Prediction and Optimization of Process Parameters for Composite Thermoforming Using a Machine Learning Approach.

Thermoforming is a process where the laminated sheet is pre-heated to the desired forming temperature before being pressed and cooled between the molds to give the final formed part. Defects such as wrinkles, matrix-smear or ply-splitting could occur if the process is not optimized. Traditionally, for thermoforming of fiber-reinforced composites, engineers would either have to perform numerous physical trial and error experiments or to run a large number of high-fidelity simulations in order to determine satisfactory combinations of process parameters that would yield a defect-free part. Such methods are expensive in terms of equipment and raw material usage, mold fabrication cost and man-hours. In the last decade, there has been an ongoing trend of applying machine learning methods to engineering problems, but none for woven composite thermoforming. In this paper, two applications of artificial neural networks (ANN) are presented. The first is the use of ANN to analyze full-field contour results from simulation so as to predict the process parameters resulting in the quality of the formed product. Results show that the developed ANN can predict some input parameters reasonably well from just inspecting the images of the thermoformed laminate. The second application is to optimize the process parameters that would result in a quality part through the objectives of minimizing the maximum slip-path length and maximizing the regions of the laminate with a predesignated shear angle range. Our results show that the ANN can provide reasonable optimization of the process parameters to yield improved product quality. Overall, the results from the ANNs are encouraging when compared against experimental data. The image analysis method proposed here for machine learning is novel for composite manufacturing as it can potentially be combined with machine vision in the actual manufacturing operation to provide active feedback to ensure quality products.

A computational study of the EN 1078 impact test for bicycle helmets using a realistic subject-specific finite element head model.

In the present study, the free fall impact test in accordance with the EN1078 standard for certification of bicycle helmets is replicated using numerical simulations. The impact scenario is simulated using an experimentally validated, patient-specific head model equipped with and without a bicycle helmet. Head accelerations and intracranial biomechanical injury metrics during the impacts are recorded. It is demonstrated that wearing the bicycle helmet during the impact reduces biomechanical injury metrics, with the biggest reduction seen in the metric for skull fracture.

Feasibility of using computer simulation to predict the postoperative outcome of the minimally invasive Nuss procedure: Simulation prediction vs. postoperative clinical observation.

The Nuss procedure is the most minimally invasive and commonly used surgical correction for pectus excavatum (PE) by using a prebent pectus bar to elevate the deformed chest wall. However, there are some complications associated with this procedure such as postoperative pain as well as surgical uncertainties because of human judgment. It is therefore important to understand the biomechanical effect of the pectus bar on PE thoraces undergoing an operation to alleviate the postoperative pain as well as to improve surgical outcome. The current study incorporated the finite element method (FEM) to simulate the entire Nuss procedure including the flipping process of the pectus bar on a preoperative PE patient-specific thorax model, in conjunction with comparison against the postoperative CT scans. The mid-sagittal sternovertebral elevation was found to be within 5.32 mm, whereas the transverse sternal deviations ranged from 1.59 to 3.02 mm. The average discrepancy between the predicted contour and postoperative CT contour was approximately 3%. On a different note, the stress and strain distributions largely concurred with reported findings. High bilateral stress was seen to occur at the back of ribs near the vertebral column, and particularly over the second to fifth ribs, whereas the greatest strain was found to be confined to the regions of costal cartilages. It is evident that the FEM is a feasible and robust approach in predicting the outcome of the mechanical surgical procedure. This contributes to the future development of a predictive tool incorporated in surgical planning to enhance surgical management of PE.

Face shield design against blast-induced head injuries.

Blast-induced traumatic brain injury has been on the rise in recent years because of the increasing use of improvised explosive devices in conflict zones. Our study investigates the response of a helmeted human head subjected to a blast of 1 atm peak overpressure, for cases with and without a standard polycarbonate (PC) face shield and for face shields comprising of composite PC and aerogel materials and with lateral edge extension. The novel introduction of aerogel into the laminate face shield is explored and its wave-structure interaction mechanics and performance in blast mitigation is analysed. Our numerical results show that the face shield prevented direct exposure of the blast wave to the face and help delays the transmission of the blast to reduce the intracranial pressures (ICPs) at the parietal lobe. However, the blast wave can diffract and enter the midface region at the bottom and side edges of the face shield, resulting in traumatic brain injury. This suggests that the bottom and sides of the face shield are important regions to focus on to reduce wave ingress. The laminated PC/aerogel/PC face shield yielded higher peak positive and negative ICPs at the frontal lobe, than the original PC one. For the occipital and temporal brain regions, the laminated face shield performed better than the original. The composite face shield with extended edges reduced ICP at the temporal lobe but increases ICP significantly at the parietal lobe, which suggests that a greater coverage may not lead to better mitigating effects.

Effect of helmet liner systems and impact directions on severity of head injuries sustained in ballistic impacts: a finite element (FE) study.

The current study aims to investigate the effectiveness of two different designs of helmet interior cushion, (Helmet 1: strap-netting; Helmet 2: Oregon Aero foam-padding), and the effect of the impact directions on the helmeted head during ballistic impact. Series of ballistic impact simulations (frontal, lateral, rear, and top) of a full-metal-jacketed bullet were performed on a validated finite element head model equipped with the two helmets, to assess the severity of head injuries sustained in ballistic impacts using both head kinematics and biomechanical metrics. Benchmarking with experimental ventricular and intracranial pressures showed that there is good agreement between the simulations and experiments. In terms of extracranial injuries, top impact had the highest skull stress, still without fracturing the skull. In regard to intracranial injuries, both the lateral and rear impacts generally gave the highest principal strains as well as highest shear strains, which exceed the injury thresholds. Off-cushion impacts were found to be at higher risk of intracranial injuries. The study also showed that the Oregon Aero foam pads helped to reduce impact forces. It also suggested that more padding inserts of smaller size may offer better protection. This provides some insights on future's helmet design against ballistic threats.

Investigation of the relationship between facial injuries and traumatic brain injuries using a realistic subject-specific finite element head model.

In spite of anatomic proximity of the facial skeleton and cranium, there is lack of information in the literature regarding the relationship between facial and brain injuries. This study aims to correlate brain injuries with facial injuries using finite element method (FEM). Nine common impact scenarios of facial injuries are simulated with their individual stress wave propagation paths in the facial skeleton and the intracranial brain. Fractures of cranio-facial bones and intracranial injuries are evaluated based on the tolerance limits of the biomechanical parameters. General trend of maximum intracranial biomechanical parameters found in nasal bone and zygomaticomaxillary impacts indicates that severity of brain injury is highly associated with the proximity of location of impact to the brain. It is hypothesized that the midface is capable of absorbing considerable energy and protecting the brain from impact. The nasal cartilages dissipate the impact energy in the form of large scale deformation and fracture, with the vomer-ethmoid diverging stress to the "crumpling zone" of air-filled sphenoid and ethmoidal sinuses; in its most natural manner, the face protects the brain. This numerical study hopes to provide surgeons some insight in what possible brain injuries to be expected in various scenarios of facial trauma and to help in better diagnosis of unsuspected brain injury, thereby resulting in decreasing the morbidity and mortality associated with facial trauma.

Conventional and complex modal analyses of a finite element model of human head and neck.

This study employs both the traditional and the complex modal analyses of a detailed finite element model of human head-neck system to determine modal responses in terms of resonant frequencies and mode shapes. It compares both modal responses without ignoring mode shapes, and these results are reasonably in agreement with the literature. Increasing displacement contour loops within the brain in higher frequency modes probably exhibits the shearing and twisting modes of the brain. Additional and rarely reported modal responses such as 'mastication' mode of the mandible and flipping mode of nasal lateral cartilages are identified. This suggests a need for detailed modelling to identify all the additional frequencies of each individual part. Moreover, it is found that a damping factor of above 0.2 has amplifying effect in reducing higher frequency modes, while a diminishing effect in lowering peak biomechanical responses, indicating the importance of identifying the appropriate optimised damping factor.

Development of a finite element head model for the study of impact head injury.

This study is aimed at developing a high quality, validated finite element (FE) human head model for traumatic brain injuries (TBI) prediction and prevention during vehicle collisions. The geometry of the FE model was based on computed tomography (CT) and magnetic resonance imaging (MRI) scans of a volunteer close to the anthropometry of a 50th percentile male. The material and structural properties were selected based on a synthesis of current knowledge of the constitutive models for each tissue. The cerebrospinal fluid (CSF) was simulated explicitly as a hydrostatic fluid by using a surface-based fluid modeling method. The model was validated in the loading condition observed in frontal impact vehicle collision. These validations include the intracranial pressure (ICP), brain motion, impact force and intracranial acceleration response, maximum von Mises stress in the brain, and maximum principal stress in the skull. Overall results obtained in the validation indicated improved biofidelity relative to previous FE models, and the change in the maximum von Mises in the brain is mainly caused by the improvement of the CSF simulation. The model may be used for improving the current injury criteria of the brain and anthropometric test devices.

Impact of complex blast waves on the human head: a computational study.

Head injuries due to complex blasts are not well examined because of limited published articles on the subject. Previous studies have analyzed head injuries due to impact from a single planar blast wave. Complex or concomitant blasts refer to impacts usually caused by more than a single blast source, whereby the blast waves may impact the head simultaneously or consecutively, depending on the locations and distances of the blast sources from the subject, their blast intensities, the sequence of detonations, as well as the effect of blast wave reflections from rigid walls. It is expected that such scenarios will result in more serious head injuries as compared to impact from a single blast wave due to the larger effective duration of the blast. In this paper, the utilization of a head-helmet model for blast impact analyses in Abaqus(TM) (Dassault Systemes, Singapore) is demonstrated. The model is validated against studies published in the literature. Results show that the skull is capable of transmitting the blast impact to cause high intracranial pressures (ICPs). In addition, the pressure wave from a frontal blast may enter through the sides of the helmet and wrap around the head to result in a second impact at the rear. This study recommended better protection at the sides and rear of the helmet through the use of foam pads so as to reduce wave entry into the helmet. The consecutive frontal blasts scenario resulted in higher ICPs compared with impact from a single frontal blast. This implied that blast impingement from an immediate subsequent pressure wave would increase severity of brain injury. For the unhelmeted head case, a peak ICP of 330 kPa is registered at the parietal lobe which exceeds the 235 kPa threshold for serious head injuries. The concurrent front and side blasts scenario yielded lower ICPs and skull stresses than the consecutive frontal blasts case. It is also revealed that the additional side blast would only significantly affect ICPs at the temporal and parietal lobes when compared with results from the single frontal blast case. By analyzing the pressure wave flow surrounding the head and correlating them with the consequential evolution of ICP and skull stress, the paper provides insights into the interaction mechanics between the concomitant blast waves and the biological head model.

Development and validation of two subject-specific finite element models of human head against three cadaveric experiments.

Head injury, being one of the main causes of death or permanent disability, continues to remain a major health problem with significant socioeconomic costs. Numerical simulations using the FEM offer a cost-effective method and alternative to experimental methods in the biomechanical studies of head injury. The present study aimed to develop two realistic subject-specific FEMs of the human head with detailed anatomical features from medical images (Model 1: without soft tissue and Model 2: with soft tissue and differentiation of white and gray matters) and to validate them against the intracranial pressure (ICP) and relative intracranial motion data of the three cadaver experimental tests. In general, both the simulated results were in reasonably good agreement with the experimental measured ICP and relative displacements, despite slight discrepancy in a few neutral density targets markers. Sensitivity analysis showed some variations in the brain's relative motion to the material properties or marker's location. The addition of soft tissue in Model 2 helped to damp out the oscillations of the model response. It was also found that, despite the fundamental anatomical differences between the two models, there existed little evident differences in the predicted ICP and relative displacements of the two models. This indicated that the advancements on the details of the extracranial features would not improve the model's predicting capabilities of brain injury.