You are on page 1of 29

Composite Structures 263 (2021) 113684

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Review

Composite structures subjected to underwater explosive loadings:


A comprehensive review
Phuong Tran a,⇑, Changlang Wu a, Michael Saleh b, Luiz Bortolan Neto b, H. Nguyen-Xuan c,d,⇑, A.J.M. Ferreira e
a
Department of Civil & Infrastructure Engineering, RMIT University, Melbourne, VIC 3000, Australia
b
Australian Nuclear Science & Technology Organisation, Lucas Heights, NSW 2234, Australia
c
CIRTech Institute, Ho Chi Minh City University of Technology (HUTECH), Ho Chi Minh City 700000, Viet Nam
d
Department of Architectural Engineering, Sejong University, 209 Neungdong-ro, Gwangjin-gu, Seoul 05006, Republic of Korea
e
Departamento de Engenharia Mecanica, Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, Porto 4200-465, Portugal

A R T I C L E I N F O A B S T R A C T

Keywords: Composites materials and structures are increasingly used to replace conventional materials in civilian and
Underwater explosions defence‐related maritime transportation and infrastructure such as naval vessels, submarines, civilian ships,
Composite damage and oil platforms for its better performance‐to‐weight ratio and electro‐magnetic signature control.
Numerical modelling However, when subjected to under water explosions (UNDEX), navel composite structures experience highly
Post‐mortem analysis
nonlinear deformations and damages. Such transient deformation phenomena of composites and associated
multiscale damages have been a subject of research for many years. This review aims to provide historical
and methodological overviews of significant research and contributions in this area over the last 20 years from
experimental programs, modelling approaches, post‐mortem analysis techniques, analytical approximation and
recently emerging area of data‐led predictive simulations. UNDEX event is often described by a series of events
including (a) the formation of the arriving shock wave, (b) the attenuation of the initial shock wave, (c) devel-
opment of cavitation due to the reflected tension wave from free surface or the structural obstacles, (d) fluid‐
structure interaction‐induced deformation and associated (e) cavitation coalescence and collapse. Such inter-
connected dynamic events and their influences on the behaviours of composite structures are subjected to
extensive research and therefore summarised in this review work to highlight state‐of‐the‐art field and
laboratory‐scaled experimental programs including investigations on low temperature and cavitation’s influ-
ences. Furthermore, the ongoing increase in the computing power and the development of advanced numerical
methods have made it possible for multiscale and multi‐physics simulations capturing the complex fluid
dynamics associated with UNDEX. Over ten different modelling approaches, hydrocodes and their hybrid com-
bination are summarised and discussed for potential applications. Review on current computational
approaches also reveals the shortcomings of predictive modelling due to unavoidable simplifications, empirical
assumptions on limited experimental data. Therefore, this work also provides a brief discussion on how data‐
led modelling approach such as artificial neural networks or deep learning, which is based largely on experi-
mental data, could provide powerful assistance to analytical and deterministic numerical analysis.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Review on experimental programs for UNDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1. Full-scale and large-scale experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2. Laboratory-scale experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.1. Water-filled shock tube . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2. Water-filled chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.3. Spherical pressure vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

⇑ Corresponding authors at: CIRTech Institute, Ho Chi Minh City University of Technology (HUTECH), Ho Chi Minh City 700000, Viet Nam (H. Nguyen-Xuan); Department of Civil &
Infrastructure Engineering, RMIT University, Melbourne, VIC 3000, Australia (Phuong Tran).

https://doi.org/10.1016/j.compstruct.2021.113684
Received 16 August 2020; Revised 21 January 2021; Accepted 1 February 2021
Available online 11 February 2021
0263-8223/© 2021 Elsevier Ltd. All rights reserved.
P. Tran et al. Composite Structures 263 (2021) 113684

2.2.4. Composites subject to extreme loading at low temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10


3. Post-mortem analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1. Non-destructive methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2. Destructive methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4. Numerical simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1. Finite Element Method (FEM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2. Coupled methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.3. Data-led modelling technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Declaration of Competing Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

1. Introduction sumption. In naval vessels, non‐magnetic properties of composite


structures allow their use against maritime mines, and low radar
Nowadays, composite materials have been widely used in the mar- cross‐section increases the stealth of warship [3,9].
ine industry, such as important components and machinery of naval Application of composites in marine structures could date back to
vessels, civilian ships, and oil platforms [1]. The increasing popularity the 1940s, during which time period U.S. Navy employed composite
of composite attributes to its advantages over traditional metallic materials in small personnel boats. Afterwards, there started to be
materials, including relatively low density, excellent marine environ- small leisure crafts and offshore components constructed using com-
ment resistance, good fatigue resistance, ease of transportation, non‐ posites [3,15]. Since the mid‐1980s, maritime composites have been
magnetic, high strength to weight and stiffness to weight ratios [2–9]. considerably used for naval applications [9]. Table 1 summarises com-
Conventionally, steel, aluminium and wood were used for ship con- posite applications in marine industry after World War I. Given the
struction [10–14]. When exposed to the marine environment, issues role of naval ships in war, they are supposed to resist high‐intensity
involving these traditional materials could be overcome by substitut- dynamic load [16], including contact/non‐contact underwater explo-
ing with composites. For example, weight has always been a primary sions (UNDEX) from torpedos, bombs, mines, and high‐velocity impact
concern when designing marine structures using steel. Lighter struc- from storm, water slamming, and collisions.
tural weight of surface ships and submarines could increase payload Table 2 summarises typical incidents that happened to different
carrying capacity, while the reduced topside weight of offshore maritime structures subject to different types of extreme loadings in
platforms could reduce weight requirement for supporting structures the marine environment.
and installation fee [7]. Composite decks could generate a weight sav- As summarised in Table 2, explosive and high velocity impact from
ing of 25% than steel, thereby translating a reduction of 4.5% fuel con- military attacks, such as torpedo, bomb, gunfire, collision, and other

Table 1
Composites application in marine structures.

Applications Time Composite Representative Comments Refs.


materials/ structures
structures

Naval vessels Piping for destroyers 1940 s FRP Higgins Boat Up to 15 m length and 20 tonnes in displacement due to [17–21]
Minesweeper – early poor fabrication quality and low stiffness of hulls
Landing crafts 1950
Patrol boats Since GFRP KNM Skjold (Royal Rarely longer than 20 m due to low hull girder stiffness; [9,22–24]
early Norwegian Navy) Not applicable to offshore patrol operation
1960 s Smyge MPC2000
(Swedish Navy)
Hovercrafts Since FRP Landing Craft Air Lightweight; Higher cost than aluminium alloy; Improved [25–27]
1970 s Cushion 100 (U.S. operation and maintenance
Navy)
Mine-countermeasure Since GFRP HMS Wilton (Royal Mostly over 50 m with displacement at exceeding 600 [28–32]
vessels (MCMV) 1973 Navy) tonnes; Local buckling and underwater shock resistance
Corvettes Since FRP Visby corvette (Royal Longest naval vessels type; [33–35]
1990 s Swedish Navy) Higher cost than built with steel
Submarines Fairwater Since FRP Collin’s class Lightweight; Lower cost due to moulding into complex [17-
Sonar dome 1940 s submarines (Royal shapes without machining; Corrosion resistance 19,22,7,36,37]
Masts Rudders Australian Navy)
Casings
Offshore Low-pressure pipes Since GFRP/ Davy and Bessemer Corrosion resistance; Lightweight; Ease of transport; High [38–44]
Diesel storage tanks 1970 s CFRP monopod platforms specific strength/stiffness; Good thermal insulation;
Lube tanks (United Kingdom) Excellent damping & fatigue performance
Utility tanks FRP Fire resistance [45,46]
High-performance tubes (phenolic)
Cable ladders and trays
Risers
Civilian and Fishing trawlers Since FRP Low cost [47–50]
commercial Pilot boat the (especially
crafts Submersibles 1950 s GFRP)
Fast ferries

Note: FRP refers to fibre reinforced plastics. GFRP refers to glass fibre reinforced polymer. CFRP refers to carbon fibre reinforced polymer.

2
Table 2

P. Tran et al.
Typical marine incidents due to extreme loadings.

Time Event/ Structure name Use purpose Built year, builder Units Cause Location Damage to vessel Casualty Loss
Historical involved
Fatal Nonfatal
event

Sep 1914 World War I HMS Aboukir, Armoured 1898, Fairfield Royal Navy Torpedoes from U- North Sea Sinking 1459 – –
HMS Hogue, cruiser Shipbuilding 9
HMS Cressy
Sep 1915 SM U-28 U-boat 1912,Kaiserliche Werft Imperial Torpedo induced North Cape, Norway Sinking 39 0 –
German explosion
Navy
Dec 1941 World War II HMS Prince of Battleship 1937, Cammell Laird Royal Navy Land-based South China Sea Sinking 840 –
Wales and Company bombers and
HMS Repulse Battlecruiser 1915, John Brown & torpedo bombers
Company, of the Imperial
Japanese Navy
USS California Battleship 1916, Mare Island Naval U.S. Navy A pair of Pearl Harbour Ship was moderately damaged, 98 61 –
(BB-44) Shipyard torpedoes and a and electrical system was disabled.
bomb
Oct 1944 USS Albert W. Destroyer 1942, Charleston Navy Gunfire Leyte Gulf, Philippines Ship suffered 22 hits, many by six- 38 104 –
Grant Yard inch shells. Fires broke out, and
the ship lost steering control and
all power.
May 1945 USS Stormes Destroyer 1944, Todd Pacific Hit by a Japanese Yomitan, Okinawa Ship was on fire, and sea water 21 15 –
(DD 780) Shipyards kamikaze aircraft poured through holes in the hull.
May 1958 Antisubmarine USS Stickleback Submarine 1944, Mare Island Naval Collision with a U. Hawaiian Islands Sinking and lost. It was found in 0 –
warfare (SS-415) Shipyard S. destroyer escort 2020.
exercise
Aug 1964 Gulf of Tonkin North P4 torpedo – Democratic Fired by a Gulf of Tonkin Three boats were damaged. 4 6 –
3

incident Vietnamese boats Republic destroyer and


torpedo boats of Vietnam aircraft carrier
(T-333, T-336,
T-339)
May 1987 Iran-Iraq War USS Stark (FFG- Frigate 1979, Todd Pacific U.S. Navy Two Iraqi Exocet Off Saudi Arabia, Hull and superstructures were 37 21 $142
31) Shipyards missile Persian Gulf destroyed million
July 1988 – Piper Alpha Oil production 1973 Occidental Oil rig explosion North Sea Platform was completely 167 Unmentioned £1.7
platform Petroleum and resulting oil destroyed. billion
(Caledonia) and gas fires
Limited
Aug 2000 Naval exercise K-141 Kursk Nuclear- 1990, Soviet Russian Explosions of A large hole in the hull and 118 0 –
powered Navy military shipyards Navy hydrogen sinking.
submarine peroxide-fuelled
torpedo
Nov 2001 – Nego Kim Bulk carrier 1985, Japan Saratoga Explosion of Dampier, Australia Tank was ruptured 8 Unmentioned –
Shipping topside water
Ltd ballast tank

Composite Structures 263 (2021) 113684


Jan 2007 – MSC Napoli Container ship 1991, South Korea Metvale Storm (bow English Channel A crack in one side and a flooded 0 0 £120
Limited slamming and hull engine room million
whipping)
Apr 2010 Deepwater Deepwater Semi- 2001, South Korea Transocean Explosion and Macondo Prospect oil Sinking 11 17 $560
Horizon oil Horizon submersible subsequent fire field, U.S. million
spill mobile
offshore
drilling rig
July 2016 – HMS Ambush Nuclear 2003, BAE Systems Royal Navy Collision with a Strait of Gibraltar Conning tower and sonar 0 0 £2.1
fleet submarine Submarine Solutions merchant ship equipment were damaged. million

(continued on next page)


P. Tran et al. Composite Structures 263 (2021) 113684

accidents, including oil rigs explosion, storm, and bow slamming

million
would lead to severe damages to marine structures, machines
Loss

$27


(Fig. 1) and massive casualty. Therefore, it is crucial to design for
explosion and high‐velocity impact resistance of marine structures.
Composites are brittle [51], which leads to little energy absorption
Nonfatal

and sudden structural failure subject to shock loading [52,53]. More-


over, blast load causes nonlinear structural deformation and multi‐
3

1
Casualty

faceted failure mechanisms in composites, which are more complex


Fatal

than steel and difficult to predict [54]. Damage modes of composites

27
2

0
include matrix cracking, delamination, fibre fracture, skin‐core

Tourist boat sank at the bottom of


debonding, and core crushing, which occur at a much lower shock
side superstructure and starboard
Severe damage to both starboard

Vessel lost control of engine and


steering, sinking after few days.
wave pressure than damaging high‐strength steel panels [3,268].
Minor damage to port bow

Therefore, the response of maritime composite structures against


Helicopter was damaged.

side below the waterline

shock waves generated by an underwater explosion (UNDEX) needs


the Danube River. to be better understood.
Damage to vessel

and bulbous bow

In addition to blast‐resistant issue, the use of maritime composites


has also been hindered by the lack of proper design methods [55]. So
far, methods for the design of composite materials against blast, shock
wave, and collision situations have not been well developed, and no
standard has been established [9]. Many naval developers employ
the same techniques, which are adopted for metal designs, to design
composites, thereby reducing their performance [56]. As a conse-
North Sea (north of

Danube, Budapest,

quence, the International Maritime Organization does not practically


Bergen) Norway

allow the use of composite materials for large commercial vessels in


its Safety of Life at Sea regulations [15,57]. Thus, it is of great neces-
Hungary
Location

Red Sea

sity to develop rigorous standards and smart tools for composite design
Japan

subject to UNDEX [9].


Early before World War I, experiments have been conducted to
explosives at high

investigate composite response to UNDEX. Between World War I and


Collision with a
tanker Sola TS
Collision with

World War II, researchers started to develop theoretical frameworks


Strike from a

packed with
small vessel

cruise ship

and come up with optimal design charts for maritime composites.


Collison

These experimental and theoretical research works contributed to


Cause

speed

improved features and new designs for military ships, such as the Mid-
way class in the United States and the Yamato class in Japan. Mean-
Royal Saudi

while, some studies focused on the understanding of underwater


Norwegian

Panorama
U.S. Navy
involved

explosion event and its associated phenomena, such as theoretical


Royal
Units

Navy

Navy

Deck

studies of gas bubble hydrodynamics and pressure field [58–61]. Dur-


ing World War II, more powerful weapons appeared, such as non‐
1949, the former Soviet

contact mines and air‐attacks. The wide‐spread damage, particularly


1993, Bath Iron Works

2006, Navantia, Spain

to ship equipment and machinery, motivated researchers to enhance


2008, South Korea
Built year, builder

1981, Arsenal de

their understanding of the effects associated with UNDEX. The major


Lorient, French

naval powers, such as Great Britain, USA, Germany, and Japan,


increased their research efforts to improve the protection of their war-
Union

ships by not only conducting experiments, establishing theoretical


framework, but also performing numerical simulations. After World
War II, further investigations regarding nuclear‐weapon explosion
Container ship

[62] were conducted, which demonstrated the importance of under-


Use purpose

Tourist boat

standing UNDEX effect and structure damage mechanisms. Since the


Destroyer
Frigate

Frigate

1970s, an increasing number of papers and reports on experiments,


numerical simulations, and theoretical analysis have been published.
In order to gain insights into composite response subject to UNDEX
MV ACX Crystal
Structure name

Ingstad (F313)
USS Fitzgerald

and provide useful information to future studies, this paper presents a


HNoMS Helge
Al Madinah

historical review on the response of composites under such conditions.


Hableány

Different methods have been explored and developed over the past few
decades, providing design suggestions for warships. Four different
research approaches including experiments, post‐mortem inspections,
numerical modelling, and theoretical analysis are reviewed showing
USS Fitzgerald
Yemen's Civil

and MV ACX

how researchers have investigated the complex deformation and fail-


Sinking of
Historical

Hableány
Collision

ure behaviours of composite structures subjected to UNDEX.


Crystal
Event/

event

War


Table 2 (continued)

2. Review on experimental programs for UNDEX


May 2019

Early before World War I, experimental methods had been per-


Nov 2018
Jun 2017
Jan 2017

formed from full‐scale and large‐scale explosive testings to


Time

laboratory‐controlled scale. Most full‐scale and large‐scale experi-

4
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 1. (a) ‘USS STARK (FFG-31)’ listing to port after being hit by two Iraqi Exocet missiles within 30 s; (b) Damage to hull and superstructure of ‘USS STARK (FFG-
31)’; (c) ‘USS Stickleback (SS-415)’ collided with ‘USS Silverstein’; (d) ‘USS Fitzgerald’ after collision.

ments were conducted by defence organizations [63], while some Qiankun and Gangyi [109], Arora et al. [110,111], Reid et al. [65],
were published in papers and reports [53,64,65]. These experiments and Rolfe et al. [66] are among the researchers who reported the
are precise to assess shock resistance of maritime structures as they results of full‐scale and large‐scale underwater blast experiments in
are close to real situation. However, they are costly and, in order to the literature. To study the UNDEX process, a range of explosives
be conducted, require many precautions and safety measures. There- and analogues are used, including TNT [132], PE4 [133], C4
fore, laboratory‐scale experiments were more commonly used at [134–137], HBX‐1 [138–140].
research centres and universities [66]. Laboratory‐scale testings simu- In the 1980s, Royal Australian Navy (RAN) conducted a three‐
lated UNDEX using various apparatus, including water‐filled shock phase underwater shock testing to evaluate the GFRP/foam sandwich
tubes [67–95], water‐filled chambers [96–103], spherical pressure composite hull of the newly proposed minehunter [107] in an approx-
vessels [99–103] and cubic tank [98]. Table 3 summarises experimen- imate ten‐year time period. Hall [53,106] reported the results of a full‐
tal methods that have been employed in previous studies. scale experiment that had been conducted as one phase of the series of
The physical phenomena associated with UNDEX is extremely com- experiments. A full‐scale midship section of the hull was submerged in
plex and involves many aspects including high temperatures and pres- a water‐filled quarry site and tested subject to explosion from certain
sures, transient states, extensive distortions, and multi‐medium flow distance. High‐speed instrumentation techniques were employed dur-
[104] The resulting loading on a submerged structure subjected to ing the experiment to measure both the shock wave magnitude and
an UNDEX is time‐dependent and can be summarised into two distinct structural response, including acceleration, velocity and strain at pre-
phases: determined structural locations. Results demonstrated that GFPR/
PVC foam sandwich composite hull could satisfy the requirements of
1) The initial transient shockwave pressure load and associated RAN, which was then supported by the third phase trials on another
cavitation reloading. minehunter, HMAS Shoalwater, in 1987 [108]. In 1999, the second
2) Subsequent shockwaves and fluid impulse due to bubble pulsa- vessel of the Huon class Minehunter of RAN, HMAS Hawkesbury,
tion and jetting. Cavitation effects may also be associated with was also tested subject to UNDEX [65]. Details are not reported due
each of these pulses. to sensitive information.
In 2011, Qiankun and Gangyi [109] conducted a full‐scale under-
As illustrated in Fig. 2 the pressure experienced by a submerged water shock experiment on a ship section. A spherical charge, which
structure fluctuates periodically as the gas bubble migrates to the free was made of 6 kg TNT, was placed at the depth of 2.5 m and stand‐
surface with initial pressure peak due to the transient shockwave sus- off distance of 15.12 m (Fig. 3a). The acceleration time histories were
tained over a short period, followed by pulsating bubble pressure measured by five accelerometers. Fig. 3b shows one of the acceleration
peaks as bubble migrates to surface. histories, which was recorded by the accelerometer mounted at the
yellow point in Fig. 3a. This shock trial was aimed to compare with
2.1. Full-scale and large-scale experiments and validate a numerical simulation method. Agreements were found
between experimental results and numerical predictions of
Full‐scale and large‐scale UNDEX testings are costly, and results are acceleration.
difficult to measure. However, it is necessary to conduct these studies Later, Arora et al. [110,111] developed two different setups to con-
to validate smaller (less expensive and more controllable) experimen- duct large‐scale underwater blast testings on glass‐fibre reinforced
tal setups, numerical simulations and analytical models. Hall [53], polymer (GFRP) skinned composite sandwich panels and GFRP tubular

5
P. Tran et al.
Table 3
Summary of experimental methods.

Different experimental setup Composite materials/structures Loading simulation Instrument and measurements Refs.

Full/Large scale GFRP sandwich panels with different foam core materials; An explosive charge with predetermined High-speed instrumentation to determine shockwave [53,106–108]
A full-scale midsection of hull made of GFRP/foam sandwich characteristics magnitude, structural acceleration, velocity and strain.
composites.
A full-scale ship section with 2.5 m length, 1.2 m width, and A spherical charge made of 6 kg TNT with 15.12 m Accelerometers to gain acceleration profiles for [109]
1.2 m height stand-off distance predetermined points on tested section.
GFRP and CFRP sandwich composite 1 kg spherical charges made of Plastic Explosive type Strain gauges to study failure mechanisms. Pressure gauges to [110]
4 (PE4) at 1 m stand-off measure the shockwave pressure.
GFRP/SAN foam sandwich panel 1 kg/0.5 kg C4 explosive charge at 6 m below water Strain gauges and high-speed camera (with DIC) to monitor [111]
GFRP tubular laminate surface at different stand-off surface strain
A GFRP composite-hulled minehunter and GFRP composite T- A series of six explosive charges at various stand-off Velocity meters, accelerometers to obtain velocity time [65]
joint history.
GFRP/SAN foam sandwich panel 1.28 kg TNT equivalent charge at stand-off distance of Pressure gauges to measure reflected and side-on pressures. [66]
CFRP/SAN foam sandwich panel 1m Strain gauges and high-speed camera to record response
Lab scale Water- Monolithic CFRP plates; GFRP/PVC foam sandwich Generate peak pressures between 15 and 200 MPa High-speed photographic and laser interferometric [75-84]
filled composites; Aluminium/PVC foam sandwich plates diagnostics to study damage and failure mechanisms.
shock GFRP curved composite panels Conical shock tube (CST) facility, peak shock A 3D DIC system to record transient response of panels [68-72]
tube pressures from 10.3 MPa to 20.1 MPa
Compound steel-foam sticker on piston in contact High-speed photography to measure cavitation, deflection [85–90]
GFRP and CFRP plates with pressurised water, generating pressures up to time history and breaking/closing fronts propagation
20 MPa
Stainless steel monolithic plate; GFRP/PVC foam sandwich A flyer plate impacted piston at the velocity of 300 m/ Shadow Moiré measurement combined with high-speed [91–95]
composite; Martensitic and austenitic steel alloys; stainless s to produce an exponentially decaying pressure with photography to measure impact velocity, pressure wave
steel alloy with honeycomb core peak value above 100 MPa history, deformation and fractures
Water- GFRP laminates 5.8 g – 300 g PE4 at 0.3 m – 1.3 m stand-off distance, Norland Prowler digital waveform analyser and pressure [96,97,112–114]
filled generating maximum pressures between 8 MPa and transducers to record and measure shock wave respectively
chambers 133 MPa
6

Spherical CFRP composite hollow cylinders, CFRP composite panels Nitrogen gas into air pocket to increase pressure High-speed camera and pressure sensors to record pressure [99–103,115]
pressure after prolonged exposure to saline inside vessel at 0.083 MPa/min; RP-85 charges, time history. 3D DIC to capture real-time full-field
vessel equivalent to 1.778 g TNT displacements and velocities.
Cubic GFRP composite plates with polyurea coating An RP-503 charge, including 454 mg RDX and Pressure transducers to measure pressure field; High-speed [98,116]
tank 167 mg PETN, equivalent to 1.5 g TNT photography coupled with DIC to capture response of plate
back-face.
Extreme Drop Short carbon fibre reinforced PEEK composite; CFRP Drop weight tower generated different level of kinetic The cell on impactor to calculate its displacement time [117-124]
loading tests weight laminates/tubes; woven Kevlar/epoxy laminates; GFRP energy to impact composites at temperatures lowest history. Post-mortem analysis using SEM, ultrasonic pulse-
at low test laminates; CFRP/PVC foam sandwich composite to −196 °C echo to examine the damages in specimen
temperature Charpy 3D multi-axial warp knitted GFRP composites; bulk The pendulum hammer impacted specimens at Post-mortem analysis using SEM, ultrasonic pulse-echo to [125–127]
impact metallic glass matrix composites; GFRP laminates different energy level at temperatures lowest to examine the damages in specimen
test −196 °C
SHPB GFRP/foam sandwich composites; 3D braided basalt/epoxy The incident bar increased the strain rate, while the Dynamic stress-strain curves were obtained for foam core and [128–130]
test composite transmission GFRP face sheets respectively.
Bar increased the transmitted signal intensity.
Ballistic Taped and woven CFRP laminates The projectile impacted specimen at various velocities Two photoelectric cells to measure the impact velocity. C- [131]
test up to 525 m/s. Temperature ranged from −150 °C to scan images to quantify the damage.

Composite Structures 263 (2021) 113684


−20 °C
Blast test GFRP/foam sandwich composites Shock tube apparatus generated an incident pressure Pressure transducers to measure the shock wave pressure. [129]
of 0.9MPa at −40 °C (core). High-speed camera (with DIC) to record the real time
displacement of specimen.

Note: DIC refers to Digital Image Correlation; GFRP refers to glass fibre reinforced polymer; CFRP refers to carbon fibre reinforced polymer. PEEK refers to polyether-ether-ketone. SEM refers to scanning electron
microscopy; SHPB refers to Split Hopkinson Pressure Bar.
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 2. Dynamics of UNDEX event illustrating surface observations, bubble migration and pressure history [105].

Fig. 3. (a) Schematics of full-scale experiment (b) acceleration time history obtained from the accelerometer (yellow dot in (a)) [109].

laminates. The first setup is shown in Fig. 4a [111]. Target panels were that there are also research using laboratory‐scale air blast to examine
bonded in a steel frame and bolted in a base frame such that the effect the performance of maritime composite structures [267].
of the blast wave wrapping around the target could be reduced. This
configuration provided an enclosed volume in which air‐backed or 2.2.1. Water-filled shock tube
water‐backed conditions can be reproduced. It was a heavy structure Water‐filled shock tubes were developed with small variations for
that replicated the boundary conditions in the hull of a ship. Strain different scopes. For example, LeBlanc and Shukla developed a shock
gauges were mounted along centre lines of the panel to measure the tube in which the initial shock wave was generated by a small explo-
strain. Fig. 4b [111] is an image taken during the explosion, showing sive charge [68–72,74], while other researchers used gas guns to gen-
bubble venting 3.5 m height to the atmosphere. erate the initial impulse [75–95]. Some of these configurations
Another full‐scale experiment was conducted by Rolfe et al. [66] in allowed for different situations, such as air‐backed or water‐backed
2017. They performed a failure analysis of composite sandwich panels conditions [75–90].
with different skin materials subject to both underwater and air explo- In 2014, LeBlanc and Shukla [69] utilised a conical shock tube to
sions. GFRP and CFRP were tested respectively. Detail setup and pho- generate impulsive loadings to examine the response of air‐backed
tos during the experiment are shown in Fig. 5. and fully coupled panels. The major apparatus was a conical water‐
filled shock that was horizontally mounted as shown in Fig. 6. The
2.2. Laboratory-scale experiment rigid walls of the tube were used to confine the expansion of the pres-
sure to simulate a conical sector of the pressure field. Coombs and
For laboratory‐scale experiments, the setups that have been used to Thornhill [141] demonstrated that the free field pressure wave distri-
simulate UNDEX mainly include water‐filled shock tube, water‐filled bution could be replicated if the walls of the tube are rigid enough. An
chamber, spherical pressure vessel, and cubic tank. It should be noted amplification factor was applied to relate shock tube pressure to free

7
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 4. (a) Schematic diagrams of a large-scale underwater blast experiment setup to evaluate water blast resistance of GFRP composites; (b) photo of water
surface during the experiment, which threw water into air [111].

field pressure, which was defined as the ratio of the volume of a spher- during the loading phase. This system also included pressure transduc-
ical charge and the conical charge as depicted by [142]: ers and a high‐frequency data acquisition system to record the trans-
1 mitted impulses.
AF ¼ ; ð2Þ Variants of USLS are shown in Fig. 8, which were also developed by
sin2 4α
Avachat and Zhou [78,83]. Fig. 8a presents a setup called ‘Dynacomp’
where AF denotes amplification factor, and α denotes the cone angle. [78]. The sandwich structures were comprised of two aluminium pla-
For this setup, α was 2.6°. The explosive charge was detonated at the tens and foam cores with different configurations. The composite
left end of tube, which generated peak shock pressures from structure was exposed to water on one side, while the aluminium pla-
10.3 MPa to 20.1 MPa. In order to capture the transient response, ten on the other side rested on a force transducer. The transmitted
two different measurement methods were applied. Strain gauges were force was measured by force transducer, and the compressive strain
positioned on the back face of composites to record strains during of foam core was recorded by a high‐speed camera (Fig. 8c and
experiments, and Digital Image Correlation (DIC) technique was used Fig. 8d). It should be noted that only air‐backed conditions can be sim-
to record and analyse the displacement of face sheet. ulated using this setup. Fig. 8b shows another variant of USLS, which
Another laboratory‐scale apparatus is called Underwater Shock was used to evaluate UNDEX resistance of axis‐symmetrically clamped
Loading Simulator (USLS), which was developed by Avachat and Zhou round plates [83]. This fully clamped circular plate was obtained by
[75–84]. It was a system that could generate impulsive loadings of dif- blocking the circular specimen with adhesive bonding between the
ferent intensities and simulate submerged structures subjected to specimen and the annular tube face near one of the two tube sides.
underwater blast loading. This facility was composed of a gas gun, a The other side was then filled with water and sealed with a piston.
shock tube filled with water and a containment chamber, in which The back face of the composite was supported by a clamping ring.
the sandwich structure was positioned. Gas gun accelerated a projec- Slightly different, Espinosa and his research group [91–95] devel-
tile, which impacted a thin flyer plate and generated a planar pressure oped an experimental setup based on scaled fluid structure interaction
wave. The sandwich structures can be either air‐backed or water‐ (FSI) to simulate the shock wave effects on a submerged structure. This
backed (Fig. 7). Moreover, high‐speed photographic and laser interfer- facility is similar to the USLS setup established by Avachat and Zhou
ometric techniques were part of this system, which allowed for the [75–82,84] (Fig. 7). The difference is that the water‐filled anvil tube
measurement of velocities and deformations of projectile and panels was shaped with a divergent configuration. This diffusion was closed

Fig. 5. (a) Photographs during the full-scale UNDEX experiment on composite panels; (b) arrangement for composite panel and explosive charge [66].

8
P. Tran et al. Composite Structures 263 (2021) 113684

water‐backed freestanding plates (Fig. 10b) and fully clamped plates


(Fig. 10c). For all three setups, pressure was generated by the impact
of a compound projectile and measured by piezoelectric pressure
transducers inside the tubes. High‐speed cameras were used to record
both the composites and water response (such as cavitation shown in
Fig. 10d, e, f).
Facilities mentioned above focus on simulating FSI, hence captur-
ing composite response subject to the simulated underwater shock
loads, including deflection and velocity histories and core compres-
sion. Thus, the load‐carry capacity, energy‐dissipation capability and
failure mechanisms of different sandwich composites subject to under-
water shock load are evaluated. Structures, which are shaped both flat
and curved, backed by air and water, free‐standing and fully clamped,
could be examined using different water‐filled shock tube experimen-
tal setups. The testings, together with computational results, offer a
promising approach to improving the capabilities of maritime struc-
tures regarding underwater blast resistance [75–84]. In addition to
structural response, it is worth mentioning that the transparent shock
Fig. 6. (a) Schematics of conical shock tube; (b) photo of setup from the tube system [85–90] allow the observation of shock wave propagation
breech end; (c) photo of setup from the muzzle end [69]. and bubble dynamics. However, the composites that have been exam-
ined are limited within particular structure configurations, specific
combinations of core and face sheet materials and core sizes using
by a piston on one side and specimen on the other side, as shown in water‐filled shock tube facilities. Further studies should be conducted
Fig. 9a. Pressure generation and load transmission process were the for different geometries, sizes and materials of sandwich composites
same as in USLS [75–82,84]. Photographs of outside and inside exper- [75–84].
imental set up are shown in Fig. 9b and Fig. 9c, respectively. A scaling
factor was highlighted in this method to keep FSI the same as the full‐ 2.2.2. Water-filled chamber
scale situation. In this case, the same fraction of the far field momen- A water‐filled chamber is another way that has been used to simu-
tum I o must be transmitted to the composite plate. A scaling factor K late explosion in a controlled way. In the 1990s, Mouritz and co‐
was used to scale down the panel thickness according to: workers [96,112–114] performed a series of experiments on GFRP
Dfullscale laminates in a water‐filled pit (Fig. 11) to simulate underwater shock
K¼ ; ð4Þ wave generated by mine, torpedo exploding. This setup was consisted
Dexperiment
of a steel cylinder confined within a concrete slab. Thin plastic sheets
where Dfullscale and Dexperiment represent plate diameters for a full‐scale containing air bubbles were applied on the internal walls to reduce the
dimension and in a laboratory experiment respectively. The deflection internal reflection of shock waves following the explosion. The pit had
histories of the panels were assessed using shadow Moiré and recorded a diameter of 1.85 m and a depth of 2.0 m. The explosive charge could
by high‐speed photography [91–95]. generate maximum pressure from 8 MPa to 133 MPa. The tested lam-
In the 2010 s, Schiffer and Tagarielli [85–90] developed a transpar- inates were clamped at the bottom of the structure between two sheets
ent shock tube that allowed direct observation of fluid and structure of rubber. Both air‐backed and water‐backed configurations could be
response. This setup was composed of a transparent tube made of realized. Pressure‐time responses were measured using pressure trans-
acrylic with an internal diameter of 27 mm and a length of 2 m. The ducers mounted on the laminate surface, and another was suspended
transparent tube allowed for the observation of both structural and at the same stand‐off distance of the coupon freely in the water. Other
fluid responses, which could be recorded using high‐speed cameras. recording instrument included strain gauges and high‐speed cameras
Three different tube configurations are presented in Fig. 10, including (with Digital Image Correlation software). However, measuring trans-
unsupported air‐backed freestanding plates (Fig. 10a), unsupported mitted reaction forces or observing the experimental process and GFRP

Fig. 7. (a) Schematic of the underwater blast testing facilities USLS for a simply-supported and air-backed composite (b) schematic of simply-supported and water-
backed composite; (c) photograph of experimental setup [79].

9
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 8. Schematics of USLS setup to evaluate UNDEX resistance of: (a) sandwich structures with foam cores of different configurations [78]; (b) clamped round
panels; [83] (c) high-speed images showing deformation of composite with PVC foam core (ρ ¼ 200kg=m3 ); (d) high-speed images showing deformation of
composite with PVC foam core (ρ ¼ 60kg=m3 ) [78].

Fig. 9. Schematics of: (a) water-filler anvil camber with a divergent configuration; (b) photographs of the outside facility; (c) photographs of the inside facility
[91].

response during the blast was not possible. It should be noted that located outside windows at an offset angle of 17°. Images were anal-
strong safety measures are required when using this experimental ysed using 3D DIC software to study displacements and velocities of
setup. composite sample and fluid response. Fig. 12b exhibits high‐speed
images, showing cavitation and bubble collapse during the explosion.
2.2.3. Spherical pressure vessel In 2020, Javier et al. [115] improved this system by introducing
The spherical pressure vessel is another type of setup for UNDEX two more windows on the right mid‐span of the pressure vessel. One
experiment (Fig. 12a), which was developed by Pinto, Gupta and more lighting equipment and high‐speed camera were located outside
Shukla [99‐103]. This spherical vessel had a diameter of 2.1 m, which these two windows. This apparatus was employed to conduct UNDEX
could provide a constant hydrostatic pressure up to 6.89 MPa. Win- experiment on CFRP composite plates, which were exposed to saline
dows located on mid‐span of pressure vessel allowed for illumination for a long time.
and observation of composite response during the experiment. The
composite tube was sealed using aluminium head caps and suspended 2.2.4. Composites subject to extreme loading at low temperature
in the centre of the vessel. The vessel was filled with filtered water and Maritime structure is likely faced with low temperature environ-
pressurized by introducing nitrogen gas into a pocket. Measuring and ment when it operates in the arctic region. Such harsh condition is
recording instrument included high‐pressure blast transducers, high‐ quite challenging, which should be carefully considered for the design
speed cameras with 3D DIC technique. Two high‐speed cameras were of marine structures regarding damage tolerance at low temperature

10
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 10. Schematics of shock tube configured underwater blast simulator to observe UNDEX response of: (a) unsupported air-backed freestanding plates (b)
unsupported water-backed freestanding plates [90]; (c) fully clamped plates [89]; High-speed images showing: (d) onset of cavitation and propagation of breaking
front (BF); (e) propagation of BF; (f) closing front (CF) at fluid-structure interface [90].

Fig. 11. Schematics of underwater blast testing facilities to investigate the development of microstructural damage in GFRP: (a) cross-sectional view of the water-
filled chamber (b) cross-sectional view of GFRP laminates and boundary conditions [97].

[117]. Some core materials show brittle behaviours at low tempera- sandwich composite at low temperature. Testings were conducted at dif-

ture, which makes sandwich composites susceptible to extreme load- ferent temperatures, including room temperature (23 C), standard freez-
 
ings. When colliding with ice particles in the cold water, the ing temperature (0 C), average temperature in the arctic region (30 C),

damages below the surface of composite structure could be extremely and the lowest temperature recorded in the arctic region (70 C). The
significant, including matrix cracking, delamination, fibre breakages, climate chamber was connected to a liquid nitrogen tank to generate
face sheet and core interface debonding, and core crushing. low temperature environment for the composite specimen. Results
In order to study the damage behaviour of composite structure sub- implied that the behaviour of CFRP/foam composite was remarkably
ject to extreme loadings at low temperature, several types of experi-  
affected at 30 C and70 C. More brittle behaviour, reduced strength
ments were performed, including drop weight testings [117–124],
of face sheet, and increased structural stiffness were revealed at low tem-
Charpy impact testings [125–127], Split Hopkinson Pressure Bar appa-
perature. Severe damages, including fibre breakage, delamination, face
ratus (SHPB) [128–130], ballistic testings [131], and blast testings
sheet and core interface debonding, and core crushing, were observed
[129].
after the experiment using Micro‐CT. It is worth noting that the debond-
Drop weight test is the most commonly used method. Elamin et al.
ing area was dramatically increased at low temperature, which was
[117] employed a drop tower machine (Fig. 13a) with a climate chamber
attributed to the weaker matrix interface in such environment.
to study the impact behaviour and failure mechanisms of CFRP/foam

11
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 12. (a) Schematics of spherical pressure vessel setup to observe composite and fluid behaviours during UNDEX experiment; (b) high-speed images showing
the cavitation and bubble collapse during explosion [99].

Charpy impact test was a method utilised to study the energy imen in a climate chamber. This chamber provided cryogenic
absorption of GFRP laminates. The experimental setup employed by environment by allowing the liquid nitrogen flow inside and the fan
Li et al. [125] is shown in Fig. 13b, which aimed to study the impact constantly recirculate. Fig. 14b presents the details of the chamber.
properties and failure mechanisms of composite at room temperature Composite specimens were tested at three different temperatures
    
(20 C) and liquid nitrogen temperature (196 C) respectively. For (20 C,60 C,150 C) for comparison. Failure mechanisms were
the experiments at liquid nitrogen temperature, the specimen was observed, and damage areas were identified by C‐scan. It was demon-
placed in a container filled with liquid nitrogen for more than two strated that the low temperature had a negative effect on high‐velocity
hours before testing. The absorbed energy in the composites at differ- impact response of taped CFRP laminates, whereas it showed little
ent temperatures was compared, and the failure mechanisms were effect on damage severity once the impact velocity was beyond the bal-
identified using a scanning electron microscope. The increased impact listic limit.
energy indicated improved impact resistance of specimen at liquid Split Hopkinson Pressure Bar (SHPB) apparatus, which utilises
 
nitrogen temperature (196 C) than room temperature (20 C). stress wave propagation in elastic bars to generate dynamic load on
Authors explained this improvement as the results of the increased target structure [143], has also been used at low temperature to cap-
stiffness and strength of the resin matrix and the enhanced bonding ture the high rate stress‐strain response of composite structures
effect between fibre and matrix at low temperature. However, they [128–130].

also pointed out that at a temperature of 196 C brittle failure feature To author’s best knowledge, the only blast test on composite struc-
became more apparent. tures at low temperature was performed by Gupta and Shukla [129] in
Compared with low‐velocity drop weight test, a high‐velocity 2012. Similar as one type of the UNDEX laboratory‐scale experiment,
impact trial in low‐temperature environment was conducted by they employed a shock tube apparatus to generate blast loading
López‐Puente et al. [131]. This experiment aimed to examine the effect (Fig. 15a). The shock load was replicated by pressurizing the driver
of low temperature on impact damage extension on CFRP laminates. section. The GFRP/foam sandwich composites were tested at three dif-
The setup is shown in Fig. 14(a). They employed a gas gun to acceler- ferent temperatures (−40 °C, 22 °C, 80 °C). To achieve the target low‐
ate a spherical‐shaped steel projectile and impact the composite spec- temperature, the composite specimen was cooled down to −70 °C

Fig. 13. (a) Drop weight testing facility to study the impact behaviour and failure modes of CFRP/foam sandwich composite at low temperature [117] (b) Charpy
impact testing facility to study the energy absorption of composite structures [125].

12
P. Tran et al. Composite Structures 263 (2021) 113684

using the cooling system shown in Fig. 15b and quickly placed inside could be captured by four types systems, whereas only the transparent
the shock tube apparatus. The blast test was conducted when the tem- shock tube system allows direct observation of the entire process dur-
perature of core dropped to −40 °C. During the experiment, high‐ ing experiment. With the aid of high‐speed camera (with DIC), beha-
speed camera with DIC technique was utilised to record the real time viours of bubble and effects of FSI are also captured. However, the
displacement as shown in Fig. 15c. Results indicated that the tested real case is always more complicated than the controlled laboratory
composites experienced larger displacement and severer damages at environment. Also, the geometry and boundary conditions of the sam-
low temperature than at room temperature. The main damage modes ples tested in laboratory are much simpler than the composites used in
observed after experiment were core cracking and face sheet/core maritime structures. Further experimental studies in this field could
debonding. focus on investigation of composites with more complex geometries
Full‐scale and large‐scale experiments, as reported in literatures, and made of different materials. Additionally, near field denotation
could capture the shockwave pressure magnitude and composite of explosive charges could also be studied extensively regarding its
response, including acceleration, velocity, strain, and failure modes. localized effect.
An approximately real simulation of in‐service conditions is high-
lighted [110]. However, only a limited number of full‐scale and 3. Post-mortem analysis
large‐scale UNDEX testings have been performed and data from
these experiments are quantitatively limited (no repeat experiment) Damage modes that composites experience are more complicated
[111]. This attributes to the costly and time‐consuming manufacture compared to conventional materials. Common failure modes include
of sample structures, challenging and expensive experimental setups. matrix cracking, core crushing, fibre fracture, debonding, delamina-
To this regard, computational model of UNDEX experiments is rec- tion, fibre pull‐out, and compound failures including kink bands
ommended as a design tool to predict and evaluate composite beha- (buckling delamination due to kinking of fibres under compression)
viours [66]. and barely visible impact damage (BVID) cone of fracture. Post‐
Laboratory‐scale experiment aims to investigate behaviour of com- mortem analysis has been widely used to study damage mechanisms
posite structures during UNDEX event using small facilities. It requires of composite structures after blast testing. Different technologies have
less preparation time, lower cost, whereas it is repeatable, safer and been employed in different methods, such as ultrasonic pulse‐echo,
more controllable than full‐scale and large‐scale experiments. The four Lock‐in/Infrared thermography, X‐ray, and optical microscopy.
types of experimental setups mentioned above applied different work- Advancements in damage detection technology have facilitated accu-
ing principles and captured slightly different characteristics during the rate predictions of the behaviour of marine composite structures under
UNDEX testings. The shock tube and chamber systems could examine extreme loads and aggressive environments. Since the number of stud-
both air‐backed and water‐backed samples, while the spherical vessel ies regarding composite response to UNDEX is limited, Table 4 sum-
and cubic tank systems are only for water‐backed and air‐backed marises approaches that have been employed to study failure modes
respectively. Both shock wave pressure history and structure response

Fig. 14. (a) Schematics of experimental setup to examine the effect of low temperature on impact response of CFPR laminates (b) photos of climate chamber to
provide a low temperature environment for composite specimens [131].

13
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 15. (a) Schematics of the shock tube apparatus to study the temperature effect on dynamic response of maritime composites subject to blast load, including
driver section, driven section, muzzle section and test section [144] (b) cooling system to obtain target temperature for composite specimen (c) high-speed images
showing the real time deformations of composite at room temperature (22 °C) and low temperature (40 °C) respectively [129].

Table 4
Summary of post-mortem analysis methods to observe damage modes of composites.

Different post-mortem analysis Composite materials/ Instrument Observed damage modes References
structures

Non-destructive Visual GFRP/PVC sandwich High-speed Delamination, matrix cracking, fibre matrix [79,83,99,101]
inspection panel cameras debonding,
fibre pull-out and intralaminar cracking
Strain-based GFRP composite joints; Fibre optic strain Delamination, crack extension [145–150]
method aluminium/foam sensors
sandwich beam;
Ultrasonic pulse- Symmetrical and D9500TM C-SAMs Inter-laminar delamination, visible cracks [92,131,151–154]
echo method asymmetrical GFRP/ apparatus
PVC foam sandwich
composite; CFRP
laminates
Vibration-based CFRP laminates Polytec PSV-400 Delamination, cracks [155–157]
method non-contact
scanning laser
Doppler
Vibrometer
Electromagnetic CFRP laminates Electromagnetic Barely visible impact damage (BVID), [158–161]
inspection coupled spiral subsurface defects and internal micro-cracks
inductors (CSI)
Lock-in/Infrared GFRP; CFRP patch; Infrared camera; Subsurface defects, widespread weft yarn [162,163 164-166]
thermography boron composite patch Pulsed cracking, matrix delamination; Debonding
thermography and delamination in bonded composite
(PT); Transient joints
thermography
(TT)
Both non-destructive Visual GFRP/PVC sandwich High-resolution Delamination, matrix cracking, debonding, [72,79,82,83,95,167]
and destructive inspection panel; GFRP laminates digital camera fibre pull-out and intralaminar cracking
with/without polyurea
coating
X-radiography GFRP/SAN foam μCT scanner, CT Debonding between front face sheet and [66,117,153,168–173]
sandwich and CFRP/ scanner core, delamination, matrix cracking,
SAN foam sandwich through-thickness
panels; CFRP rupture
laminates
Microscopy GFRP and CFRP Optical Delamination, matrix cracking, fibre failure,
laminates, GFRP/PVC microscope, SEM core crushing, through-thickness
foam sandwich panel, Rupture, fibre matrix debonding, z-binder
hybrid metal yarn bridging the delamination crack
composite plates
[83,92,95,99,101,118,122,125–127,
131,151,153,167,170]

Note: GFRP refers to glass fibre reinforced plastics. CFRP refers to carbon fibre reinforced plastics. DIC refers to Digital Image Correlation.

14
P. Tran et al. Composite Structures 263 (2021) 113684

of composites resulted from air and water blast event, impact action destructively observe failure modes of monolithic and hybrid metal/-
and environmental corrosion. composite laminates after UNDEX tests. Fig. 17(d, e) are high‐
As summarised in Table 4, post‐mortem analysis could be achieved resolution micrographs, showing interlaminar delamination, fibre
using both non‐destructive and destructive damage detection methods cracking, matrix cracking, fibre‐matrix debonding observed using
[92]. In previous studies, both destructive and non‐destructive meth- SEM.
ods were employed. Herein, several examples of using different tech- The ultrasonic pulse‐echo method is another non‐destructive tech-
nologies are given in the context of UNDEX. nique that utilises ultrasonic waves to detect defects in materials. The
scanning data could be presented in numbers of different formats. C‐
3.1. Non-destructive methods scan is one of the most commonly used formats, which provides a plan
view of the tested specimen. Latourte et al. [92] employed a C‐scan
Non‐destructive techniques (NDT) include many methods that can pulse‐echo technique, using a D9500TM C‐SAMs apparatus, to exam-
detect damage location, shape and size without destruction of the ine the performance of both monolithic and sandwich composite pan-
structure, which is economically advantageous [174]. For post‐ els subject to underwater impulsive loading. Bending induced
mortem analysis in composite subject to UNDEX, high‐speed camera delamination of the tested panel was found as a bright ring on the mar-
(with DIC technique), optical microscope, scanning electron micro- gin in Fig. 18a, while the darker patches corresponded to delamination
scope (SEM), C‐scan and X‐ray CT scanner have been employed. It within the mapped gate. In addition to delamination, the pulse‐echo
should be noted that microscopy and X‐ray could also be used in a technique also detected the visible cracks represented by a cross‐
destructive way (refer to Section 3.2 for further detail). shape feature in Fig. 18b. It should be noted that this C‐scan detection
High‐speed camera (with DIC technique) has been widely used to process was conducted in a water‐immersed environment.
track the displacement time histories of the viewable surface of target X‐radiography, including conventional tomography and micro‐
structure [175]. DIC technique was calibrated in underwater explosive tomography, could be used both destructive and non‐destructive dam-
experiments by Gupta et al. [176], showing accurate measurement of age detection. Rolfe et al. [168] carried out a post‐blast analysis of
displacements and velocities. Pinto and co‐authors [99,101] investi- GFRP and CFRP composite structures subject to full‐scale underwater
gated types and magnitudes of damages on composite structures sub- blast testing by using a μCT scanner. Deboning between front‐face
ject to underwater shock load by both high‐speed camera (with DIC sheets and cores were found dominant, while core crushing was also
technique) and optical microscopy. Fig. 16 presents three different observed to different degrees for GFRP and CFRP, as shown in
modes of damages, while the longitudinal cracking is identified dom- Fig. 18c, d respectively.
inant regarding tube failure.
In 2011, LeBlanc and Shukla [72] assessed laminate modification 3.2. Destructive methods
effects by observing the visible damage using a high‐resolution cam-
era. It can be seen in Fig. 17a, b, c, for all three laminates, that the To improve the reliability and accuracy of non‐destructive damage
primary damage mode is delamination with minimal matrix and fibre detection [92], destructive methods have been employed in some stud-
cracking. Still, the authors pointed out that damages could not be ies. However, destructive detection methods are typically time‐
separated regarding the blast loading phase and secondary deforma- consuming and confined to local damage observation [178]. The tech-
tion phase. Similarly in 2013, Avachat and Zhou [82] took post‐ nique employed in composite subject to UNDEX involves high resolu-
mortem photographs to investigate damage mechanisms of a mono- tion camera, microscopes and CT‐scanner.
lithic composite structure since it was thin and failure modes were Early in 1994, Mouritz et al. [96] conducted post‐mortem analysis
difficult to identify using high‐speed imaging. Delamination, matrix of GRP laminates using optical microscope and SEM after underwater
cracking, fibre pull‐out and rupture could be found in the magnified shock tests. The tested samples were analysed firstly non‐destructively
photos. and then cut, polished and etched for microstructural damage identifi-
According to Zhou et al. [177], scanning electron microscope cation using SEM. Fig. 19a and Fig. 19b are the micrographs taken dur-
(SEM) is a sound instrument for microstructure morphology analysis. ing the non‐destructive phase, revealing that glass fibres could stop or
Compared with the optical microscope, SEM can provide more deflect cracks within the polymer matrix, thereby carrying most of the
detailed information attributed to its higher magnification and greater tensile loads and improving the underwater blast resistance of lami-
focus depth. In damage detection, SEM could be used both destructive nates. Fig. 19(c) is the micrograph after sectioning, which reveals
and non‐destructive. Avachat and Zhou [83] employed SEM to non‐ extensive delamination between glass piles.

Fig. 16. Post-mortem images of (a) filament-wound and (b) roll-wrapped composite specimens: (A) through-thickness longitudinal crack, (B) interfibrillar cracks,
(C) fractured fibres [99].

15
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 17. Damage of three laminates after tests (a) base line, (b) equivalent thickness baseline with a glass veil between plies, (c) baseline with polyurea [72]; SEM
micrographs of the composite specimen after UNDEX test at two different magnifications, showing (d) interlaminar delamination, (e) fibre cracking, matrix
cracking and fibre-matrix debonding [83].

Wei et al. [95] compared, what were observed using optical micro- lithic composite panel with partially propagating to mid‐span. The
scope and high‐resolution camera after underwater impulsive tests on cross‐section of a foam core sandwich panel was examined with mag-
monolithic and sandwich composites, with their proposed numerical nified photographs (Fig. 20b). Core crushing and cracking were
simulation results. As shown in micrographs (Fig. 20a), delamination observed at mid‐span, while delamination, fibre cracking and matrix
patterns could be clearly identified at the clamped edge of the mono- failure were found at clamped edges.

Fig. 18. (a) C-scan for solid panel subject to underwater impulsive loading, (b) C-scan for waterside face sheet of the panel [92]; X-Ray CT scan showing deboning
between front face-sheet and core of (c) CFRP and (d) GFRP after underwater blast testing [168].

16
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 19. Scanning electron micrographs showing (a) a crack stopped at glass fibre; (b) a crack waved between glass fibres; (c) extensive delamination between two
glass piles [96].

Fig. 20. Optical micrographs of cross-section of: (a) monolithic panel after impulse test, showing delamination pattern, (b) sandwich panel with magnified photos
at edges and mid-span, showing foam crushing, delamination and shear failure [95]; (c) Cross-section of AISI 304 stainless steel honeycomb sandwich panels after
blast loading, (d) amplified cross-section at panel centre, (e) amplified cross-section near panel edge [94].

Mori et al. [94] examined both the cross‐section of sandwich panels • X‐ray CT: features 3D visualisation of internal damages and quanti-
and fractures on face sheets of the examined composite after experi- tative analysis, but strongly rely on variations in material density
ment to identify failure mechanisms of such structures subject to and thickness.
UNDEX. Face sheet examination was conducted using a non‐ • Ultrasonic pulse‐echo: allows detection of small and deep internal
destructive method, whereas the cross‐section was studied using a flaws with only one surface of sample accessible. Size, shape, orien-
destructive method. The tested sampled was cut into half by water‐ tation of damages could be identified. However, it cannot be
jet machining and the cross‐section was observed in high‐resolution applied to irregular, inhomogeneous or very thin composites.
images. As exhibited in Fig. 20 (c, d, e) core crushing occurred in • SEM: it is capable of damage quantitative analysis, providing high‐
the centre and the peripheries of the sample. resolution, magnified and three‐dimensional images of nanostruc-
X‐ray CT scanning has also been employed as a destructive post‐ tures and microstructures. However, it is limited to small size sam-
mortem analysis method. Recently, Rolfe et al. [66] used X‐ray CT ple, and particularly sensitive to external magnetic fields.
scanning to evaluate damages of composite panels with GFRP/CFRP
skins with graded foam core after underwater blast tests. Panels were In future studies, other advanced techniques, such as lock‐in/
cut into three sections, and a certain thickness of material was infrared thermography, could be used as the tool to examine the dam-
removed before examination. Fig. 21 shows the 3D reconstructed age modes of composite structures. Moreover, post‐mortem analysis
GFRP panels after testing with cross‐sectional scans as well. Front face cannot identify during which time period of the UNDEX the damage
sheet debonding, core crushing and core cracking could be easily actually happens. To this regard, the damage modes captured in
observed from CT scans. The post‐blast analysis revealed that panels post‐mortem analysis could be compared with numerical simulations
with GFRP skins suffered less damage, but severer core crushing com- to help further investigate the progress of damage development.
pared to CFRP skins.
Both destructive and non‐destructive methods could identify the 4. Numerical simulations
typical failure mechanisms of composite, including delamination,
matrix cracking, fibre matrix debonding, fibre pull‐out, core crushing, Numerical simulations aim to provide deeper insights into how
and through‐thickness rupture. The techniques employed in UNDEX fluid and composites respond during UNDEX and interact with each
cases are limited to X‐ray, microscopy, ultrasonic pulse‐echo, high‐ other. Numerical simulations of composites subjected to UNDEX
speed camera (with DIC technique) and high‐resolution camera as mainly focused on fluid structure interaction (FSI), which date back
summarised. Pros and cons of CT scanner, ultrasonic pulse‐echo and to the early 1940 s. Interaction between the motion of surrounding
SEM are listed as follows: fluid and structure has a remarkable effect on the stress within the

17
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 21. 3D reconstructed CFRP composite panel from 2D X-ray CT scans, showing the damaged composite after underwater blast testing [66].

structure [179]. In previous studies, the Finite Element method (FEM) et al. [91] performed extensive FEM simulations by the L‐L model to
has been widely used, including Lagrangian model, Eulerian Lagran- gain insight into effect of divergent diffuser and nonlinearity in the
gian model and Acoustic Lagrangian model. Meanwhile, some coupled response of water at the imposed pressures. The pressure wave propa-
methods have also been validated capable of simulating FSI, such as gation in the water and its distribution at the fluid‐solid interface was
coupled Boundary Element Method (BEM)/FEM and coupled investigated. Results indicated that the L‐L method could accurately
Smoothed Particle Hydrodynamics (SPH)/FEM. Table 5 gives a sum- capture history‐dependent material properties.
mary of the models and major software employed by previous Later, Avachat and Zhou [77] carried out FE simulations using
research, but is also pertinent to highlight examples of other packages Lagrangian framework to investigate the relationship between the
can also be used for UNDEX simulation: RADIOSS [180], AUTODYNE thickness of face sheet and the response of underwater blast loading
[181], MSC/Dytran [132,182–184], NASTRAN/USA/CFA [185], on sandwich structures. By performing FE simulations, they found
MARC [186,187], NISA [188], Trident [189,190], Chinook that there existed an optimal design of face sheet thickness that
[135,137], PRONTO [191] and User‐Developed Code [192,193]. can maximise dissipated energy and minimise structural deformation
The following discussions aim to provide key development and at the same time. Additionally, they pointed out that the L‐L model
applications of each numerical technique. It should be emphasised that could produce FSI and accurately simulate impulses and pressures.
all numerical models a based on a set of assumptions and simplifica- Fig. 22a shows the finite element mesh for the constructed FE model,
tions and therefore there is no perfect modelling technique and each and Fig. 22b presents the simulation results for equivalent plastic
technique is suitable for a class of problems. Besides the deterministic strain distribution in the sandwich core. Similarly, Schiffer and
modelling tools, recently there is sufficient interest in developing data‐ Tagarielli [89] also performed FE simulations based on the Lagran-
led predictive models such as neural network, machine learning or gian framework to study response of circular composite plates subject
deep learning. to UNDEX.
However, the L‐L method was found difficult when dealing with
4.1. Finite Element Method (FEM) large deformation problems [219,220]. To eliminate the element dis-
tortion problem of L‐L model, another FEM method, Eulerian‐
FEM regarding simulation of fluid–structure interaction (FSI) dur- Lagrangian technique (E‐L method), has been introduced. The E‐L
ing UNDEX mainly includes simplified models, Lagrangian framework model combines the advantages of both the Eulerian and the Lagran-
(L‐L method), Eulerian‐Lagrangian framework (E‐L method) and gian formulations. Two popular frameworks have been proposed as
Acoustic and Lagrangian framework (A‐L model). Simplified models E‐L model in order to simulate FSI during UNDEX, i.e., coupled E‐L
[87,88] were conducted based on certain assumptions, such as roller model (CEL) and arbitrary Lagrangian‐Eulerian model (ALE).
constraints and infinite boundary distance. Simulations were con- CEL was proposed and validated by Xie and et al. [194] and applied
ducted to investigate one‐dimensional response of sandwich plates to UNDEX analysis of multi‐layered composite plates in a 2D axisym-
and found consistent with experimental results. The other three FEM metric model. The multi‐phase compressible Eulerian solver is coupled
approaches have been more commonly utilised to simulate FSI during the Lagrangian solid solver using an interface capturing method which
UNDEX events [95]. The L‐L method is to simulate both water and updates the interface conditions of the fluid (pressure, density, normal
composite structures with the Lagrangian framework [91]. Espinosa velocity) into characteristic equations for fluid and solid motion, gen-

18
Table 5

P. Tran et al.
Summary of numerical simulation employed for composite structures subject to UNDEX.

Numerical simulation method Captured Fluid behaviour Domain description Software Composite material/ Comment Reference
feature structure

FEM Simplified FSI Mie-Gruneisen EOS with a - ABAQUS/ Metallic sandwich Assumptions were made to study [87,88]
model linear Hugoniot relation Explicit plates with cores of one-dimensional response.
different materials and
configurations
L-L model Fluid domain and solid domain are both Stainless steel Capture history-dependent [77,89,91]
described using the Lagrangian monolithic plates; material properties; difficulty with
framework. circular composite; large deformation problems
FRPsandwich plates
with polymer foam core
E-L CEL Mie-Gruneisen EOS Fluid domain is described using the LS-DYNA GFRP and CFRP Eliminate element distortion [116]
model Eulerian framework, while solid domain is laminates problem; capture nonlinear FSI;
Barotropic Tait’s EOS; bilinear described by the Lagrangian framework. ABAQUS/ Steel/foam core overestimate the effect of matrix [194-196]
EOS Explicit sandwich structures cracking and rupture.
Mie-Gruneisen EOS GFRP cylindrical [80,84]
specimen and sandwich
structure
Compressible Tillotson EOS DYSMAS CFRP composite plates [115]
after prolonged
exposure to saline
ALE Mie-Gruneisen EOS Fluid and solid domains are same as CEL. A LS-DYNA CFRP and GFRP Similar to CEL, but ALE is more [98,199–201]
third domain, called reference domain, laminates; GFRP efficient in solving large
allows nodes of the discretisation move in sandwich structures deformation problems than CEL
space independently of the material [197]. with hard foam core [198].
A-L model Inviscid, linear and Fluid is described as an acoustic medium, ABAQUS/ GRP monolithic panels Accurately simulate FSI and able [95,109,202–209]
compressible fluid behaviours while solid domain is described by the Explicit and sandwich to capture different failure
19

are assumed. Lagrangian framework. structures with PVC mechanisms of composite


foam core, partially structures
submerged ship hull
CFD Shock Linear wave theory, The domain volume is filled with ANSYS – Not precise enough, expensive [210,211]
waves Navier–Stokes equations tetrahedral cells and then improved by ICEM CFD
refinement, local mesh adaption.
SPH Fluid Mie-Gruneisen EOS SPH particles – A rigidly confined Simulate large deformation and [132,212–214]
dynamics, water chamber fracture of complex structure;
structural track movement of fluid-structure
damage, interface; not suitable for a long-
FSI time duration.
Coupled method Coupled FSI, Mie-Gruneisen EOS SPH particles and the Lagrangian Finite A rigidly confined Well-suited for load transmission [191,212]
SPH/FEM bubble formulation. element water chamber, flat from UNDEX to nearby structures
code plate, hollow cylinder and large deformation
PRONTO
Couple FSI, FSI is modelled only at the Solid domain is described using LS- A partially/fully Interaction between explosive gas [200,215-218]
BEM/FEM structural wettedsurface on the structures, Lagrangian formulation, while DYNA3D submerged composite and surrounding water is not

Composite Structures 263 (2021) 113684


analysis while the surrounding fluid fluid–structure interface is modelled using and ship section; GFRP properly considered so that
elements need not be modelled as a DAA boundary. boundary sandwich structures prediction of peak pressure is
[215]. element with hard foam core inaccurate.
code USA

Note: FEM refers to the Finite Element method. L-L refers to Lagrangian-Lagrangian. E-L refers to Eulerian-Lagrangian. A-L refers to Acoustic and Lagrangian. ALE refers to Arbitrary Lagrangian-Eulerian. CEL refers to
Coupled Eulerian-Lagrangian. CFD refers to Computational Fluid Dynamics. SPH refers to Smoothed Particle Hydrodynamics. FSI refers to fluid structure interaction. EOS refers to the Equation of State. BEM refers to
Boundary Element Method. USA refers to Underwater Shock Analysis. DAA refers to Doubly Asymptotic Approximation.
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 22. (a) The quarter-symmetric model, showing finite element mesh using L-L framework to simulate FSI in shock tube setup, to study effect of face sheet
thickness on energy absorption; (b) equivalent plastic strain distribution in the core [77].

Fig. 23. Coupled E-L (CEL) framework simulate FSI for experimental setup of: (a) water-filled shock tube [80]; (b) cubic tank [116].

erating the coupled FSI. A narrow band of ghost fluid cells are applied et al. [116] and Javier et al. [115] employed CEL to simulate FSI for
to the solid‐side of the interface which is updated with extrapolated the cubic tank (Fig. 23b) and the spherical pressure vessel experimen-
values from the fluid‐ solid side for a more accurate FSI in all domains. tal setups respectively.
On the basis of previous work, Xie and his research group Compared to CEL, according to Bakroon et al. [198], ALE is more
[195,196] employed this strongly coupled E‐L model (CEL) to investi- efficient in dealing with large deformation problems. ALE method
gate the transient response of sandwich structures subject to UNDEX. has an outstanding feature of dynamic remeshing, which mitigates sig-
Results indicated that CEL model could capture nonlinear FSI involv- nificant distortion issues involved in the pure Lagrangian method. In
ing strong shock wave, cavitation, and bubble dynamics. Similarly, a 2016, LeBlanc et al. [98] employed the ALE model to simulate FSI of
fully coupled 2D axisymmetric CLE study by Wang et al. [221] consid- cubic tank UNDEX experimental setup, as mentioned in Section 2.2.4.
ers the boundary effects of UNDEX at a free surface and solid bound- The numerical simulations were found a high level of correlation to
ary. The resultant rarefaction wave at the boundary demonstrates experimental results, thereby demonstrating the capability of ALE
multi‐phase applicability of the CLE method, highlighting the cavita- regarding FSI simulation. Supportively, Özarmut et al. [200] con-
tion regions in the fluid domain. ducted simulation using ALE framework to study the response of
Avachat and Zhou [80] also employed the CEL framework in ABA- FRP sandwich with hard foam core subject to underwater shock load.
QUS software to capture the extreme deformation of both monolithic The ALE model and meshing details are shown in Fig. 24a, while the
and sandwich composites subject to underwater impulsive loading. assumed planar wave generated by explosion is shown in Fig. 24b.
The CEL computational framework for the shock tube experimental However, capturing the interaction between fluid and solid
setup, which is shown in Fig. 23a, was aimed to investigate deforma- involves the Eulerian mesh extending beyond the original fluid
tion mechanisms and energy dissipation in composites. They found domain, thereby leading to issues to achieve a fine mesh. Then, a
that CEL model allowed Eulerian material interacting with Lagrangian coarse Eulerian mesh results in stiff interface between the fluid and
elements so that FSI could be captured. Additionally, they pointed out solid domains [95]. Therefore, the Acoustic and Lagrangian frame-
that the Eulerian formulation could accurately simulate the explosive work (A‐L model) was introduced by Marcus [222] as a third FEM
pressure waves and cavitation at the interface. On this basis, Qu method regarding FSI simulation. In A‐L model, water is modelled
et al. [84] employed the CEL framework to investigate collapse and using acoustic element, while solid is modelled using Lagrangian for-
damage evolution of composite structures under large impulsive load mulation. The interface was modelled using surface‐based tie con-
in the water. Results showed that this numerical model slightly over- straints. McShane et al. [206], in 2007, conducted numerical
estimated the effect of matrix cracking and rupture. Recently, Matos simulation using A‐L frameworks to evaluate UNDEX resistance of

20
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 24. (a) ALE model with magnified meshing constructed using LS- DYNA to simulate FSI during UNDEX to evaluate its suitability regarding capture FRP
response; (b) shock wave propagation in two-dimensional ALE model [200].

Fig. 25. (a) cross-section view of mesh in A-L model to simulate FSI for divergent configuration shock tube setup; (b) deflection history of sandwich panel
predicted by A-L model [95].

both monolithic and sandwich composites. Similar studies have been FSI using A‐L model (Fig. 26a), which shows the deformation of com-
conducted since then to capture the FSI for the experimental setups posites subject to UNDEX.
mentioned in Section 2 [74,95,109,202–204,207–209,223,224].
In 2013, Wei et al. [95] employed A‐L model to study the failure 4.2. Coupled methods
mechanisms of sandwich composite panels subject to UNDEX
(Fig. 9). Fig. 25a schematically shows the detailed meshing, deforma- In addition to FEM, some coupling methods, such as Boundary Ele-
tion and failure modes. The deflections predicted by A‐L model are ment Method (BEM)/FEM and Smoothed Particle Hydrodynamics
shown in Fig. 25b, which was found to slightly overestimated matrix (SPH)/FEM were also utilised to simulate FSI for evaluation of com-
damage. To improve model accuracy, authors suggested establishing posite UNDEX resistance.
a new rate‐dependent progressive degradation model for composites BEM is a more efficient tool for shock load simulation than FEM
and interfaces [95,202]. when the non‐contact underwater explosion is far from the structure
Supportively, Tran et al. [203,204] investigated the performance of [215]. In the late 1970 s and early 1980 s, Geers and his group
bio‐inspired composite structures subjected to UNDEX by simulating [217,225] proposed an approximate technique in BEM, doubly asymp-

21
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 26. (a) Schematics of A-L model to simulate the FSI in divergent shock tube experimental setup to understand deformation and failure mechanisms of
composites (b) Side view of composite deformation obtained in numerical results [203].

Fig. 27. (a) FE model with USA code to simulate FSI in LS-DYNA (b) pressure time histories of the hard foam core obtained in three different numerical methods
[200].

BEM and FEM improves the level of accuracy and efficiency of FSI
problems.
In the past several decades, McCoy and Sun [179], Gong and Lam
[218], Hung et al. [215], and Özarmut et al. [200] are among the
researchers who used BEM/FEM method to conducted numerical sim-
ulations of FSI in UNDEX. Recently, Özarmut et al. [200] compared the
suitability of L‐L, ALE, and BEM/FEM. The established BEM/FEM
model is presented in Fig. 27a, where fluid was only modelled at the
interface. The capability of shock wave simulation of BEM/FEM
method was evaluated by comparing the pressure time histories of
the hard foam core from all three methods (Fig. 27b). Authors con-
cluded that BEM/FEM method, compared to other two methods, could
reduce the computational time if the modelling of a significant amount
of water could be avoided.
In addition to BEM/FEM, the Smoothed Particle Hydrodynamics
SPH/FEM is another coupled numerical simulation method. SPH is a
meshless Lagrangian technique, which offers a solution to distortion
problems attributed to the Lagrangian formulation and also diffusion
problems associated with Eulerian formulation [226]. Early in 1995,
Swegle and Attaway [191] coupled SPH into FE code, PRONTO, to
Fig. 28. (a) SPH/FEM mesh for submerged cylinder subject to UNDEX, with evaluate feasibility of coupled SPH/FEM method regarding FSI in
magnified SPH mesh region; (b) material deformation plots during explosion UNDEX. Fig. 28a presents the simulation mesh for a submerged cylin-
[191]. der subject to UNDEX. Explosive and nearby water was simulated
using SPH, while the rest of water was hex elements and cylinder pipe
was using FEM. Results showed that SPH/FEM is capable of transmit-
totic approximation (DAA), to determine the impact loading on a par- ting loading to structure and capturing large deformation (Fig. 28b)
tially submerged structure by an incident shock wave. DAA has an out- subject to underwater explosion. Nevertheless, results also revealed
standing feature that only the fluid on a wetted structure surface that this coupled method was not promising to simulate FSI consider-
should be modelled instead of simulating the entire fluid medium, ing the late time effect caused by the acceleration of gravity and
which is often used in FEM simulations. A few years later, Everstine buoyancy.
and Henderson [216] coupled FE model of the structure with the BE Similarly, Ming et al. [212] proposed an SPH/FEM coupling
model of the fluid on interface to simulate FSI. The combination of method to simulation FSI for a flat plate subject to contact UNDEX

22
P. Tran et al. Composite Structures 263 (2021) 113684

Fig. 29. (a) SPH/FEM coupling procedure; (b) FSI simulation model using coupled SPH/FEM to investigate the damage process of a ship hull subject to UNDEX;
(c) damage of the hull and (d) the focused cabin at time of 15 ms [212].

by applying ‘glue’ treatment at the interface. The procedures of cou- This jet can be potent and can destroy the ship hull. Depending on
pling SPH and FEM is exhibited in Fig. 29a, while the established the exact location of the bubbles, the proximity to the free surface
FSI model is shown in Fig. 29b. Researchers observed the damage pro- and the influence of gravity, the jet may impact on the structure or
cess of a ship hull. Fig. 29c and Fig. 29d show the damages at the time miss it completely. Artificial viscosity is another characteristic that
of 15ms. could be captured using numerical simulation [87,228]. It smears off
When dealing with FSI, each numerical method has its own shock waves in several meshes. Large coefficients of artificial viscosity
strengths and limitations as listed here: make peak pressure lower than the actual value, while small artificial
viscosity coefficients cannot suppress serious oscillations after peak
• L‐L method: it has difficulty in large deformation problems regard- pressure [87,228]. In addition, damage modes of composites and FSI
ing element distortion. are also vital features that have been widely investigated in previous
• E‐L method: it combines the advantages of both the Eulerian and numerical studies. Table 6 summaries various UNDEX characteristics
the Lagrangian formulations, which eliminates distortion but has by using different modelling tools.
problems with a coarse Eulerian mesh.
• A‐L method: it is capable of accurately simulating cavitation, FSI 4.3. Data-led modelling technique
and depicting different damage modes of composites.
• Coupled BEM/FEM method: it only requires modelling of the fluid As summarised in Table 5, numerical models could capture FSI,
on fluid‐structure interface, thereby suitable for large distance non‐ structure response, shockwave propagation and behaviour of gas bub-
contact UNDEX event. However, the interaction between explosive ble. However, results from numerical simulations are still approxima-
gas and surrounding water is not properly considered so that pre- tions as FSI phenomenon are highly complex. The mathematical
diction of peak pressure is inaccurate. modelling of dynamic events has always been very challenging due
• Coupled SPH/FM method: it is well‐suited for load transmission to complex nonlinear governing equations, which lead into singulari-
from UNDEX to nearby structures and large deformation problem, ties in the formulation making it difficult to obtain analytical solutions.
whereas it is not promising to simulate FSI considering the late Numerical FSI simulations often introduce simplifications that allow
time effect caused by the acceleration of gravity and buoyancy. for satisfactorily accurate solutions with the disadvantage of requiring
long lead‐times until completion, which might not suitable for time‐
The physical processes of the underwater explosion near bound- critical applications. Alternative approaches have tapped onto
aries are incredibly complex, which involves various complicated machine learning paradigms to develop novel rapid assessment tools.
issues such as the shock wave propagation, water–air or water–struc- The introduction of more powerful microcomputers in the mid‐
ture interaction [221]. In terms of fluid behaviour, cavitation has been 1990s led to the popularisation of machine learning tools to classify
investigated by many scientists as blast loading of submerged struc- and analyse large datasets and to predict the outcomes of physical
tures results in complex water cavitation phenomena [227]. Numerical events. Broadly speaking, machine learning can be divided into super-
analyses show that an underwater bubble placed near a structure vised and unsupervised algorithms. The former being more suitable to
would, in many cases, develop a jet directed towards this structure. regression applications, and the latter to classification and pattern

23
P. Tran et al. Composite Structures 263 (2021) 113684

Table 6
Summary of captured characteristics in different numerical simulations of UNDEX.

Captured feature Numerical Modelling details Comment References


simulation
method

Fluid behaviour Gas Cavitation formation FEM Set tensile failure of Cavitation initiates at a [73,87–90,95,109,132,133,202,
bubble zero stress in ABAQUS/ finite distance from FS 206,221,223,227,229-231]
Explicit. interface;
Detail of cavitation strongly
influences pressure history
on sandwich structure.
Bubble BEM Pressure field inside High calculating efficiency [232-237]
hydrodynamics bubble is assumed [232]
homogeneous and
related to bubble
volume variations with
an appropriate
equation of state.
Bubble-bubble and ALE Bubble is assumed Most commonly used [238-241]
bubble-wall interaction; incondensable steam
violent bubble motion and state is assumed to
be gamma law (gamma
of 1.3) in ALE3D
software
Interaction between ALE, Coupled Refer to Fig. 26 – Costly and time-consuming [191,242 98,116]
bubble and deformed SPH/FEM Figure 32
plate
Incompressibility and BEM, ALE, Rayleigh–Plesset model Flow rate is low, suitable for [239,244]
irrotationality Finite volume [243] the fluid domain in FSI.
method
Artificial viscosity Finite In ANSYS/AUTODYN, Large coefficients of [87,228]
Volume recommended values artificial viscosity make
method for linear and quadratic peak pressure lower than
(Euler) viscosity coefficients the actual value, while small
are 0.2 and 0.1, one cannot suppress serious
respectively. oscillations after peak
pressure.
Composite structure FEM Hashin’s damage Fibre damage in tension and [80,87,92,95,202–204,207,245–247]
(Lagrangian model [205] compression, matrix
formulation) damage in tension and
compression could be
captured.
Extended Gurson Void volume fracture could [93]
damage model [207] be captured Table 5.
SPH/coupled Mostly focus on soil under –
FEM and SPH impact/ballistic scenarios
Fluid structure interaction ALE, SPH, Refer to Table 5 Interaction between the [77,80,84,87,89,91,95,98,115,
coupled SPH/ motion of surrounding fluid 116,194,195,202-204,212,214]
FEM, couple and structure has a
BEM/FEM remarkable effect on
stresses within the structure
[179].

Note: FEA stands for finite element analysis. ALE stands for Arbitrary Lagrangian-Eulerian. L-L stands for Lagrangian-Lagrangian. E-L stands for Eulerian-
Lagrangian. A-L stands for Acoustic and Lagrangian. CFD stands for Computational Fluid Dynamics. SPH stands for Smoothed Particle Hydrodynamics. FSI stands
for fluid structure interaction, BEM stands for Boundary Element Method.

recognition. Applications focused on modelling physical phenomena imental datasets had to be surpassed with usage of so‐called synthetic
have relied on supervised algorithms, such as Artificial Neural Net- data – i.e. data furnished from validated computational simulations.
works (Multilayer Perceptron), Support Vector Machines, and logistic Validated numerical models can provide complementary data to train
regression. An extensive description of all machine learning paradigms and improve the accuracy of machine learning implementations but
available is outside the scope of this paper and the reader is referred to they lack the stochastic nature inherent to experimental test trials.
comprehensive textbooks – e.g. [248–250]. Stochastic results introduce scatter (noise) to training population,
Cost, resource limitations and complexity restrict the experimental which may assist the machine learning model in recognising trends
blast testing to a low data environment with many features poorly and developing confidence intervals [258]. Synthetic data has been
quantified since data collection is usually difficult and localised. As a combined to machine learning models to evaluate the propagation of
result, only recently datasets from field tests started to be utilised for air blast loadings in built environments [256,257,259–262] and to
training machine learning tools for modelling dynamic events. Such estimate damage of air blast loadings on structural elements
studies used machine learning for evaluating the performance of [258,263,264].
spaced aluminium armour (Whipple shields) to hypervelocity impacts Surprisingly, not many studies have employed machine learning
from spherical projectiles [251–254] and to estimate the damage algorithms to describe UNDEX settings. One case study employed
caused by localised blast loadings on mild‐steel plates [255]. Initial Probabilistic Neural Networks to evaluate the quality of data his-
applications geared towards modelling dynamic events were intro- tograms from shock qualification tests for a floating shock plat-
duced in the mid‐2000 s [256,257] where limited availability of exper- form subjected to underwater explosion [265]. A more recent

24
P. Tran et al. Composite Structures 263 (2021) 113684

investigation used a simple 1D analytical formulation to model the tions must be employed however if bubble formation, pulsation,
response of a structure submerged in water subjected to far‐field collapse jetting, or rarefaction‐induced cavitation are to be
UNDEX loading and utilised the results to train a Back‐ considered.
propagating Neural Network [266], but it lacked a comprehensive • Current trends in FE analysis of UNDEX employ techniques cou-
description and in‐depth performance analysis of the machine pling the fluid dynamics solver to structural solver. These tech-
learning model. niques include the coupled FEM/BEM and coupled Eulerian‐
Imperfect and incomplete description of physics is a well‐known Lagrangian, along with the ALE technique. Each of these techniques
limitation of supervised machine learning tools, since they are not a has advantages and disadvantages, with the number of finite ele-
first principles approach. Whilst allowing for immediate construction ments, mesh refinement and computational resources required to
of suitable models, these models usually cannot be generalised and handle high‐resolution regions influencing computational effi-
are constrained by the training population bounds. Other limitations ciency, especially when considering gas bubble collapse.
may include overfitting of the training data and the development of • SPH techniques are in increasing usage for UNDEX consideration,
extremely complex models, making it difficult to understand the con- primarily due to the “mesh‐less” nature of the formulation and abil-
nections between inputs and outputs [250]. ity to handle significant deformations.
All things considered, the combination of UNDEX experiments, • Blast analysis is highly dependent on mesh sensitivity and correct
analytical and numerical models, and machine learning techniques implementation of mapping techniques will likely alleviate some
can produce powerful frameworks for estimating the mechanical of the large computational penalties associated with the analysis.
response of structures to dynamic loadings, reducing associated costs
and delivery time. As an example, appropriate machine learning mod- Further increases in the fidelity of the experimental testing, and the
els could be used for swiftly selecting potential prototypes, which incorporation of multi‐scale modelling techniques, will ultimately
could then be further investigated with in‐depth numerical modelling increase the uptake of composites in the maritime space. Closer inte-
and experiments. gration of artificial intelligence and machine learning techniques
may also aid in the accelerated development of more novel solution
5. Conclusion e.g. design of optimal multi‐layered configuration or auxetic structure.
Machine learning may also allow for the interpolation of sparse exper-
The UNDEX event involves a combination of extremely complex imental data sets, when supplemented with analytical or numerical
phenomena whose impact on a marine structure or vessel can be catas- modelling results.
trophic. In order to accommodate such events in the design of future
vessels and structures, proper understanding of the processes and Declaration of Competing Interest
interconnectivity each of these phenomena is needed. This will ulti-
mately allow for more efficient and informed design choices and set The authors declare that they have no known competing financial
appropriate operational limits. While metallic materials continue to interests or personal relationships that could have appeared to influ-
be the dominant choice in naval vessels, composite materials are ence the work reported in this paper.
increasingly prominent with the advances in their manufacturability
allowing for application as both functional and structural materials. References
The UNDEX phenomena is broadly characterised by shock propaga-
[1] Baley C et al. Application of interlaminar tests to marine composite. Appl Compos
tion, FSI (Fluid–structure interaction) dynamics with the concomitant Mater 2004;11(2):99–126.
bubble motion and cavitation and multi‐faceted structural response in [2] Sutherland L. A review of impact testing on marine composite materials: Part I-
the maritime composites. The incident pressures of UNDEX experi- Marine impacts on marine composites. Compos Struct 2018;188:197–208.
[3] Mouritz, A.P., Progress Toward Explosive Blast-Resistant Naval Composites, in
ments, in comparison to air blast, are often orders of magnitude Explosion Blast Response of Composites. 2017.
higher, with the added complexity of high fidelity and high‐speed data [4] Smith, C.S., Design of marine structures in composite materials. 1990: Elsevier
collection in a submerged environment. The flow stress in the marine London.
[5] Ballinger, C.J.C.E.A., Structural FRP composites. 1990. 60(7): p. 63-65.
composite structures can be described through both empirical and
[6] Chalmers D. Experience in design and production of FRP marine structures. Mar
mechanistic constitutive models, which also take into account the Struct 1991;4(2):93–115.
damage evolution. Delamination, matrix cracking, fibre matrix [7] Chalmers, D.J.M.s., The potential for the use of composite materials in marine
debonding, fibre pull‐out and intralaminar cracking are widely observ- structures. 1994. 7(2-5): p. 441-456.
[8] Smith, C. Applications of fibre reinforced composites in marine technology. in
able using both destructive and non‐destructive analysis, as outlined in Compos-Stand Test and Des, Conf Proceeding, National Physics Laboratory,
the previous chapters. Teddington, England, April 8-9, 1974. 1974.
The validity of the modelling technique reviewed is contingent on [9] Mouritz A et al. Review of advanced composite structures for naval ships and
submarines. Compos Struct 2001;53(1):21–42.
appropriate experimental validation. Thus, the current review aimed [10] Nandakumar C. Environmental impact of non metallic hull ships. IEEE; 2012.
to provide a gestalt of the various research elements and to establish [11] Dong, P., et al., A math-based design-for-produceability evaluation of titanium
an analysis pathway between small scale testing and the combat sur- applications in ship hull structures. 2013. 120(1): p. 299-305.
[12] Løseth R, Sekkesæter G, ValsgÔrd S. Economics of high-tensile steel in ship hulls.
vivability of vessels during their in‐service operation. The application Mar Struct 1994;7(1):31–50.
of numerical modelling was found to be a key tool in the understand- [13] Heinz, D., et al., Application of advanced materials for ship
ing of the UNDEX problem and has allowed for accelerated interroga- construction–Experiences and problems. 2000. 51(6): p. 407-412.
[14] Eyres, D.J., Ship construction. 2006: Elsevier.
tion of the phenomena. It is clear that the progress in numerical [15] Job S. Why not composites in ships? Reinf Plast 2015;59(2):90–3.
methods has allowed researchers to accurately simulate the FSI for [16] Gopalakrishnan, S. and Y. Rajapakse, Blast Mitigation Strategies in Marine
both surface and submerged vessels. From the reviewed literature, cur- Composite and Sandwich Structures. 2017: Springer.
[17] Fried N, Graner W. Durability of reinforced-plastic structural materials in marine
rent “state‐of‐the‐art” practices relevant to the numerical analysis of
service. Marine Technol SNAME News 1966;3(03):321–7.
the UNDEX process can be summarised as: [18] Greene, E., Use of fiber reinforced plastics in the marine industry. 1990.
[19] McKenzie AM, Stark H. Progress on naval use of reinforced plastic piping. J Am
• The entire UNDEX process considering detonation, shock propaga- Soc Naval Eng 1953;65(1):57–70.
[20] Johnson H et al. Modelling impact damage in marine composite panels. Int J
tion, and bubble formation can be considered using multi‐physics Impact Eng 2009;36(1):25–39.
FE hydrocodes. The ability to consider multi‐material phase transi- [21] Chen N-Z, Guedes Soares C. Ultimate strength and reliability of composite
material structures. Mar Technol Eng 2011;2:817–40.

25
P. Tran et al. Composite Structures 263 (2021) 113684

[22] Lemiere, Y.J.I., The evolution of composite materials in submarine structures. [59] Lamb, H., XXIII. The early stages of a submarine explosion. The London,
1992: p. 441-449. Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 1923. 45
[23] Huang G, Sun H. Effect of water absorption on the mechanical properties of glass/ (266): p. 257-265.
polyester composites. Mater Des 2007;28(5):1647–50. [60] Butterworth, S., Report on the Theoretical Shape of the Pressure-Time Curve and
[24] Arun, K., et al., Damage characterisation of glass/textile fabric polymer hybrid on the Growth of the Gas Bubble. Compendium on Underwater Explosions
composites in sea water environment. 2010. 31(2): p. 930-939. Research. 1.
[25] PAXHIA, V. LCAC-from test craft to production design. in 7th Marine Systems [61] Hillair H. Experiments on the pressure wave thrown out by submarine
Conference. 1983. explosions. Admiralty: Department of Scientific Research and Experiment; 1919.
[26] Amiruddin, A., et al., Analysis of glass fibre reinforced epoxy composite [62] Glasstone, S. and P.J. Dolan, Effects of nuclear weapons. 1977, Department of
hovercraft hull base. 2008. 29(7): p. 1453-1458. Defense, Washington, DC (USA); Department of Energy, Washington, DC (USA).
[27] Aiman A et al. A study on fiber glass composite hovercraft hull with different wall [63] Keil A. The response of ships to underwater explosions. DTIC Document; 1961.
thickness using generative structural analysis. Int Rev Mech Eng 2012;6 [64] Lyson J. The underwater explosion response of a GRP mine countermeasures
(6):1307–11. vessel. J Naval Sci 1979;5:182–90.
[28] Brown, D., Design considerations for MCMV. 1990. [65] Reid, W.D., et al., Experimental Techniques and Testing of Lightweight Naval
[29] Chalmers, D., R. Osborn, and A. Bunney, Hull construction of MCMVs in the Structures Against Weapons Effects, in Explosion Blast Response of Composites.
United Kingdom. 1984. 2017.
[30] Dixon, R., B. Ramsey, and P. Usher. Design and build of the GRP hull of HMS [66] Rolfe E et al. Failure analysis using X-ray computed tomography of composite
Wilton. in ROYAL institution of naval symposium on GRP ship construction sandwich panels subjected to full-scale blast loading. Compos B Eng
(LONDON). 1972. 2017;129:26–40.
[31] Sjogren, J. and C. Celsing, Swedish development of MCMV-hull design and [67] Gauch E, LeBlanc J, Shukla A. Response of preloaded thin composite panels
production. 1984. subjected to underwater explosive loading. Comput Struct 2012;112:342–53.
[32] Mathijsen D. Now is the time to make the change from metal to composites in [68] LeBlanc J, Gardner N, Shukla A. Effect of polyurea coatings on the response of
naval shipbuilding. Reinf Plast 2016;60(5):289–93. curved E-glass/vinyl ester composite panels to underwater explosive loading.
[33] Lindblom F. Use of composites in the Visby-class stealth corvette. Proceedings of Compos B Eng 2013;44(1):565–74.
the Conference on Marine Composites, 2003. [69] LeBlanc J, Shukla A. Underwater explosive response of submerged, air-backed
[34] Lonno A, Hellbratt S. Use of carbon fibre in a 63M high speed vessel, YS2000, for composite materials: experimental and computational studies. In: Blast
the Swedish Navy. Sandwich Constructions 3, Proceedings of 3rd International Mitigation. Springer; 2014. p. 123–60.
Conference on Sandwich Constructions, 1996. [70] LeBlanc J, Shukla A. Dynamic response and damage evolution in composite
[35] Remy, J., et al., 2006 corvette Z06 carbon fiber structural composite panels- materials subjected to underwater explosive loading: an experimental and
design, manufacturing and material development considerations. 2005: p. 137- computational study. Compos Struct 2010;92(10):2421–30.
141. [71] LeBlanc J, Shukla A. Dynamic response of curved composite panels to underwater
[36] Dabrowski Z, Dziurdz J, Skorski W. Fault Diagnosis of Carbon Fibre Composite explosive loading: experimental and computational comparisons. Compos Struct
Masts. In: Engineering Asset Management. Springer; 2006. p. 990–7. 2011;93(11):3072–81.
[37] Edwards AK et al. Examination of F/A-18 honeycomb composite rudders for [72] LeBlanc J, Shukla A. Response of E-glass/vinyl ester composite panels to
disbond due to water using through-transmission ultrasonics. Ultragarsas underwater explosive loading: Effects of laminate modifications. Int J Impact
“Ultrasound” 2011;66(2):36–44. Eng 2011;38(10):796–803.
[38] Ochoa, O.O., M.M.J.C.S. Salama, and Technology, Offshore composites: [73] LeBlanc J, Shukla A. Underwater explosion response of curved composite plates.
Transition barriers to an enabling technology. Composites Science and Compos Struct 2015;134:716–25.
Technology, 2005. 65(15-16): p. 2588-2596. [74] LeBlanc J, Shukla A. Response of polyurea-coated flat composite plates to
[39] Tamarelle, P. and C. Sparks. High-performance composite tubes for offshore underwater explosive loading. J Compos Mater 2015;49(8):965–80.
applications. in Offshore technology conference. 1987. Offshore Technology [75] Avachat, S. and M. Zhou. Dynamic response of composite sandwich structures
Conference. subjected to underwater impulsive loads: experiments and simulations in 16th
[40] Williams, J.G. Composite material offshore corrosion solutions. in International international conference on composite structures (ICCS-16). 2011. Porto: FEUP.
Workshop on Corrosion Control of Marine Structures and Pipelines. Galveston, [76] Avachat, S. and M. Zhou. Dynamic response of submerged composite sandwich
TX. 1999. structures to blast loading. in IMPLAST. 2010. Providence, Rhode Island, USA.
[41] Yousif B, Ku H. Suitability of using coir fiber/polymeric composite for the design [77] Avachat S, Zhou M. Effect of facesheet thickness on dynamic response of
of liquid storage tanks. Mater Des 2012;36:847–53. composite sandwich plates to underwater impulsive loading. Exp Mech 2012;52
[42] Lopato P, Chady T, Rebello JM. Terahertz time domain inspection of (1):83–93.
composite coatings for corrosion protection. Stud Appl Electromagn Mech [78] Avachat S, Zhou M. Compressive response of sandwich plates to water-based
2011;35:417. impulsive loading. Int J Impact Eng 2016;93:196–210.
[43] Dharsini SP et al. Fea studies on the interfacial behavior of epoxy-cfrp composites. [79] Avachat S, Zhou M. High-speed digital imaging and computational modeling of
J Civ Eng Res 2014;4:26–30. dynamic failure in composite structures subjected to underwater impulsive loads.
[44] Prabhakar K et al. A review of mechanical and tribological behaviour of polymer Int J Impact Eng 2015;77:147–65.
composite materials. IOP Conference Series: Materials Science and [80] Avachat S, Zhou M. Response of cylindrical composite structures to underwater
Engineering. IOP Publishing; 2018. impulsive loading. Procedia Eng 2014;88:69–76.
[45] Allison, D., A. Marchand, and R.J.M.S. Morchat, Fire performance of composite [81] Avachat S, Zhou M. Novel experimental and 3D multiphysics computational
materials in ships and offshore structures. 1991. 4(2): p. 129-140. framework for analyzing deformation and failure of composite laminates
[46] Hale J et al. High temperature failure envelopes for thermosetting composite subjected to water blasts. Int J Impact Eng 2017.
pipes in water. Plast, Rubber Compos 2000;29(10):539–48. [82] Avachat S, Zhou M. Experimental analysis of dynamic deformation and damage
[47] Marsh G. Applying appropriate composites technology. Reinf Plast 2005;49 in composite sandwich structures subjected to underwater impulsive loads.
(9):22–9. Dynamic Behavior of Materials, Volume 1. Springer; 2013. p. 275–86.
[48] Selvaraju S, Ilaiyavel S. Applications of composites in marine industry. J Eng Res [83] Avachat S, Zhou M. High-speed digital imaging and computational modeling of
Stud 2011;2(2):89–91. hybrid metal-composite plates subjected to water-based impulsive loading. Exp
[49] Gu H. Behaviours of glass fibre/unsaturated polyester composites under seawater Mech 2016;56(4):545–67.
environment. Mater Des 2009;30(4):1337–40. [84] Qu T, Avachat S, Zhou M. Response of cylindrical composite structures subjected
[50] Kootsookos A, Mouritz A. Seawater durability of glass-and carbon-polymer to underwater impulsive loading: experimentations and computations. J Eng
composites. Compos Sci Technol 2004;64(10–11):1503–11. Mater Technol 2017;139(2):021020.
[51] Slater, J.J.M.s., Selection of a blast-resistant GRP composite panel design for [85] Schiffer, A. and V. Tagarielli, Observations and Numerical Modeling of the
naval ship structures. 1994. 7(2-5): p. 417-440. Response of Composite Plates to Underwater Blast, in Explosion Blast Response of
[52] Slater J. Selection of a blast-resistant GRP composite panel design for naval ship Composites. 2017.
structures. Mar struct 1994;7(2–5):417–40. [86] Tagarielli, V. and A. Schiffer, The response to underwater blast, in Dynamic
[53] Hall D. Examination of the effects of underwater blasts on sandwich composite Deformation, Damage and Fracture in Composite Materials and Structures, V.V.
structures. Compos Struct 1989;11(2):101–20. Silberschmidt, Editor. 2016, Woodhead Publishing. p. 279.
[54] Diyaroglu C et al. Peridynamic modeling of composite laminates under explosive [87] Schiffer A, Tagarielli V. One-dimensional response of sandwich plates to
loading. Compos Struct 2016;144:14–23. underwater blast: fluid-structure interaction experiments and simulations. Int J
[55] Mouring SE. Composites for naval surface ships. Marine Technology Society Impact Eng 2014;71:34–49.
1998;32(2):41. [88] Schiffer A, Tagarielli V. The one-dimensional response of a water-filled double
[56] Junhou P, Shenoi R. Examination of key aspects defining the performance hull to underwater blast: experiments and simulations. Int J Impact Eng
characteristics of out-of-plane joints in FRP marine structures. Compos A Appl Sci 2014;63:177–87.
Manuf 1996;27(2):89–103. [89] Schiffer A, Tagarielli V. The response of circular composite plates to underwater
[57] Edition SC. Consolidated text of the International Convention of Safety of Life at blast: Experiments and modelling. J Fluids Struct 2015;52:130–44.
Sea, 1974, and its Protocol of 1988: articles, annexes and [90] Schiffer, A. and V.L. Tagarielli. The response of rigid plates to blast in deep water:
certificates. London: IMO; 2004. fluid–structure interaction experiments. in Proc. R. Soc. A. 2012. The Royal
[58] Ramsauer C. Die Massenbewegung des Wassers bei Unterwasserexplosionen. Ann Society.
Phys 1923;377(20):265–84.

26
P. Tran et al. Composite Structures 263 (2021) 113684

[91] Espinosa HD, Lee S, Moldovan N. A novel fluid structure interaction experiment [128] Dutta, P.K., Low-temperature compressive strength of glass-fiber-reinforced
to investigate deformation of structural elements subjected to impulsive loading. polymer composites. 1994.
Exp Mech 2006;46(6):805–24. [129] Gupta S, Shukla A. Blast performance of marine foam core sandwich composites
[92] Latourte F et al. Failure mechanisms in composite panels subjected to underwater at extreme temperatures. Exp Mech 2012;52(9):1521–34.
impulsive loads. J Mech Phys Solids 2011;59(8):1623–46. [130] Pan Z, Gu B, Sun B. Experimental investigation of high-strain rate properties of 3-
[93] Latourte F et al. Design and identification of high performance steel alloys for D braided composite material in cryogenic field. Compos B Eng 2015;77:379–90.
structures subjected to underwater impulsive loading. Int J Solids Struct 2012;49 [131] López-Puente J, Zaera R, Navarro C. The effect of low temperatures on the
(13):1573–87. intermediate and high velocity impact response of CFRPs. Compos B Eng 2002;33
[94] Mori L et al. Deformation and fracture modes of sandwich structures subjected to (8):559–66.
underwater impulsive loads. J Mech Mater Struct 2007;2(10):1981–2006. [132] Liu MB et al. Smoothed particle hydrodynamics for numerical simulation of
[95] Wei X et al. Three-dimensional numerical modeling of composite panels underwater explosion. Comput Mech 2003;30(2):106–18.
subjected to underwater blast. J Mech Phys Solids 2013;61(6):1319–36. [133] Brett JM, Yiannakopolous G. A study of explosive effects in close proximity to a
[96] Mouritz A, Saunders D, Buckley S. The damage and failure of GRP laminates by submerged cylinder. Int J Impact Eng 2008;35(4):206–25.
underwater explosion shock loading. Composites 1994;25(6):431–7. [134] Veldman, R.L., J. Ari-Gur, and C.J. Folkert. Effects of Pre-Pressurization on Plastic
[97] Mouritz, A.P., Underwater Explosive Blast Response of Fiberglass Laminates, in Deformation of Blast-Loaded Square Aluminum Plates. in 8th International LS-
Explosion Blast Response of Composites. 2017. DYNA Conference. 2004. Dearborn, Michigan.
[98] LeBlanc J et al. Near field underwater explosion response of polyurea coated [135] Riley, M., Modeling of Gas Bubble Behaviour and Loading on a Rigid Target due
composite plates. Exp Mech 2016;56(4):569–81. to Close-Proximity Underwater Explosions - Comparison to Test Data Conducted
[99] Pinto M, Gupta S, Shukla A. Study of implosion of carbon/epoxy composite at DRDC Suffield. 2010, Defence R & D Canada - Atlantic.
hollow cylinders using 3-D digital image correlation. Compos Struct [136] Lee, J.M., G. Rude, and G.T. Paulgaard, Loading Regimes for Close-Proximity
2015;119:272–86. Underwater Explosions, in Proceedings of the 81st Shock and Vibration
[100] Pinto M, Gupta S, Shukla A. Hydrostatic implosion of GFRP composite tubes Symposium. 2010: Orlando, FL.
studied by digital image correlation. J Pressure Vessel Technol 2015;137 [137] Riley M et al. Loading on a rigid target from close proximity underwater
(5):051302. explosions. Shock Vib 2012;19(4):555–71.
[101] Pinto M, Shukla A. Dynamic collapse mode evolution in carbon composite tubes. [138] Kalavalapally R, Penmetsa R, Grandhi R. Multidisciplinary optimization of a
Extreme Mech Lett 2015;3:55–8. lightweight torpedo structure subjected to an underwater explosion. Finite Elem
[102] Pinto M, Shukla A. Mitigation of pressure pulses from implosion of hollow Anal Des 2006;43(2):103–11.
composite cylinders. J Compos Mater 2016;50(26):3709–18. [139] Zong Z, Zhao Y, Li H. A numerical study of whole ship structural damage
[103] Pinto M, Shukla A. Shock-initiated buckling of carbon/epoxy composite tubes at resulting from close-in underwater explosion shock. Mar Struct 2013;31:24–43.
sub-critical pressures. Exp Mech 2016;56(4):583–94. [140] Sprague MA, Geers TL. A spectral-element method for modelling cavitation in
[104] Cole, R.H., Underwater explosions. 1948: Princeton university press. transient fluid–structure interaction. Int J Numer Meth Eng 2004;60
[105] International, S., et al. Proceedings of the 18th International Ship and Offshore (15):2467–99.
Structures Congress. Volume 2 Volume 2. [141] Coombs A, Thornhill C. An underwater explosive shock gun. J Fluid Mech
[106] Hall, D. and B. Robson, A review of the design and materials evaluation 1967;29(02):373–83.
programme for the GRP/foam sandwich composite hull of the RAN minehunter. [142] Poche, L.B. and J.F. Zalesak, Development of a Water-Filled Conical Shock Tube
1984. 15(4): p. 266-276. for Shock Testing of Small Sonar Transducers by Simulation of the Test
[107] Hatt, D., et al., Minehunter Inshore Shock Trials. Phase 2. A Limited Shock Trial Conditions for the Heavyweight Test of MIL-S-901D (NAVY). 1992, DTIC
for Developing Trial Methodology and Logistics. 1988, MATERIALS RESEARCH Document.
LABS ASCOT VALE (AUSTRALIA). [143] Chen, W.W. and B. Song, Split Hopkinson (Kolsky) bar: design, testing and
[108] Elischer, P. and J. Howe, Australia’s Shock Testing Capability. Dynamic Loading applications. 2010: Springer Science & Business Media.
in Manufacturing Service: Preprints of Papers, 1993: p. 149. [144] LeBlanc J et al. Shock loading of three-dimensional woven composite materials.
[109] Qiankun, J. and D.J.I.J.o.I.E. Gangyi, A finite element analysis of ship sections Compos Struct 2007;79(3):344–55.
subjected to underwater explosion. 2011. 38(7): p. 558-566. [145] Silva-Muñoz RA, Lopez-Anido RA. Structural health monitoring of marine
[110] Arora H et al. Full-Scale Air and Underwater-Blast Loading of Composite Sandwich composite structural joints using embedded fiber Bragg grating strain sensors.
Panels, in Explosion Blast Response of Composites. Woodhead Publishing; 2017. Compos Struct 2009;89(2):224–34.
[111] Arora H, Hooper P, Dear J. The effects of air and underwater blast on [146] Measures, R.M., Structural monitoring with fiber optic technology. 2001:
composite sandwich panels and tubular laminate structures. Exp Mech 2012;52 Academic Press, Inc.
(1):59–81. [147] Kuang K et al. Damage monitoring in aluminum-foam sandwich structures based
[112] Mouritz A. The effect of underwater explosion shock loading on the fatigue on thermoplastic fibre-metal laminates using fibre Bragg gratings. Compos Sci
behaviour of GRP laminates. Composites 1995;26(1):3–9. Technol 2005;65(11–12):1800–7.
[113] Mouritz A. The damage to stitched GRP laminates by underwater explosion shock [148] Herszberg I et al. Damage assessment and monitoring of composite ship joints.
loading. Compos Sci Technol 1995;55(4):365–74. Compos Struct 2005;67(2):205–16.
[114] Mouritz A. The effect of underwater explosion shock loading on the flexural [149] Li H et al. Health monitoring of marine composite structural joints using fibre
properties of GRP laminates. Int J Impact Eng 1996;18(2):129–39. optic sensors. Compos Struct 2006;75(1–4):321–7.
[115] Javier C, LeBlanc J, Shukla A. Hydrothermally degraded carbon fiber/epoxy [150] Jones R, Galea S. Health monitoring of composite repairs and joints using optical
plates subjected to underwater explosive loading in a fully submerged fibres. Compos Struct 2002;58(3):397–403.
environment. Mar Struct 2020;72:102761. [151] Huang W et al. Dynamic response of circular composite laminates subjected to
[116] Matos H et al. Underwater nearfield blast performance of hydrothermally underwater impulsive loading. Compos A Appl Sci Manuf 2018;109:63–74.
degraded carbon–epoxy composite structures. Multiscale Multidiscip Model [152] Wooh S-C, Wei C. A high-fidelity ultrasonic pulse-echo scheme for detecting
Experim Des 2018;1(1):33–47. delaminations in composite laminates. Compos B Eng 1999;30(5):433–41.
[117] Elamin M, Li B, Tan K. Impact damage of composite sandwich structures in arctic [153] Kerber A et al. Explosive blast damage resistance of three-dimensional textile
condition. Compos Struct 2018;192:422–33. composites. Compos A Appl Sci Manuf 2017;100:170–82.
[118] Kara M et al. Impact behavior of carbon fiber/epoxy composite tubes reinforced [154] Hosur, M., et al., Estimation of impact-induced damage in CFRR laminates
with multi-walled carbon nanotubes at cryogenic environment. Compos B Eng through ultrasonic imaging. 1998. 31(5): p. 359-374.
2018;145:145–54. [155] Zhang Z et al. Vibration-based inverse algorithms for detection of delamination in
[119] Ma H-L et al. Impact properties of glass fiber/epoxy composites at cryogenic composites. Compos Struct 2013;102:226–36.
environment. Compos B Eng 2016;92:210–7. [156] Cawley P, Adams RD. A vibration technique for non-destructive testing of fibre
[120] Salehi-Khojin A et al. The role of temperature on impact properties of composite structures. J Compos Mater 1979;13(2):161–75.
Kevlar/fiberglass composite laminates. Compos B Eng 2006;37(7–8):593–602. [157] Paolozzi A, Peroni I. Detection of debonding damage in a composite plate
[121] Yu YH, Nam S, Lee D. Cryogenic impact resistance of chopped fiber reinforced through natural frequency variations. J Reinf Plast Compos 1990;9(4):369–89.
polyurethane foam. Compos Struct 2015;132:12–9. [158] Li Z et al. Damage evaluation of carbon-fibre reinforced polymer composites
[122] Garcia-Gonzalez D et al. Low temperature effect on impact energy absorption using electromagnetic coupled spiral inductors. Adv Compos Lett 2015;24(3).
capability of PEEK composites. Compos Struct 2015;134:440–9. 096369351502400303.
[123] Taraghi I, Fereidoon A, Taheri-Behrooz F. Low-velocity impact response of woven [159] Salski, B., W. Gwarek, and P. Korpas. Non-destructive testing of carbon-fiber-
Kevlar/epoxy laminated composites reinforced with multi-walled carbon reinforced polymer composites with coupled spiral inductors. in 2014 IEEE MTT-
nanotubes at ambient and low temperatures. Mater Des 2014;53:152–8. S International Microwave Symposium (IMS2014). 2014. IEEE.
[124] Gómez-del Rıo T et al. Damage in CFRPs due to low velocity impact at low [160] Salski B, Gwarek W, Korpas P. Electromagnetic inspection of carbon-fiber-
temperature. Compos B Eng 2005;36(1):41–50. reinforced polymer composites with coupled spiral inductors. IEEE Trans Microw
[125] Li, D.-s., et al., Charpy impact properties and failure mechanism of 3D MWK Theory Tech 2014;62(7):1535–44.
composites at room and cryogenic temperatures. 2014. 62: p. 37-47. [161] Salski B et al. Non-destructive testing of carbon-fibre-reinforced polymer
[126] Roberts S et al. Cryogenic Charpy impact testing of metallic glass matrix materials with a radio-frequency inductive sensor. Compos Struct
composites. Scr Mater 2012;66(5):284–7. 2015;122:104–12.
[127] Shokrieh, M.M., M.A. Torabizadeh, and A. Fereidoon, Dynamic failure behavior [162] Montanini R, Freni F. Non-destructive evaluation of thick glass fiber-reinforced
of glass/epoxy composites under low temperature using Charpy impact test composites by means of optically excited lock-in thermography. Compos A Appl
method. 2011. Sci Manuf 2012;43(11):2075–82.

27
P. Tran et al. Composite Structures 263 (2021) 113684

[163] Montesano J, Fawaz Z, Bougherara H. Non-destructive assessment of the fatigue [198] Bakroon, M., et al., Multi-material arbitrary Lagrangian-Eulerian and coupled
strength and damage progression of satin woven fiber reinforced polymer matrix Eulerian-Lagrangian methods for large deformation geotechnical problems.
composites. Compos B Eng 2015;71:122–30. Numerical Methods in Geotechnical Engineering, 2018. : p. 8.
[164] Genest M et al. Pulsed thermography for non-destructive evaluation and damage [199] Donea J, Giuliani S, Halleux J-P. An arbitrary Lagrangian-Eulerian finite element
growth monitoring of bonded repairs. Compos Struct 2009;88(1):112–20. method for transient dynamic fluid-structure interactions. Comput Meth Appl
[165] Avdelidis N et al. A thermographic comparison study for the assessment of Mech Eng 1982;33(1–3):689–723.
composite patches. Infrared Phys Technol 2004;45(4):291–9. [200] Özarmut, B., et al., Fluid-Composite Structure-Interaction in Underwater Shock
[166] Avdelidis N, Moropoulou A, Riga ZM. The technology of composite patches and Simulations. 2019.
their structural reliability inspection using infrared imaging. Prog Aerosp Sci [201] Trevino, T., Applications of arbitrary lagrangian eulerian (ALE) analysis approach
2003;39(4):317–28. to underwater and air explosion problems. 2000, NAVAL POSTGRADUATE
[167] Gardner N et al. Blast response of sandwich composites: effect of core gradation, SCHOOL MONTEREY CA.
pre-loading, and temperature. In: Blast Mitigation. Springer; 2014. p. 279–330. [202] Wei X et al. A new rate-dependent unidirectional composite model–application to
[168] Rolfe E et al. X-ray CT analysis after blast of composite sandwich panels. Proc Eng panels subjected to underwater blast. J Mech Phys Solids 2013;61(6):1305–18.
2016;167:176–81. [203] Tran P, Ngo TD, Mendis P. Bio-inspired composite structures subjected to
[169] Lambrineas, P., et al., X-ray computed tomography examination of inshore underwater impulsive loading. Comput Mater Sci 2014;82:134–9.
minehunter hull composite material. 1991. 24(4): p. 207-213. [204] Tran P, Ngo TD, Mendis P. Underwater impulsive loading-induced dynamic
[170] Gargano A et al. Effect of fibre-matrix interfacial strength on the explosive blast failures of monolithic composite panel. Applied Mechanics and Materials. Trans
resistance of carbon fibre laminates. Compos Sci Technol 2017;138:68–79. Tech Publ; 2014.
[171] Schilling, P.J., et al., X-ray computed microtomography of internal damage in [205] Bachynski EE, Motley MR, Young YL. Dynamic hydroelastic scaling of the
fiber reinforced polymer matrix composites. 2005. 65(14): p. 2071-2078. underwater shock response of composite marine structures. J Appl Mech 2012;79
[172] Garcea S, Wang Y, Withers P. X-ray computed tomography of polymer (1).
composites. Compos Sci Technol 2018;156:305–19. [206] McShane, G., V. Deshpande, and N. Fleck, The underwater blast resistance of
[173] Bull D et al. Three-dimensional assessment of low velocity impact damage in metallic sandwich beams with prismatic lattice cores. 2006.
particle toughened composite laminates using micro-focus X-ray computed [207] Motley MR, Young YL, Liu Z. Three-dimensional underwater shock response of
tomography and synchrotron radiation laminography. Compos A Appl Sci composite marine structures. J Appl Mech 2011;78(6):061013.
Manuf 2013;52:62–9. [208] Tilbrook M, Deshpande V, Fleck N. Underwater blast loading of sandwich beams:
[174] Adams, R. and P.J.N.i. Cawley, A review of defect types and nondestructive regimes of behaviour. Int J Solids Struct 2009;46(17):3209–21.
testing techniques for composites and bonded joints. 1988. 21(4): p. 208-222. [209] Hayman B. Underwater Explosion Response of Sandwich Structures with
[175] Sutton, M.A., J.J. Orteu, and H. Schreier, Image correlation for shape, motion and Compliant Cores. In: Blast Mitigation Strategies in Marine Composite and
deformation measurements: basic concepts, theory and applications. 2009: Sandwich Structures. Springer; 2018. p. 23–52.
Springer Science & Business Media. [210] Rahaman M, Zheng K, Akimoto H. Numerical investigation of bow slamming on
[176] Gupta, S., et al., Study of dynamic underwater implosion mechanics using digital ships with large flare. Proceedings of MARTEC 2010.
image correlation. Proceedings of the Royal Society A: Mathematical, Physical [211] Rigas F, Sklavounos S. Experimentally validated 3-D simulation of shock waves
Engineering Sciences, 2014. 470(2172): p. 20140576. generated by dense explosives in confined complex geometries. J Hazard Mater
[177] Zhou W et al. Fundamentals of scanning electron microscopy (SEM). In: Scanning 2005;121(1–3):23–30.
microscopy for nanotechnology. Springer; 2006. p. 1–40. [212] Ming F et al. Damage characteristics of ship structures subjected to shockwaves of
[178] Yin J-F, Bai Q, Zhang B. Methods for detection of subsurface damage: a review. underwater contact explosions. Ocean Eng 2016;117:359–82.
Chin J Mech Eng 2018;31(1):41. [213] Xiong-liang Y, Juan C, Zhang A. Computation of the blast loading of two-
[179] McCoy R, Sun C. Fluid-structure interaction analysis of a thick-section composite dimensional underwater explosion using SPH method. J Harbin Eng Univ
cylinder subjected to underwater blast loading. Compos Struct 1997;37 2010;31(10):1303–10.
(1):45–55. [214] Messahel R, Souli M. SPH and ALE formulations for fluid structure coupling.
[180] Langrand B et al. Submarine hull integrity under blast loading. Int J Impact Eng Comput Model Eng Sci 2013;96(6):435–55.
2009;36(8):1070–8. [215] Hung C, Hsu P, Hwang-Fuu J. Elastic shock response of an air-backed plate to
[181] A-man Z, Wen-shan Y, Xiong-liang Y. Numerical simulation of underwater underwater explosion. Int J Impact Eng 2005;31(2):151–68.
contact explosion. Appl Ocean Res 2012;34:10–20. [216] Everstine GC, Henderson FM. Coupled finite element/boundary element
[182] Li J, Rong J-L. Experimental and numerical investigation of the dynamic response approach for fluid–structure interaction. J Acoust Soc Am 1990;87(5):1938–47.
of structures subjected to underwater explosion. Eur J Mech B Fluids [217] Geers TL, Felippa CA. Doubly asymptotic approximations for vibration analysis of
2012;32:59–69. submerged structures. J Acoust Soc Am 1983;73(4):1152–9.
[183] Liu MB et al. Computer simulation of high explosive explosion using smoothed [218] Gong S, Lam K. Transient response of floating composite ship section subjected to
particle hydrodynamics methodology. Comput Fluids 2003;32(3):305–22. underwater shock. Compos Struct 1999;46(1):65–71.
[184] Liu MB, Liu GR, Lam KY. Comparative study of the real and artificial detonation [219] Benson DJ. A new two-dimensional flux-limited shock viscosity for impact
models in underwater explosions. Electron Model 2003;25(2):113–24. calculations. Comput Methods Appl Mech Eng 1991;93(1):39–95.
[185] Shin YS, Santiago LD. Surface ship shock modeling and simulation: two- [220] Fedkiw RP. Coupling an Eulerian fluid calculation to a Lagrangian solid
dimensional analysis. Shock Vib 1998;5(2):129–37. calculation with the ghost fluid method. J Comput Phys 2002;175(1):200–24.
[186] Sanal Z. Nonlinear analysis of pressure vessels: some examples. Int J Press Vessels [221] Wang, G., et al., Investigation of the shock wave propagation characteristics and
Pip 2000;77(12):705–9. cavitation effects of underwater explosion near boundaries. 2014. 46: p. 40-53.
[187] Ramajeyathilagam K, Vendhan CP, Rao VB. Non-linear transient dynamic [222] Marcus MS. A finite-element method applied to the vibration of submerged
response of rectangular plates under shock loading. Int J Impact Eng 2000;24 plates. J Ship Res 1978;22(2).
(10):999–1015. [223] Schiffer A, Tagarielli V. The dynamic response of composite plates to underwater
[188] Radha P, Rajagopalan K. Ultimate strength of submarine pressure hulls blast: theoretical and numerical modelling. Int J Impact Eng 2014;70:1–13.
with failure governed by inelastic buckling. Thin-Walled Struct 2006;44 [224] McMeeking RM et al. An analytic model for the response to water blast of
(3):309–13. unsupported metallic sandwich panels. Int J Solids Struct 2008;45(2):478–96.
[189] Jing, L., et al., Finite element modelling of collapse experiments of ring stiffened [225] Geers TL. Doubly asymptotic approximations for transient motions of submerged
cylinders with simulated corrosion damage, in Warship 2008: Naval Submarines structures. J Acoust Soc Am 1978;64(5):1500–8.
9, R.I.o.N. Architects, Editor. 2008: Glasgow, Scotland, UK. [226] Swegle J et al. An analysis of smoothed particle hydrodynamics. Albuquerque,
[190] Gannon, L., Submarine Pressure Hull Collapse Considering Corrosion and NM (United States): Sandia National Labs; 1994.
Penetrations. 2010, Defence R&D Canada - Atlantic. [227] Schiffer A, Tagarielli V. Underwater blast loading of water-backed sandwich
[191] Swegle JW, Attaway SW. On the feasibility of using Smoothed Particle plates with elastic cores: theoretical modelling and simulations. Int J Impact Eng
Hydrodynamics for underwater explosion calculations. Comput Mech 1995;17 2017;102:62–73.
(3):151–68. [228] Xin C, Gengguang X, Liu K. Numerical simulation of underwater explosion loads.
[192] Zhang N, Zong Z, Zhang W. Dynamic response of a surface ship structure Trans Tianjin Univ 2008;14(1):519–22.
subjected to an underwater explosion bubble. Mar Struct 2014;35:26–44. [229] Kira A, Fujita M, Itoh S. Underwater explosion of spherical explosives. J Mater
[193] Sprague MA, Geers TL. A spectral-element/finite-element analysis of a ship-like Process Technol 1999;85(1–3):64–8.
structure subjected to an underwater explosion. Comput Methods Appl Mech Eng [230] McShane G, Deshpande V, Fleck N. Underwater blast response of free-standing
2006;195(17–18):2149–67. sandwich plates with metallic lattice cores. Int J Impact Eng 2010;37
[194] Xie W, Liu Z, Young YL. Application of a coupled Eulerian-Lagrangian method to (11):1138–49.
simulate interactions between deformable composite structures and compressible [231] Zhang Z et al. Underwater explosion of cylindrical charge near plates: analysis of
multiphase flow. Int J Numer Meth Eng 2009;80(12):1497–519. pressure characteristics and cavitation effects. Int J Impact Eng
[195] Young YL, Liu Z, Xie W. Fluid-structure and shock-bubble interaction effects 2018;121:91–105.
during underwater explosions near composite structures. J Appl Mech 2009;76 [232] Wang L et al. Study on load characteristics of underwater explosion using RKDG-
(5). LS-DGF and BEM. Shock Vib 2015;2015.
[196] Liu Z, Xie W, Young YL. Numerical modeling of complex interactions between [233] Harris P, Verma A, Chakrabarti R. Interaction of an explosion bubble with a fixed
underwater shocks and composite structures. Comput Mech 2009;43(2):239–51. rigid structure. Int J Numer Meth Fluids 1999;29(4):389–96.
[197] Askes, H., et al., An ALE formulation based on spatial and material settings of [234] Klaseboer E, Khoo B, Hung K. Dynamics of an oscillating bubble near a floating
continuum mechanics. Part 2: Classification and applications. 2004. 193(39-41): structure. J Fluids Struct 2005;21(4):395–412.
p. 4223-4245.

28
P. Tran et al. Composite Structures 263 (2021) 113684

[235] Pearson A et al. Bubble interactions near a free surface. Eng Anal Boundary Elem [253] Ryan S, Thaler S. Artificial Neural Networks for Characterizing Whipple Shield
2004;28(4):295–313. Performance. Procedia Engineering, The 12th Hypervelocity Impact Symposium
[236] Wang Q et al. Nonlinear interaction between gas bubble and free surface. Comput 2013;58:31–8.
Fluids 1996;25(7):607–28. [254] Ryan S, Thaler S, Kandanaarachchi S. Machine learning methods for
[237] Zhang A, Yao X, Yu X. The dynamics of three-dimensional underwater explosion predicting the outcome of hypervelocity impact events. Expert Syst Appl
bubble. J Sound Vib 2008;311(3–5):1196–212. 2016;45:23–39.
[238] Abe A et al. Numerical study of underwater explosions and following bubble [255] Bortolan Neto L et al. Rapid assessment of structures subjected to localised blast.
pulses. AIP Conference Proceedings. American Institute of Physics; 2007. 25th International Symposium on Military Aspects of Blast and Shock, 2018.
[239] Barras G et al. Numerical simulation of underwater explosions using an ALE [256] Remennikov AM, Mendis PA. Prediction of airblast loads in complex
method. The pulsating bubble phenomena 2012;41:53–66. environments using artificial neural networks. WIT Press; 2006.
[240] Menon, S., Experimental and Numerical Studies of Underwater Explosions. 1996, [257] Remennikov AM, Rose TA. Predicting the effectiveness of blast wall barriers
GEORGIA INST OF TECH ATLANTA SCHOOL OF AEROSPACE ENGINEERING. using neural networks. Int J Impact Eng 2007;34(12):1907–23.
[241] Szymczak, W.J.A.-P.-P., Computations and experiments of pressure loadings from [258] Bortolan Neto L et al. Rapid mechanical evaluation of quadrangular steel plates
underwater explosion bubbles. 1996. 351: p. 17-28. subjected to localised blast loadings. Int J Impact Eng 2020;137:103461.
[242] Link, R., et al. Cfd modeling of close proximity underwater explosions. in 73rd [259] Bewick B, Flood I, Chen Z. A neural-network model-based engineering tool for
Shock and Vibration Symposium, Newport, USA. 2002. blast wall protection of structures. Int J Protect Struct 2011;2(2):159–76.
[243] Lamb, H., Hydrodynamics. 1924: University Press. [260] Flood I, Bewick BT, Dinan RJ. A New Method for Very Fast Simulation of Blast
[244] Hughes TJ, Liu WK, Zimmermann TK. Lagrangian-Eulerian finite element Wave Propagation in Complex Built Environments. Nottingham University Press;
formulation for incompressible viscous flows. Comput Meth Appl Mech Eng 2010.
1981;29(3):329–49. [261] Flood, I., B.T. Bewick, and E. Rauch, Rapid Simulation of Blast Wave Propagation
[245] Avachat S, Zhou M. Effect of core density on deformation and failure in sandwich in Built Environments Using Coarse-Grain Based Intelligent Modeling Methods:.
composites subjected to underwater impulsive loads. Int J Multiphys 2016;6(3). 2011, Defense Technical Information Center: Fort Belvoir, VA.
[246] Bellur-Ramaswamy R et al. Numerical study of composite panels subjected to [262] Flood I et al. Modeling blast wave propagation using artificial neural network
underwater blasts. Dynamic Behavior of Materials, Volume 1. Springer; 2011. p. methods. Adv Eng Inf 2009;23(4):418–23.
169–70. [263] Ibrahim A, Salim H, Flood I. Damage prediction for RC slabs under near-field
[247] Lemmen P, Meijer G-J, Rasmussen EA. Dynamic behavior of composite ship blasts using artificial neural network. Int J Protect Struct 2011;2(3):315–32.
structures (DYCOSS) failure prediction tool. 70th shock and vibration [264] Stewart LK, Morrill KB. Residual capacity prediction of blast-loaded steel columns
symposium, 1999. using physics-based fast running models. Int J Safety Security Eng 2015;5
[248] Bishop, C.M., Pattern recognition and machine learning. Information science and (4):289–303.
statistics. 2006, New York: Springer. 738. [265] Teferra K et al. Mapping model validation metrics to subject matter expert scores
[249] Vapnik, V., The Nature of Statistical Learning Theory. 2 ed. Information Science for model adequacy assessment. Reliab Eng Syst Saf 2014;132:9–19.
and Statistics. 2000, New York: Springer-Verlag. [266] Zhang M et al. Machine-learning prediction of underwater shock loading on
[250] Grosan, C. and A. Abraham, Intelligent Systems. Intelligent Systems Reference structures. Computation 2019;7(4):58.
Library, ed. J. Kacprzyk and L.C. Jain. Vol. 17. 2011, Berlin, Germany: Springer. [267] Alexander Gargano, Raj Das, Adrian P. Mouritz. Explosive Blast Response of
[251] Ryan, S., S. Kandanaarachchi, and K. Smith-Miles, Support Vector Machines for Marine Sandwich Composites. Advances in Thick Section Composite and
Characterising Whipple Shield Performance. Procedia Engineering, the 13th Sandwich Structures. Springer; 2020. p. 121–44. In press.
Hypervelocity Impact Symposium (HVIS 2015), 2015. 103: p. 522-529. [268] Mouritz Adrian P. Advances in understanding the response of fibre-based
[252] Ryan S, Thaler S. Artificial neural networks for characterising Whipple shield polymer composites to shock waves and explosive blasts. Composites Part A:
performance. Int J Impact Eng 2013;56:61–70. Applied Science and Manufacturing 2019;125. https://doi.org/10.1016/
j.compositesa.2019.105502. In press.

29

You might also like