Published online: 18 Mar 2013.

Transcription

1 This article was downloaded by: [Texas Technology University], [James Yang] On: 09 January 2015, At: 10:23 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK Computer Methods in Biomechanics and Biomedical Engineering Publication details, including instructions for authors and subscription information: Simulation-based assessment for construction helmets James Long a, James Yang a, Zhipeng Lei a & Daan Liang b a Department of Mechanical Engineering, Texas Tech University, Lubbock, TX 79409, USA b Department of Construction Engineering, Texas Tech University, Lubbock, TX 79409, USA Published online: 18 Mar Click for updates To cite this article: James Long, James Yang, Zhipeng Lei & Daan Liang (2015) Simulation-based assessment for construction helmets, Computer Methods in Biomechanics and Biomedical Engineering, 18:1, 24-37, DOI: / To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content ) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at

2 Computer Methods in Biomechanics and Biomedical Engineering, 2015 Vol. 18, No. 1, 24 37, Simulation-based assessment for construction helmets James Long a, James Yang a *, Zhipeng Lei a and Daan Liang b a Department of Mechanical Engineering, Texas Tech University, Lubbock, TX 79409, USA; b Department of Construction Engineering, Texas Tech University, Lubbock, TX 79409, USA (Received 21 June 2012; final version received 5 February 2013) In recent years, there has been a concerted effort for greater job safety in all industries. Personnel protective equipment (PPE) has been developed to help mitigate the risk of injury to humans that might be exposed to hazardous situations. The human head is the most vulnerable to impact as a moderate magnitude can cause serious injury or death. That is why industries have required the use of an industrial hard hat or helmet. There have only been a few articles published to date that are focused on the risk of head injury when wearing an industrial helmet. A full understanding of the effectiveness of construction helmets on reducing injury is lacking. This paper presents a simulation-based method to determine the threshold at which a human will sustain injury when wearing a construction helmet and assesses the risk of injury for wearers of construction helmets or hard hats. Advanced finite element, or FE, models were developed to study the impact on construction helmets. The FE model consists of two parts: the helmet and the human models. The human model consists of a brain, enclosed by a skull and an outer layer of skin. The level and probability of injury to the head was determined using both the head injury criterion (HIC) and tolerance limits set by Deck and Willinger. The HIC has been widely used to assess the likelihood of head injury in vehicles. The tolerance levels proposed by Deck and Willinger are more suited for finite element models but lack wide-scale validation. Different cases of impact were studied using LSTC s LS-DYNA. Keywords: impact simulation; experiment method; construction helmet; industrial helmet; head injury prediction 1. Introduction The head is the most critical area of a human body. Severe trauma to the head can lead to death or long-term disability. In fact, emergency rooms accept and treat 1 million Americans a year for traumatic brain injury (TBI). Out of these patients, 50,000 die; 80,000 90,000 experience long-term disability and the rest of them are hospitalised and survive (Goldsmith 2001). The major causes of TBI include vehicle accidents, violence, falls, sports and industrial incidents (Goldsmith 2001). A reduction in head injury could save thousands of lives and minimise the total number of people who survive but live with disabilities. Much research has been devoted to the topic of head injury. Understanding the mechanics and biomechanics of head injury is vital for engineers and scientists to mitigate such risks. The studies of head injury biomechanics can be divided into four groups: experimental, analytical, numerical and regulatory (Goldsmith 2001), where the analytical study refers to the theoretical model and the numerical method refers to the finite element method (FEM). Experimental and regulatory studies are a key to utilising numerical results to aid in the design process. Experimental results help researchers validating analytical and numerical solvers. Experiments are also vital to obtain material behaviour simulated in FE solvers. Most devices designed to reduce head injury need to receive approval of the regulatory body governing the use of a particular device. An important PPE in mitigating head injury has been the helmet. Helmets not only prevent the skull from being perforated but also dampen the force of the impact object transmitted to the wearer. Hard hats first saw wide spread use in 1931 on the Hoover Dam Project (Hoppe 2004). In the USA, employers must follow Occupational Safety and Health Administration, or OSHA, regulation and ensure that their employees wear head protection if any of the following conditions apply: Objects might fall from above and strike them on the head; they might bump their heads against objects such as exposed pipes and beams; or there is a possibility of accidental head contact with electrical hazards (OSHA R, 2003). Hard hats are also regulated by the American National Standards Institute (ANSI). The American National Standard for Industrial Head Protection or ANSI Z is referenced in the research conducted. The ANSI standards for impact only dictate two types of helmets, Type I and Type II. Type I helmets must reduce the force of impact on only the top of the head and Type II must reduce the force of impact on the top and sides of the head. ANSI creates a minimum level of protection that an industrial helmet should provide to the wearer. Not all helmets provide the same level of protection, and designs are constantly evolving to make more protective and comfortable helmets. In addition, if designers can assess helmet performance in the early design stage, it can potentially save time and money. This paper presents FE *Corresponding author. james.yang@ttu.edu q 2013 Taylor & Francis

3 Computer Methods in Biomechanics and Biomedical Engineering 25 models and injury prediction methods to accurately predict injury and enable helmet designers to study the impact of changes made to the geometry and materials. Previous studies have been conducted using FEMs to study the effects of impact on users of helmets. Simple low element count FEMs of human heads were developed in the late 1970s and 1980s. With improved computer technology in the 1990s, Dimasi et al. (1991), Mendis (1992) and Ruan et al. (1993) developed more complex head models. Dimasi et al. (1991) simulated damped and undamped impacts of car crashes for comparing the difference between the damped and undamped models. Mendis (1992) analysed brain stress and strain to correlate axonal injury intensity. Ruan et al. (1993) was able to demonstrate the countercoup phenomenon that the injury occurs on the side opposite of the area that is impacted. Computers have continued to advance and allow everincreasing complex FEMs to be developed by researchers. Two main helmet FE models dominate most of the literature. The first is the University Louis Pasteur, or the ULP, model. This model was developed by Willinger et al. (1999). This model has been used in most of the literature associated with Willinger et al. (2000). The ULP model has evolved since the mid-1990s to incorporate more complex material models for the different sections of the head. The ULP model has also been validated with experimental data from Nahum et al. (1977) (Willinger et al. 1999). Pinnoji and Hahajan borrowed the ULP model to model helmets that included foam liners (Pinnoji and Hahajan 2007). The other widely used model was developed by Horgan and Gilchrist (2003). This model is visually and structurally similar to the ULP model. Forero Rueda et al. (2011) utilised this model for FE modelling for an equestrian helmet that also included foam liners. It has been modified by Yan and Pangestu (2011) to develop an even more complex and accurate FE model. FEM head models have been applied to more than lowspeed impact cases. The model developed for use in this paper was originally obtained by Yang and Dai (2010). The original study utilised simulation to assess the risk of rear effect in ballistic helmets (Yang and Dai 2010). A ballistic helmet is able to stop handgun bullets and rifle bullets in some cases; however, the shell of the helmet is still deformed and this deformation can cause a contact between the inside of the helmet and the head. This contact may cause head tissue injury known as rear effect. Aare and Kleiven (2007) also studied the human response of a bullet impacting a ballistic helmet. The helmet in their study included the use of a suspension system that rested the helmet on the human head. This paper is based on previous study (Long et al., 2012). A basic understanding of the human anatomy is required for the set-up of the FE model. This paper will start with a brief overview of human head anatomy. Next, the two head injury criteria used will be briefly described. The two head criteria are the head injury criterion (HIC) and tolerance limits set by Deck and Willinger. Afterwards, the proposed FE model will be explained in great detail. This includes initial set-up and the material models utilised to accurately predict the reaction of the helmet and the head under impact. The head model is then compared with the Willinger et al. model for validation. Next, a study is conducted to compare the two injury criteria and determine the probability of injury to various levels. Another study is conducted comparing change in helmet geometry and its effectiveness on mitigating injury. Two different helmet geometries are compared. This paper will end with a conclusion. 2. Human head anatomy The accurate FE model of the human head is the key for simulation-based assessment and it depends on how well we understand the human head anatomy. The human head consists of three components: the bony skull, the skin and other soft tissue covering the skull and the contents of the skull (Pike 1990). There are three main sections surrounding the human brain (Figure 1). These main parts include the scalp, skull and the meninges. The scalp is stretched over the outer surface of the skull. The scalp has an average thickness of 3 6 mm and is composed of five anisotropic layers (Goldsmith 2001). These parts from descending order are as follows: (a) the skin with hairy coverings; (b) the layer of tela subcutanea, a loose, fiberous connective tissue that attaches the skin to the deeper structures; (c) the aponeurotic layer, a fiberous membrane constituting flattened tendon connecting the frontal and occipital muscles; (d) a loose subaponeurotic layer of connective tissue and (e) the pericranium, a tough vascular membrane, also designated as the subpericranial layer proximate to the skull (Goldsmith 2001). The next main part of materials, the skull is a more uniform and rigid structure. The skull has an average thickness of mm (Goldsmith 2001). The skull encloses the entire brain except for an opening at the bottom for the spinal cord. The final part consists of three sub parts. It has an average thickness of 2.5 mm. The first sub-part, the dura, which is located below the skull is tough, dense, inelastic and an anisotropic membrane consisting of connective tissue. Between the dura and the second part, the arachnoid, is a space. This space is referred to as the subdural space. The arachnoid is a delicate nonvascular membrane with interconnected trabecular fibres. The arachnoid trabecular fibres connect to the next final layer, the pia. The pia is white fibrous tissue that is attached to the surface of the brain. There is another space in between the arachnoid and pia parts. This space is referred to as the subarachnoid space. This space is occupied by water-like fluid known as the cerebrospinal fluid (CSF). The CSF provides damping and cushions the brain in

4 26 J. Long et al. Figure 1. Anatomy of the human head. (Patel and Goswami 2012). impact situations. The CSF is also produced in cavities of the brain and circulates through the spinal canal and perivascular space (Goldsmith 2001). The brain is divided into three main sections. The largest fraction being the cranium is divided into two convoluted hemispheres. The section that connects the brain and the spinal cord is the brain stem. The final section the cerebellum is where higher level functions are concentrated. All of the sections of the brain are separated by dura mater and coated with pia and arachnoid layers. The human head is a complex structure explained in previous section. However, it is impossible to model an exact human head in the FE model. We have to simplify the model to be possible to use any numerical solver to simulate the impact. 3. Injury criteria Patel and Goswami (2012) summarised the head injury criteria. Two different injury criteria are utilised in this paper: the HIC and diffuse axonal injury (DAI). The HIC is a widely accepted measure of the likelihood of head injury. The HIC is based only on translational acceleration and is originally developed for use in automotive crash test dummies. The HIC is defined as follows: HIC ¼ max ð 1 t¼t1! 2:5 ðt 1 2 t 0 Þ a t dt ; t 1 2 t 0 t¼t 0 where t 0 and t 1 are the beginning and end times of the portion of the acceleration time pulse being examined. The integral account for the duration of the acceleration and an iterative search found the time interval (t 0,t 1 )to maximise the HIC score (Shorten and Himmelsbach 2003). A HIC score correlates to a probability for a level of injury. The HIC levels of injury are as follows: Minor head injury is a skull trauma without loss of consciousness, fracture of nose or teeth and superficial face injuries. Moderate head injury is a skull trauma with or without dislocated skull fracture and brief loss of consciousness. Critical head injury is a cerebral contusion, loss of consciousness for more than 12 h with intracranial haemorrhaging and other neurological signs (Prasad and Mertz 1985). A HIC score of 1000 represents the safe limit of human tolerance, above which the risk of a serious head injury is non-zero. In the sports surfacing world, HIC scores are the primarily determinant of playground surfacing, shock attenuation performance. Other terms of surfacing shock attenuation use a 200-g max limiting performance criterion; on that basis it approximates the HIC limit (Shorten and Himmelsbach 2003). Figure 2 shows examples of Expanded Prasad Mertz Curves and the relationship between the HIC score of a head impact and the probability of an injury. Critics argue that rotational acceleration influences head injury as well. Also, the HIC does not distinguish between specific mechanisms of injury in the head. Despite the limitations of the HIC, it is most validated to date.

5 Computer Methods in Biomechanics and Biomedical Engineering 27 Figure 2. Probability of specific head injury level for a given HIC score (Canadian Playground Advisory Inc.). Deck and Willinger have proposed head injury criteria based on tolerance limits for separate parts of the human head (Deck and Willinger 2008). Deck and Willinger reconstructed 68 known head impact conditions that occurred in motorcyclist, American football and pedestrian accidents. The study concluded with proposed limits for injury mechanisms. These mechanics included moderate and severe DAI, skull fracture and subdural haematoma. The proposed limits for 50% chance of injury are classified by Von Mises stress (pressure) and skull stain energy: a mild DAI, the value for Von Mises stress recorded in the brain is 26 kpa; a severe DAI, the value for Von Mises stress recorded in the brain is 33 kpa for skull fracture, the value is 865 mj of skull strain energy and the minimum amount of CSF pressure for a subdural haematoma is 135 kpa (compression). It is important to note that these values are from the max value calculated for any element within the part involved in the injury mechanism (Deck and Willinger 2008). The detailed explanation can be referred in Table 1 in Deck and Willinger (2008). Table 1. Material models utilised in LS-DYNA to simulate the human head. The skin material properties are from Yan and Pangestu (2011) and the skull and brain material properties are from Willinger et al. (1999). Material type Human head material property r (kg m 23 ) E (MPa) V K (MPa) Scalp Elastic N/A Skull Elastic N/A Brain Viscoelastic N/A Note: r, density; K, bulk modulus; E, Young s modulus; V, Poisson s ratio 4. Finite element model The model shown in Figures 3 and 4 is the 3D FE representation of a human head which is reconstructed from cross-sectional images of the Visible Human Project Dataset (Visual Human Project). The volumetric meshing was performed in CFD-GEOM (Version 2009, CFD Research Corporation, Huntsville, AL). The skin, skull and brain are modelled with solid elements. The outer layer of human skin, which includes the scalp, consists of 128,061 elements and 25,798 nodes. The skull is modelled as a single layer with 44,938 elements and 11,828 nodes. The brain is also modelled as a homogenous structure with 33,786 elements and 6009 nodes. The mass of the head is 8.37 kg. Yan and Pangestu (2011) assumed that the behaviour of the scalp is elastic. Yan and Pangestu s material model of the skin is utilised in this paper and is listed in Table 1. In order to maximise the effectiveness of the tolerance limits proposed by Deck and Willinger, this model will use the material models employed by Deck and Willinger for Figure 3. Different views of the human FE model: (a) the outer skin surface; (b) the middle part, the skull; (c) the brain; (d) crosssectional side view; (e) front view; (f) top view. Figure 4. Comparison between the head FE model and anatomy: (a) The cross-sectional view of the FE model; (b) the cross-sectional view of the human head.

6 28 J. Long et al. Figure 5. Different views of the helmet model: (a) top view of the helmet; (b) perspective view of the helmet; (c) top view of the straps of the suspension system; (d) bottom view of the helmet, straps are coloured black; (e) side top view of the helmet, set to transparent to display the straps; (f) side view of the straps. the skull and brain. Deck and Willinger assumed that the skull is a single elastic surface, see Table 1 for further details. The skin and skull are elastic material. The brain is a viscoelastic material. The brain viscoelastic response due to shear behaviour is modelled by the following equation: GðtÞ ¼G 1 þðg 0 2 G 1 Þ exp ð2btþ; where G 0 is the dynamic shear modulus and has a value of 528 kpa. G 1 is the static shear modulus and has a value of 168 kpa. The final variable, b, is a decay constant and has a value of m/s. The constants of the viscoelastic shear behaviour are applied to the viscoelastic material in LS- DYNA. The remaining part of the FE model is the industrial hard hat. The hard hat includes an outer shell along with two straps that are in place for the suspension system. The suspension system is crucial to the effectiveness of the helmet. According to ANSI Z , the suspension is connected to the harness and acts as an energyabsorbing mechanism. Also, this harness should leave a 1.25-in. (3.175 cm) gap between the suspension and the inner helmet shell. The shell and strap pieces are modelled as shell elements. The helmet shell consists of 3752 elements with 1878 nodes. Each strap contains 2916 elements and 1708 nodes. The mass of the helmet is kg (Figure 5). 4Two more different materials are employed for the plastic response of the helmet shell and straps, see Table 1. Industrial hard hats are usually molded from high-density polymers or thermoplastics. Sabic s Ultem ATX 100, a common thermoplastic for impact and a popular additive for hard hat construction, was chosen to represent the helmet shell (Sabic Inc.). The material responses of plastics are dependent on strain rate. To capture this phenomena, the piecewise linear plasticity was chosen in LS-DYNA, MAT-024 Piecewise Linear Plasticity. Three stress strain curves at different strain rates are provided from the manufacture of Ultem (Sabic Inc.). The material tests conducted by Sabic Inc. are given in engineering stress and engineering strain, as shown in Figure 6. These values must be converted to true stress and true strain and is shown in Figure 7 as the input for LS-DYNA. Engineering stress strain curves do not give a true indication of the deformation characteristics. This is because engineering stress and strain relationship is calculated from original dimensions of the test specimens. Because the specimen dimensions change throughout the test, engineering measures are not the representative of the actual stress and strain of the material. True stress, s t,isdefinedasthe ratio of the instantaneous applied force to the instantaneous cross-sectional area instead of the initial conditions. The true stress is expressed in terms of engineering stress, s e,and engineering strain, 1 e, see equations listed below (Arriaga et al. 2007). s t ¼ s e ð1 þ 1 e Þ; 1 t ¼ ln ð1 þ 1 e Þ: There is even less published information on the materials used for the suspension system. Some common additives include nylon and plastics, such as Ultem Based on Sabic Inc. experience, the material properties

7 Computer Methods in Biomechanics and Biomedical Engineering Ultem ATX-100 Tensile Stress Strain Test Figure 6. Figure 7. True stress (MPa) Stress (MPa) of the suspension is similar to the additives, a plastic kinematic material card was developed from known material properties of Ultem 1000, see Table 2 (Sabic Inc.). The material MAT-003 Plastic Kinematic is best suited to represent the straps for limited material information given for the helmet suspension system Pull speed: 5 mm/min Pull speed: 50 mm/min Pull speed: 500 mm/min Strain (%) Stress strain test at three different pull speeds for Ultem ATX 100 in terms of engineering stress and strain. Ultem ATX-100 True Stress Strain for LS -DYNA Pull speed: 5 mm/min Pull speed: 50 mm/min Pull speed: 500 mm/min True strain (mm/mm) Stress strain curves that have been converted to true stress and strain. The first step of simulating the model is to properly place the helmet on the head. An initial simulation is run to achieve the proper gap between the suspension and the helmet shell. This step also molds the suspension to the human head and represents the wearer placing the helmet on his head. Note that the straps of the suspension system Table 2. Material models utilised in LS-DYNA to simulate the industrial helmet. Helmet material Material type r (kg m 23 ) E (MPa) V s Y (MPa) 1 F (m m 21 ) Helmet shell Piecewise linear plastic Helmet straps Plastic kinematic Notes: r, density; s Y, yield stress; E, Young s modulus; 1 F, failure strain; V, Poisson s ratio. Sabic informs that these material properties are for selection purposes only and that the user is responsible for their own material testing.

8 30 J. Long et al. Figure 8. The straps of the suspension system are constrained to four appropriate locations on the helmet: (a) a top view of the four constrained locations of the helmet (circled in red); (b) an inside view of one of the constrained node locations. are constrained to the locations of the helmet that house the connectors for the suspension system (Figure 8). The helmet is set to a location that is moved 85 mm towards the human head in the Z-direction. This is the proper location that the wearer of a hard hat would have their helmet positioned shown in Figure 9 according to construction workers experience. There is at least cm in the gap between the head and helmet. 5. Model validation Before the developed head FE model for construction helmet assessment is used, it is critical to validate it. In this work, to validate the head FE model, the same loading and boundary conditions in Willinger et al. (1999) were used. The selected outputs from this model will be compared with those data in the literature. The proposed head FE model was compared with the ULP head model. The proposed model followed the same validation process developed by Willinger et al. (1999). Willinger et al. (1999) compared their ULP model against the experimental data to validate the model. The replication experiments were conducted by Nahum et al. (1977). The experiment used a human cadaver. The ULP model was impacted in the front of the skull with a 5.6 kg rigid cylinder with an initial velocity of 6.3 m/s with the head model unconstrained. The parameters and boundary conditions for validation in this paper are shown in Figure 10. Figures 11 and 12 show the pressure response of one front location in the proposed model and the ULP test. In general, the proposed model has similar results for the total brain behaviour and even captures nearly the same maximum pressure in all of the recorded element sections. The proposed model mainly does dampen but not as quickly as the ULP model, see Figures 11 and 12. The proposed model also sees a larger countercoup brain pressure than the ULP model, see Figure 13. The ULP model records an HIC score of 744, while the proposed model HIC score is 629. One of the main differences between the two models is that the ULP model includes a layer of CSF. This could be the cause for the more dampened response in the ULP model. The two head models have different geometries, which is another possible factor for variation in the results. Other sources of error could be prorogated from approximated conditions in the validation of simulation conducted by Willinger et al. (1999). Some of these approximated parameters include impactor dimensions and material properties. The proposed model replicates most of the behaviour of the ULP model. More accurate behaviour of the head model is a consideration of future research. Note that Figures 11 and 12 are from the same simulation but from different plots of results. To keep consistent with Willinger et al. s (1999) validation examples, we showed the same simulation times as 4 and 3 ms in Figures 11 and 12, respectively. Most of the simulations conducted were run through the High Performance Computing Center at Texas Tech, or HPCC. The HPCC provides several clusters of the use of any researcher at Texas Tech. The HPCC cluster connected to run LS-DYNA is Janus. Janus is a 22-node Figure 9. Different steps of the helmet being placed on the head; (a) the starting point of the helmet, before the simulation begins; (b) around the half-way point; (c) the final placement of the helmet.

9 Computer Methods in Biomechanics and Biomedical Engineering 31 Figure 10. Both head models undergo same impact conditions. The head model is rotated 458 and impacted in the front with a velocity 6.3 m/s: (a) the proposed head FE model; (b) the ULP model by Willinger et al. (1999). Figure 11. Contour plots of the Von Mises stress in both simulations around 4 ms: (a) the proposed model; (b) the ULP model. Figure 12. Contour plots of pressure in both simulations around 3 ms: (a) the proposed model; (b) the ULP model. cluster running Windows HPC server. There are four different nodes with the capability of running simulations in LS-DYNA. Each node is running an Intel Xeon E5450 at 3.00 GHz. Each node is also equipped with 16 GB of RAM. Each simulation usually experiences a run time of 6 12 h. 6. Simulation examples 6.1 Different falling objects For studying top impact two falling objects were chosen. The first object was a cylindrical bar, with similar material properties of steel. The steel bar has a weight of 2 kg. The second object chosen was a wood board with 5 kg mass

10 32 J. Long et al. x Frontal pressure Pressure (Pa) ULP model Proposed model Time (ms) Figure 13. The recorded pressure in a front element for both the ULP model and the proposed model. and dimensions of 90 mm 90 mm 160 mm (Goldsmith 1975). Both objects can be seen in Figure 14. The results for front and side impacts did not show any significant risk of injury. The worst possible case considered was if a human is moving 3 m/s, much faster than most of the individuals. The highest HIC value was less than one, the highest Von Mises stress calculated was 3.77 kpa and the highest strain energy in the skull was 2 mj. As stated by OSHA, a hard hat is designed to protect employees from bumping their heads. If a worker is properly wearing a hard hat, it is assumed that they should be protected from the most typical injuries. Fall injuries were not considered because they are out of the scope of designed requirements for most hard hats. For top impact, the only limitation on speed is what height an object may fall. Simulations for top impact were conducted until an HIC score of 1000 was reached or until the helmet experienced a catastrophic failure in the case of the 5 kg board. The results are listed in Tables 3 and 4. Figures show the summary of results. Figure 14. The two cases of top impact: (a) the impactor is a 2-kg cylinder; (b) the impactor is a 5-kg board. The 2-kg steel cylinder has a 50% chance of causing a mild DAI around an impact speed of 8.5 m/s, a severe DAI around an impact speed of 11 m/s, a skull fracture at 15 m/s and an HIC score of 1000 at 18.5 m/s. The 5-kg wooden board has a 50% chance of causing a mild DAI around an impact speed of 6 m/s, a severe DAI around an impact speed of 8 m/s and a skull fracture at 13 m/s. However, the 5-kg wooden board never reached an HIC score of The helmet fails within the first few milliseconds of impact, at an impact speed of 18 m/s. This catastrophic failure cause the simulation to abort and an HIC score cannot be tabulated. The 5 kg does cause an HIC score of 903 at 17 m/s. The HIC gives a generalised description of different injury mechanisms that may occur at the different levels of injury. For example, a moderate head injury as described previously may or may not include skull fracture. When Table 3. Results from the 2-kg cylinder object impacting the top of the helmet. V I (m/s) s VM (kpa) Steel object (2 kg) U S (mj) HIC score HIC(d) Max HIC Note: V I, impact velocity; s VM, Max brain Von Mises stress; U S, skull strain energy.

11 Computer Methods in Biomechanics and Biomedical Engineering 33 Table 4. helmet. V I (m/s) Results from the 5-kg board impacting the top of the s VM (kpa) Wood board (5 kg) U S (mj) HIC score HIC(d) Max HIC N/A a N/A a N/A a a The simulation did not run to completion due to the catastrophic failure of the helmet. Figure 15. Brain von mises stress (kpa) kg weight 5 kg board 50% severe DAI 50 % mild DAI studying Figure 2, 50% chance of moderate injury occurs around an HIC score of 600. Now comparing the results from the two different falling objects, an HIC score of 600 is reached when the 2-kg weight has a velocity around 15 m/s and the 5-kg board has a velocity around 14 m/s. To compare with the tolerance limits, the 50% chance of skull fracture is examined. Deck and Willinger (2008) stated that a 50% chance of skull fracture occurs when the skull strain energy reaches 865 mj. For each falling object, a 50% chance of skull fracture occurs when the 2-kg weight has an impact velocity around 15 m/s and when the 5-kg board has an impact velocity of around 13 m/s. Both injury criteria predict very similar results. The advantage of using the tolerance limits is that the injuries to the head can be localised. However, the tolerance limits do not predict a more critical injury than skull fracture. Skull fracture being a severe injury may or may not lead to death. The HIC can predict more critical injuries and leaves researchers with more insight into the Max Von Mises stress in a brain element to predict DAI. Skull strain energy (mj) kg weight 5 kg board 50% skull fracture injury Velocity of impactor (m/s) Velocity of impactor (m/s) Figure 16. Max element strain energy to predict skull fracture injury.

12 34 J. Long et al kg weight 5 kg board HIC score Figure 17. HIC scores from various cases of top impact. (a) Velocity (m/s) Time (ms) injuries studied. Relying solely on an HIC score has its disadvantages as well. The HIC couples all of the sections of the human head and does not provide researchers with the enough detailed information to localise the injury mechanism. For studying head injury and determining the effectiveness of design improvements to helmets, it is advantageous for researchers to use both injury criteria. In each simulation, the impactor was defined an initial velocity. When it impacted the helmet, its velocity was reduced and its momentum was transferred to the helmet and the human head. For example, a 2-kg cylinder with an initial velocity of 8 m/s hit the top of the helmet. After the simulation, Figure 18(a) gives the time history of the impactor velocity; Figure 18(b) gives the time history of the impact force between the cylinder and the helmet. With a high Young s modulus (205 GPa), the cylinder had negligible deformation during the impact simulation. 6.2 Effectiveness of different construction helmets A study was conducted to determine the effectiveness of implanting changes to the geometry. Different helmet geometries were studied. The first helmet is the same Velocity of impactor (m/s) (b) Force (N) Time (ms) Figure 18. A 2.2 kg cylinder with an initial velocity, 8 m/s, hit the top of the helmet: (a) the time history of the impactor velocity; (b) the time history of the impacting force between the cylinder and the helmet. helmet used in the previous study, see Figure 5. The previous hard hat featured three ridges on the upper portion of the helmet and is referred to as the ridged helmet. The second helmet studied consisted of a smooth top, see Figure 19. This helmet is referred to as the smooth helmet. The smooth helmet is slightly wider. The helmet heights are identical. Both helmets are compared in Figure 20. The same straps molded to the head for the first study were reused and constrained to the smooth helmet. The helmet shell and straps used the same materials described for the first study. The straps were also constrained to the smooth helmet in the same locations as the ridged helmet. Both helmets were impacted with the 2-kg weight that was used previously. The simulations were conducted until both the tolerance limit and HIC failure criteria were reached. For an impact velocity up to 13 m/s, there is little difference in the response of the human head. Fifty percent chance of severe and mild DAIs is reached at nearly the same impact velocity, see Figure 21. When examining the plot of skull strain energy, Figure 22, the 50% chance of skull fracture occurs at nearly the same impact velocity. When approaching the HIC score of 1000, the smooth helmet suddenly shows a period of rapid acceleration

13 Computer Methods in Biomechanics and Biomedical Engineering 35 helmet never experienced catastrophic failure which causes a spike in the recorded values. Figure 19. Different views of the smooth helmet: (a) front view; (b) side view; (c) bottom view; and (d) top view. towards the maximum set of HIC score, shown in Figure 21. The strain energy in the skull also displays a similar rise after an impact speed of 15 m/s for the falling object shown in Figure 23. This is caused by the sudden failure of the helmet shell. The difference between smooth and ridged helmet, when an HIC score of 1000 is reached, is only around 2 m/s. It is important to note that the ridged Figure Discussion Note that the helmet suspension utilised for both helmets in this study is one of the simpler and basic designs in the market. Newer suspension systems include more straps, foam liners for the helmet shell and rear suspension ratchet that constrains the helmet to the back of the head as well. These newer systems could cause a dramatic increase in the impact velocity required to reach the different thresholds of injury. In this study, modelling validation was achieved by comparing with the ULP model results because the ULP model was validated through experiments. However, the proposed and the ULP models have different geometries and material properties. These factors can bring errors. Construction helmet design includes the strap design of suspension system. In this study, we chose only one type of strap material. However, for the same helmet geometry, different strap materials may have different head injury situations. Injury criteria used in this study are HIC and DAI. In the ULP model, the Von Mises stress for moderate DAI is 27 kpa and 39 kpa for severe DAI. The skull strain for skull fracture is 833 mj (Deck and Willinger 2008). The smooth helmet is set-up transparent to show the difference in geometry: (a) front view; (b) side view and (c) top view. Brain von mises stress (kpa) Ridged helmet Smooth helmet 50% severe DAI 50% mild DAI Velocity of impactor (m/s) Figure 21. Brain Von Mises stress for top impact with 2-kg weight for the wearer of the smooth helmet or the ridged helmet.

14 36 J. Long et al Ridged helmet Smooth helmet 1100 HIC score Figure 22. Figure However, the head injury tolerance limits from accidents reconstruction are as follows: a brain pressure reaching 200 kpa is an indicator for brain contusions, oedema and haematoma. A brain Von Mises stress reaching 18 kpa is an indicator for moderate neurological injuries. A brain Von Mises stress reaching 38 kpa is an indicator for severe neurological injuries. A global strain energy of the brain skull interface reaching 5.4 J is an indicator for subdural haematoma and subarachnoidal bleeding. A global strain energy of the skull reaching 2.2 J is an indicator for skull fractures (Baumgartner et al. 2007, 2009). The simulation model and accidents reconstruction limits are obviously different. In this paper, we used the ULP simulation limits because we validated the proposed model with the ULP model. The future work includes a comprehensive experiment study either by cadavers or by manikins (Hybrid Dummies) to validate the proposed simulation model to ensure that the model is accurate in the construction helmet work conditions. We will also carry out comprehensive construction helmet design examples Velocity of impactor (m/s) The HIC score for top impact with 2-kg weight for the head form with the ridged or smooth helmet. Skull strain energy (mj) 4,000 3,500 3,000 2,500 2,000 1,500 1, Ridged helmet Smooth helmet 50% skull fracture injury Velocity of impactor (m/s) Strain energy for top impact with 2-kg weight in the skull with the smooth helmet or ridged helmet. based on available helmets in the market and evaluate the effect of different strap materials. 8. Conclusion With the data collected using a FE model and the tolerance levels proposed by Deck and Willinger (2008) and tabulating an HIC score, it is possible to predict injury in wearers of construction hard hats. Not only can the threshold of injury be calculated but also the mechanism of injury can be predicted. The results from the two criteria s for head injury are not an exact match. This could be caused by the limitations of the HIC and/or the limited amount of cases reconstructed by Deck and Willinger. The proposed method for assessing the effectiveness for helmet design can be applied to more than just simple changes helmet shell geometry. Different suspensions designs can be tested. Also, different materials for the suspension system and for the helmet shell can be test as well. Future work includes different applications to the

15 Computer Methods in Biomechanics and Biomedical Engineering 37 head model. Different helmets such as firefighter, football and bicycle helmets can be applied to the head form to test different applications. Future work also includes improvements to the head form. The addition of a CSF layer could improve the validation results. Additional applications and improvements to the head form will aid designers in building more effective helmets in the future. References Aare M, Kleiven S Evaluation of head response to ballistic helmet impacts using the finite element method. Int J Impact Eng. 34: American National Standards Institute, Inc Z American National Standard for Industrial Head Protection. Arlington, VA: International Safety Equipment Association. Arriaga A, Lazkano JM, Pagaldai R, Zaldua AM, Hernandez R, Atxurra R, Chrysostomou A Finite-element analysis of quasi-static characterization tests in thermoplastic materials: experimental and numerical analysis results correlation with ANSYS. Polym Test. 26: Baumgartner D, Marjoux D, Willinger R, Carter E, Neal-Sturgess C, Guerra L, Martinez L, Hardy R Pedestrian safety enhancement using numerical methods. In: Proceedings of the 20th International Technical Conference on the Enhanced Safety of Vehicles Conference, Lyon, France, June 18 21, National Highway Traffic Safety Administration, Paper Baumgartner D, Willinger R, Raul JS Finite element modeling of the human head and application to forensic medicine. 19ème Congrès Français de Mécanique; 2009 août 24 28; Marseille. Canadian Playground Advisory Inc. Risk of head injury and HIC scores. Available from: Deck C, Willinger R Improved head injury criteria based on head FE model. Int J Crashworthiness. 13(6): Dimasi F, Marcus J, Eppinger R D anatomic brain model for relating cortical strain to automobile crash loading. Proceedings of the International Technical Conference on Experimental Safety Vehicles. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration, Vol. 2. p Forero Rueda MA, Cui L, Gilchrist MD Finite element modeling of equestrian helmet impacts exposes the need to address rotational kinematics in future helmet designs. Comput Meth Biomech. Biomed. Eng. 14(12): Gilchrist A, Mills NJ Construction site workers helmets. J Occup Accid. 9: Goldsmith W Construction helmet response under severe impact. J Constr Div. 101(2): Goldsmith W The state of head injury biomechanics past, present and future. Part 1. Crit Rev Biomed Eng. 29: Hoppe A From the hard-boiled hat to today s skull bucket: a history of hard hats [Internet] Bullard Inc. Galway, Ireland. Available from: face/head_protection/hard_hat_history/ [Accessed 2012 Jan 10]. Horgan TJ, Gilchrist MD The creation of threedimensional finite element models for simulating head impact biomechanics. Int J Crashworthines. 8(4): Long J, Lei Z, Yang J, Liang D Helmet risk assessment for top and side impact in construction sectors. Proceedings of the 4th International Conference on Applied Human Factors and Ergonomics, San Francisco, CA, July Mendis K Finite Element modeling of the brain to establish diffuse axonal injury criteria [Ph.D. dissertation]. Ohio State University. Nahum AM, Smith R, Ward CC Intracranial pressure dynamics during head impact. Proceedings of the 21st Stapp Car Crash Conference held in Detroit MI. Occupational Safety and Health Administration. OSHA R 2003 Personal Protective Equipment Patel A, Goswami T Biomedical engineering comparison of intracranial pressure by lateral and frontal impacts validation of computational model. In: Goswami, Cao, editors. Injury and skeletal biomechanics. InTech. Open Access. ISBN Pike JA Automotive safety: anatomy, injury, testing and regulation. Warrendale, PA: SAE International. Pinnoji K, Puneet M Finite element modeling of helmeted head impact under frontal loading. Sadhana. 32(4): Prasad P, Mertz HJ The position of the United States delegation to the ISO working group on the use of HIC in the automotive environment. SAE Technical Paper No Ruan SJ, Tawfik KB, King AI Finite element modeling of direct head impact. SAE Technical Paper No Sabic Inc. Sabic Innovative Plastics, Material data sheets [Internet]. Houston, Texas, USA. Available from: com/gepapp/eng/datasheetinter/dswinter [Accessed 2012 Jan 10]. Shorten MR, Himmelsbach JA Sports surfaces and the risk of traumatic brain injury. In: Nigg BM, Cole GK, Stefanyshyn DJ, editors. Sports surfaces. Calgary, AB: University of Calgary. p Visible Human Project. Available from: gov/research/visible/visisble_human.html Willinger R, Diaw BM, Kang HS Three-dimensional human head finite-element model validation against two experimental impacts. Ann Biomed Eng. 27(3): Willinger R, Diaw BM, Kang HS Finite element modeling of skull fractures caused by direct impact. Int J Crashworthiness. 5(3): Yan W, Pangestu O A modified human head model for the study of impact head injury. Comput Methods Biomech Biomed Eng. 14(12): Yang J, Dai J Simulation-based assessment of rear effect to ballistic helmet impact. Comput-Aided Des Applic. 7(1):59 73.