Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Development of a Head Injury Criteria-Compliant Aircraft Seat by Design of Experiments

Development of a Head Injury Criteria-Compliant Aircraft Seat by Design of Experiments aerospace Article Development of a Head Injury Criteria-Compliant Aircraft Seat by Design of Experiments 1 2 2 , 1 Giuseppe Lamanna , Amalia Vanacore , Michele Guida * , Francesco Caputo , 2 3 3 Francesco Marulo , Bonaventura Vitolo and Salvatore Cicatiello Department of Industrial and Information Engineering, University of Campania Luigi Vanvitelli, Aversa, 81100 Caserta, Italy Department of Industrial Engineering, University of Naples “Federico II”, 80125 Napoli, Italy Geven, Zona Asi Boscofangone, 80035 Nola, Italy * Correspondence: michele.guida@unina.it Received: 18 June 2019; Accepted: 30 August 2019; Published: 3 September 2019 Abstract: This paper deals with the redesign of an aircraft passenger seat, placed at the first seat row, which was not compliant with Federal Aviation Regulations FAR 25.562 “Emergency landing dynamic conditions” regulation (due to a high value for the Head Injury Criterion (HIC)) and related guidelines. Starting from an accurate analysis of some results obtained via an experimental seat sled test, a numerical procedure was developed in order to improve the passenger safety with respect to head injury. Specifically, the proposed numerical procedure, using the advantages of a Finite Element (FE) model and a Design of Experiment (DoE) approach for simulation modeling, was aimed at identifying a new design solution to avoid the impact between the passenger ’s head and the bulkhead. The redesign of the passenger seat was validated against an experimental test carried out at Geven S.p.A. Company by demonstrating, consequently, the compliance of the modified seat-belt system with the regulations. Keywords: multibody systems; finite element analysis; aircraft seat; design of experiment; head injury 1. Introduction Crashworthiness requirements drive the design of most automotive and aerospace structural components in order to improve their passive safety performance and, consequently, to protect occupants against injuries and death during a crash event. Design e orts are increasingly focussed on satisfying these requirements, even if passive safety criteria lead to a small increase in the structure’s weight. Many research studies involving experimental tests and numerical studies are carried out to verify that the proposed design solutions respect crashworthiness requirements and to predict the injury level for occupants in case of an accident [1,2]. In the aeronautical sector, the sled is considered as an appropriate system to check the protection of occupants, and a crashworthy seat has to guarantee that the consequences of occupant injuries are not fatal [3]. Currently, aeronautical regulations consider the experimental test to be the only possible solution to certify an aircraft seat. For lighter aircraft (take-o weight of 5670 kg or less), two tests are required [4]: the first test is based on the loads transferred to the occupant during a forward deceleration; the second test considers the downward deceleration. For heavier aircraft (take-o weight above 5670 kg) [5], a test related to the deformation of the floor is required in addition to the previous two tests. Whatever the airplane category, it is necessary to evaluate the entity of the damage in the presence of a head impact. For this purpose, the experimental test for component certification can take advantage of the high level of technology introduced in the current Anthropometric Test Devices (ATDs), which allow the recording of the occupant’s response (i.e., kinematical as well as Aerospace 2019, 6, 95; doi:10.3390/aerospace6090095 www.mdpi.com/journal/aerospace Aerospace 2019, 6, 95 2 of 12 dynamical parameters) during the experimental simulation of a crash event [6]. However, it is evident that the fitting and the implementation of these tests are time-, space- and cost-demanding, and thus it is not a ordable to experimentally test several configurations of a proposed design solution. For all these reasons, the development of a numerical model capable of reproducing such complex experimental tests, according to the Certification by Analysis paradigm, seems to be the best strategy for the development of a crashworthy design solution. The main advantage provided by numerical methods consists in the possibility to identify an optimal virtual solution that can be validated via experimental tests, avoiding the “trial and error” approach typically adopted by companies and, consequently, reducing time and cost for development. Of course, the development of a reliable numerical model requires many e orts in dealing with the underlying assumptions and hypotheses as well as in the necessary availability of high computational power, complemented by high skills in modeling. Among the several structural components that can significantly contribute to occupant safety, the seat plays a key role in reducing the loads transferred to passengers during the crash as well as in reducing the probability of the passenger ’s head impact against the bulkhead, in the case of the first seat row. In [6], the authors propose a multibody model of the aircraft seat structure for the simulation of a 16-g compliance with the Head Injury Criteria (HIC) requirement. This test involves an impact against a bulkhead developed and analyzed by using an ad hoc algorithm implemented in MATLAB code, as a 2D system of rigid bodies interconnected by springs and joints. Lankarani, in [7], focuses on the design and development of bulkheads evaluating various honeycomb materials for HIC attenuation, rather than on high costs and schedule overruns due to the development and certification of aircraft seats. The aim of this paper is to define a numerical procedure to improve the design of an aircraft passenger seat, considering passive safety as the main goal. For this purpose, di erent numerical methods can be used, including Finite Element (FE) and Multibody (MB) models. MB numerical models are generally adopted to evaluate the kinematics of the dummy (representing the occupant) and its interactions with both the restraint systems, if available, and the seat cushion, with no possibility whatsoever to obtain any results about the structural behavior of the seat frame. These models exhibit the convenience of allowing quick modifications of the analyzed configuration and obtaining reliable results—within the recalled limits—with very short runtimes. Numerical FE models are, instead, required to obtain information about the structural behavior of the seat. Even if the developing time of these models is considered acceptable for current design time scheduling, the same cannot be said for runtimes. In particular, runtimes are very long for applications of dynamical type and for the use of anthropomorphic dummies, whose models prove to be very complex and therefore require very accurate discretization and high mesh densities. Some solutions involving the coupling of both methods are proposed in literature [8,9]. The research activity presented in the present paper makes use of the FE and MB methods either by an internal explicit code (hybrid) [8] or independent code (coupling) [9], with a di erence in terms of computational cost and correlation with experimental data. Specifically, the research starts from an established FE model presented by authors in previous papers [8]. This FE model allows the description of the kinematics of a passenger as well as the injuries that a passenger seated at the first row of an aircraft may su er in a frontal impact against the bulkhead. A corresponding biomechanical head injury index was calculated and compared with the extreme one considered according to the current regulations. The investigation was also experimentally carried out by launching an aircraft seat, equipped with an ATD, against a bulkhead at the required speed and the corresponding acceleration/deceleration profile. The experimental data were used to calculate a damage index, as described below, useful for the evaluation of the level of injuries a ecting di erent body parts directly involved in the impact or just subjected to high inertia loads. The good level of accuracy of the developed FE model was demonstrated by comparing numerical results to experimental data. Aerospace 2019, 6, 95 3 of 12 Aerospace 2018, 5, x FOR PEER REVIEW 3 of 11 According to the experimental test presented in [8], the calculated HIC (Head Injury Criterion) According to the experimental test presented in [8], the calculated HIC (Head Injury Criterion) parameter, representing the head injury, was higher than the limit provided by AC 25.862 and SAE parameter, representing the head injury, was higher than the limit provided by AC 25.862 and SAE 8049b because the head of the dummy contacted the rigid bulkhead. The analysis of both numerical 8049b because the head of the dummy contacted the rigid bulkhead. The analysis of both numerical results and experimental data suggested that this problem could be related to the sti ness characteristics results and experimental data suggested that this problem could be related to the stiffness of some seat components, which have been thus selected as design factors to be further investigated. characteristics of some seat components, which have been thus selected as design factors to be further Based on the validated FE model and the adoption of an ecient experimental design, new numerical investigated. Based on the validated FE model and the adoption of an efficient experimental design, new experiments numericwer al exp e carried eriment out s were in or cder arrito ed find out in a new order design to find solution a new des forig the n sol seat utiframe on forto thimpr e seaove t frame the passenger passive safety. All analyses were carried out using Ls-Dyna software. In the following to improve the passenger passive safety. All analyses were carried out using Ls-Dyna software. In th sections, e follow the ing baseline sections, experimental the baseline exp study eriment , the aFE l stud model, y, the the FE strategy model, th for e strate planning gy for the plannin numerical g the experiments and the obtained results are described and discussed. numerical experiments and the obtained results are described and discussed. 2. Baseline Study and Dynamic Testing 2. Baseline Study and Dynamic Testing A metallic seat frame fabricated in aluminum alloy was used in the dynamic sled test. To maximize A metallic seat frame fabricated in aluminum alloy was used in the dynamic sled test. To the energy transferred to the head, no yaw was given to the seat in these tests. The experimental setup maximize the energy transferred to the head, no yaw was given to the seat in these tests. The of the seat, shown in Figure 1, is representative of a typical airline economy class seat with the seat experimental setup of the seat, shown in Figure 1, is representative of a typical airline economy class back fixed in the upright position. seat with the seat back fixed in the upright position. The seat setback distance—defined as the horizontal distance between the seat reference point The seat setback distance—defined as the horizontal distance between the seat reference point (i.e., the intersection point between the seat back and the seat pan) and the outer surface of the (i.e., the intersection point between the seat back and the seat pan) and the outer surface of the bulkhead—was fixed at 583 mm (23 inches). bulkhead—was fixed at 583 mm (23 inches). Figure 1. Experimental setup. Figure 1. Experimental setup. The seat cushions used during the test consisted of foam with the static deformation under the passenger weight set to a maximum penetration of 5 mm; a typical polyester seat belt was used. The The seat cushions used during the test consisted of foam with the static deformation under the pass sample enger we full-scale ightsled set to a m test setup aximum with pe the net bulkhead ration of is 5 mm shown ; a typ in Figur ical po e 2 ly . e A ster se triaxial at belt wa accelerometer s used. was The mounted at the center of gravity (c.g.) of the ATD head to determine the resultant head acceleration. sample full-scale sled test setup with the bulkhead is shown in Figure 2. A triaxial accelerometer was mounted at the center of gravity (c.g.) of the ATD head to determine the resultant head acceleration. Aerospace 2019, 6, 95 4 of 12 Aerospace 2018, 5, x FOR PEER REVIEW 4 of 11 Aerospace 2018, 5, x FOR PEER REVIEW 4 of 11 Figure 2. Full-scale sled test setup. Figure 2. Full-scale sled test setup. Figure 2. Full-scale sled test setup. The experimental test sessions described in this section were conducted at the Impact Dynamics The experimental test sessions described in this section were conducted at the Impact Dynamics The experimental test sessions described in this section were conducted at the Impact Dynamics Laboratory of Geven S.p.A. with the aim of measuring the head accelerations of a Hybrid-II Labor Laboratory atory o of f Gev Geven en S S.p.A. .p.A. w with ith th the e ai aim m of of me measuring asuring th the e he head ad acce accelerations lerations o of f a a Hy Hybrid-II brid-II Anthropomorphic Test Dummy (ATD), as specified in Certification Specifications CS 25.562, [5], so Anthropomorphic Test Dummy (ATD), as specified in Certification Specifications CS 25.562, [5], so Anthropomorphic Test Dummy (ATD), as specified in Certification Specifications CS 25.562, [5], as to support the development of the FE model and to assess the performances of the new seat design as so to as support to support the develop the development ment of thof e FE the mo FE del model and to and assess to assess the per the formance performances s of the new of the seat new design seat solution. solution. design solution. The experimental tests were carried out using a sled deceleration system pneumatically Th The e experimental experimental tests testwer s we ere carried carried out using out ua sisled ng a deceleration sled deceler system ation pneumatically system pneumatic activated ally activated like a “sling”. The seat with a dummy is fixed on the sled, and both together are launched activated like a “sling”. The seat with a dummy is fixed on the sled, and both together are launched like a “sling”. The seat with a dummy is fixed on the sled, and both together are launched at an at an assigned speed. The race is stopped by a mechanical brake (decelerator system) made with a at assigned an assigne speed. d spee The d.race The is rastopped ce is stop by ped a mec by a hanical mechabrake nical br (decelerator ake (deceler system) ator system made ) made with a with lower a lower carbon steel wire, which is properly assembled in order to have a g-peak compliant with the lower carbon steel wire, which is properly assembled in order to have a g-peak compliant with the carbon steel wire, which is properly assembled in order to have a g-peak compliant with the rules. The rules. The number, position, and length of each steel wire can affect the deceleration pulse shape. rul number es. Th , e position, number, and posi length tion, and of each length steel of wir eae ch can stee a l ect wire thecan deceleration affect the pulse decele shape. ration This pulse HIC shape test . This HIC test can be performed in two different setup configurations: This HIC test can be performed in two different setup configurations: can be performed in two di erent setup configurations:  First row test, in which the seat is placed in front a rigid bulkhead at a fixed distance in order to  First row test, in which the seat is placed in front a rigid bulkhead at a fixed distance in order to First row test, in which the seat is placed in front a rigid bulkhead at a fixed distance in order to simulate a typical first row installation inside the cabin (Figure 2). In this case, the main purpose simulate a typical first row installation inside the cabin (Figure 2). In this case, the main purpose simulate a typical first row installation inside the cabin (Figure 2). In this case, the main purpose is to verify the head contact, while the obtained HIC value can be considered only as a reference is to verify the head contact, while the obtained HIC value can be considered only as a reference is to verify the head contact, while the obtained HIC value can be considered only as a reference value since the installed bulkhead is not the real cabin installation. value since the installed bulkhead is not the real cabin installation. value since the installed bulkhead is not the real cabin installation.  Row to row test, in which two seat rows are fixed on the sled at a proper distance. The aim of  Row to row test, in which two seat rows are fixed on the sled at a proper distance. The aim of Row to row test, in which two seat rows are fixed on the sled at a proper distance. The aim of the the test is measuring the HIC value during the head impact in order to evaluate the potential the test is measuring the HIC value during the head impact in order to evaluate the potential test is measuring the HIC value during the head impact in order to evaluate the potential injury injury related to the design of the seat backrest including mounted equipment (monitor, rear injury related to the design of the seat backrest including mounted equipment (monitor, rear related to the design of the seat backrest including mounted equipment (monitor, rear table etc.). table etc.). table etc.). As for the United States Code of Federal Regulations (CFR), a triangular deceleration pulse with a As for the United States Code of Federal Regulations (CFR), a triangular deceleration pulse with peak of 16 g and a rise time of 90 ms was targeted for the sled tests. The ideal pulse shape and the As for the United States Code of Federal Regulations (CFR), a triangular deceleration pulse with a peak of 16 g and a rise time of 90 ms was targeted for the sled tests. The ideal pulse shape and the a actual peak sled of 16 test g and deceleration a rise time pu of lse 90 for ms the was two targeted baseline for tests the sled are te shown sts. Th in e ide Figur al puls e 3. e Pr sh oper ape seat and belt the actual sled test deceleration pulse for the two baseline tests are shown in Figure 3. Proper seat belt actu installation al sled te requir st deceler ed aatio testn rig puls able e for to th guarantee e two baselin the corr e test ect s position are shown of in airFigur craft/e belt 3. Proper interface seat points belt installation required a test rig able to guarantee the correct position of aircraft/belt interface points insta withllation respect re to quire the d seat. a test rig able to guarantee the correct position of aircraft/belt interface points with respect to the seat. with respect to the seat. Figure 3. Deceleration pulse. Figure 3. Deceleration pulse. Figure 3. Deceleration pulse. Aerospace 2019, 6, 95 5 of 12 Aerospace 2018, 5, x FOR PEER REVIEW 5 of 11 All devices were equipped with accelerometers and load cells to measure forces and acceleration All devices were equipped with accelerometers and load cells to measure forces and acceleration affecting the most sensitive human body parts. Particular attention was paid to the acceleration of the a ecting the most sensitive human body parts. Particular attention was paid to the acceleration of the head and to the loads transmitted to the lower limbs of the dummy in order to verify the accuracy of head and to the loads transmitted to the lower limbs of the dummy in order to verify the accuracy of the FE model proposed in the previous papers [9,10] whose main characteristics are briefly described the FE model proposed in the previous papers [9,10] whose main characteristics are briefly described in the next section. in the next section. 3. Numerical Models 3. Numerical Models The full seat FE model consisting of 105,226 elements and 151,219 nodes is shown in Figure 4. The full seat FE model consisting of 105,226 elements and 151,219 nodes is shown in Figure 4. All All structural seat components were modelled by considering aluminum alloy material, whereas structural seat components were modelled by considering aluminum alloy material, whereas foam foam material was considered for cushions, ref. [11]. material was considered for cushions, ref. [11]. Figure 4. Finite Element (FE) model. Figure 4. Finite Element (FE) model. For each material, an elasto-plastic model was selected; the constitutive curves of each material are For each material, an elasto-plastic model was selected; the constitutive curves of each material shown in Figure 5. The aluminum alloy adopted is an elasto-plastic material with kinematic hardening are shown in Figure 5. The aluminum alloy adopted is an elasto-plastic material with kinematic (model n. 24 Ls-dyna’s material library), and failure is defined based on the plastic strain. The foam hardening (model n. 24 Ls-dyna’s material library), and failure is defined based on the plastic strain. material considered is a rubber-like foam of polyurethane. It is a simple one-parameter model with a The foam material considered is a rubber-like foam of polyurethane. It is a simple one-parameter fixed Poisson’s ratio of 0.25. model with a fixed Poisson’s ratio of 0.25. The deceleration pulse was applied to the nodes of both the bulkhead and the seat fixed to the The deceleration pulse was applied to the nodes of both the bulkhead and the seat fixed to the slide. Gravity and initial velocity were applied to all parts of the model. slide. Gravity and initial velocity were applied to all parts of the model. Figure 5. Sigma-epsilon curves. Figure 5. Sigma-epsilon curves. The head acceleration curve, filtered according to [5], obtained from the FE model was compared against the head acceleration curve obtained from the sled test; both curves are shown in Figure 6. Aerospace 2018, 5, x; doi: FOR PEER REVIEW www.mdpi.com/journal/aerospace Aerospace 2019, 6, 95 6 of 12 The head acceleration curve, filtered according to [5], obtained from the FE model was compared against the head acceleration curve obtained from the sled test; both curves are shown in Figure 6. Aerospace 2018, 5, x FOR PEER REVIEW 6 of 11 Figure 6. Experimental-numerical Head Injury Criterion (HIC) comparison. Figure 6. Experimental-numerical Head Injury Criterion (HIC) comparison. A good agreement was achieved in terms of both acceleration peak and curve trend. Based on A good agreement was achieved in terms of both acceleration peak and curve trend. Based on the the numerical and experimental acceleration curves, the respective HIC values were calculated as numerical and experimental acceleration curves, the respective HIC values were calculated as follows: follows: 8 9 " Z # 2.5 > t > 2 . > > < = HIC = max (t t ) a(t)dt (1) > > 2 1 = max>( − ) () > (1) : ; (t t ) ( 1 ) t − 1 where a(t) is the resultant head acceleration measured in g, and t1 and t2 are the extremes of the where a(t) is the resultant head acceleration measured in g, and t and t are the extremes of the 1 2 integration interval containing the head acceleration peak, measured in seconds. integration interval containing the head acceleration peak, measured in seconds. Despite the good agreement between the acceleration peaks of the numerical and experimental Despite the good agreement between the acceleration peaks of the numerical and experimental head acceleration curves, the calculated HIC values were different. The difference in the HIC values head acceleration curves, the calculated HIC values were di erent. The di erence in the HIC values can be explained taking into account that HIC is calculated choosing the extremes of the integration can be explained taking into account that HIC is calculated choosing the extremes of the integration interval considering the whole acceleration head curve. Actually, as shown by Figure 7, the numerical interval considering the whole acceleration head curve. Actually, as shown by Figure 7, the numerical and experimental curve trends are quite similar, but the corresponding areas are different. and experimental curve trends are quite similar, but the corresponding areas are di erent. The validated FE model was used to conduct a second round of numerical experiments to The validated FE model was used to conduct a second round of numerical experiments to investigate the stiffness characteristics of some seat components to obtain useful information for re- investigate the sti ness characteristics of some seat components to obtain useful information for designing the seat to avoid head–bulkhead impact. Only a single modification was introduced in the re-designing the seat to avoid head–bulkhead impact. Only a single modification was introduced in FE model: the belt was replaced with a Y-belt (Figure 7). the FE model: the belt was replaced with a Y-belt (Figure 7). Figure 7. New FE model with modified belt. Aerospace 2018, 5, x FOR PEER REVIEW 6 of 11 Figure 6. Experimental-numerical Head Injury Criterion (HIC) comparison. A good agreement was achieved in terms of both acceleration peak and curve trend. Based on the numerical and experimental acceleration curves, the respective HIC values were calculated as follows: = max( − ) () (1) ( − ) where a(t) is the resultant head acceleration measured in g, and t1 and t2 are the extremes of the integration interval containing the head acceleration peak, measured in seconds. Despite the good agreement between the acceleration peaks of the numerical and experimental head acceleration curves, the calculated HIC values were different. The difference in the HIC values can be explained taking into account that HIC is calculated choosing the extremes of the integration interval considering the whole acceleration head curve. Actually, as shown by Figure 7, the numerical and experimental curve trends are quite similar, but the corresponding areas are different. The validated FE model was used to conduct a second round of numerical experiments to investigate the stiffness characteristics of some seat components to obtain useful information for re- designing the seat to avoid head–bulkhead impact. Only a single modification was introduced in the Aerospace 2019, 6, 95 7 of 12 FE model: the belt was replaced with a Y-belt (Figure 7). Figure 7. New FE model with modified belt. Figure 7. New FE model with modified belt. 4. Design of Numerical Experiments Numerical experiments were planned according to a Plackett–Burman (PB) design [12] that is a two-level non-regular orthogonal design used as a highly resource-ecient strategy for planning experiments under the assumption of negligible interactions between design factors. In recent years, the PB design has been proposed as an ecient strategy for factor screening; this was also realized in the specialized literature on crashworthiness. M. Hatami [13] used the Design of Experiment (DoE) method to perform optimization, setting the partial shape of a variable turbocharger using several parameters, and verified the e ect. Zhang et al. in [14] approached the optimization design of a motorcycle engine. Tarlochan and Faridz [15] considered the potential factors that could contribute to the frontal crash performance with the use of a 12-run PB design; Pradeep Kanna et al. [16] discussed vehicle structure behavior in a roof crush considering 16 factors in a 20-run PB design; for crashworthiness optimization of a vehicle body, Hou et al. [17] used a 20-run PB to study 19 factors on three responses; for vehicle side impact, Hou et al. [18] used a 20-run PB to study 19 factors and then 15 factors; Wang et al. [19] implemented a crashworthiness analysis of a vehicle door based on a PB design. Lin and Draper [20–22] showed that the projection of the PB design into a lower dimensional space corresponding to k important factors leads to the identification of a helpful additional run. For the PB design adopted in our study, addressing the head–bulkhead impact in relation to the seat frame and the belt, nine factors were taken into account: X and Y coordinates of the Anchor point (AP and AP , respectively); the thicknesses of the Rear beam (D ), Rear leg frame (D ), Rear leg web X Y 1 2 (D ), Shock absorber (D ) and Reinforcing beam (D ); the width of the Rear leg frame (D ) and the 3 4 5 6 length of the Reinforcing beam (D ). For cases of eight to eleven factors, the number of experiments, for a PB plan, is 12. However, in order to arrange the plan, all the eleven factors are necessary. The last two columns are two dummy factors (dumm and dumm , respectively). 1 2 Two di erent levels are representative of the factors for the given experiment; one is relative to its high level and the other one denotes the factor at its low level. The design factors and their ranges are reported in Table 1 and shown in Figure 8. The response variable is a binary performance indicator representing the CONTACT/NO-CONTACT between the passenger head and the bulkhead. Aerospace 2019, 6, 95 8 of 12 Table 1. Design factors and ranges. Code Design Factor Range (mm) AP Anchor Point (coord_X) 26  45 AP Anchor Point (coord_Y) 44  88 D Rear beam (thick.) 2  3 D Rear leg frame (thick.) 3  4.5 D Rear leg web (thick.) 1.8  2.7 D Shock abs (thick.) 2  3.2 D Rear leg frame (width) 18  20 D Reinforcing beam (thick.) 1  1.5 D Reinforcing beam (length) 150  250 dumm1 DUMMY 0  1 dumm2 DUMMY 0  1 Aerospace 2018, 5, x FOR PEER REVIEW 8 of 11 Figure 8. Design factors. Figure 8. Design factors. The results of the second round of numerical experiments, reported in Figure 9, show that the head–bulkhead impact is only avoided at run 2 with a distance of 2 mm between the top point of the passenger head and the bulkhead. Figure 9. Results of the second round of numerical experiments. Figure 9 shows the head displacements in two selected frames for both the preliminary and modified FE models, respectively. According to Figure 10, the red displayed head trajectory is representative of the head–bulkhead impact in the preliminary FE model, whilst the green one indicates the lack of contact in the new model. It must be noticed that the developed FE model is very time-demanding, requiring about 20 h for each run. This suggests the development of a new model, able to take advantages from both the FE and the Multibody (MB) (less time-consuming) approaches. In a previous paper, [8], the authors demonstrated that computational costs are reduced by up to 2 h using a hybrid FE-MB model, which simultaneously implements MB models, for system components whose deformations do not influence the dynamic system responses and for which only kinematic aspects must be investigated (e.g., anthropomorphic dummies), and FE models for the other system components. Aerospace 2018, 5, x FOR PEER REVIEW 8 of 11 Aerospace 2019, 6, 95 9 of 12 Figure 8. Design factors. Figure 9. Results of the second round of numerical experiments. Figure 9. Results of the second round of numerical experiments. Figure 9 shows the head displacements in two selected frames for both the preliminary and Figure 9 shows the head displacements in two selected frames for both the preliminary and modified FE models, respectively. modified FE models, respectively. According to Figure 10, the red displayed head trajectory is representative of the head–bulkhead According to Figure 10, the red displayed head trajectory is representative of the head–bulkhead impact in the preliminary FE model, whilst the green one indicates the lack of contact in the new impact in the preliminary FE model, whilst the green one indicates the lack of contact in the new model. model. It must be noticed that the developed FE model is very time-demanding, requiring about 20 It must be noticed that the developed FE model is very time-demanding, requiring about 20 h for each h for each run. This suggests the development of a new model, able to take advantages from both the run. This suggests the development of a new model, able to take advantages from both the FE and the FE and the Multibody (MB) (less time-consuming) approaches. In a previous paper, [8], the authors Multibody (MB) (less time-consuming) approaches. In a previous paper, [8], the authors demonstrated demonstrated that computational costs are reduced by up to 2 h using a hybrid FE-MB model, which that computational costs are reduced by up to 2 h using a hybrid FE-MB model, which simultaneously simultaneously implements MB models, for system components whose deformations do not implements MB models, for system components whose deformations do not influence the dynamic influence the dynamic system responses and for which only kinematic aspects must be investigated system responses and for which only kinematic aspects must be investigated (e.g., anthropomorphic (e.g., anthropomorphic dummies), and FE models for the other system components. dummies), and FE models for the other system components. Aerospace 2018, 5, x FOR PEER REVIEW 9 of 11 Figure 10. Head displacements for preliminary (on the left) and new (on the right) FE models. Figure 10. Head displacements for preliminary (on the left) and new (on the right) FE models. 5. Numerical-Experimental Comparison An experimental test was carried out to verify the competence of the new restraint system as well as the proposed seat modifications. Figure 11 shows the new sled test configuration with details of the new Y-belt. Figure 11. HIC front row experimental test. The resultant deceleration profile applied to the sled is the same as the one used for the baseline experimental test described in Section 2 and shown in Figure 2. All baseline experimental conditions remained unchanged. The result of the new experimental test was that the passenger head did not impact against the bulkhead, as predicted by the numerical simulation. Specifically, the distance between the bulkhead and the top point of the passenger head (at the maximum value of the head path due to a typical 16- g forward inertial loading condition) was experimentally evaluated to be 555 mm (21.8 inches). By avoiding the impact, it was possible to respect the HIC limit and to provide a passenger seat design configuration compliant with American FAR and European CS25.562 “Emergency landing dynamic conditions” regulation and related guidelines. Aerospace 2018, 5, x FOR PEER REVIEW 9 of 11 Aerospace 2019, 6, 95 10 of 12 Figure 10. Head displacements for preliminary (on the left) and new (on the right) FE models. 5. Numerical-Experimental Comparison 5. Numerical-Experimental Comparison An experimental test was carried out to verify the competence of the new restraint system as well An experimental test was carried out to verify the competence of the new restraint system as as the proposed seat modifications. Figure 11 shows the new sled test configuration with details of the well as the proposed seat modifications. Figure 11 shows the new sled test configuration with details new Y-belt. of the new Y-belt. Figure 11. Figure 11. HIC HIC front front rowr e ow xpeexperimental rimental test. test. The The resultant resultant deceleration deceleration pro prfile ofile applie applied d to th to e the sledsled is the issa the mesame as the as one the used one for used the b for asethe line baseline experimental test described in Section 2 and shown in Figure 2. All baseline experimental conditions experimental test described in Section 2 and shown in Figure 2. All baseline experimental conditions remained unchanged. remained unchanged. The result of the new experimental test was that the passenger head did not impact against the The result of the new experimental test was that the passenger head did not impact against the bulkhead, as predicted by the numerical simulation. Specifically, the distance between the bulkhead bulkhead, as predicted by the numerical simulation. Specifically, the distance between the bulkhead and the top point of the passenger head (at the maximum value of the head path due to a typical 16- and the top point of the passenger head (at the maximum value of the head path due to a typical g forward inertial loading condition) was experimentally evaluated to be 555 mm (21.8 inches). By 16-g forward inertial loading condition) was experimentally evaluated to be 555 mm (21.8 inches). By avoiding the impact, it was possible to respect the HIC limit and to provide a passenger seat design avoiding the impact, it was possible to respect the HIC limit and to provide a passenger seat design configuration compliant with American FAR and European CS25.562 “Emergency landing dynamic configuration compliant with American FAR and European CS25.562 “Emergency landing dynamic conditions” regulation and related guidelines. conditions” regulation and related guidelines. 6. Conclusions This paper deals with the re-design of an aircraft passenger seat in order to make it certifiable according to CS 25.562 “Emergency landing dynamic conditions” regulation and related guidelines. The results of previous experimental investigations and an established FE model for the simulation of a seat sled test demonstrated that the passenger seat did not comply with the regulations (providing an excessively high HIC value). Starting from such results, a new numerical procedure is presented in this paper to improve the passenger safety. Specifically, a highly resource-ecient strategy for planning numerical experiments is adopted for gathering useful information to redesign both the seat frame primary structure and the restraint system (belt) in order to avoid the impact between the passenger ’s head and the bulkhead. Working numerically on the belt anchor points, thicknesses of the rear beam, rear leg frame, rear leg web, shock absorber and reinforcing beam, width of the rear leg frame and length of the reinforcing beam, it was possible to propose a redesigned passenger-seat system able to avoid the impact under a deceleration of 16 g. The numerical solution was then validated experimentally at Geven S.p.A. laboratory. Further investigation will be carried out in order to apply a DoE approach, based on a new numerical model combining both the MB and FE methods, in order to further reduce the computational costs. In this way, it will be possible to investigate more quickly the response of the passenger-restraint system. Aerospace 2019, 6, 95 11 of 12 Author Contributions: Conceptualization, M.G.; Methodology, F.C.; G.L.; Validation, G.L. and M.G. Investigation, M.G.; Resources, B.V. and S.C.; Data Curation, A.V. and F.M.; Writing-Original Draft Preparation, G.L. and M.G.; Writing-Review & Editing, G.L. and M.G.; Supervision, F.M., F.C., A.V. Funding: This research received no external funding. Conflicts of Interest: The authors declare no conflict of interest. References 1. ICAO (International Civil Aviation Organization) Safety Report. 2017. Available online: https://www.icao. int/safety/Documents/ICAO_SR_2017_18072017.pdf (accessed on 31 December 2017). 2. National Transportation Safety Board, Review of Aircraft Accident Data TSB/ARA-14/01 PB2014-101453. 2014. Available online: https://www.ntsb.gov/investigations/data/Documents/ARA1401.pdf (accessed on 31 December 2011). 3. Lankarani, H.M. Current Issues Regarding Aircraft Crash Injury Protection. In Crashworthiness of Transportation Systems: Structural Impact and Occupant Protection; NATO ASI Series (Series E: Applied Sciences); Ambrósio, J.A.C., Pereira, M.F.O.S., da Silva, F.P., Eds.; Springer: Dordrecht, The Netherland, 1997; Volume 332. [CrossRef] 4. European Aviation Safety Agency. Certification specifications for Normal, Utility, Aerobatic, and Commuter Category Aeroplanes; CS-23.562 Emergency Landing Dynamic Conditions, Amendment 5 (17 March 2017); European Aviation Safety Agency: Cologne, Germany, 2017. 5. European Aviation Safety Agency. Certification Specifications for Large Airplanes; CS-25.562 “emergency landing dynamic conditions, Amendment 18 22 (1 June 2016). Amendment 18; European Aviation Safety Agency: Cologne, Germany, 2016. 6. Olschinka, C.; Schumacher, A. Dynamic Simulation of Flight Passenger Seats. In Proceedings of the 5th LS-Dyna Conference, Anwenderforum, Ulm, Germany, 12 October 2006. 7. Lankarani, H. Design and Fabrication of a Head Injury Criteria-Compliant Bulkhead; Report: DOT/FAA/AR-02/98; U.S. Department of Transportation Federal Aviation Administration: Washington, DC, USA, 2002. 8. Guida, M.; Manzoni, A.; Zuppardi, A.; Caputo, F.; Marulo, F.; de Luca, A. Development of a multibody system for crashworthiness certification of aircraft seat. Multibody Syst. Dyn. 2018, 44, 191–221. [CrossRef] 9. di Napoli, F.; de Luca, A.; Caputo, F.; Marulo, F.; Guida, M.; Vitolo, B. Mixed FE-MB methodology for the evaluation of passive safety performances of aeronautical seats. Int. J. Crashworthiness 2019, 24, 314–325. [CrossRef] 10. Caputo, F.; de Luca, A.; Marulo, F.; Guida, M.; Vitolo, B. Numerical-experimental assessment of a hybrid FE-MB model of an aircraft seat sled-test. Int. J. Aerosp. Eng. 2018, 7. [CrossRef] 11. Hooper, S.J.; Henderson, M.J. Development and Validation of an Aircraft Seat Cushion Component Test—Volume I; Technical Report DOT/FAA/AR-05/5, I; U.S. Department of Transportation Federal Aviation Administration Oce of Aviation Research: Washington, DC, USA, 2005. 12. Plackett, R.L.; Burman, J.P. The Design of Optimum Multifactorial Experiments. Biometrika 1946, 33, 305–325. [CrossRef] 13. Hatami, M.; Cuijpers, M.C.M.; Boot, M.D. Experimental optimization of the vanes geometry for a variable. Energy Convers. Manag. 2015, 106, 1057–1070. [CrossRef] 14. Li, Y.; Zhang, S. Research on the Optimization Design of Motorcycle Engine Based on DOE Methodology. Procedia Eng. 2017, 174, 740–747. 15. Tarlochan, F.; Faridz, A. Application of Plackett-Burman Experimental Design through HyperStudy for Sensitivity Study in Frontal Crash Performance. Simulation Driven Innovation with Enterprise Simulation, HTC2009. 2009. Available online: https://www.altairatc.com/india/previous-events/atc/2009/HTC09/OS_ 10_Plackett-Burman_through_HyperStudy_sensitivity_study_in_Frontal_crash_performance_Proton.pdf (accessed on 14 June 2009). 16. Pradeep Kanna, K.A.R.; Ganesh Shanbhag; Satish Kumar, B. DOE Screening Study for Roof Crush Simulation, Simulation driven Innovation. 2010. Available online: https://www.altairatc.com/india/previous-events/ atc/2010/HTC2010/O_15_DOE_Screening_Study_for_Roof_Crush_Simulation_Mahindra.pdf (accessed on 27 April 2010). Aerospace 2019, 6, 95 12 of 12 17. Hou, S.; Dong, D.; Ren, L.; Han, X. Multivariable crashworthiness optimization of vehicle body by unreplicated saturated factorial design. Struct. Multidiscip. Optim. 2012, 46, 891–905. [CrossRef] 18. Hou, S.; Liu, T.; Dong, D.; Han, X. Factor Screening and Multivariable Crashworthiness Optimization for Vehicle Side Impact by Factorial Design. Struct. Multidiscip. Optim. 2014, 49, 147–167. [CrossRef] 19. Wang, X.; Lou, W.L.; Liu, H.X.; Liu, D.L. Crashworthiness Design Analysis of Vehicle Door Using Simulation Experiment Design and Multi-Objective Genetic Algorithm. Key Eng. Mater. Trans Tech Publ. 2014, 579, 234–240. [CrossRef] 20. Lin, D.K.J.; Draper, N.R. Projection Properties of Plackett and Burman Designs. Technometrics 1992, 34, 423–428. [CrossRef] 21. Lin, D.K.J.; Draper, N.R. Screening Properties of Certain Two-Level Designs. Metrika 1995, 42, 99–118. [CrossRef] 22. Lin, D.K.J.; Draper, N.R. Characterizing Projected Designs: Repeat and Mirror-Image Runs. Commun. Stat. Theory Methods 1995, 24, 775–795. © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Aerospace Multidisciplinary Digital Publishing Institute

Development of a Head Injury Criteria-Compliant Aircraft Seat by Design of Experiments

Free
Download PDF
12 pages

Loading next page...
 
/lp/multidisciplinary-digital-publishing-institute/development-of-a-head-injury-criteria-compliant-aircraft-seat-by-C00fThlrH9
Publisher
Multidisciplinary Digital Publishing Institute
Copyright
© 1996-2019 MDPI (Basel, Switzerland) unless otherwise stated Terms and Conditions Privacy Policy
ISSN
2226-4310
DOI
10.3390/aerospace6090095
Publisher site
See Article on Publisher Site

Abstract

aerospace Article Development of a Head Injury Criteria-Compliant Aircraft Seat by Design of Experiments 1 2 2 , 1 Giuseppe Lamanna , Amalia Vanacore , Michele Guida * , Francesco Caputo , 2 3 3 Francesco Marulo , Bonaventura Vitolo and Salvatore Cicatiello Department of Industrial and Information Engineering, University of Campania Luigi Vanvitelli, Aversa, 81100 Caserta, Italy Department of Industrial Engineering, University of Naples “Federico II”, 80125 Napoli, Italy Geven, Zona Asi Boscofangone, 80035 Nola, Italy * Correspondence: michele.guida@unina.it Received: 18 June 2019; Accepted: 30 August 2019; Published: 3 September 2019 Abstract: This paper deals with the redesign of an aircraft passenger seat, placed at the first seat row, which was not compliant with Federal Aviation Regulations FAR 25.562 “Emergency landing dynamic conditions” regulation (due to a high value for the Head Injury Criterion (HIC)) and related guidelines. Starting from an accurate analysis of some results obtained via an experimental seat sled test, a numerical procedure was developed in order to improve the passenger safety with respect to head injury. Specifically, the proposed numerical procedure, using the advantages of a Finite Element (FE) model and a Design of Experiment (DoE) approach for simulation modeling, was aimed at identifying a new design solution to avoid the impact between the passenger ’s head and the bulkhead. The redesign of the passenger seat was validated against an experimental test carried out at Geven S.p.A. Company by demonstrating, consequently, the compliance of the modified seat-belt system with the regulations. Keywords: multibody systems; finite element analysis; aircraft seat; design of experiment; head injury 1. Introduction Crashworthiness requirements drive the design of most automotive and aerospace structural components in order to improve their passive safety performance and, consequently, to protect occupants against injuries and death during a crash event. Design e orts are increasingly focussed on satisfying these requirements, even if passive safety criteria lead to a small increase in the structure’s weight. Many research studies involving experimental tests and numerical studies are carried out to verify that the proposed design solutions respect crashworthiness requirements and to predict the injury level for occupants in case of an accident [1,2]. In the aeronautical sector, the sled is considered as an appropriate system to check the protection of occupants, and a crashworthy seat has to guarantee that the consequences of occupant injuries are not fatal [3]. Currently, aeronautical regulations consider the experimental test to be the only possible solution to certify an aircraft seat. For lighter aircraft (take-o weight of 5670 kg or less), two tests are required [4]: the first test is based on the loads transferred to the occupant during a forward deceleration; the second test considers the downward deceleration. For heavier aircraft (take-o weight above 5670 kg) [5], a test related to the deformation of the floor is required in addition to the previous two tests. Whatever the airplane category, it is necessary to evaluate the entity of the damage in the presence of a head impact. For this purpose, the experimental test for component certification can take advantage of the high level of technology introduced in the current Anthropometric Test Devices (ATDs), which allow the recording of the occupant’s response (i.e., kinematical as well as Aerospace 2019, 6, 95; doi:10.3390/aerospace6090095 www.mdpi.com/journal/aerospace Aerospace 2019, 6, 95 2 of 12 dynamical parameters) during the experimental simulation of a crash event [6]. However, it is evident that the fitting and the implementation of these tests are time-, space- and cost-demanding, and thus it is not a ordable to experimentally test several configurations of a proposed design solution. For all these reasons, the development of a numerical model capable of reproducing such complex experimental tests, according to the Certification by Analysis paradigm, seems to be the best strategy for the development of a crashworthy design solution. The main advantage provided by numerical methods consists in the possibility to identify an optimal virtual solution that can be validated via experimental tests, avoiding the “trial and error” approach typically adopted by companies and, consequently, reducing time and cost for development. Of course, the development of a reliable numerical model requires many e orts in dealing with the underlying assumptions and hypotheses as well as in the necessary availability of high computational power, complemented by high skills in modeling. Among the several structural components that can significantly contribute to occupant safety, the seat plays a key role in reducing the loads transferred to passengers during the crash as well as in reducing the probability of the passenger ’s head impact against the bulkhead, in the case of the first seat row. In [6], the authors propose a multibody model of the aircraft seat structure for the simulation of a 16-g compliance with the Head Injury Criteria (HIC) requirement. This test involves an impact against a bulkhead developed and analyzed by using an ad hoc algorithm implemented in MATLAB code, as a 2D system of rigid bodies interconnected by springs and joints. Lankarani, in [7], focuses on the design and development of bulkheads evaluating various honeycomb materials for HIC attenuation, rather than on high costs and schedule overruns due to the development and certification of aircraft seats. The aim of this paper is to define a numerical procedure to improve the design of an aircraft passenger seat, considering passive safety as the main goal. For this purpose, di erent numerical methods can be used, including Finite Element (FE) and Multibody (MB) models. MB numerical models are generally adopted to evaluate the kinematics of the dummy (representing the occupant) and its interactions with both the restraint systems, if available, and the seat cushion, with no possibility whatsoever to obtain any results about the structural behavior of the seat frame. These models exhibit the convenience of allowing quick modifications of the analyzed configuration and obtaining reliable results—within the recalled limits—with very short runtimes. Numerical FE models are, instead, required to obtain information about the structural behavior of the seat. Even if the developing time of these models is considered acceptable for current design time scheduling, the same cannot be said for runtimes. In particular, runtimes are very long for applications of dynamical type and for the use of anthropomorphic dummies, whose models prove to be very complex and therefore require very accurate discretization and high mesh densities. Some solutions involving the coupling of both methods are proposed in literature [8,9]. The research activity presented in the present paper makes use of the FE and MB methods either by an internal explicit code (hybrid) [8] or independent code (coupling) [9], with a di erence in terms of computational cost and correlation with experimental data. Specifically, the research starts from an established FE model presented by authors in previous papers [8]. This FE model allows the description of the kinematics of a passenger as well as the injuries that a passenger seated at the first row of an aircraft may su er in a frontal impact against the bulkhead. A corresponding biomechanical head injury index was calculated and compared with the extreme one considered according to the current regulations. The investigation was also experimentally carried out by launching an aircraft seat, equipped with an ATD, against a bulkhead at the required speed and the corresponding acceleration/deceleration profile. The experimental data were used to calculate a damage index, as described below, useful for the evaluation of the level of injuries a ecting di erent body parts directly involved in the impact or just subjected to high inertia loads. The good level of accuracy of the developed FE model was demonstrated by comparing numerical results to experimental data. Aerospace 2019, 6, 95 3 of 12 Aerospace 2018, 5, x FOR PEER REVIEW 3 of 11 According to the experimental test presented in [8], the calculated HIC (Head Injury Criterion) According to the experimental test presented in [8], the calculated HIC (Head Injury Criterion) parameter, representing the head injury, was higher than the limit provided by AC 25.862 and SAE parameter, representing the head injury, was higher than the limit provided by AC 25.862 and SAE 8049b because the head of the dummy contacted the rigid bulkhead. The analysis of both numerical 8049b because the head of the dummy contacted the rigid bulkhead. The analysis of both numerical results and experimental data suggested that this problem could be related to the sti ness characteristics results and experimental data suggested that this problem could be related to the stiffness of some seat components, which have been thus selected as design factors to be further investigated. characteristics of some seat components, which have been thus selected as design factors to be further Based on the validated FE model and the adoption of an ecient experimental design, new numerical investigated. Based on the validated FE model and the adoption of an efficient experimental design, new experiments numericwer al exp e carried eriment out s were in or cder arrito ed find out in a new order design to find solution a new des forig the n sol seat utiframe on forto thimpr e seaove t frame the passenger passive safety. All analyses were carried out using Ls-Dyna software. In the following to improve the passenger passive safety. All analyses were carried out using Ls-Dyna software. In th sections, e follow the ing baseline sections, experimental the baseline exp study eriment , the aFE l stud model, y, the the FE strategy model, th for e strate planning gy for the plannin numerical g the experiments and the obtained results are described and discussed. numerical experiments and the obtained results are described and discussed. 2. Baseline Study and Dynamic Testing 2. Baseline Study and Dynamic Testing A metallic seat frame fabricated in aluminum alloy was used in the dynamic sled test. To maximize A metallic seat frame fabricated in aluminum alloy was used in the dynamic sled test. To the energy transferred to the head, no yaw was given to the seat in these tests. The experimental setup maximize the energy transferred to the head, no yaw was given to the seat in these tests. The of the seat, shown in Figure 1, is representative of a typical airline economy class seat with the seat experimental setup of the seat, shown in Figure 1, is representative of a typical airline economy class back fixed in the upright position. seat with the seat back fixed in the upright position. The seat setback distance—defined as the horizontal distance between the seat reference point The seat setback distance—defined as the horizontal distance between the seat reference point (i.e., the intersection point between the seat back and the seat pan) and the outer surface of the (i.e., the intersection point between the seat back and the seat pan) and the outer surface of the bulkhead—was fixed at 583 mm (23 inches). bulkhead—was fixed at 583 mm (23 inches). Figure 1. Experimental setup. Figure 1. Experimental setup. The seat cushions used during the test consisted of foam with the static deformation under the passenger weight set to a maximum penetration of 5 mm; a typical polyester seat belt was used. The The seat cushions used during the test consisted of foam with the static deformation under the pass sample enger we full-scale ightsled set to a m test setup aximum with pe the net bulkhead ration of is 5 mm shown ; a typ in Figur ical po e 2 ly . e A ster se triaxial at belt wa accelerometer s used. was The mounted at the center of gravity (c.g.) of the ATD head to determine the resultant head acceleration. sample full-scale sled test setup with the bulkhead is shown in Figure 2. A triaxial accelerometer was mounted at the center of gravity (c.g.) of the ATD head to determine the resultant head acceleration. Aerospace 2019, 6, 95 4 of 12 Aerospace 2018, 5, x FOR PEER REVIEW 4 of 11 Aerospace 2018, 5, x FOR PEER REVIEW 4 of 11 Figure 2. Full-scale sled test setup. Figure 2. Full-scale sled test setup. Figure 2. Full-scale sled test setup. The experimental test sessions described in this section were conducted at the Impact Dynamics The experimental test sessions described in this section were conducted at the Impact Dynamics The experimental test sessions described in this section were conducted at the Impact Dynamics Laboratory of Geven S.p.A. with the aim of measuring the head accelerations of a Hybrid-II Labor Laboratory atory o of f Gev Geven en S S.p.A. .p.A. w with ith th the e ai aim m of of me measuring asuring th the e he head ad acce accelerations lerations o of f a a Hy Hybrid-II brid-II Anthropomorphic Test Dummy (ATD), as specified in Certification Specifications CS 25.562, [5], so Anthropomorphic Test Dummy (ATD), as specified in Certification Specifications CS 25.562, [5], so Anthropomorphic Test Dummy (ATD), as specified in Certification Specifications CS 25.562, [5], as to support the development of the FE model and to assess the performances of the new seat design as so to as support to support the develop the development ment of thof e FE the mo FE del model and to and assess to assess the per the formance performances s of the new of the seat new design seat solution. solution. design solution. The experimental tests were carried out using a sled deceleration system pneumatically Th The e experimental experimental tests testwer s we ere carried carried out using out ua sisled ng a deceleration sled deceler system ation pneumatically system pneumatic activated ally activated like a “sling”. The seat with a dummy is fixed on the sled, and both together are launched activated like a “sling”. The seat with a dummy is fixed on the sled, and both together are launched like a “sling”. The seat with a dummy is fixed on the sled, and both together are launched at an at an assigned speed. The race is stopped by a mechanical brake (decelerator system) made with a at assigned an assigne speed. d spee The d.race The is rastopped ce is stop by ped a mec by a hanical mechabrake nical br (decelerator ake (deceler system) ator system made ) made with a with lower a lower carbon steel wire, which is properly assembled in order to have a g-peak compliant with the lower carbon steel wire, which is properly assembled in order to have a g-peak compliant with the carbon steel wire, which is properly assembled in order to have a g-peak compliant with the rules. The rules. The number, position, and length of each steel wire can affect the deceleration pulse shape. rul number es. Th , e position, number, and posi length tion, and of each length steel of wir eae ch can stee a l ect wire thecan deceleration affect the pulse decele shape. ration This pulse HIC shape test . This HIC test can be performed in two different setup configurations: This HIC test can be performed in two different setup configurations: can be performed in two di erent setup configurations:  First row test, in which the seat is placed in front a rigid bulkhead at a fixed distance in order to  First row test, in which the seat is placed in front a rigid bulkhead at a fixed distance in order to First row test, in which the seat is placed in front a rigid bulkhead at a fixed distance in order to simulate a typical first row installation inside the cabin (Figure 2). In this case, the main purpose simulate a typical first row installation inside the cabin (Figure 2). In this case, the main purpose simulate a typical first row installation inside the cabin (Figure 2). In this case, the main purpose is to verify the head contact, while the obtained HIC value can be considered only as a reference is to verify the head contact, while the obtained HIC value can be considered only as a reference is to verify the head contact, while the obtained HIC value can be considered only as a reference value since the installed bulkhead is not the real cabin installation. value since the installed bulkhead is not the real cabin installation. value since the installed bulkhead is not the real cabin installation.  Row to row test, in which two seat rows are fixed on the sled at a proper distance. The aim of  Row to row test, in which two seat rows are fixed on the sled at a proper distance. The aim of Row to row test, in which two seat rows are fixed on the sled at a proper distance. The aim of the the test is measuring the HIC value during the head impact in order to evaluate the potential the test is measuring the HIC value during the head impact in order to evaluate the potential test is measuring the HIC value during the head impact in order to evaluate the potential injury injury related to the design of the seat backrest including mounted equipment (monitor, rear injury related to the design of the seat backrest including mounted equipment (monitor, rear related to the design of the seat backrest including mounted equipment (monitor, rear table etc.). table etc.). table etc.). As for the United States Code of Federal Regulations (CFR), a triangular deceleration pulse with a As for the United States Code of Federal Regulations (CFR), a triangular deceleration pulse with peak of 16 g and a rise time of 90 ms was targeted for the sled tests. The ideal pulse shape and the As for the United States Code of Federal Regulations (CFR), a triangular deceleration pulse with a peak of 16 g and a rise time of 90 ms was targeted for the sled tests. The ideal pulse shape and the a actual peak sled of 16 test g and deceleration a rise time pu of lse 90 for ms the was two targeted baseline for tests the sled are te shown sts. Th in e ide Figur al puls e 3. e Pr sh oper ape seat and belt the actual sled test deceleration pulse for the two baseline tests are shown in Figure 3. Proper seat belt actu installation al sled te requir st deceler ed aatio testn rig puls able e for to th guarantee e two baselin the corr e test ect s position are shown of in airFigur craft/e belt 3. Proper interface seat points belt installation required a test rig able to guarantee the correct position of aircraft/belt interface points insta withllation respect re to quire the d seat. a test rig able to guarantee the correct position of aircraft/belt interface points with respect to the seat. with respect to the seat. Figure 3. Deceleration pulse. Figure 3. Deceleration pulse. Figure 3. Deceleration pulse. Aerospace 2019, 6, 95 5 of 12 Aerospace 2018, 5, x FOR PEER REVIEW 5 of 11 All devices were equipped with accelerometers and load cells to measure forces and acceleration All devices were equipped with accelerometers and load cells to measure forces and acceleration affecting the most sensitive human body parts. Particular attention was paid to the acceleration of the a ecting the most sensitive human body parts. Particular attention was paid to the acceleration of the head and to the loads transmitted to the lower limbs of the dummy in order to verify the accuracy of head and to the loads transmitted to the lower limbs of the dummy in order to verify the accuracy of the FE model proposed in the previous papers [9,10] whose main characteristics are briefly described the FE model proposed in the previous papers [9,10] whose main characteristics are briefly described in the next section. in the next section. 3. Numerical Models 3. Numerical Models The full seat FE model consisting of 105,226 elements and 151,219 nodes is shown in Figure 4. The full seat FE model consisting of 105,226 elements and 151,219 nodes is shown in Figure 4. All All structural seat components were modelled by considering aluminum alloy material, whereas structural seat components were modelled by considering aluminum alloy material, whereas foam foam material was considered for cushions, ref. [11]. material was considered for cushions, ref. [11]. Figure 4. Finite Element (FE) model. Figure 4. Finite Element (FE) model. For each material, an elasto-plastic model was selected; the constitutive curves of each material are For each material, an elasto-plastic model was selected; the constitutive curves of each material shown in Figure 5. The aluminum alloy adopted is an elasto-plastic material with kinematic hardening are shown in Figure 5. The aluminum alloy adopted is an elasto-plastic material with kinematic (model n. 24 Ls-dyna’s material library), and failure is defined based on the plastic strain. The foam hardening (model n. 24 Ls-dyna’s material library), and failure is defined based on the plastic strain. material considered is a rubber-like foam of polyurethane. It is a simple one-parameter model with a The foam material considered is a rubber-like foam of polyurethane. It is a simple one-parameter fixed Poisson’s ratio of 0.25. model with a fixed Poisson’s ratio of 0.25. The deceleration pulse was applied to the nodes of both the bulkhead and the seat fixed to the The deceleration pulse was applied to the nodes of both the bulkhead and the seat fixed to the slide. Gravity and initial velocity were applied to all parts of the model. slide. Gravity and initial velocity were applied to all parts of the model. Figure 5. Sigma-epsilon curves. Figure 5. Sigma-epsilon curves. The head acceleration curve, filtered according to [5], obtained from the FE model was compared against the head acceleration curve obtained from the sled test; both curves are shown in Figure 6. Aerospace 2018, 5, x; doi: FOR PEER REVIEW www.mdpi.com/journal/aerospace Aerospace 2019, 6, 95 6 of 12 The head acceleration curve, filtered according to [5], obtained from the FE model was compared against the head acceleration curve obtained from the sled test; both curves are shown in Figure 6. Aerospace 2018, 5, x FOR PEER REVIEW 6 of 11 Figure 6. Experimental-numerical Head Injury Criterion (HIC) comparison. Figure 6. Experimental-numerical Head Injury Criterion (HIC) comparison. A good agreement was achieved in terms of both acceleration peak and curve trend. Based on A good agreement was achieved in terms of both acceleration peak and curve trend. Based on the the numerical and experimental acceleration curves, the respective HIC values were calculated as numerical and experimental acceleration curves, the respective HIC values were calculated as follows: follows: 8 9 " Z # 2.5 > t > 2 . > > < = HIC = max (t t ) a(t)dt (1) > > 2 1 = max>( − ) () > (1) : ; (t t ) ( 1 ) t − 1 where a(t) is the resultant head acceleration measured in g, and t1 and t2 are the extremes of the where a(t) is the resultant head acceleration measured in g, and t and t are the extremes of the 1 2 integration interval containing the head acceleration peak, measured in seconds. integration interval containing the head acceleration peak, measured in seconds. Despite the good agreement between the acceleration peaks of the numerical and experimental Despite the good agreement between the acceleration peaks of the numerical and experimental head acceleration curves, the calculated HIC values were different. The difference in the HIC values head acceleration curves, the calculated HIC values were di erent. The di erence in the HIC values can be explained taking into account that HIC is calculated choosing the extremes of the integration can be explained taking into account that HIC is calculated choosing the extremes of the integration interval considering the whole acceleration head curve. Actually, as shown by Figure 7, the numerical interval considering the whole acceleration head curve. Actually, as shown by Figure 7, the numerical and experimental curve trends are quite similar, but the corresponding areas are different. and experimental curve trends are quite similar, but the corresponding areas are di erent. The validated FE model was used to conduct a second round of numerical experiments to The validated FE model was used to conduct a second round of numerical experiments to investigate the stiffness characteristics of some seat components to obtain useful information for re- investigate the sti ness characteristics of some seat components to obtain useful information for designing the seat to avoid head–bulkhead impact. Only a single modification was introduced in the re-designing the seat to avoid head–bulkhead impact. Only a single modification was introduced in FE model: the belt was replaced with a Y-belt (Figure 7). the FE model: the belt was replaced with a Y-belt (Figure 7). Figure 7. New FE model with modified belt. Aerospace 2018, 5, x FOR PEER REVIEW 6 of 11 Figure 6. Experimental-numerical Head Injury Criterion (HIC) comparison. A good agreement was achieved in terms of both acceleration peak and curve trend. Based on the numerical and experimental acceleration curves, the respective HIC values were calculated as follows: = max( − ) () (1) ( − ) where a(t) is the resultant head acceleration measured in g, and t1 and t2 are the extremes of the integration interval containing the head acceleration peak, measured in seconds. Despite the good agreement between the acceleration peaks of the numerical and experimental head acceleration curves, the calculated HIC values were different. The difference in the HIC values can be explained taking into account that HIC is calculated choosing the extremes of the integration interval considering the whole acceleration head curve. Actually, as shown by Figure 7, the numerical and experimental curve trends are quite similar, but the corresponding areas are different. The validated FE model was used to conduct a second round of numerical experiments to investigate the stiffness characteristics of some seat components to obtain useful information for re- designing the seat to avoid head–bulkhead impact. Only a single modification was introduced in the Aerospace 2019, 6, 95 7 of 12 FE model: the belt was replaced with a Y-belt (Figure 7). Figure 7. New FE model with modified belt. Figure 7. New FE model with modified belt. 4. Design of Numerical Experiments Numerical experiments were planned according to a Plackett–Burman (PB) design [12] that is a two-level non-regular orthogonal design used as a highly resource-ecient strategy for planning experiments under the assumption of negligible interactions between design factors. In recent years, the PB design has been proposed as an ecient strategy for factor screening; this was also realized in the specialized literature on crashworthiness. M. Hatami [13] used the Design of Experiment (DoE) method to perform optimization, setting the partial shape of a variable turbocharger using several parameters, and verified the e ect. Zhang et al. in [14] approached the optimization design of a motorcycle engine. Tarlochan and Faridz [15] considered the potential factors that could contribute to the frontal crash performance with the use of a 12-run PB design; Pradeep Kanna et al. [16] discussed vehicle structure behavior in a roof crush considering 16 factors in a 20-run PB design; for crashworthiness optimization of a vehicle body, Hou et al. [17] used a 20-run PB to study 19 factors on three responses; for vehicle side impact, Hou et al. [18] used a 20-run PB to study 19 factors and then 15 factors; Wang et al. [19] implemented a crashworthiness analysis of a vehicle door based on a PB design. Lin and Draper [20–22] showed that the projection of the PB design into a lower dimensional space corresponding to k important factors leads to the identification of a helpful additional run. For the PB design adopted in our study, addressing the head–bulkhead impact in relation to the seat frame and the belt, nine factors were taken into account: X and Y coordinates of the Anchor point (AP and AP , respectively); the thicknesses of the Rear beam (D ), Rear leg frame (D ), Rear leg web X Y 1 2 (D ), Shock absorber (D ) and Reinforcing beam (D ); the width of the Rear leg frame (D ) and the 3 4 5 6 length of the Reinforcing beam (D ). For cases of eight to eleven factors, the number of experiments, for a PB plan, is 12. However, in order to arrange the plan, all the eleven factors are necessary. The last two columns are two dummy factors (dumm and dumm , respectively). 1 2 Two di erent levels are representative of the factors for the given experiment; one is relative to its high level and the other one denotes the factor at its low level. The design factors and their ranges are reported in Table 1 and shown in Figure 8. The response variable is a binary performance indicator representing the CONTACT/NO-CONTACT between the passenger head and the bulkhead. Aerospace 2019, 6, 95 8 of 12 Table 1. Design factors and ranges. Code Design Factor Range (mm) AP Anchor Point (coord_X) 26  45 AP Anchor Point (coord_Y) 44  88 D Rear beam (thick.) 2  3 D Rear leg frame (thick.) 3  4.5 D Rear leg web (thick.) 1.8  2.7 D Shock abs (thick.) 2  3.2 D Rear leg frame (width) 18  20 D Reinforcing beam (thick.) 1  1.5 D Reinforcing beam (length) 150  250 dumm1 DUMMY 0  1 dumm2 DUMMY 0  1 Aerospace 2018, 5, x FOR PEER REVIEW 8 of 11 Figure 8. Design factors. Figure 8. Design factors. The results of the second round of numerical experiments, reported in Figure 9, show that the head–bulkhead impact is only avoided at run 2 with a distance of 2 mm between the top point of the passenger head and the bulkhead. Figure 9. Results of the second round of numerical experiments. Figure 9 shows the head displacements in two selected frames for both the preliminary and modified FE models, respectively. According to Figure 10, the red displayed head trajectory is representative of the head–bulkhead impact in the preliminary FE model, whilst the green one indicates the lack of contact in the new model. It must be noticed that the developed FE model is very time-demanding, requiring about 20 h for each run. This suggests the development of a new model, able to take advantages from both the FE and the Multibody (MB) (less time-consuming) approaches. In a previous paper, [8], the authors demonstrated that computational costs are reduced by up to 2 h using a hybrid FE-MB model, which simultaneously implements MB models, for system components whose deformations do not influence the dynamic system responses and for which only kinematic aspects must be investigated (e.g., anthropomorphic dummies), and FE models for the other system components. Aerospace 2018, 5, x FOR PEER REVIEW 8 of 11 Aerospace 2019, 6, 95 9 of 12 Figure 8. Design factors. Figure 9. Results of the second round of numerical experiments. Figure 9. Results of the second round of numerical experiments. Figure 9 shows the head displacements in two selected frames for both the preliminary and Figure 9 shows the head displacements in two selected frames for both the preliminary and modified FE models, respectively. modified FE models, respectively. According to Figure 10, the red displayed head trajectory is representative of the head–bulkhead According to Figure 10, the red displayed head trajectory is representative of the head–bulkhead impact in the preliminary FE model, whilst the green one indicates the lack of contact in the new impact in the preliminary FE model, whilst the green one indicates the lack of contact in the new model. model. It must be noticed that the developed FE model is very time-demanding, requiring about 20 It must be noticed that the developed FE model is very time-demanding, requiring about 20 h for each h for each run. This suggests the development of a new model, able to take advantages from both the run. This suggests the development of a new model, able to take advantages from both the FE and the FE and the Multibody (MB) (less time-consuming) approaches. In a previous paper, [8], the authors Multibody (MB) (less time-consuming) approaches. In a previous paper, [8], the authors demonstrated demonstrated that computational costs are reduced by up to 2 h using a hybrid FE-MB model, which that computational costs are reduced by up to 2 h using a hybrid FE-MB model, which simultaneously simultaneously implements MB models, for system components whose deformations do not implements MB models, for system components whose deformations do not influence the dynamic influence the dynamic system responses and for which only kinematic aspects must be investigated system responses and for which only kinematic aspects must be investigated (e.g., anthropomorphic (e.g., anthropomorphic dummies), and FE models for the other system components. dummies), and FE models for the other system components. Aerospace 2018, 5, x FOR PEER REVIEW 9 of 11 Figure 10. Head displacements for preliminary (on the left) and new (on the right) FE models. Figure 10. Head displacements for preliminary (on the left) and new (on the right) FE models. 5. Numerical-Experimental Comparison An experimental test was carried out to verify the competence of the new restraint system as well as the proposed seat modifications. Figure 11 shows the new sled test configuration with details of the new Y-belt. Figure 11. HIC front row experimental test. The resultant deceleration profile applied to the sled is the same as the one used for the baseline experimental test described in Section 2 and shown in Figure 2. All baseline experimental conditions remained unchanged. The result of the new experimental test was that the passenger head did not impact against the bulkhead, as predicted by the numerical simulation. Specifically, the distance between the bulkhead and the top point of the passenger head (at the maximum value of the head path due to a typical 16- g forward inertial loading condition) was experimentally evaluated to be 555 mm (21.8 inches). By avoiding the impact, it was possible to respect the HIC limit and to provide a passenger seat design configuration compliant with American FAR and European CS25.562 “Emergency landing dynamic conditions” regulation and related guidelines. Aerospace 2018, 5, x FOR PEER REVIEW 9 of 11 Aerospace 2019, 6, 95 10 of 12 Figure 10. Head displacements for preliminary (on the left) and new (on the right) FE models. 5. Numerical-Experimental Comparison 5. Numerical-Experimental Comparison An experimental test was carried out to verify the competence of the new restraint system as well An experimental test was carried out to verify the competence of the new restraint system as as the proposed seat modifications. Figure 11 shows the new sled test configuration with details of the well as the proposed seat modifications. Figure 11 shows the new sled test configuration with details new Y-belt. of the new Y-belt. Figure 11. Figure 11. HIC HIC front front rowr e ow xpeexperimental rimental test. test. The The resultant resultant deceleration deceleration pro prfile ofile applie applied d to th to e the sledsled is the issa the mesame as the as one the used one for used the b for asethe line baseline experimental test described in Section 2 and shown in Figure 2. All baseline experimental conditions experimental test described in Section 2 and shown in Figure 2. All baseline experimental conditions remained unchanged. remained unchanged. The result of the new experimental test was that the passenger head did not impact against the The result of the new experimental test was that the passenger head did not impact against the bulkhead, as predicted by the numerical simulation. Specifically, the distance between the bulkhead bulkhead, as predicted by the numerical simulation. Specifically, the distance between the bulkhead and the top point of the passenger head (at the maximum value of the head path due to a typical 16- and the top point of the passenger head (at the maximum value of the head path due to a typical g forward inertial loading condition) was experimentally evaluated to be 555 mm (21.8 inches). By 16-g forward inertial loading condition) was experimentally evaluated to be 555 mm (21.8 inches). By avoiding the impact, it was possible to respect the HIC limit and to provide a passenger seat design avoiding the impact, it was possible to respect the HIC limit and to provide a passenger seat design configuration compliant with American FAR and European CS25.562 “Emergency landing dynamic configuration compliant with American FAR and European CS25.562 “Emergency landing dynamic conditions” regulation and related guidelines. conditions” regulation and related guidelines. 6. Conclusions This paper deals with the re-design of an aircraft passenger seat in order to make it certifiable according to CS 25.562 “Emergency landing dynamic conditions” regulation and related guidelines. The results of previous experimental investigations and an established FE model for the simulation of a seat sled test demonstrated that the passenger seat did not comply with the regulations (providing an excessively high HIC value). Starting from such results, a new numerical procedure is presented in this paper to improve the passenger safety. Specifically, a highly resource-ecient strategy for planning numerical experiments is adopted for gathering useful information to redesign both the seat frame primary structure and the restraint system (belt) in order to avoid the impact between the passenger ’s head and the bulkhead. Working numerically on the belt anchor points, thicknesses of the rear beam, rear leg frame, rear leg web, shock absorber and reinforcing beam, width of the rear leg frame and length of the reinforcing beam, it was possible to propose a redesigned passenger-seat system able to avoid the impact under a deceleration of 16 g. The numerical solution was then validated experimentally at Geven S.p.A. laboratory. Further investigation will be carried out in order to apply a DoE approach, based on a new numerical model combining both the MB and FE methods, in order to further reduce the computational costs. In this way, it will be possible to investigate more quickly the response of the passenger-restraint system. Aerospace 2019, 6, 95 11 of 12 Author Contributions: Conceptualization, M.G.; Methodology, F.C.; G.L.; Validation, G.L. and M.G. Investigation, M.G.; Resources, B.V. and S.C.; Data Curation, A.V. and F.M.; Writing-Original Draft Preparation, G.L. and M.G.; Writing-Review & Editing, G.L. and M.G.; Supervision, F.M., F.C., A.V. Funding: This research received no external funding. Conflicts of Interest: The authors declare no conflict of interest. References 1. ICAO (International Civil Aviation Organization) Safety Report. 2017. Available online: https://www.icao. int/safety/Documents/ICAO_SR_2017_18072017.pdf (accessed on 31 December 2017). 2. National Transportation Safety Board, Review of Aircraft Accident Data TSB/ARA-14/01 PB2014-101453. 2014. Available online: https://www.ntsb.gov/investigations/data/Documents/ARA1401.pdf (accessed on 31 December 2011). 3. Lankarani, H.M. Current Issues Regarding Aircraft Crash Injury Protection. In Crashworthiness of Transportation Systems: Structural Impact and Occupant Protection; NATO ASI Series (Series E: Applied Sciences); Ambrósio, J.A.C., Pereira, M.F.O.S., da Silva, F.P., Eds.; Springer: Dordrecht, The Netherland, 1997; Volume 332. [CrossRef] 4. European Aviation Safety Agency. Certification specifications for Normal, Utility, Aerobatic, and Commuter Category Aeroplanes; CS-23.562 Emergency Landing Dynamic Conditions, Amendment 5 (17 March 2017); European Aviation Safety Agency: Cologne, Germany, 2017. 5. European Aviation Safety Agency. Certification Specifications for Large Airplanes; CS-25.562 “emergency landing dynamic conditions, Amendment 18 22 (1 June 2016). Amendment 18; European Aviation Safety Agency: Cologne, Germany, 2016. 6. Olschinka, C.; Schumacher, A. Dynamic Simulation of Flight Passenger Seats. In Proceedings of the 5th LS-Dyna Conference, Anwenderforum, Ulm, Germany, 12 October 2006. 7. Lankarani, H. Design and Fabrication of a Head Injury Criteria-Compliant Bulkhead; Report: DOT/FAA/AR-02/98; U.S. Department of Transportation Federal Aviation Administration: Washington, DC, USA, 2002. 8. Guida, M.; Manzoni, A.; Zuppardi, A.; Caputo, F.; Marulo, F.; de Luca, A. Development of a multibody system for crashworthiness certification of aircraft seat. Multibody Syst. Dyn. 2018, 44, 191–221. [CrossRef] 9. di Napoli, F.; de Luca, A.; Caputo, F.; Marulo, F.; Guida, M.; Vitolo, B. Mixed FE-MB methodology for the evaluation of passive safety performances of aeronautical seats. Int. J. Crashworthiness 2019, 24, 314–325. [CrossRef] 10. Caputo, F.; de Luca, A.; Marulo, F.; Guida, M.; Vitolo, B. Numerical-experimental assessment of a hybrid FE-MB model of an aircraft seat sled-test. Int. J. Aerosp. Eng. 2018, 7. [CrossRef] 11. Hooper, S.J.; Henderson, M.J. Development and Validation of an Aircraft Seat Cushion Component Test—Volume I; Technical Report DOT/FAA/AR-05/5, I; U.S. Department of Transportation Federal Aviation Administration Oce of Aviation Research: Washington, DC, USA, 2005. 12. Plackett, R.L.; Burman, J.P. The Design of Optimum Multifactorial Experiments. Biometrika 1946, 33, 305–325. [CrossRef] 13. Hatami, M.; Cuijpers, M.C.M.; Boot, M.D. Experimental optimization of the vanes geometry for a variable. Energy Convers. Manag. 2015, 106, 1057–1070. [CrossRef] 14. Li, Y.; Zhang, S. Research on the Optimization Design of Motorcycle Engine Based on DOE Methodology. Procedia Eng. 2017, 174, 740–747. 15. Tarlochan, F.; Faridz, A. Application of Plackett-Burman Experimental Design through HyperStudy for Sensitivity Study in Frontal Crash Performance. Simulation Driven Innovation with Enterprise Simulation, HTC2009. 2009. Available online: https://www.altairatc.com/india/previous-events/atc/2009/HTC09/OS_ 10_Plackett-Burman_through_HyperStudy_sensitivity_study_in_Frontal_crash_performance_Proton.pdf (accessed on 14 June 2009). 16. Pradeep Kanna, K.A.R.; Ganesh Shanbhag; Satish Kumar, B. DOE Screening Study for Roof Crush Simulation, Simulation driven Innovation. 2010. Available online: https://www.altairatc.com/india/previous-events/ atc/2010/HTC2010/O_15_DOE_Screening_Study_for_Roof_Crush_Simulation_Mahindra.pdf (accessed on 27 April 2010). Aerospace 2019, 6, 95 12 of 12 17. Hou, S.; Dong, D.; Ren, L.; Han, X. Multivariable crashworthiness optimization of vehicle body by unreplicated saturated factorial design. Struct. Multidiscip. Optim. 2012, 46, 891–905. [CrossRef] 18. Hou, S.; Liu, T.; Dong, D.; Han, X. Factor Screening and Multivariable Crashworthiness Optimization for Vehicle Side Impact by Factorial Design. Struct. Multidiscip. Optim. 2014, 49, 147–167. [CrossRef] 19. Wang, X.; Lou, W.L.; Liu, H.X.; Liu, D.L. Crashworthiness Design Analysis of Vehicle Door Using Simulation Experiment Design and Multi-Objective Genetic Algorithm. Key Eng. Mater. Trans Tech Publ. 2014, 579, 234–240. [CrossRef] 20. Lin, D.K.J.; Draper, N.R. Projection Properties of Plackett and Burman Designs. Technometrics 1992, 34, 423–428. [CrossRef] 21. Lin, D.K.J.; Draper, N.R. Screening Properties of Certain Two-Level Designs. Metrika 1995, 42, 99–118. [CrossRef] 22. Lin, D.K.J.; Draper, N.R. Characterizing Projected Designs: Repeat and Mirror-Image Runs. Commun. Stat. Theory Methods 1995, 24, 775–795. © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Journal

Aerospace – Multidisciplinary Digital Publishing Institute

Published: Sep 3, 2019

There are no references for this article.