and logistic regression, in order to see the influence of the method on the results. The injury risk curves constructed by certainty method were presented within the CHILD Workshop, the results of all methods are presented in the document.

One of the aims of the CHILD project was to develop a sled test procedure for determining the effectiveness of Child Restraint Systems (CRSs) in frontal impacts and that was representative of the accidents in the CHILD database. This report is providing details on the CREST frontal test procedure and on the work conducted in the CHILD project to ensure that the testing conditions were reasonably similar in CREST and CHILD accident reconstructions to be still the best compromised to evaluate by the performance of a single sled test the efficiency of child restraint systems in frontal impacts. The test procedure was expected to specify the geometrical arrangement of the test seat, together with the impact conditions, which needed to be replicated. Although the CREST procedure did not aim first to be used in CRS certification legislation, it is based on ECE Regulation 44 (R44; Economic Commission for Europe; 1981, as amended). However, changes to the R44 procedure were proposed in order to meet the objectives of the CREST project. The profile for the CREST deceleration corridor was developed around the pulses recorded during accident reconstructions, which lead to a higher plate for the deceleration of the sled, with a change in velocity of 55 km/h and a stopping distance of 650 to 700mm. Modifications were proposed for the test bench shape in order to make it more representative recent cars, this included an anti-sub-marining device. The CHILD frontal test procedure is based on the use the Q-series dummies, which were to be used in the evaluation of the injury criteria that had been developed in another CHILD project Work-Package. Whilst differences may have existed between the geometry of the vehicles in the CREST and CHILD accident databases, these were not considered within the development of a CHILD frontal sled test procedure. The differences were thought to be of only very minor consequence at the CRS and occupant level. Instead, the key element was considered to be the deceleration pulse. As noted with the development of the CREST test procedure, the accidents in the CHILD accident database tended to be quite severe on account of the selection criteria used and the acceleration profile to be typical of modern vehicles. This means that the accident cases are not representative of the whole real-world accident situation. On this basis, the acceleration profile developed for use in the CHILD project, would not be comparable to the levels of the current legislation used for the certification of child restraints. A few of the accident reconstruction cases conducted in the CHILD project had considerably different deceleration pulses compared with the CREST corridor profile. The most significant of these differences were explained through consideration of the accident scenario. In the majority of cases, the reconstruction pulse was close to the corridor. However, even for these cases, the pulses did not meet the corridor exactly and a general trend was observed for the decrease from peak value to occur later than was suggested by the corridor. Following consideration of the test conditions and in particular the impact pulse, and on the basis that only little differences would be observed at the CRS and occupant level in sled tests using either the CREST acceleration profile or a new CHLD profile, the CREST corridor can be considered to represent the CHILD accident reconstruction cases. It was accepted that some differences would be expected in the occupant response at this time, the CHILD project partners considered that little difference would be observed up to this point. Therefore peak excursion and instrumentation values used in injury risk prediction would show little difference in tests conducted with either the CREST or a CHILD acceleration profile, assuming that other conditions were consistent. It is then possible to recommend that the CREST frontal impact test procedure would be appropriate for investigating the injury criteria being developed in another CHILD WP.

TUB developed a validated numerical model of the newborn child dummy Q0. This model can be used for occupant simulation tasks to analyse and improve the performance of child restraint systems. The model uses FEA techniques and was validated by using results of component tests of all body segments. Until now a preliminary version for the use with LS-Dyna is available. The MADYMO version is completed too.

Two related presentations will be made at the Protection of Children in Cars 4th International Conference, in Munich on the 7-8 December 2006. The first presentation is 'CHILD: Analysis of CHILD data related to frontal impacts', Kirk, A. and Grant, R., VSRC Loughborough University, Lesire, P. LAB, Johansson, M., Chalmers University of Technology. The second is 'CHILD: Analysis of CREST and CHILD data related to side impacts'. Lesire, P. LAB, Herve, V., CEESAR, Kirk, A., VSRC, Loughborough University. The audience of this conference is made up of representatives from the scientific community, the automotive industry, child restraint manufacturers and others involved in child safety. The information presented will be available to inform the development of vehicles, child restraint systems and education and awareness campaigns to improve the safety of children in cars.

The analysis report has two main objectives, to describe the information available from the accident data collection activity of the CHILD project (and its predecessor, CREST), and to provide an initial analysis of the available child accident data. During the CHILD project, a total of 264 accident cases involving restrained children have been gathered from France, Spain, Germany, Italy, Sweden and the UK, according to a common and well defined methodology, including a sampling plan. Injuries are coded according to the Abbreviated Injury Scale (AIS) (AAAM, 1990). When added to the 405 CREST accident cases, there is now a total of 669 cases. The cases found in the CHILD dataset are not proportionally representative of the accident situation across Europe, or in individual countries. Case selection is made using criteria that favour more severe cases, both in terms of injury severity and impact severity, although cases that combine low injury severity and high impact severity are also included. The report includes an overview of the CHILD and CREST datasets, an analysis of restrained children in frontal impacts using the CHILD dataset and an analysis of side impacts using the combined CHILD and CREST database. This analysis report provides the most recent and comprehensive European study of accidents involving restrained children in cars. The report is available to inform the scientific community and the public about the circumstances in which restrained children are injured in more severe car accidents.

The CHILD reconstruction database is a universal tool to store, manage and analyse data of all crash tests and sled test performed in CREST and CHILD. It is a relational database that allows the storage of all information related to the accident, the general test conditions, vehicles (static and dynamic data) and on child dummies. For each vehicle involved in a test, the database contains information related to vehicle characteristics, test severity (EES, angle,...) and static deformations measured on it. A large number of photographs of the vehicle (before and after test) can be stored. A lot of information for each child dummy is available, including the list of injuries sustained by the child in the original accident, the main environment of the dummy, the restraint system used for the test, the list of sensors on the child dummy for the test, and pictures of the dummy and its restraint system. The software includes following features: Visualization of data (pictures, films, and curves) from crash tests, component tests, or simulation and user is guided through convivial screens when entering or consulting data with the software. In order to make the tool easier to use and to ensure a good coherence between different users, index tables have been created for user interface to enter information related to test laboratories, test engineers, car manufactures and models, and numbering of tests (new crash or component tests). This tool is offering the possibility to import ISO MME channel data with automatic calculation of a3ms, HIC36, HIC15 and the minimum/maximum values, import or export selected cases to another CHILD database (data exchange), delete cases and offer filters to select a case. In addition to all data gathered during the CHILD project, these tools have been designed in order to be able to contain all results of all tests performed in the previous EC CREST project (56 reconstructions and about 100 sled tests). The analysis of data is possible in the reconstruction database, data can be filtered and graphical visualisation of AIS and measurement values is offered. Reports can be automatically generated by a simple click and the edition of general data for crash and component test, pictures, main results and a comparison between real accident data and crash or component test data is done. The developed software is a tool to analyse the results of crash test with child dummies involved. It was designed to fit to the demands of the CHILD partners. Test was performed in more than ten different European test laboratories and the software has allowed that all data was entered in the reconstruction database by the database manager. This tool is easy to use and give plenary satisfactions to the partners who have used it. It can be considered as a good basis for crash data storage, management and visualisation for future research works in similar context.

The database for collection and analysis of real accidents was developed with Microsoft Access 2000. It is a relational database that allows the storage of all information related to the accident divided in three main categories: general information of the accident, vehicles data and occupants' data. For each vehicle involved in the accident the database contains information related to: vehicle characteristics, accident reconstruction results (EES, delta velocity,...) and static deformations measured on it. A large number of photographs of the vehicle (and of accident site) can be stored in order to better understand the behaviour of the vehicle during the accident and as well a lot of information for each occupant either adult or child, including anthropometric data, list of injuries coded according to an International coding system. In order to allow to the other workpackages the use of these data, several details related to the child environment in the car are also stored in the database: front seat position, seatback inclination, restraint conditions, airbag type and status, misuse of child restraint system (CRS), description of CRS, damage of CRS, appropriate use, and many pictures of both the seating place in the car and of the CRS to permit a better evaluation. In order to guarantee the coherence of data introduced into the database, a database software tool of management was developed with Microsoft Visual Basic 6. This software tool was designed mainly to analyse all data stored into the database and see one complete accident in each component or to perform statistical queries or case selection according to criteria.

When the CHILD project started, a newborn dummy was not yet available for use in accident reconstructions. TNO was responsible for the development of the Q0-dummy, a dummy representing a newborn child as part of the Q-series. The Q0 dummy is based on up-to-date biomechanical and anthropometrical data and knowledge. It has a total body mass of 3.4kg and a biofidelity requirement for head and neck have been derived. These two body parts are the most vulnerable of a newborn in a car. The Q0 is provided with extensive instrumentation possibilities: head, thorax and pelvis accelerometers and an upper neck load-cell. A limited test program has proven the durability, repeatability and biofidelity of this dummy. At the moment three prototypes are ready for the reconstruction work of the CHILD program. An evaluation phase will run parallel with the reconstruction work.

The results of the comparative testing of different test procedures show that the original CHILD procedure is not sensible as the dummy readings are quite low and there are almost no possibilities to differentiate between different qualities of seats. The comparison of the both modified CHILD procedures (high-g and high-g and fixed door) shows that the simulated intrusion has almost no influence on the dummy readings in the non worst-case configuration therefore it seems to be unfair to use a test procedure with intrusion, when it does not influence the results significantly. The fixed door derivate of the CHILD procedure and the modified NPACS procedures seem to be the best choice based on the test results mentioned above. Both represent an adequate severity level and allow distinguishing between different seats. The main advantage of the fixed door procedure is the simple set-up, which promises high reproducibility. However, based on the experience from the NPACS project, where the ADAC test procedure (fixed door procedure but in an angle of 80°) was compare with car and hinged door tests, it is reasonable to expect a completely different situation when testing RF CRS. The NPACS protocol is currently subject to ISO standardisation. During this process it is very likely that the intrusion velocity will be reduced compared to the NPACS settings. With respect to harmonisation it is reasonable to propose a side impact test procedure, which is already in use. As the CHILD proposal is meant to form, as base for legislation and NPACS is a consumer test, there are good reasons to reduce the severity level compared to NPACS. For these reasons the modified TUB/NPACS procedure with reduced angular velocity was selected as the CHILD side impact test procedure.

A 3 YOC neck FE model developed by ULP in RADIOSS platform was transfer to MADYMO platform and combined with Q3 dummy head model. The validity of the head-neck model was evaluated with the Q3 dummy head neck mini sled tests done at TUB in loading conditions for neck flexion, lateral bending and extension. The kinematics and calculated parameters from head-neck model were compared with measured acceleration of head CG, neck force and moment at OC joint in three directions. A parametric study of 3YOC neck FE model was carried out for different input crash pulse levels, the resultant head CG acceleration and Nij criteria are used for analysis of the head and neck injury risk. The 3YOC neck FE model is expected to be used for prediction of neck injury risks.

In order to identify injury mechanisms and to establish injury risk curves, accident analyses are carried out and afterwards accident reconstructions performed in laboratories. Crash test conditions such as equivalent energy speed (EES), overlap, angulations and body vehicle heights are assessed by experts and partly based on empirical methods. Moreover, particularly in the case of injured children using CRS's, parameters such as adjustment of belt or harness and especially misuses are difficult to determine. Consequently, it seemed necessary to develop methods aiming at eliminating those approximations leading to weak correlations. It is the case of accident speed, overlap and angulations, which have an effect on the car(s) deformations and consequently on the loads sustained by the occupants. Over the years, a lot of effort has been devoted to increase the accuracy of the evaluation of these accident parameters from accident scene evidences (see for instance McHenry et al., 2003 or Moser et al., 2003). But in the evaluation of the quality of the reconstruction, the deformation sustained by the vehicle(s) in the reconstruction cannot play an important role, either because they are calculated by simulation or they are considered globally. It seems that no systematic approach based on the study of the deformation of well- identified vehicle structural points has been tried so far. The use of this new method requires that vehicle deformations has been measured on the crushed parts, compared and submitted to basic statistical functions such as average value, standard deviation and variance in order to establish a quality score of the reconstruction. This method is intended to help experts to assess the quality of the reconstruction of a real world accident in terms of correlation of dissipated energy between a vehicle involved in a real world accident and its homologue used for the reconstruction. For this purpose, a "Reconstruction Quality Score" based on the deformations of the main relevant vehicle body parts -- longitudinal members, damper housing, A pillar, foot well, etc, is calculated. There is an infinite number of ways to calculate a reconstruction quality from the comparison of vehicle deformations. The present software, after an extensive comparison of various candidate indicators and score weighing methods, led to define a composite score based on the absolute values of the deformation differences, the relative values of the deformation differences. Weighing factors depending on the deformation variability at each considered point and depending on its position with respect to the impact point. For each point, a score is computed from an absolute and a relative deformation indicators and then, all scores are weighted and mixed in order to give a global quality score. At present, a software developed under Java and a database developed under MySQL are available for the partners of the CHILD consortium for frontal crash analysis. Further development will be undertaken for the lateral crash configuration

Risultati finali

Condividi questa pagina

Scarica