1 Introduction

In various football codes, players utilise studded footwear to increase their traction with the playing surface. While this increase in traction reduces the risk of slipping, excessive traction can cause foot fixation within the playing surface and an increased risk of lower limb non-contact injuries, such as anterior cruciate ligament (ACL) tears [1]. A tear to the ACL can be detrimental to athletes on various levels from large economic setbacks, to an increased risk of future ACL injuries or osteoarthritis later in life [2]. It has been estimated that 70 to 80% of these ACL tears in athletes occur in non-contact situations whereby large reaction forces are transmitted through a closed kinematic chain within the lower limb [3].

The primary role of the ACL is to restrict anterior movement of the tibia with respect to a stationary femur as well as provide a secondary restraint to the internal rotation of the tibia [4]. The exact mechanism of an ACL tear is debated within literature with Koga et al. [5] proposing the quadriceps drawer hypothesis, and others highlighting the importance of external tibial rotation [6], knee valgus and hyper-extension [7]. Ireland [8] suggests a ‘position of no return’ as the primary mechanism for ACL injury whereby the athlete’s knee ‘collapses’ inward and the ACL to tear on its medial side. Despite the uncertainty around injury mechanism, it is evident that there are different anatomical positions of the lower limb that can lead to ACL injury.

In soccer, players wear boots that incorporate studs to increase their contact area with the playing surface to increase traction and improve performance. A study by Sterzing et al. [9] highlighted the impact studs have on playing performance, showing a 3% improvement time for an agility course between two different outsole configurations. While an increase in traction can lead to performance benefits, excessive traction has also been linked to higher injury risk. Studies from Torg et al. [10] and more current investigations from Wannop et al. [11] showed a link between increased traction and a higher risk of lower limb injuries. Thus, due to its connection to player performance and injury risk, boot/surface traction has been widely researched, most commonly with a rotational movement [1, 12,13,14,15].

When performing these mechanical traction tests, researchers often utilise a fixed footform that is either rotated or translated on a testing apparatus at a constant speed with a consistent load. To simplify the testing mechanism, the position of the foot and the angle of contact are typically fixed throughout the trial. In some traction tests, researchers adopt a neutral testing angle, whereby the foot remains flat relative to the playing surface and the entire outsole is in contact with the playing surface [15, 16], while others have opted for forefoot contact only by applying plantarflexion to the ankle joint [12, 17,18,19]. Video analysis of ACL injuries in soccer and rugby players performed by Della Villa et al. [20] and Della Villa et al. [21] highlighted two key points during the movement that resulted in injury. The first is the initial contact (IC) which Cochrane et al. [22] described as a failure point as the leg is usually fully extended, and the second point is the injury frame (IF). The IF occurs about 40 ms after the IC, where there is anterior translation and internal rotation of the tibia and when the ground reaction force is assumed to be at its peak [23]. In research involving male soccer players and rugby union players, it was found that at the IF, all of the rugby players and 89% of the soccer players had a flat foot, and, at IC, 51% of rugby players and 48% of soccer players landed on their heel [20, 21]. This mechanism of heel striking is consistent within the literature for female soccer players, players in the Australian Football League as well as American Football and male and female American basketball players [22, 24,25,26]. Despite the frequency of heel strike during IC of an ACL injury, it is rarely investigated during traction analysis.

Previous traction literature rarely investigates different foot angles for each direction of movement of the footform relative to the playing surface. The aim of this investigation was to observe the effect of the ankle angle on the translational traction on both a natural and artificial grass playing surface for different movement directions. As a player’s foot was generally flat during IF [20, 21], suggesting that this may be the point of highest load, it was hypothesised that the translational traction force would decrease as the angle of attack increased as there is less contact area between the studded outsole and the playing surface.

2 Methodology

2.1 Testing apparatus

To examine the translational traction of soccer boots at different foot approach angles, a custom-built testing apparatus was used. The apparatus, shown in Fig. 1, was developed at the University of Adelaide and has been used in previous traction research [27]. The apparatus utilised an electric hoist (winch) to drag a sliding table across a system of linear bearings at a speed of 4 m/min. This table was attached, via a four-bar linkage, to a mounting plate that housed a footform. A normal vertical force of 148 N was applied to this mounting plate via barbell weights (100 N in weight, remaining 48 N applied by mounting plate and accessories) and the traction force was recorded using an S-Type load cell (model STS Metric; SunScale Inc; combined error of 0.025% of the output) at a rate of 150 Hz. Prior to use, the load cell and apparatus was calibrated and validated. The load cell was calibrated with a series of known masses from 1 to 50 kg as this was the estimated range of loads for the experimental procedure. The validity of the apparatus was confirmed by comparing dynamic friction coefficient values of common materials, in a dry, unlubricated setting, against the values obtained by Serway [28] (Table A1).

Fig. 1
figure 1

The testing apparatus used. A (left) shows the mounting plate and sliding table and B (right) shows the hoist

Full size image

The footform was taken from a full body 3D scan of a male from the database presented by Mitsuhashi et al. [29]. The foot was 3D printed using fused acrylonitrile butadiene styrene (ABS) filament using the Flashforge Creator Pro (Flashforge Technology Co. Zhejiang, China, 2011). The length and width of the footform was 270 mm and 95 mm, respectively. The footform, shown in Fig. 2, was connected to a removable ankle joint via a series of pegs and dowels to ensure a secure fit between the interchangeable ankle joints and the footform. During each trial, the footform was visually monitored to ensure minimal movement of the ankle joint. Trials were discarded and the joint reinforced using thicker dowels if the relative movement between the ankle joint and the mounting platform or the ankle joint and the footform had shifted more than 5 mm from the initial position to the end position during the trial. This check between the start and end position was measured with digital callipers after each trial. It was this separate ankle joint that was altered to allow different angles of the foot relative to the playing surface to be tested during this investigation.

Fig. 2
figure 2

The complete footform and plate (left), unassembled footform (central) used in its neutral position as well as the footform and mounting plate in the 30° plantarflexion position (right)

Full size image

2.2 Surfaces

Both natural grass and artificial grass playing surfaces were utilised. The playing surfaces belonged to soccer clubs who compete in the top semi-professional division in South Australia and the artificial grass surface passed FIFA quality assessment for match play. Due to the challenges with consistency over time involved with traction testing on natural grass [27], the artificial grass playing surface was chosen as the primary surface for investigation. The artificial grass playing surface was a third-generation artificial grass soccer surface consisting of a Max S yarn with a blade length of 50 mm, a styrene-ethylene-butylene-styrene performance infill, a quartz sand supporting infill and a shockpad draining and shock absorption system. The natural grass playing surface consisted of a Kikuyu base with Ryegrass support with a moderate thatch layer. The moisture content of this natural grass playing surface was between 13 and 15% during the testing period and the grounds underwent regular maintenance by ground staff. Testing locations on each pitch were examined to avoid high-wear areas (e.g., the centre circle of penalty spot) as well as any sprinkler units or other interferences.

2.3 Boot selection

To ensure consistency between tests, one boot was utilised for the tests. Due to its popularity, the boot chosen was the Nike Tiempo Legend 9 Elite Firm Ground boot [30]. The specific boot was the right boot of a US size 9.

The boot was secured to the footform using the lacing system provided. Before each trial, the boot was laced up to its maximum capacity. While this was seen as excessive compared to normal use, it limited movement of the boot on the footform.

2.4 Testing procedure

The boot was translated in four different directions as shown in Fig. 3 on both a natural and an artificial grass playing surface with varying ankle angles. As there is less spatial and temporal variation on an artificial grass playing surface [27], a more detailed range of ankle angles were used on artificial grass. While the angle modifications made to the 3D footform do not directly translate to the complex, multiplanar movements of the ankle, they were performed along planes that correspond with plantarflexion and dorsiflexion, as well as calcaneal inversion and eversion, especially when the applied normal load is vertical. This approach is similar to Thomson et al. [17] who tested different outsole types at 20° plantarflexion. To limit variability within the data, for each direction of movement, one degree of freedom at the ankle joint was altered. For posterior movement (Fig. 3A), ankle plantarflexion was examined, ankle dorsiflexion was altered during anterior movement (Fig. 3B), calcaneal inversion explored during medial movement (Fig. 3C) and finally the calcaneal eversion angle varied during lateral movement (Fig. 3D).

Fig. 3
figure 3

The four different movement directions (and their associated angles) and an example of how the boot’s orientation was altered for each movement

Full size image

Ten trials were taken for each ankle joint angle and movement type on the chosen playing surface. The boot was examined for any significant wear between trials. If there was no significant damage present, the same boot was used again to ensure limited quality changes between two separate boots. As anterior movement of the forefoot was designed to simulate a sudden deceleration, varying levels of dorsiflexion were investigated to simulate the heel strike shown at IC during ACL injury analysis [20]. Posterior movement replicated an athlete pushing off; thus, ankle plantarflexion was altered in this movement. The varying degrees of movement were limited between 0° and 30° based on the data from Schick et al. [31]. Their study found that 53% of ACL injuries in the national football league (NFL) occurred with 0° to 30° of plantarflexion while 40% of injuries occurred with 0° to 30° of dorsiflexion. The lateral movement experienced in a plant-and-cut injury scenario was coupled with sub-talar pronation and calcaneal eversion as 60% of injuries to male soccer players incurred some level of calcaneal eversion. Finally, the medial movement of the leg was used to simulate a cross-cut manoeuvre whereby the athlete lands on the lateral portion of their foot with a degree of sub-talar supination and calcaneal inversion. The movement types, coupled with an example of the angle used for investigation can be seen in Fig. 3. The range of calcaneal inversion and calcaneal eversion angle choice was between 0° and 15° as this was the range obtained at IC from 10 ACL injuries to female basketballers and handball players [23]. The values tested for each movement on each playing surface are described in Table 1, where 0° indicates no movement along the associated anatomical plane for the specific ankle movement. The movement of the boot was visually observed by researchers at each test angle to determine what studs interacted with the playing surface for each movement direction.

Table 1 The various angles used for each movement on both natural and artificial grass
Full size table

2.5 Data processing and analysis

Due to inconsistencies in growth rate, irregular maintenance and possible diseases, natural grass playing surfaces will often exhibit inconsistencies in shear resistance and penetration resistance [32], which can affect traction [33]. To account for these inconsistencies in translational traction on natural grass [27], the reporting methodology proposed by Loud et al. [27] was utilised to process and analyse the data. This methodology suggests taking a prescribed number of trials at the same location on the playing surface, isolating the initial trial as an indicator of injury risk, then taking the mean of the remaining trials to form a comparison between the test variable/s. The initial trial was reported in Figs. 4, 5, 6, 7, however, should only be used for an injury-based analysis as it provides the absolute maximum value of traction generated.

Fig. 4
figure 4

The mean tractional coefficient values at different dorsiflexion angles of the boot moving in the anterior direction on natural and artificial grass. Error bars indicate the standard deviation in the maximum TC from trials 2–10

Full size image
Fig. 5
figure 5

The mean tractional coefficient values at different plantarflexion angles of the boot moving in the posterior direction on natural and artificial grass. Error bars indicate the standard deviation in the maximum TC from trials 2–10

Full size image
Fig. 6
figure 6

The mean tractional coefficient values at different calcaneal inversion angles of the boot moving in the medial direction on natural and artificial grass. Error bars indicate the standard deviation in the maximum TC from trials 2–10

Full size image
Fig. 7
figure 7

The mean tractional coefficient values at different calcaneal eversion angles of the boot moving in the lateral direction on natural and artificial grass. Error bars indicate the standard deviation in the maximum TC from trials 2–10

Full size image

A custom MATLAB script was written to convert the traction force recorded by the load cell into a dimensionless traction coefficient (TC) for the duration of each trial. The peak TC for each trial was then recorded as a predictor of the potential injury risk for each loading condition [23, 34]. This use of the peak translational traction was also adopted by Wannop et al. [35] and Thomson et al. [17]. In their analysis of translational traction against injury type, Wannop [36] utilised the peak TC to group different boots into ‘low’, ‘mid’, and ‘high’ levels of traction with ‘high’ traction boots resulting in more moderate and severe injuries compared to the other traction groups. The maximum TC for each of the nine remaining trials for each loading condition was collated in terms of surface type, movement type and the ankle joint angle before undergoing a 3-way ANOVA test in SPSS Statistics 27 (IBM, New York, US). A significance level of 0.05 was used. To measure the effect size of each variable (i.e., the magnitude of the impact of each variable), the Partial Eta-Squared variable was used with 0.01 denoting a small effect size, 0.06 for a medium effect size and 0.14 for a large effect size [37]. Distribution of data was found to be non-normal through the Shapiro–Wilk Test and thus the non-parametric Levene’s median test for the homogeneity of variance was used [38]. As the data fulfilled the remaining assumptions, a 3-way ANOVA test was run to determine the impact of angle of attack, surface type and movement type. The results of this analysis are in the Online Resource 1. As the hypothesis suggested a linear trend between angle and translational traction, a Pearson’s correlation analysis was performed in SPSS for each movement and surface type.

3 Results

The significance level shown in Table A2 demonstrates that all three testing factors, as well as their combinations, are significant in terms of translational traction. The Partial Eta-Squared values in Table A2 highlight that Surface Type and Movement Type separately have the most significant impact on traction values. These, have been explored previously in Loud et al. [27]. In this work, it was hypothesised that the translational traction would decrease as the tested angles (plantarflexion, dorsiflexion, calcaneal inversion and calcaneal eversion) increased due to a decrease in contact area between the boot and the playing surface compared to the neutral angle position. The neutral position used in this investigation was a position whereby the sole of the foot was to lie at 90° to the longitudinal axis of the ankle and there was no rotation about the sagittal axis. As the tested angles increased, this raised the sole of the foot in a particular direction depending on the specific angle. As the sole gradually lifted from the surface, the contact area between the boot outsole and the surface would decrease, thus lowering the translational traction. For 75% of these loading conditions there was no significant linear trend as indicated by the Pearson’s correlation. As the results presented in Figs. 4, 5, 6, 7 do not show a direct linear correlation between angle and TC, a one-way ANOVA was carried out to determine the statistical significance of the different approach angles. A post-hoc Tukey test was also carried out to determine the statistical significance of various angles for each movement and surface type. As there was a tendency for players who sustained an ACL injury to be flat-footed at the assumed point of failure [20, 21], the neutral position was used as a baseline for the Tukey analysis for each movement [20, 24].

Figures  4, 5, 6, 7 show the varying levels of traction of each angle for the four various movements. Each Figure shows the mean of the maximum TC from trials 2–10 for each loading condition as well as the initial trial. Where differences were significant, indications of the level of significance are shown between the neutral angle and the various angles that were tested. Specific details of the Tukey analysis, as well as the specific traction values themselves, that compare each individual test angle within the data can be found in Online Resource 1.

Anterior movements (Fig. 4) on artificial grass followed a bimodal trend, with peaks in traction occurring at 5° and 20° of dorsiflexion while there was significantly less traction at 30° dorsiflexion angle. Similarly, for posterior movements on artificial grass as shown in Fig. 5, there was a bimodal trend, however, with this movement, larger TC values were observed around 15 and 20° plantarflexion, as well as the neutral position of 00 plantarflexion. Both medial and lateral movements, Figs. 6 and 7 respectively, showed substantially larger levels of traction at increased angles of calcaneal inversion and calcaneal eversion on artificial grass.

4 Discussion

The studs on soccer boots have a direct impact on the traction a player can generate. While it has been shown that the size and shape of the studs has a direct impact on this traction [39], the results presented in Figs.  4, 5, 6, 7 highlight that there is not a direct linear trend between ankle angle and traction. This non-linear interaction indicates that the relationship between stud and traction depends on the stud shape relative to the direction of movement, and its ability to move through the playing surface. Thus, further comparison of individual angles for each movement and surface type was needed.

At 30° of dorsiflexion, there was significantly less translational traction for anterior movement than in the neutral angle position. This steep angle allowed for only the two rear studs of the boot to fully penetrate the playing surface and while these two studs are the longest on the outsole, the significant reduction in contact area led to a decrease in TC. While this decrease in TC could potentially lead to an increased risk of the player slipping, the transition from a lower traction value at IC to a higher TC at IF could provide the player with a larger window to correctly align their lower limb to avoid injury. If a player experiences similar traction levels at IC than they would at IF, then there is potential for minimal changes in player biomechanics between IC and IF. This notion of a decreased risk of ACL injury due to a decrease in TC beyond 30° of dorsiflexion is supported by data from the NFL where, of all the recorded ACL injuries, none of them incurred an ankle dorsiflexion angle of between 30° and 60° [31]. The decrease in TC at 7.5° and 10° of dorsiflexion could be explained by the shape and position of the rear studs as shown in Fig. 8. At these angles of dorsiflexion, the forefoot studs do not interact with the playing surface, leaving only the rearfoot studs in contact with the playing surface. The two foremost studs are shaped such that they can efficiently move through the playing surface with minimal resistance. As the dorsiflexion angle is further increased, the two foremost studs are no longer in contact with the playing surface, resulting in the two rearmost studs providing all the traction for the boot. The orientation of these studs, which provides a large contact area normal to the direction of movement, results in an increase in TC shown at 20° dorsiflexion as these studs engage in the placing surface before finally reducing the traction at 30°. This reduction in traction could be caused by the first row of studs (indicated in blue in Fig. 8) riding over the surface at this higher dorsiflexion angle, rather than engaging in it which likely caused the increase in traction experienced at 20° of dorsiflexion.

Fig. 8
figure 8

The rear section of the right boot tested, highlighting the direction of movement for the anterior tests and the shape of the rear studs (Colour figure online)

Full size image

A peak in the TC for posterior movement was found at 15° of plantarflexion. At this angle, there was maximum penetration from all studs on the forefoot while there was still some contribution from the heel studs. At 30° of plantarflexion, only the four studs at the front of the boot penetrated the playing surface. Two of these studs are circular in shape and the two at the toe of the boot are bladed, designed to aid in the propulsion of the player during acceleration. It has been found, however, that athletes exhibit plantarflexion angles of about 26° when performing high acceleration sprints [40]. This suggests that the positioning and geometry of the forefoot studs may not provide the required level of traction for athletes to perform a high acceleration sprint at the optimum plantarflexion angle. Data from the NFL shows that for higher angles of plantarflexion, there is a decreased risk of ACL injury among athletes which could possibly be linked to the decrease in TC at 30° [31].

Medial movements (Fig. 6), while not commonly associated with ACL injury risk, simulate a cross-cut movement performed by an athlete and a lateral foot strike, as indicated in Fig. 3D. Based on a video analysis of male professional soccer players, and female Australian Football players, this cross-cut movement was found during 5% of the ACL injuries experienced by the soccer players [41] and 5.8% of the ACL ruptures of the female Australian Football players [42]. While following a near-normal distribution, rather than a linear model, the data failed the hypothesis that an increase in calcaneal inversion angle would lead to a decrease in TC. Instead, an increase in calcaneal inversion angle saw a significant increase in TC for each magnitude of angle tested. The shape of the studs plays a significant role in this increase in traction as shown in Fig. 9. When the foot is in a neutral position, it is speculated that the medial studs are shaped to disperse dirt and debris from the playing surface in an efficient manner with minimal resistance. As the calcaneal inversion angle increases, the contact area of the medial studs decreases as they are lifted higher, decreasing their contact area, forcing the lateral studs to provide most of the traction. These studs, however, are shaped with flat surfaces on their medial side such that they resist motion in the medial direction and increase the overall traction of the boot, despite the overall contact area decreasing. The flat-fronted shape of the lateral studs with regards to medial movement increases the TC until the 7.5° inversion angle. After this angle, the medial studs lift further from the playing surface, engaging less which leads to a decrease in TC for inversion angles beyond 7.5°, before the decrease in contact area of the lateral studs caused a decrease in TC.

Fig. 9
figure 9

The engagement profile of the outsole configuration of the right boot based on the direction of movement that highlight the medial studs (red) and the lateral studs (blue) (Colour figure online)

Full size image

A plant-and-cut movement is one of the most common injury mechanisms for soccer players [20]. This movement involves the foot moving laterally, and often with a degree of calcaneal eversion. In fact, from video analysis of professional male soccer players worldwide, 60% of ACL injuries occurred with some degree of calcaneal eversion [41] and 94% of ACL injuries that occurred in female Australian Football players did so by having their leg in an antero-lateral position with a degree of knee valgus [42]. Like the medial data in Fig. 6, the lateral data, shown in Fig. 7, did not support the hypothesis that an increase in eversion angle would lead to a decrease in lateral TC. Again, this can be explained with the asymmetric shape of the individual studs, shown in Fig. 9, whereby an increase in the eversion angle, thus raising the lateral studs beyond ground contact, isolates the medial studs that are more resistant to motion for that direction. Similarly, to the medial movement, a peak occurs at approximately 5 to 7.5° of eversion before the TC declines as the boot now begins to slide over the playing surface, rather than engage in it as the medial studs are unable to effectively penetrate the surface as this higher calcaneal eversion angle. This phenomenon of the boot sliding over the surface was first proposed by Heidt et al. [43] and termed ‘crow-hopping’. Given that female soccer players tend to land from a vertical jump with 6.67° of eversion on average [44], this landing, coupled with the peak TC experienced for this angle of eversion, could lead to potential lower limb injury as a player attempts to accelerate post landing.

As demonstrated by Loud et al. [27], natural grass undergoes temporal changes when performing traction testing. To mitigate the impact this had on the results, most of this investigation took place on an artificial grass playing surface. The reporting procedure suggested by Loud et al. [27] was also used for all loading conditions whereby the initial trial was removed from the statistical analysis, which removed any ‘new-ground’ bias and produced a more uniform result.

While this present study was able to observe the impact of ankle position on translational traction, there were still some limitations present. The present research was also limited in terms of the normal load applied to the apparatus and the translational speed of the boot. A normal load of 148N was applied to the footform, a load significantly lower than normal loads experienced by athletes during movement. Despite the lower load, full stud penetration for the boot was achieved at the neutral foot position on both natural and artificial grass. Normal load and traction has been discussed before with researchers identifying a linear relationship between TC and normal load, with no change in the relation of the test variables, often outsole configuration, at each normal load tested [10, 18, 45,46,47]. While the applied speed of 4 m/min is below typical running speeds of players in a match, the speed is similar to those used by Cooper [48]. While the sampling speed of 150 Hz is low relative to the 2000 Hz used by Wannop and Stefanyshyn [18] and the 500 Hz used by Shorten et al. [49] and Thomson et al. [12], it is higher than the 50 Hz used by Keshvari et al. [50] and comparable to the 250 Hz used by Kuhlman et al. [51] and Cooper [48] when considering the number of trials and the trial length. The value of 150 Hz for this research was chosen such that the data could be compared to biomechanical analysis undertaken by Loud et al. [52] to determine key traction values at specific time points for athlete testing. To control the number of variables, the size and style of the boot was limited in this experiment. While this allowed research to be focussed on the ankle position, the results produced are specific to the boot and size tested, thus further analysis with different shapes and sizes of boots is needed, particularly those designed for female players. A limitation for this investigation was the inability of the apparatus to test beyond 30° of plantarflexion and dorsiflexion as well as 15° of calcaneal inversion and eversion. While inversion and eversion angles at IC that exceed 15° are uncommon amongst players, angles in the sagittal plane, particularly plantarflexion angles between 30° and 60° have been observed in some (about 5%) ACL injury cases in NFL players [31]. However, as already mentioned, there is a tendency amongst soccer players who rupture their ACL to do so in a dorsiflexed position, where angles rarely exceed 30°.

5 Conclusion

Due to the link between traction and potential ACL injury, the tractional properties of soccer boots have been widely researched, however, most of the studies adopt a neutral foot position. The purpose of this investigation was to examine the impact of ankle angle on the translational TC for various movements. The results highlighted the role of stud profile and orientation with regards to traction coefficient. For anterior movements, the orientation of the rear studs had a significant impact on the degree of traction. At higher dorsiflexion angles, these two studs forced playing surface material towards the centreline of the boot, allowing it to compact and form a barrier. For medial and lateral movements, the asymmetric nature of the ‘circular’ forefoot studs as well as the orientation of the rear studs, caused an increase in traction with an increase in inversion/eversion angle. From these findings, players can understand what levels of traction their boots provide at different angles in the ankle joint. Players can be encouraged to land in positions that minimise the TC compared to a neutral angle of 0° flexion to increase the likelihood of lower-limb alignment and a smooth transition between the IC and potential IF.