Numerical investigation of the rider's head injury in typical single-electric self-balancing scooter accident scenarios

Abstract

As the use of electric self-balancing scooters (ESSs) increases, so does the number of related traffic accidents. Because of the special control method, mechanical structure and driving posture, ESSs are prone to various single-vehicle accidents, such as collisions with fixed obstacles and falls due to mechanical failures. In various ESS accident scenarios, the rider's head injury is the most frequent injury type. In this study, several typical single-ESS accident scenarios are reconstructed via computational methods, and the risk of riders' head/brain injury is assessed in depth using various injury criteria. Results showed that two types of ESSs (solo- and two-wheeler) do not have clear differences in head kinematics and head injury risks; the head kinematics (or falling posture) and ESS accident scenario exhibit a distinct effect on the head injury responses; half of the analysed ESS riders have a 50% probability of skull fracture, a few riders have a 50% risk of abbreviated injury scale (AIS) 4+ brain injury, and none has a diffuse axonal injury; the ESS speed plays an important role in producing the head/brain injury in ESS riders, and generally, higher ESS speed generates higher level of predicted head injury parameters. These findings will provide theoretical support for preventing head injury among ESS riders and data support for developing and legislating ESSs.

1. Introduction

Recently, with the rapid development of battery and control technologies, electric self-balancing scooters (ESSs) are favoured by consumers due to their affordable prices, portability and flexibility [1,2]; however, a series of safety concerns have aroused researchers’ interest [3,4]. The control method, mechanical structure and driving posture of ESSs differ from those of conventional bicycles or electric two-wheelers [5]. Head injury was the most common form of injury. An examination of a series of real-world ESS accident cases showed that 44% of the reported cases had clear head injuries [6]. Through analysis of clinical records of ESS rider-related injuries, Roider et al. [4] found that the fracture of radial head was the most common injury. Some researchers have conducted in-depth research on ESS riders' head injuries using computational biomechanics methods. Xu et al. [7] used MADYMO multi-rigid body analysis software to analyse the severity of head injuries of ESS riders, cyclists and pedestrians in side-collision scenarios and found that the ESS riders’ risk of head injury was less than that of pedestrians. Shang et al. [8] used a multi-body (MB) and finite-element (FE) method to analyse the injury due to the collision between ESS riders' head and car windshield through the reconstruction of ESS collision accident; their results showed that the severity of riders’ head injury had a significant relationship with head linear velocity, angular velocity and collision position. Xu et al. [9] used the same method to conduct an in-depth study on the injury due to the collision between the head of ESS riders and the road surface, and the results showed that the risk of head injury positively correlated with speed. In summary, existing research is limited to accident scenarios where ESSs collide with vehicles, and these analysed scenarios are substantially different from the main collision patterns reflected in real accident data: Boniface et al. [10] reviewed a series of injury cases related to the ESS use and found that the riders fell off the device in 41 out of 44 cases and no vehicle crash-related case was observed. Furthermore, Trivedi et al. [11] assessed the emergency reports of many hospitals and found that the probability of single-vehicle accidents was high for standing electric scooters, of which falls accounted for 80.2%, collisions with fixed obstacles accounted for 11%, and collisions with other vehicles accounted for only 8.8%. Similarly, a recent report by Parliamentary Advisory Council for Transport Safety also revealed that an amount of single e-scooter accidents happened, and the e-scooter riders were prone to fall in such accidents [12]. As ESS is one kind of standing electric scooter, the findings obtained from these two studies [11,12] could serve as references to understand the typical ESS accident scenarios. Therefore, conducting research targeted at head injuries in single-ESS accident scenarios is necessary.

Nowadays, research on the injury of vulnerable road users (VRUs) mainly focuses on pedestrians [13–15], cyclists [16–18] and motorcyclists [19–21], and methods such as in-depth investigation of real-world traffic accidents [22–25] and numerical simulation analysis [26–29] are frequently used. With the continuous improvement in computing efficiency and the rapid reduction in computing cost, computational biomechanics models have gradually become the mainstream research approach [30–33]. Specifically for tissue-level brain injury investigation, the computational biomechanical analysis of the head of VRUs is mainly based on the reconstruction of accidents with MB models, and the head boundary conditions are input into the head-only FE model to predict the brain injury response [34,35]. This method uses the advantages of both the MB (high efficiency of calculation and model adjustment) and FE (ability to access brain injury responses) models. However, such a method underestimates the head/brain injury risk because the influence of other segments on head injury response during a collision is ignored [36,37].

Furthermore, when focusing on the evaluation of VRU head injury risk, continuous efforts have been made to establish various head injury criteria. The application of the existing criteria is mainly based on intracranial pressure such as coup pressure and contrecoup pressure [30,38], kinematics of the head centre of gravity (CG) such as head injury criterion (HIC), head impact power (HIP), rotational injury criterion (RIC) and brain injury criterion (BrIC) [39–41], brain tissue deformation such as maximum principal strain (MPS) and cumulative strain damage measure that depend on the brain deformation predicted using FE head models [38,42,43]. These criteria were widely used for the assessment of human body injury due to various impact loadings such as vehicle crash and sport events [38,42,44,45], in which the evaluation and improvement of the helmet is a critical topic [46–50]. However, the investigation of the head injury of the ESS riders was inadequate; particularly, research dedicated to the brain injury of the ESS riders has, to the best of the authors' knowledge, not been reported in the literature before.

Therefore, adopting a more sophisticated human body model and more comprehensive head injury evaluation indices for the investigation on the ESS riders' head/brain injury is necessary. In this study, a systematic and comprehensive summary of single-ESS accident scenarios through the collection and analysis of accident videos and literature research is presented. A previously built FE-MB-coupled pedestrian model [36] is introduced to numerically reconstruct the accident scenarios and extract head kinematics and injury responses, using various head/brain injury evaluation criteria, to conduct an in-depth investigation of the risk of ESS riders' head injury. The findings of this research will provide theoretical support for preventing head injury among ESS riders and data support for developing and legislating ESS.

2. Methods and materials

The widely used MB system dynamic software package MADYMO v. 7.7 [51] was employed to create the numerical model of a solo- and a two-wheeled ESS in this study. Thereafter, various single-ESS accident scenario virtual reconstructions were performed and combined with the FE-MB-coupled pedestrian model [36]. Finally, the head kinematics and brain tissue responses were extracted, and various head/brain injury criteria were used to investigate the overall riders' head/brain injury risk.

2.1. Choice of accident scenario

Falling during riding is the most common form of single-ESS accident [10]. To find the causes leading to the riders falling off the ESS and choose the single-ESS accident scenarios for our study, we collected both related literature and video records of real-world accidents. According to the study by Boniface et al. [10], some cases of the ESS rider unintentionally striking a fixed object, particularly narrow objects (traffic sign poles, guardrail posts, etc.), were observed, which motivated us to include the scenario of ‘collision with traffic sign poles'. Como et al. [52] proposed a novel method for e-scooter crash testing, in which the e-scooter with a crash dummy riding it impacts with a kerb. Although ESS is different from e-scooter, we believe that analysing the scenario of ‘collision with kerbs’ would help us get better understanding of the riders' kinematics and injury response due to self-ESS accident. While for the remaining scenarios, the motivations come from the collected video records of real-world single-ESS accidents. We collected in total 56 cases on YouTube and Bilibili (one of the major Chinese video-sharing websites), in which there are 26 cases of ‘sharp turn’, 11 cases of ‘rapid acceleration’ and five cases of ‘rapid deceleration’. For these scenarios, the main causes of falling are the loss of body balance and quick directional changes in the ESS. Furthermore, there are another four cases caused by an uneven road surface (leading to the vehicle jumping and the driver separating from the vehicle), and we simplified and classified them as ‘rumble strips’, because the uneven road surface condition is complicated and hard to simulate, and there are deceleration belts everywhere in China. The single-ESS accident scenarios in this study are summarized in table 1.

Table 1.

Single-ESS accident scenarios.

accident scenario	two-wheeler	solo-wheeler
collision with traffic sign poles	√	√
rapid deceleration	√	√
rumble strips	√	√
collision with kerbs	√	√
rapid acceleration	√	√
sharp turn	√	√

2.2. Human body and electric self-balancing scooter models

The FE (head and neck) and MB (remaining segments of the human body)-coupled pedestrian model established and verified in our previous research [36] is employed in this study. The head and neck complexes used in this model were from the Total HUman Model for Safety (THUMS, version 5.02) FE human body model developed by Toyota Central R&D Laboratory based on real human measurement parameters and cadaver experiment data [53] (figure 1a). The head includes the detailed anatomical structure of brain tissue, such as grey matter, white matter and dura matter, which can reliably predict head kinematics and brain tissue injury response [53], and the material models and material properties for the head and cervical vertebrae models are as shown in table 2 [49,53]; the MB part was from the 50th percentile adult male pedestrian model by TNO (https://www.tno.nl/en/) in MADYMO software [51] (figure 1b), and this model has been widely used in automobile safety research and road traffic accident reconstruction research [7,54,55]. For more details about this FE-MB coupled model, please refer to our previous publications [36,49,56].

Figure 1.

(a) Head–neck complex from THUMS FE human body model. (b) 50th percentile adult male MB pedestrian model. (c) The coupled FE-MB pedestrian model used in this study.

Table 2.

Material parameters for the head (cerebrospinal fluid (CSF) and meninges, brain and skull) and cervical vertebrae models used in this study.

component	material model	parameters
cerebrospinal fluid and meninges		density (kg m−3)	Young's modulus (MPa)	ultimate stress (MPa)	Poisson's ratio
CSF	linear elastic	1000	0.000164	—	0.49
pia matter		1000	1.1	—	0.40
arachnoid		1000	1.1	—	0.40
dura matter		1133	70	—	0.45

brain parenchyma		density (kg m−3)	instant shear modulus G0 (Pa)	shear relaxation modulus Gi (Pa)	material coefficient α	time constants τ (s)
white matter cerebrum/cerebellum	Ogden hypervisco-elastic	1000	1100	550	−4.7	0.06
grey matter cerebrum/cerebellum		1000	850	425	−4.7	0.06

skull		density (kg m−3)	Young's modulus (MPa)	yield stress (MPa)	Poisson's ratio
frontal-cortical	elasto-plastic	2120	11 000	48	0.22
frontal-trabecular	elasto-plastic	1000	100	0.35	0.22
parietal, temporal, occipital-cortical	elasto-plastic	2120	11 000	48	0.22
parietal, temporal, occipital-trabecular	elasto-plastic	1000	1000	4.8	0.22
facial-cortical	elasto-plastic	2120	11 000	48	0.22
facial-trabecular	elasto-plastic	1000	200	0.7	0.22

cervical vertebrae		density (kg m−3)	Young's modulus (MPa)	yield stress (MPa)	Poisson's ratio
cortical bone	elasto-plastic	2000	12 000	100	0.3

The ESS studied in the current research includes two- and solo-wheeler. Ninebot is a well-known ESS brand [57]. Therefore, the brand's No. 9 two-wheeled ESS and the solo-wheeled ESS Ninebot One A1 were used as the prototypes of MB ESS models. In the constructed MB ESS models, ellipsoids were used to establish the external contour of an ESS (figure 2); the stiffness characteristics of each component are from the literature [7] (figure 3). By referring to the literature [8,9], the friction coefficient between the ESS and pedestrian/road surface was set to 0.3.

Figure 2.

The comparison of the prototype and MB model. (a) Two-wheeled ESS and (b) solo-wheeled ESS.

Figure 3.

Stiffness characteristics of the ESS components [7].

2.3. Simulation matrix

ESSs are used as leisure and entertainment riding tools as well as short-distance transportation tools in closed places. Their design speed is generally low; most commercially available ESSs have a maximum speed of 16–18 km h⁻¹. Therefore, in this study, we used 18 km h⁻¹ as the maximum speed of ESSs, and five groups evenly from 1 to 5 m s⁻¹ were used to investigate the influence of the initial speed on the head kinematics/injury responses.

The scenario of ‘sharp turn’ was modelled by applying an initial rotational velocity of around 5 rad s⁻¹ to the ESS, which was extracted from the kinematics reconstruction of a real-world ESS accident from a video (the first one at link: https://www.youtube.com/watch?v=OWv_9BrwLik). Rapid acceleration was simulated by applying the initial velocities to the ESS, inspired by an accident at 3:40 in the video at the link: https://www.youtube.com/watch?v=9zTTbzQeaFU. The modelling of ‘rapid deceleration’ scenario was conducted by setting a friction coefficient of 0.8 between the ESS and the road surface, following the approach used in the literature [58]. The deceleration belt was modelled with arc-shaped sectional dimensions of 300 mm (width, or chord) × 40 mm (arc height, or sagitta), and the kerb was modelled with a height of 200 mm that is commonly used for real kerbs in China. When colliding with a traffic sign pole, the impact location was at the transverse midpoint of the ESS [7], and the pole diameter of 76 mm followed the commonly used value. All the accident configurations of used scenarios are listed in figure 4. Therefore, a total of 60 simulations were conducted in this study.

Figure 4.

Simulation matrix for single-ESS accidents analysed in this study. V denotes the speed of the ESS; a denotes the acceleration of the ESS; and Vω denotes the angular velocity of the ESS.

2.4. Head injury criteria

HIC is widely used to assess the severity of head injuries. It is also the injury standard of the Federal Motor Vehicle Safety Standards [7]. Therefore, HIC is selected as one of the evaluation criteria for head injury of ESS riders [59].

$$\mathrm{HIC}={\left\{{\left({\displaystyle \frac{1}{(t_2-t_1}{\int }_{t_1}^{t_2}\alpha (t)\hspace{0.17em}\mathrm{d}t}\right)}^{2.5}(t_2-t_1\right\}}_{max},$$ 2.1

where t₂−t₁ may be selected to be 36 and 15 ms. The head impact duration of the ESS accident is relatively short, so HIC₁₅ (HIC with t₂ − t₁ = 15 ms) is selected as the standard for evaluating head injury [2].

The calculation of HIC only considers the linear acceleration of head collision. From previous studies, it can be concluded that head rotation acceleration is more likely to cause serious head injury [60]. Therefore, HIP is introduced as another evaluation criterion.

HIP is a head injury evaluation criterion based on the calculation of linear and rotational acceleration of the head, proposed by Newman et al. [61],

$$\begin{array}{rl}\mathrm{HIP}& =m{a}_x\int a_x(t)\hspace{0.17em}\mathrm{d}t+m{a}_y\int a_y(t)\hspace{0.17em}\mathrm{d}t+m{a}_z\int a_z(t)\hspace{0.17em}\mathrm{d}t\\ & +I_x{\alpha }_x\int {\alpha }_x(t)\hspace{0.17em}\mathrm{d}t+I_y{\alpha }_y\int {\alpha }_y(t)\hspace{0.17em}\mathrm{d}t+I_z{\alpha }_z\int {\alpha }_z(t)\hspace{0.17em}\mathrm{d}t,\end{array}$$ 2.2

where m (kg) is the head mass, a (m s⁻²) is linear acceleration, I (kg m²) represents the moment of inertia of the head, and α represents the rotational acceleration (rad s⁻²). The human head mass and moment of inertia are as follows: m = 4.5 kg, I_x = 0.016 kg m², I_y = 0.024 kg m² and I_z = 0.022 kg m².

The head centroid angular velocity is another key factor that causes brain injury [62]. The BrIC proposed by Takhounts et al. [41] is calculated based on the maximum angular velocity of the head. BrIC is defined as follows:

$$\mathrm{BrIC}=\sqrt{{\left({\displaystyle \frac{{\omega }_x}{{\omega }_{xc}}}\right)}^2+{\left({\displaystyle \frac{{\omega }_y}{{\omega }_{yc}}}\right)}^2+{\left({\displaystyle \frac{{\omega }_z}{{\omega }_{zc}}}\right)}^2},$$ 2.3

where ω_x, ω_y and ω_z are the maximum angular velocities of the head's CG on the x-, y- and z-axis, respectively; ω_xc, ω_yc and ω_zc are the corresponding critical angular velocities, which are, respectively, 66.3, 53.8 and 41.5 rad s⁻¹ [41].

The MPS is widely used to predict the occurrence probability of diffuse axonal injury (DAI) [41]. The relevant formula is shown below [63].

$$p={\displaystyle \frac{1}{(1+e^{-3,759\times MPS+3.286})}.}$$ 2.4

Here it should be pointed out that the BrIC was developed based on a different detailed head computational model: the simulated injury monitor (SIMon) model (the specific parameters ω_xc, ω_yc and ω_zc in equation (2.3) are derived using SIMon model). Thus, this criterion is not applicable to other models, including THUMS model used in the current study, as different head models will produce different brain deformation, and this problem has been verified in the literature [50]. Similarly, this case also applies to equation (2.4). However, we believe that the introduction of BrIC and MPS-based DAI risk would provide important references to quantitatively compare the ESS riders' head/brain injury predictions for different ESS types (solo- and two-wheeled ESSs) and different accident scenarios, and investigate the influence of the ESS speed and riders’ falling postures on head/brain injury. Related results are presented in the electronic supplementary material, appendices B, C and D.

3. Results

3.1. Head kinematics of the rider

3.1.1. Kinematics classification

In this study, ESS riders exhibited the following four typical types of kinematic responses when a single-ESS accident occurred.

• (i) The rider's head directly collides with a fixed obstacle and the road surface successively. The head impacts twice and contacts the road surface first at the head back (the scenario of ‘collision with traffic sign poles,’ figure 5, referred to as ‘Kinematics 1’). After the rider's head directly collides with the traffic sign pole, the rider and ESS stop at the same position due to the interception effect of the pole. The rider loses balance and sits on the ESS, and the buttocks are in contact with the ESS. Finally, the rider leans backward because of inertia, causing the head to land on the road surface. In this case, the head collided with the traffic sign pole and road surface, respectively.
• (ii) The rider jumps forward and falls. The head impacts once on the face, with the road surface (the scenarios of ‘rapid deceleration’, ‘rumble strips’ and ‘collision with kerbs’, figure 6, referred to as ‘Kinematics 2’). The riders leaped forward because the rider and ESS were only frictionally connected with their feet during riding. Therefore, when the sudden change of the ESS motions caused the friction between the two to decrease, or the ESS speed sharply drops, the ESS and rider had sliding friction, and the rider separated from the ESS, fell forward and the face collided with the road surface. As shown in figure 6, when the human body separated from the ESS, the feet slowly separated from the ESS due to friction, the upper body fell forward and the knees first touched the road surface; finally, the head collided with the road surface.
• (iii) The ESS drives forward with a large acceleration and causes the rider to fall backward. The head impacts once on the head back, with the road surface (the scenario of ‘rapid acceleration’, figure 7, referred to as ‘Kinematics 3’). In this scenario, the ESS moved forward at a high acceleration due to mechanical failures, operating errors, etc. The CG of the rider stayed in place, the feet moved forward with the ESS due to frictional force and the human body fell backward in a clockwise direction, making the back of the head collide with the road surface.
• (iv) The ESS makes a sharp turn, and the rider falls sideways. The side of the head hits the road surface once (the scenario of ‘sharp turn’, figure 8, referred to as ‘Kinematics 4’). ESSs can achieve high-speed turning due to their strong flexibility, but it is difficult for riders to achieve rapid changes in the forward direction only by relying on foot friction. As shown in figure 8, the rider's feet slipped and touched the road surface, and then the hips, arms or shoulders, and head contact with the road surface in that order. Because there were multiple levels of cushioning before the head touched the road surface, the head injury of this type of fall was relatively small (refer to the subsequent analysis).

Figure 5.

Kinematics 1: the rider's head successively collided with a fixed obstacle and the road surface. The case involving a 5 ms⁻¹ speed is presented as an example.

Figure 6.

Kinematics 2: the rider leaped forward, causing a face collision with the road surface. The case involving a 5 ms⁻¹ speed is presented as an example.

Figure 7.

Kinematics 3: the rider fell backward, causing a back of the head collision with the road surface. The case involving a 5 m s⁻¹ speed is presented as an example.

Figure 8.

Kinematics 4: the rider fell sideways, causing a head collision with the road surface (but it was not the first body segment to collide). The case involving a 5 m s⁻¹ speed is presented as an example.

3.1.2. Head kinematics analysis

Figure 9 shows the comparisons of the predicted peak values of the kinematics parameters of the ESS rider's head CG between solo- and two-wheeled ESSs for various kinematics classified in §3.1.1, and the detailed kinematics parameter time histories can be seen in the electronic supplementary material, appendix A. Generally, the peak predictions for different ESS types and initial velocities fall into the same level (figure 9d).

Figure 9.

Comparisons of the linear acceleration time histories of the rider's head CG between solo- and two-wheeled ESSs in various single-ESS accidents at different initial velocities. (a) Kinematics 1, figure 5; (b) Kinematics 2, figure 6; (c) Kinematics 3, figure 7; (b) Kinematics 4, figure 8.

For Kinematics 1 (the scenario of ‘collision with traffic sign poles’, figure 5), only the difference in the peak rotational acceleration between two types of ESSs was relatively large (figure 9b). The difference in the first peak was insignificant, and that in the second peak (occurring during the collision between the rider's head and road surface) lies in the moment of collision (electronic supplementary material, figures A1a, A2a, A3a and A4a): the head collision time was slightly earlier for the solo-wheeled ESS rider than that for the two-wheeled rider because the latter lost his balance and his buttocks sat on the control rod, which interfered with the rider's fall process and delayed the collision of his head with the road surface (figure 5).

For Kinematics 2 (the scenarios of ‘rapid deceleration,’ ‘rumble strips’ and ‘collision with kerbs’, figure 6), no significant difference in the peak parameters was observed except for linear acceleration (figure 9a). The predicted kinematics values of the head CG before the rider's knee touched the road surface were extremely small; the value increased after the knee touched the road surface and abruptly reached the extreme value when the head touched the road surface (electronic supplementary material, figures A1b, A2b, A3b and A4b). Furthermore, the time history trends of all kinematics parameters for two ESS types were clearly consistent.

For Kinematics 3 (the scenario of ‘rapid acceleration’, figure 7), the observations of the difference in peak predictions between two ESS types were similar to those for Kinematics 2 (figure 9). The predicted kinematics parameters of the human body remained at a low level when the human body fell backward. When the head collided with the road surface, the value quickly increased. There was an insignificant difference in the time histories of the head kinematics using the two-wheeled ESS and only a slight difference in the collision time (electronic supplementary material, figures A1c, A2c, A3c and A4c).

For Kinematics 4 (the scenario of ‘sharp turn’, figure 8), the solo-wheeled ESS rider touched the road surface on his shoulders, thus avoiding a direct head collision with the road surface, resulting in much lower peak predicted acceleration of the head CG than occurred in the two-wheeler riders (figure 9a,b). However, the predicted peak velocities did not show clear difference (figure 9c,d). Similar findings could also be seen in the electronic supplementary material, appendix A (electronic supplementary material, figures A1d, A2d, A3d and A4d).

3.2. Assessment of rider's head injury risk

3.2.1. Injury criteria based on head kinematics

The evaluation criteria for head injury based on head kinematic response used in this study include HIC₁₅, HIP and BrIC (see §2.4). The distribution of predictions calculated based on the above criteria for the rider's head in various accident scenarios is shown in figures 10 and 11, and electronic supplementary material, appendix, figure B1. Obviously, irrespective of the adopted injury criterion, there were remarkable differences in the head injuries of riders in different accident scenarios. Specifically, the head injury risk in the sharp turn scenarios was significantly lower than those in other scenarios. In the rapid acceleration scenarios, the head HIC₁₅ and HIP values were both the largest (figures 10 and 11). As shown in the electronic supplementary material, appendix, figure B1, the BrIC values of the ESS rider in the scenarios of rapid deceleration, rumble strips and collision with kerbs were relatively larger than those in the remaining ones.

Figure 10.

HIC₁₅ distribution of ESS riders in various single-ESS accident scenarios.

Figure 11.

HIP distribution of ESS riders in various single-ESS accident scenarios.

We also investigated the head/brain injury risk of the riders when evaluated with the above-mentioned injury criteria (see figures 12a, 13a and 14a; electronic supplementary material, appendix, figure B2a), according to the related injury risk curves reported in the literature [41,65–67]. The cases with the predicted HIC₁₅ values of greater than 1000 (HIC₁₅ greater than 1000 may cause severe head injury [67], which is widely used as the head injury threshold in mainstream vehicle safety mandatory standards), greater than the thresholds corresponding to the 50% risk of skull fracture [64], and abbreviated injury scale (AIS) 4+ brain injury [65], accounted for 25%, 5% and 6.7%, respectively (see figures 12a and 13a). According to the skull fracture risk curve related to the HIP obtained by Marjoux et al. [66], when HIP = 38 kW, the occurrence probability of skull fracture was 50% (figure 14a), and in our study, the cases with a skull fracture risk of greater than 50% based on this curve accounted for 58.3% of all analysed single-ESS accidents. Similarly, 41.7% of the cases had a greater than 50% risk of AIS 4+ brain injury when evaluated with the BrIC (the BrIC threshold is 0.87 for a risk of 50%), calculated using the risk curve proposed by Takhounts et al. [41]. Overall, for the ESS riders in the analysed single-ESS accident scenarios, skull fracture is more likely to occur, compared with severe (AIS 4+) brain injury and DAI.

Figure 12.

Skull fracture risks of the riders were predicted using the injury risk curves for HIC₁₅ in the literature [64] (a), and the comparison (b) of the cumulative distribution in all scenarios for the predicted risks between solo- and two-wheeled ESS riders.

Figure 13.

AIS 4+ brain injury risks of the riders were predicted using the injury risk curves for HIC₁₅ in the literature [65] (a), and the comparison (b) of the cumulative distribution in all scenarios for the predicted risks between solo- and two-wheeled ESS riders.

Figure 14.

Skull fracture risks of the riders were predicted using the injury risk curves for HIP in the literature [66] (a), and the comparison (b) of the cumulative distribution in all scenarios for the predicted risks between solo- and two-wheeled ESS riders.

Furthermore, the cumulative distribution of all analysed single-ESS accident cases in all scenarios when evaluating the head/brain injury risk with various injury criteria was examined, and the differences between solo- and two-wheeled ESS riders were considered (see figures 12b, 13b, 14b; electronic supplementary material, appendix, figure B2b). For the injury risk calculated from HIC (figures 12b and 13b), no obvious difference was observed in the 80% of cases with relatively lower risk, whereas for the remaining 20% cases with a higher risk, the solo-wheeled ESS riders seemed to be more vulnerable. For the analysis based on HIP (figure 14b) and BrIC (electronic supplementary material, appendix, figure B2b), two types of ESSs did not show remarkable differences in head/brain injury risk.

3.2.2. Injury criteria based on brain tissue deformation

In this study, the 98th percentile maximum principal strain MPS_0.98 was used to predict the occurrence of DAI, because the peak MPS computed at a single Gauss integration point in the FE head model may be prone to modelling artefacts [49]. Figure 15 shows the brain strain cloud diagram and predicted MPS_0.98 of riders when typical single-ESS accidents occurred for both solo- and two-wheeled ESS riders. Figure 16 shows the predicted MPS_0.98 of ESS riders in different accident scenarios. Similar to the analysis on the head kinematics-based injury criteria, the calculated MPS_0.98 in the sharp turn scenarios was clearly lower than in other scenarios (figure 16).

Figure 15.

Predicted distribution of the MPS (we report the 98th percentile value of the maximum principal strain MPS_0.98) within the brain of the solo- (a) and two- (b) wheeled ESS riders for typical single-ESS accident scenarios. The figures show mid-sagittal cross-sections at the time when the MPS_0.98 was observed.

Figure 16.

MPS_0.98 distribution of ESS riders in various single-ESS accident scenarios.

4. Discussion

4.1. The influence of electric self-balancing scooter speed on head injury

In a typical VRU-to-vehicle impact accident, the vehicle speed is a key factor that affects the severity of the VRU head injury risk [9,68,69]. However, the influence of the rider's speed on head injury in a single-ESS accident has not been explored. In this study, the ESS rider's speed was divided into five levels evenly from 1 to 5 m s⁻¹, and predicted injury parameters were calculated from the analysed head/brain injury criteria (HIC₁₅, HIP, BrIC and MPS_0.98) in each accident scenario at five selected speeds (60 accident cases). We performed linear regression analysis on the criteria-based predictions of different speeds for each scenario (figure 17; electronic supplementary material, appendix, figure C1), and a correlation coefficient greater than 0.6 (R² > 0.6) reflected a significant correlation between the predicted criterion values and speed [70]. We did not address solo- and two-wheeled ESS riders separately, considering that they had not shown evident differences in head/brain injury responses (figures 12b, 13b, 14b; electronic supplementary material, appendix, figure B2b).

Figure 17.

Head injury parameters were predicted using the three injury criteria in typical single-ESS accident scenarios versus ESS speed: (a) HIC₁₅, (b) HIP and (c) MPS_0.98. For the description of the scenarios, see §2.3.

When assessing the correlation coefficient with the selected threshold, the correlations with ESS speed were more obvious for HIC₁₅ and BrIC than for HIP and MPS. Specifically, the HIC₁₅ was significantly correlated with the ESS speed in the scenarios of ‘rapid deceleration’ (R² = 0.6927) and ‘rumble strips’ (R² = 0.7595), and significant correlations also applied to BrIC in the scenarios of ‘rapid deceleration’ (R² = 0.6113), ‘rapid acceleration’ (R² = 0.6724) and ‘rumble strips’ (R² = 0.909). These two criteria produced two more correlation coefficients close to 0.6: R² = 0.402 for HIC₁₅ in ‘rapid acceleration’ and R² = 0.4424 for BrIC in ‘collision with traffic sign poles’. For the remaining criteria of HIP and MPS, all of the coefficients were lower than the predefined threshold, with the only exception being the case for HIP in the scenario of ‘rapid acceleration’ (R² = 0.4832). Particularly for MPS, the calculated coefficients were far below the threshold. One possible explanation for the above phenomenon is that the ESS speed exerted a more obvious effect on HIC and BrIC because these criteria solely rely on a single type of motion (HIC: linear, BrIC: rotational), whereas the influence tended to disappear when both linear and rotational motions of the head occurred, leading to a much more complicated head loading. Meanwhile, it is worth noting that, considering the limited statistical accuracy due to small numbers, the main objective of employing the linear regression analysis here is to help understand how the ESS travel speed influences the head injury of the rider, rather than to quantitatively investigate the correlation between the ESS travel speed and the head injury of the rider. In summary, the predicted injury criteria of the riders increased with ESS speed in single-ESS accidents (figure 17; electronic supplementary material, appendix, figure C1), except for some scenarios using HIP and BrIC (figure 17b; electronic supplementary material, appendix, figure C1). Such finding is consistent with the similar research focusing on the head injury of e-scooter rider [12,31].

4.2. The influence of falling postures on the head injury risk

Subsection 3.1.1 describes in detail the kinematic response of the rider's head and the classifications of all analysed accident cases into four types (see §3.1.1) in terms of the head kinematics observed in the selected various single-ESS accident scenarios (table 1 and figure 4).

In this section, the predicted head injury risks based on the analysed criteria (HIC₁₅, HIP, BrIC and MPS_0.98) of the ESS rider under the conditions of four head kinematics are presented in scatter plots (figure 18; electronic supplementary material, appendix, figure D1) to show the influence of falling posture on the head injury risk of ESS riders.

Figure 18.

Comparisons of the head injury risk distribution of ESS riders by head kinematics (for the detailed classifications, refer to §3.1.1) when evaluated with various injury criteria: (a) HIC and (b) HIP. Black lines are the group means.

Generally, the head impact kinematics exerted an apparent effect on head injury risks, and the Kinematics 4 (scenario of ‘sharp turn’, figure 8) produced the lowest head injury risks of all injury types. For the skull fracture, the risks from Kinematics 1/2/3 did not appear significantly different, and the differences between these risks and those from Kinematics 4 based on HIP were far more evident than on HIC (figure 18a,b). For AIS 4+ brain injury, the risks from Kinematics 1/3/4 were quite similar and considerably lower than those from Kinematics 2 (electronic supplementary material, appendix, figure D1a). The DAI risks seemed to be relatively insensitive to the change of head impact kinematics, especially considering that all the predictions fell into an extremely low range of risk (electronic supplementary material, appendix, figure D1b).

4.3. Study limitations

In this study, only the 50th percentile adult male model was used for analysis, thus the human body diversity in real accidents, such as females and minors [71,72], was not considered. Second, only the single-ESS accident was considered, while attention was not paid to the collision of the ESS with other traffic participants. In fact, when an ESS is used for commuting on public roads, it may appear on pavements, non-motorized vehicle lanes and motor vehicle lanes, so there is a possibility of collision with pedestrians, electric motorcycles or passenger cars [73]. Finally, in the event of a real-world accident, a VRU usually has an avoidance motion, such as holding the head with hands, jumping up or curling up [74]. These actions are difficult to reproduce in simulation, which may lead to a certain impact on the validity of the results [75,76].

5. Conclusion

A coupled FE-MB human body computational biomechanics model was used to explore the riding safety of ESS in typical single-ESS accident scenarios. By analysing the head kinematic and craniocerebral strain responses to comprehensively evaluate the rider's head/brain injury risks, we draw the following conclusions.

Solo- and two-wheeled ESSs did not show clear differences in the head kinematics or head injury risks, particularly the risks of AIS 4+ brain injury based on BrIC and those of DAI injury calculated from the MPS, whereas the head kinematics (or falling posture) and ESS accident scenario exhibited a distinct effect on the head injury responses. The ESS riders in the accident scenario of ‘sharp turn’ exhibited the lowest risk of head/brain injury.

Among all analysed ESS riders, half had a 50% probability of experiencing a skull fracture, only a small proportion of the riders had a 50% risk of AIS 4+ brain injury, and none had a DAI injury. Furthermore, the ESS speed played an important role in producing head/brain injuries in ESS riders, particularly when evaluated with injury criteria HIC, BrIC and MPS; generally, higher ESS speed generates higher level of predicted head injury parameters.

Our results suggest that the ESS riders who have an accident would benefit from the cushioning of their hands, shoulder and chest to lower the potential severity of collisions between their head and the road surface.

Data accessibility

The data are provided in the electronic supplementary material [77].

Authors' contributions

F.W.: conceptualization, formal analysis, funding acquisition, project administration and writing—review and editing; J.H.: data curation, formal analysis, software, validation and writing—original draft; L.H.: funding acquisition, project administration and resources; S.H.: formal analysis, software and visualization; M.W.: formal analysis, investigation, software and visualization; J.Y.: formal analysis, methodology, software and visualization; T.Z.: investigation and methodology; Q.L.: investigation, validation and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was supported by Hunan Province Natural Science Fund, China (grant no. 2021JJ30721), Department of Education of Hunan Province (grant nos. 21B0324 and 21A0193), the National Natural Science Foundation of China (grant nos. 52172399, 52005054 and 51875049) and Hunan Key Research and Development Program, China (grant no. 2020SK2099).