The tests performed used a setup for investigating injury from sand, utilising a gas-gun system modified to deliver sandy gravel aggregate at high velocities. The experimental design and procedures were carried out in compliance with the Human Tissue Act 2004. Ethical approval was granted from the local regional ethics committee at the Imperial College Healthcare Tissue Bank (ethical approval number: 17/WA/0161). Experiments were carried out using twelve male human cadaveric thigh samples with no prior relevant injury or pathology (median age 38 years, range 36–51 years). Samples were fresh frozen at − 20 °C and thawed at room temperature (21 ± 2 °C) for 24 h prior to testing.
Sand size and properties were chosen based upon NATO unclassified AEP-55 recommendations for typical sandy gravel soil granulometry . A sandy gravel aggregate size range was subsequently chosen to fall as closely as possible to the median value (2 mm, range 0.09–38 mm) of the ideally distributed particle sizes . This consisted of sandy gravel of which 100% passed a 1–2-mm sieve, with any sand subsequently passing a 1-mm sieve removed (the sand size utilised in experiments therefore ranged from 1 to 2 mm). The sand was housed within a 11-g hollow polycarbonate sabot weighed prior to, and following, loading it with sand.
The sabot-sand unit was subsequently loaded into the firing chamber of a double-reservoir gas-gun system. This system utilised a 2-L reservoir charged with air or helium and a Mylar® diaphragm firing mechanism to accelerate the sabot-sand unit down a 3-m-long, 32-mm-bore barrel . To accelerate the sabot-sand unit to the desired velocity, the reservoir section of the gas gun was charged to a predetermined firing pressure prior to release. After release, the sabot-sand unit accelerates down the barrel to exit into the target chamber, where the sabot is separated from the sand by a stainless-steel sabot-stripper. The sabot is halted at this point, whilst the sand continues to travel towards the cadaveric sample at the intended terminal velocity. In order to simulate the distribution and spread of sand ejecta as occurs following blast, two interconnecting fenestrated steel fences, separated by 10 mm and offset to one another by 50% of the diameter of each fenestration, were placed distal to the gas-gun outlet and proximal to the mount (Fig. 1a). Offsetting of the fenestrated steel fences changed the initial stream of sand delivered by the gas gun into multiple streams of differing trajectories which subsequently dispersed into a widely distributed spread of sand (Fig. 1b); this setup achieved blast propagation in three dimensions and acceleration of debris and soil ejecta in all directions, which can be considered a realistic simulation of the event.
The speed of the sand particles at the point of impact with the sample was estimated using high-speed photography (Phantom VEO710L, AMETEK, USA) at 68,000 fps. An average velocity for the sand cloud as a whole was determined based upon identifying and tracking four unique points spread across the distributed sand. These points varied in velocity and were chosen from the front, front-centre, centre, and centre-back of the peripheries of the sand spread (Fig. 1c). Cadaveric samples were divided into one of two groups: either wearing (1) UK Military Tier 1 pelvic protection  (knitted silk of 490 g/m2 areal density) and standard-issue combat trousers or (2) standard-issue combat trousers only (control group). For each individual test, a cadaveric thigh was secured in position within the target chamber. The thigh was placed in a neutral resting position with an abduction angle of 30° from the midline (Fig. 2). The Tier 1 pelvic protection was worn as a whole on the cadaveric thigh, with the sand blast directed to impact with the two layers of high-performance knitted silk protection (Fig. 3).
Following impact with the sand, samples were removed from the target chamber and taken for subsequent photography and dissection. A separate photograph was taken of each individual injury, with adjacent ruler, and the film plane of the camera parallel to the injury to avoid parallax errors. Recorded injury patterns included (1) number of injuries sustained, (2) surface area of injuries sustained (surface area per injury and total injured surface area), and (3) maximal anatomical depth of injury sustained (superficial/subcutaneous only, or deep to the subcutaneous tissues/subfascial).
Image Processing and Statistical Analysis
Photographed images were subsequently assessed with image processing software to calculate the surface area of injuries sustained. ImageJ was used for image processing calculations (National Institutes of Health, USA). Image scale was set, followed by tracing the outer edges of the zone of injury for each individual injury sustained to calculate the total surface area.
IBM SPSS was used for statistical analysis (version 26, IBM, USA). The Mann–Whitney test was used to assess significant differences in non-parametric data between groups, including number of injuries sustained and surface area of injuries. Cross-tabulation with Pearson χ2 test was used to assess significant differences in categorical variables between groups, including depth of penetration (subcutaneous only vs. deep (subfascial)).