Professional Documents
Culture Documents
Accepted Manuscript: International Journal of Impact Engineering
Accepted Manuscript: International Journal of Impact Engineering
Author: Wei Huang, Bin Jia, Wei Zhang, Xianglin Huang, Dacheng Li, Peng
Ren
PII: S0734-743X(16)30195-6
DOI: http://dx.doi.org/doi: 10.1016/j.ijimpeng.2016.04.006
Reference: IE 2679
Please cite this article as: Wei Huang, Bin Jia, Wei Zhang, Xianglin Huang, Dacheng Li, Peng
Ren, Dynamic Failure of Clamped Metallic Circular Plates Subjected to Underwater Impulsive
loads, International Journal of Impact Engineering (2016), http://dx.doi.org/doi:
10.1016/j.ijimpeng.2016.04.006.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will
undergo copyediting, typesetting, and review of the resulting proof before it is published in its
final form. Please note that during the production process errors may be discovered which could
affect the content, and all legal disclaimers that apply to the journal pertain.
1 Dynamic Failure of Clamped Metallic Circular Plates Subjected to
3 Wei Huanga,b, Bin Jiaa, Wei Zhanga*, Xianglin Huanga, Dacheng Li, Peng Rena,c
a
4 Hypervelocity Impact Research Center, Harbin Institute of Technology, Harbin 150080, P.R. China
b
5 Department of Mechanical Engineering, Michigan State University, East Lansing, MI 48824,USA
c
6 School of naval architecture & ocean engineering, Jiangsu university of science and technology,
7 Jiangsu, P.R.China
9 Highlights:
11 conducted;
13 plates;
17
18 Abstract: The dynamic response and failure of monolithic metallic plates subjected to
20 the effects of plate thickness, fluid-structure interaction parameter, and patch size of
21 loading area on deformation and failure modes in clamped solid 5A06 aluminum alloy
*
Corresponding author. Tel.: +86 451 86417978 13; fax: +86 451 86402055.
Page 1 of 32
22 plates under air-backed and water-backed loading conditions. The plates are subjected
25 capture the dynamic response of plates to make comparison with postmortem analysis.
26 Depending on the loading rate, the inelastic deformation is the primary failure mode
27 of the plates. The different linear relationships between deflection resistance and
28 applied impulse are identified experimentally, considering the influences of the effects
29 of plate thickness, fluid-structure interaction parameter, and patch size of loading area.
30 The results shows that the effect of loading area is the most influential factor on
31 transverse deflection. The results affirm that the plate under water-backed condition
32 shows a 53% reduction in the maximum plate deflection compared with the plate
34 out to facilitate the advanced study on metallic structures and provides guidance for
35 structural design.
38 1. Introduction
39 Military and civilian ship structures, such as the hull and keel structures, are
40 exposed to various environmental loadings which include high and low temperature
41 extremes, transient impulsive loads, and corrosive sea water. Additionally, the
42 structures are designed to survive from both surface and underwater explosions and
Page 2 of 32
46 attracted a great amount of interest to investigate the dynamic responses. Recently,
47 experimental and theoretical studies on the metallic and composite sandwich plates
48 have been conducted by many researchers [1-14]. Metallic solid and sandwich
55 prismatic diamond core, honeycomb core, and metal foam [4, 8, 9, 16, 17].
56 Constitutive relations have been developed for sandwich structures, accounting for
57 dynamic crush behavior of core and plasticity in constituents by Deshpande [18, 19]
58 and Xue [20]. The deformation of sandwich is divided into three phases: phase I is the
59 fluid-structure interaction which is up to the point of the first cavitation of the fluid;
60 stage II is the core compression until the front and back faces get an equal velocity,
61 and is followed by the bending and stretching of stage III. McShane et al.[16]
62 analyzed the three phases by making comparisons among the fully decoupled model,
64 interaction effect during the deformation of plates. The results indicate that the
66 transmitted momentum by 20–30% due to the continued fluid loading during the
67 whole deformation of sandwich. Fleck and Deshpande [19, 21] examined the
70 research [12] on different sandwich cores was based on the similar analytical model.
3
Page 3 of 32
71 These articles concluded that metallic sandwich structures outperform monolithic
72 plates when the deformation is dominated by bending. However, Schiffer [5, 17]
73 reported that sandwich plates may or may not outperform rigid plates of equal mass in
75 The solid metallic panels are the basic components of significant metallic
76 structures, which have been studied experimentally and theoretically for several
77 decades. Neglecting the elastic effect, Jones [22] studied the rectangular and circular
78 solid metallic panels under different loading conditions and proposed the ‘bound’
79 solution for the structural dynamic response. Considering the bending and shearing,
80 Schiffer [7] developed a model for elastic deformation of composite solid plates
83 circular and rectangular steel plates subjected to blast loads. With the increase of
84 impulsive intensities, failure modes are divided into three phases: mode I, inelastic
85 deformation, which is caused by plastically bending and stretching; mode II, tearing at
86 the supports. Plate stretching is followed by tensile rupture at the supports; mode III,
87 shearing at supports. Shear failure occurs at the supports with negligible plastic
88 deformation in the remainder of the beam. The typical discing and petalling failure
89 modes in impulsively loaded clamped plates were analyzed by Lee and Wierzbicki
90 [24, 25], and the tensile tearing modes were reminiscent mode II of failure modes for
92 investigations into the shear rupture modes (mode III) of impulsively loaded clamped
94 modes of low strength copper plates subjected to underwater blast loads. It was
95 concluded from the micrographs that the local failure mechanism is tensile necking,
4
Page 4 of 32
96 regardless of whether the macroscopic mode is petalling or shear-off. Zamani [28]
97 presented the results of analytical and experimental studies on the response of steel
98 and aluminum circular plates in two different media of air and water. A verified
100 strength, normal deflection, and intensity of impulses. Until now, detailed
101 experimental validation needs to be attempted to provide more correct analysis and
102 numerical predictions, especially in dynamic underwater situations for which the
104 The high strength-weight ratios and high stiffness-weight ratios are the
105 remarkable requirements of ship structures to resist transverse impulsive loads. Light
106 structures, such as sandwich and some aluminum alloy solid panels, outperform the
107 traditional steel plates in terms of these mechanical properties. To investigate the blast
108 resistance and failure modes of the clamped plate as a function of applied impulses,
109 the geometric property and loading configuration is important to the optimal design of
110 vessel structures [4]. Despite recent advances in understanding the dynamic response
111 of solid metallic plates, several issues remain. One is the lack of design relations that
112 quantify the dynamic response as functions of both geometric parameters and load
113 configurations. To obtain such relations, diagnostics which can provide in-situ,
115 Additionally, studies focused on plates which were in contact with water only on one
116 side and with air on the opposite side, but the plates in contact with water on both
117 sides were not considered especially in the experimental studies. The water-backed
119 The objective of this work is to identify the dynamic response of solid metallic
120 panels subjected to underwater impulsive loads experimentally. The focuses of present
5
Page 5 of 32
121 analysis are on understanding the deformation, failure modes and associated
122 mechanisms, and quantifying the blast resistance of panels as functions of plate
124 are conducted under three distinct loading conditions: (1) an air-backed condition,
125 with the plate in contact with water on the impulse side, (2) a water-backed condition,
126 with both sides of the plate in contact with water, and (3) the plates subjected to
127 impulsive loading over a central loading patch, with the loading patch size r=0.7. The
128 results are presented in normalized forms to gain insight into underlying trends that
133 impulsive loads for testing marine structures, a projectile-impact based fluid–structure
134 interaction experimental simulator was designed to measure temporal and spatial
135 evolution and failure of structures, as shown in Fig. 1. A Planar pressure pulse is
136 generated by firing a projectile at a sliding piston. In order to obtain much higher
137 intensity of underwater impulses, the dimensions of the water chamber similar to that
138 used by Zhou [10] and Deshpande [2]. Important features of this setup include the
139 ability to generate pressure waves of a wide range of intensities, the ability to simulate
140 the loading of air-backed, water-backed, changeable loading areas and integrate
142 Fig. 1 shows the fully edge clamped plates under air-backed and water-backed
143 conditions respectively. The shock tube is a 500 mm long cylinder which is
Page 6 of 32
144 horizontally mounted and filled with water. It is made of armor steel and has an inside
145 diameter of 66 mm. A thin piston plate is mounted at the front end and the specimen is
146 located at the rear end. A projectile is accelerated by the gas gun and strikes the piston
147 plate, generating a planar pressure pulse in the shock tube. According to the analysis
148 of Deshpande [2], the mass of the projectile is an important factor affecting the peak
149 pressure and decay time of planer impulse. In order to obtain two different decay
150 times of the impulses, 5mm-thick (0.13kg) and 12mm-thick (0.22kg) projectiles are
151 used respectively. Initial velocities of projectile in the range of 20~220m/s are used to
152 delineate the effect of loading rate on the deformation of the structures. This velocity
153 range corresponds to peak pressures between 10~300MPa which are captured by the
154 pressure transducers [29] mounted at the top of water tube. Monolithic aluminum
156 Air-backed condition: Clamped specimens are tested by using the air-backed
157 shock simulator sketched in Fig. 1 (a). Six equally spaced clearance holes for bolts are
158 drilled into the aluminum plates on a pitch circle of radium 65mm, to clamp the plate
159 onto the end of the water tube. Two 0.5mm-thick annular rubber rings and a
160 5mm-thick annular steel ring are used to ensure that the specimens were edge
161 clamped fully. The effective loading region of clamped air-backed plates has the same
162 radius as the water column in this configuration, R=33mm. The 3D digital imaging
163 correlation method (DIC) is used to capture the dynamic responses of monolithic
164 plates temporally and spatially. Two high-speed cameras Phantom v12.2 are put at the
165 back of the specimen directly, in appropriate degrees, to ensure the error analysis
166 conducted in the calibration is acceptable and obtains accuracy and stability results. In
167 all of the tests, the cameras are ~25º from the axial line of the specimens. The selected
168 frame rate and resolution are 33,000 frames and 384×384 pixels respectively. During
7
Page 7 of 32
169 the calibration and post analysis, the business analysis software ARAMIS is used to
171 Water-backed condition: Unlike the air-backed loading test, both sides of the
172 tested specimens are in contact with water, as shown in Fig. 1 (b). The length of back
173 water column is 150mm. The similar mounting method is used to clamp the plates
174 between the two water tubes which have identical inner radius. Two 0.5mm-thick
175 annular rubber rings with 66mm inner diameter are put at both sides of the monolithic
176 plate to ensure a good seal. The invisibility of the specimens’ dynamic deformation in
177 this condition results in that dynamic response cannot be captured by cameras.
178 Central loading condition: In this case, the plates are subjected to the air-backed
179 impulsive loading over a central loading patch, as shown in Fig. 1 (e). To obtain the
180 loading patch size 0.7, the inner diameter of steel annular ring on the back of
181 specimens is changed from 66mm to 96mm. Meanwhile, the loading area provided by
182 water remains 66mm as presented in the air-backed condition. Additionally, the
183 selected frame rate and resolution are changed to 20,000 frames and 512×512 pixels
184 respectively.
186 As noted in [19, 21, 30, 31], the classical Taylor model [32] is entirely
188 underwater blast waves for a plane wave impacting on a sliding plate, the impulse in
t /
p p0e (1)
190 where p 0 is the peak pressure of the impulse, is the decay time. According to the
191 one-dimensional wave theory revealed by Deshpande [2], the peak pressure and decay
8
Page 8 of 32
192 time are adjusted independently by varying the velocity and mass of projectile
m p (2
p0 cw wv0 ,
wcw )
194 where cw , w are the speed of sound in water and density of water, v0 is the initial
195 velocity of the projectile, mp is the areal mass of the projectile. Regarding the full
196 reflection of the shock wave, the impulse applied on the stationary plate can be
197 calculated as
t /
I0 2 p0e d t 2 p 0 (3)
0
c 1
ln (4)
1
200 where w c w / m f
. The momentum per unit area I tr a n s transmitted into the plate
I tr a n s I 0 (5)
1
202 where .
203 Parameter analysis dictates that the problem under investigation is governed by
I0 h
I , ,h (6)
wcw A R R
205 where A is area of loading, R is radius of loaded region, δ is the normal deflection of
208 Quasi-static tension and high temperature quasi-static tension tests are conducted
9
Page 9 of 32
209 with the plate material. Considering the high strain effect during the impulsive tests,
210 the effect of strain rate is performed for 5A06 aluminum alloy with Hopkinson bars.
211 Fig. 2 shows the geometry for quasi-static and dynamic tensile specimens.
212 These tests provide typical engineering stress-strain curves for different test
213 configurations, as shown in Fig. 3. In Fig. 3(a)~(b), the results are obtained from
214 standard quasi-static tensile tests on a smooth symmetric specimen at room temperature
215 and high temperatures at a nominal strain rate of 1.25×10-3/s on an Instron 5569
216 universal testing machine. As shown in Fig. 3 (a), the specimens are cut from the same
217 5A06 panel in different degrees to ensure the accuracy of tests. Additionally, an
218 extensometer with 25 mm gauge length is used to measure the engineering strain at
219 room temperature. The slight vibration of the curves in room temperature is caused by
220 the clamp of the machine. It is worth noting that 5A06 material presents an obvious
221 temperature softening effect. Fig. 3(c) shows the effect of strain rate hardening of the
222 metallic material using the SHTB. No obvious yield point is found for all of the
223 stress-strain curves, so the flow stress at 0.2% plastic strain is selected as the yield
225 Johnson-Cook constitutive model for aluminum panel: Base on the tests
226 conducted on the materials, the materials herein are identified with obvious strain
227 hardening and high ductility. The Johnson-Cook constitutive model [34], which
228 accounts for the strain-hardening, temperature softening, and strain rate dependence, is
229 used to describe the metallic material’s mechanical response. The equation is described
230 as follow:
(7
( A B e q ) (1 C ln e q ) (1 T * )
n * m
eq )
Page 10 of 32
eq /
*
232 equivalent plastic strain, eq 0 is nondimensional strain rate, eq is equivalent
234 temperature and T, Tr, Tm are current temperature, conference room temperature and
235 melt temperature of 5A06 aluminum respectively. In order to use the Johnson-Cook
236 constitutive model to characterize the aluminum alloys in the further numerical
237 studies, the mechanical parameters are achieved as the fitted Johnson-Cook
240 A parametric study was carried out, focusing on the effects of (i) loads intensity,
241 (ii) thickness of plate, (iii) fluid-structure interaction parameter, (iv)patch size of
242 central loading area, and (v) air-backed and water-backed conditions on dynamic
243 response. The objective is to quantify the transverse deflection and failure mode of
244 plates as functions of loads intensity, geometric thickness and load configuration.
245 Although several input variables were considered, for brevity, this paper focused on
246 the dynamic response of 0.5mm aluminum plates under air-backed loading condition
248 It should be noted that mechanical coupling between the tube and the water
249 column reduces the speed at which pressure pulses propagate in the fluid, cw [35]. In
250 order to quantify these effects, the pressures captured by different transducers at the
251 same tests were used to calculate the speed of sound in the water tube. Measurements
252 provided cw=1106m/s, significantly lower than the speed of sound in open water
253 1500m/s. As the decay time is controlled by the mass and velocity of projectiles,
254 5mm-thick and 12mm-thick projectiles are used to generate two different decay times:
11
Page 11 of 32
255 35us and 58us according to eq.(2). Similar to Fleck and Deshpande [21], we assume
256 that the entire impulse I0 of the shock is transmitted to the front face of the solid plate
257 to simplify the experimental analysis. In the following analysis, normalized impulses
258 according to eq.(6) are used to quantify the incident impulses. In the eq.(6), the
259 measured speed of sound in water 1106m/s and different decay times obtained by
260 different masses of strikers are used to calculate the normalized impulses. As shown
261 in Fig. 4, different pressure histories of impulsive loads are captured by pressure
262 transducers. Fig. 4 shows the pressure histories with two different decay times at
263 different initial velocity of projectiles. The predicted pressure by eq.(1) histories are in
264 good agreement with the histories in the experiments carried out with two distinct
266
268 During the experimental test, two high-speed cameras were used to capture the
270 plate subjected to normalized impulse 0.16 is shown in Fig. 5. In order to observe the
271 change more clearly, the amplitude of deflections for all points is amplified 2 times.
273 hinges emanate from the supports to the mid-span of the plate, showing the shrinking
274 of the central red area at the plots. The midpoint reaches the highest value 8.06mm at
275 0.75ms and falls to 7.71mm after the elastic recovery, as shown in Fig. 6 (a) as well.
276 The sectional deflection profile versus time of the 0.5mm plate is plotted in Fig. 6 (a).
277 The symmetric profiles of deflections caused by the flow of dynamic hinges exhibit
278 the identical process as shown in Fig. 5. It is obvious that the postmortem
12
Page 12 of 32
279 cross-section deformation of the plate shows great agreement with the ultimate profile
281 Fig. 6 (b) shows the transverse deflection history of central points on the 1.0mm
282 plates subjected to four different normalized impulsive intensities, 0.11, 0.24 and 0.25.
283 After the initially linear increase, the mid-point normal deflection remains roughly
284 constant at peak value. The histories of central deflection indicate that the effect of
285 loading rate not only affects the peak deflection, but also the response rate of
286 deflection. The response rate, as well as the maximum normal displacements,
287 increases monotonously with the increasing impulses. The panels exhibit similar rates
288 of response at the initial phase, but the responses of deflections diverge at 0.03ms and
289 0.09ms successively. This suggests that the duration of similar deflection rate
290 increases with the increase of impulsive loading. Additionally, slight elastic recoveries
291 of the deformation shown in the Fig. 5 and Fig. 6 (a) can be also observed in every
293 After strengthening the impulsive intensity, the failure modes of thin metallic
294 plates change from inelastic deformation to tearing and shearing, with crack initiation
295 at the mid-span or periphery of loading area. Three failure modes for metallic plates
296 have been identified by previous studies, which are divided into mode I for plastic
297 deformation with weak impulse, mode II for tearing at support with medium impact
298 intensity, and mode III for shearing failure at high strength impulsive loads. The
299 dynamic responses of inelastic deformation (mode I) are shown in Fig. 5 and Fig. 6,
300 as well as the quantified relation between dynamic deflection and impulsive
301 intensities. Rather than appearing at the support, the initial crack occurs at the
302 midpoint of the plate when the impulse exceeds the limited intensity for mode I.
303 High-speed photographic sequences of the dynamic deformation and crack initiation of
13
Page 13 of 32
304 a 0.5mm solid plate subjected to underwater impulse are shown in Fig. 7. Before the
305 initiation of central tearing, the flow of dynamic plastic hinges within the loading area
306 propagate toward the central region as shown in Fig. 5. When the stretch strain at the
307 mid-span exceeds the fracture strain at t=300µs, the tearing commences at the central
308 point of the plate. Then, the plate fails ultimately by the formation of four petals (as
309 shown in Fig. 8 as mode IIc). With the increasing impact velocity of the projectile,
310 tearing failure initiates at the periphery of the loading area. Within the range of
311 impulsive intensities, different failure modes for plates are presented in Fig. 8, mode I,
312 mode II, and the central tearing failure mode IIc. Fig. 9 shows the in-plane strain
313 histories of the midpoint and 1/4 radius point (near midpoint) at airside of the plates
314 under different tested conditions. The positive strain histories indicate that the failure of
315 the back side face sheets of structures is governed by the axial tension, regardless of the
319 impulsive loads is sensitive to the intensity of impulses. The normalized transverse
321 are shown in Fig. 10 (a). In this case, the thicknesses of pistons and projectiles are
322 12mm and 5mm respectively, which means the decay times are 35µs according to
323 eq.(2). Table 2 lists the results of experiments conducted to probe the effect of
324 thickness. In the table, h , and I are the non-dimensional parameters obtained from
325 eq.(6).
326 For the identical thick panels, the relationship between the normal deflections
327 and impulsive intensities of the plates exhibits a good linear relation. The increasing
14
Page 14 of 32
328 thickness strengthens the impulse resistance of the plates. Three fitting curves in Fig.
329 10 (a) show similar slopes, which indicates that the thickness affects the magnitudes
330 of plastic deformation, but it nearly has no influences on the sensitivity for impulsive
331 intensities. To obtain the effect of thickness on the deflection rate, a comparison of
332 mid-point deflection history between 1.0mm-thick and 1.5mm-thick panels subjected
333 to two different impulses is shown in Fig. 10 (b). The results show that the deflection
334 rate increases with the increasing applied impulses and the thickness of panels, which
335 indicates that the thicker panel experiences longer duration of dynamic response than
336 the thinner panels at the identical applied impulses. When the normalized impulse is
337 0.21, the deflection of the 1.5mm-thick panel moves much more rapidly than the
338 1.0mm-thick panel in the direction away from the impulsive load during the rising
339 time, and all the deflections almost keep parallel eventually. As the loading intensity
340 decreases, the rate of deflection also decreases monotonically. When 1.0mm target
341 and 1.5mm target are subjected to normalized impulses 0.11 and 0.12 respectively, the
342 initial slope of the 1.5mm target is still steeper than that of the thinner plate, even
343 though the thinner one has a much larger deflection after 0.45ms. The result of the
344 comparison shown in Fig. 10 (b) indicates that the effect of thickness promotes the
345 blast resistance of monolithic aluminum plates for it not only reduces the permanent
346 plastic deformation, but also shortens the time of fluid-structure interaction by
349 between performance parameter Z (δ/R), influencing factors x (r, , h ), and load
350 properties y ( I ). A, m, n are constants specific to each load condition. The form is
351 given by
15
Page 15 of 32
(8
m n
Z Ax y
)
353 (δ/L)ab of midpoints for all tests under air-backed conditions as a function of
354 normalized impulse ( I ) and panel thickness (h/R). The transverse deflections
355 increase as the applied impulses increases and decrease as the panel thickness
356 increases. For all impulses, the 1.5mm-thick panel exhibits better deflection resistance
357 than the other panels when subjected to the identical applied impulses. Fitting the
358 experimental data shown in Fig. 11 (b), the relationship between deflection in
359 air-backed plates (δ/R)ab, and normalized incident impulse ( I ) and plate thickness
(9
0 .4 2
( / R ) a b 0 .2 ( h )
0 .8 7
(I )
)
362 Table 3 lists the results of experiments carried out to investigate the effect of
364 discussed in section 2.1, the two distinct fluid-structure interaction parameters
365 controlled by the mass of strikers are 5.3 and 8.7 and the resulting decay times
366 correspond to 35µs and 58µs respectively. The results of the 0.5mm aluminum plates
367 subjected to the underwater impulsive loads with two distinct are shown in Fig.
368 12. The central deflection exhibits a linear increase as the loading intensity increases
369 from 0.18 to 0.35, which shows a similar slope with the plates subjected to impulses
370 with FSI parameters 5.3. Compared with the deflection for =5.3, the deflection for
371 =8.7 is ∼ 41% lower at all range of load magnitudes. When is equal to 8.7,
16
Page 16 of 32
372 petalling failure commences when the plate is subjected to normalized impulse 0.37.
373 The quantified relation caused by Taylor parameter is expressed with the impact
( 10
0 .8 4
( / R ) a b 6 .0 5
1 .0 1
(I )
)
376 The response of clamped monolithic beams to impulsive loading over a central
377 loading patch was first studied by Martin et al. [36]. They concluded that the response
378 of solid beam can be separated into three sequential phases. Two distinct patch sizes,
379 r<0.5 and r>0.5, were studied by Qiu and Deshpande [1], and the analytical model
380 was developed for the response of clamped monolithic plate and sandwich beams
381 subjected to impulses over those two kinds of central loading patches. In this paper,
382 the ratio r>0.5 was investigated experimentally by changing the inner diameter of the
383 annular clamped metallic ring. Loading over the entire span and central span, r=1 and
384 r=0.7, are compared to investigate the effect of the loading patch size, as shown in Fig.
385 1 (e). Experimental results involving in the effect of patch size are listed in Table 4.
386 Fig. 13 (a) shows transverse deflection history of midpoints when the plate is
387 subjected to the central loaded underwater impulses, where r=0.7 and =8.7. After a
388 rapid increase at an initial time of 0.48ms (phase I), the normal deflection remains a
389 constant for about 0.30 ms (phase II) followed by the peak deflection (phase III).
390 Compared with the central deflection shown in Fig. 6 (b), a new response phase
391 (phase II) is observed when the plates are subjected to impulses over a central loading
392 patch (r=0.7). The propagation of dynamic plastic hinges can explain the discrepancy
393 between those two configurations. Initial plastic hinges emanate from the edge of
17
Page 17 of 32
394 loading and opposing propagate to plate boundary and midpoint (phase I). When the
395 patch size r>0.5, the dynamic plastic hinges reach the plate boundary before arriving
396 at the midpoint. Meanwhile, the central segment moves at a constant velocity resulted
397 from the initial impact (phase II). The plastic wave continues to propagate until it
398 reaches the midpoint of the plate and results in the maximum normal plastic
399 deflection (phase III) followed by a slight elastic recovery (phase IV). The
400 propagation of the plastic wave can be observed more obviously in Fig. 14. Failure
401 modes shown by the central loaded plate also present the difference between the two
402 distinct patch size configurations, as shown in Fig. 13 (b). The ultimate deformation
403 of central loaded plate is a dome-shape profile with smaller curvature at the mid-span
404 but larger curvature near the plate boundary than that of the fully loaded plates (r=1).
405 Transverse deflection of the monolithic plate is sensitive to the patch size of
406 loading area. When the plate is subjected to centrally loaded impulse, the amplitude of
407 plastic deformation is much larger than that of panel subjected to fully loaded impulse
408 with identical intensity, as shown in Fig. 15. As the impulse increases, the transverse
409 deflection of the midpoint grows linearly, and tearing failure occurs as the normalized
410 impulse reaches 0.35. For the 0.5mm plates, the normal deflection is ~62% larger than
411 that of panel subjected to fully loaded impulses with identical intensity. Compared to
412 the normal deflection of 0.5mm plates with 1.0mm plates subjected to loads over
413 central patch r=0.7 and r=1, an obvious difference can be observed from Fig. 15 and
414 Fig. 10 (a). When the plate is fully loaded, the thicker plate only improves the
415 deflection resistance ~20%. However, this improvement increases ~51% when the
416 two kinds of plate are subjected to impulsive loads over central patch r=0.7. The
417 relationship between normalized transverse deflection, patch size and impulse can be
418 given by
18
Page 18 of 32
(1
1 .9 9
( / R ) a b 0 .9 5 r
0 .9 9
(I ) 1)
421 structures, such as keels, rudders, propeller blades and underwater pipelines. A
422 comparison of results between air-backed and water-backed panels reveals significant
424 impulse to the supports, the water-backed plate transmits most of the impulse into the
425 backside water by deformation. The presence of water on both sides of plates prevents
426 large scale bending and absorbs much more initial energy by the backside water
427 column. To quantify the differences under these two conditions, a comparative study
429 In the water-backed condition, another 150mm long water tube is horizontally
430 mounted at the backside, as shown in Fig. 1 (b). In this case, the 12mm-thick (0.22kg)
431 projectiles are selected, which means that the decay time θ and fluid-structure
432 interaction parameter Ψ are θ=58µs and Ψ=8.7 respectively. The results for
434 Fig. 16 shows the central deflection of the water-backed plate with thicknesses of
435 0.5mm and 1.0mm subjected to underwater impulsive loads. Due to the resistance of
436 backside water, the overall deflections of plates are limited significantly. Comparing
437 the central deflections under air-backed and water-backed condition, the 0.5mm-thick
438 panels undergo ~53% lower central deflections under water-backed condition than
439 that of the same thick panels under air-backed condition. As the loading intensity
440 increases from I =0.20 to I =0.45, the trends of deflection for both 0.5mm-thick and
19
Page 19 of 32
441 1.0mm-thick plates exhibit linear relationships. Under water-backed condition, the
443 small at lower impulsive intensity, but it increases rapidly with the increase of
444 impulses. The relationship among normalized deflection under water-backed plates,
1 .2 5 ( 12 )
( / L ) w b 0 .0 2 ( h )
2 .6 4
(I )
446 The presence of water not only restricts the transverse deflection but also affects
447 the failure mechanism of the clamped plates. Fig. 17 exhibits a cross-section profile of
448 the 0.5mm panel subjected to normalized impulse 0.28 under water-backed condition,
449 which differs from the sectional deflection profiles under air-backed condition
451 circumferential bulge in the direction away from the impulsive loading at the plate
452 support area and a dented deformation at the circumferential boundary of loads. The
453 central area presents a conical deformation which is different from all of the deform
454 profiles presented in the air backed tests. The slightly corrugated-shape deformation is
455 observed at all postmortem images obtained from water-backed tests. Once the panel
456 is subjected to impulse, the incident energy is converted into the deformation energy
457 in the plate and then transmits most of the energy into backside water in a short
458 amount of time. The plate compresses the backside water by transverse deformation at
459 the initial time. Due to the incompressibility of the water, the compressive wave
460 transmitted from the plate impinges on the plate again after it is reflected from the
461 boundary of the backside water tube, to make an overall reverse deflection on the
462 plates. The process of deformation and reverse deformation of plates lead to the
20
Page 20 of 32
464 4. Concluding remarks
465 Dynamic response and failure of 5A06 aluminum alloy solid plates subjected to
466 water-based impulsive loadings have been evaluated experimentally in this paper. The
467 effect of panel thickness, fluid-structure interaction parameter, loading patch size and
468 intensity of underwater impulse on the dynamic deformation, failure modes and
469 associated mechanisms of the solid plate in air-backed and water-backed conditions
471 This study has yielded experimental data on the dynamic response of blast
473 dynamic deformation of solid plates indicates that thickness of panels contributes
474 limited influences on the response of plates. The response rate increases with the
475 increase of applied impulses and thickness of panels. In the central loading condition
476 (patch size r=0.7), an additional response phase is exhibited by the panels due to the
477 different propagations of dynamic plastic hinges, which is the primary factor on distinct
478 profile of deformation. The failure modes of solid panels includes inelastic deformation,
479 tearing at central region and tearing at supports with the increasing impulsive loads.
481 considering the effect of plate thickness, Taylor’s FSI parameter, and patch size of
482 loading under the air-backed condition. The parallel linear relations for different
483 thicknesses indicate that the transverse deflections increase as the applied impulses
484 increases and decrease as the panel thickness increases. The thickness of panel
485 influences only the amplitudes of deflection rather than the sensitivity to impulsive
486 intensities. The loading patch size involved in this paper is verified as the most
487 sensitive effect on the deflection resistance of panels. In the central loading condition,
21
Page 21 of 32
488 the peak deflection increases ~62% for 0.5mm panels, and discrepancy between
489 0.5mm-thick and 1.0mm-thick panels has been amplified from 20% to 51%. The
491 fluid-structure interaction parameter increases from 5.3 to 8.7, e.g. the deflections of
493 Due to the bending resistance of backside water, solid panels experience much
494 lower transverse deflection under water-backed condition than the panels under
495 air-backed condition, e.g. the deflections of 0.5mm panels decrease ~53% under
496 water-backed condition. The discrepancy of deflections between 0.5mm and 1.0mm
497 shows an increasing trend within the range of impulses rather than the almost parallel
498 relation under air-backed conditions. The cross-section profile of the plate subjected
500 the periphery of loads and a conical deformation at the mid-point, which are different
504 parameter and patch size of loading, are investigated respectively, as shown in Table 6.
505 The insight obtained in this paper provides guidelines for the design of metallic
507 consideration. Finally, it should be noted that the relationships described in this paper
508 are applicable considering the structural attributes and loading conditions.
509 Acknowledgments
510 The authors would like to thank the National Natural Science Foundation of China
Page 22 of 32
512 References
513 [1] X. Qiu, V.S. Deshpande, N.A. Fleck, Impulsive loading of clamped monolithic and sandwich
514 beams over a central patch, Journal of the Mechanics and Physics of solids, 53 (2005)
515 1015–1046.
516 [2] V.S. Deshpande, A. Heaver, N.A. Fleck, An underwater shock simulator, Proc. R. Soc. Lon.,
517 Ser-A 462 (2006) 1021–1041.
518 [3] Z. Xue, J.W. Hutchinson, Preliminary assessment of sandwich plates subject to blast loads, Int. J.
519 Mech. Sci. , 45 (2003) 687–705
520 [4] S. Avachat, M. Zhou, Response of submerged metallic sandwich structures to underwater
521 impulsive loads, Journal of Mechanics of Materials and Structures, 10 (2015 ) 17–41.
522 [5] A. Schiffer, V.L. Tagarielli, The response of rigid plates to blast in deep water: fluid–structure
523 interaction experiments, Proceedings of the royal sociaty: A, (2014).
524 [6] A. Schiffer, V.L. Tagarielli, The one-dimensional response of a water-filled double hull to
525 underwater blast: experiments and simulations, International Journal of Impact Engineering,
526 63 (2014) 177–187.
527 [7] A. Schiffer, V.L. Tagarielli, The dynamic response of composite plates to underwater blast:
528 theoretical and numerical modelling, International Journal of Impact Engineering, 70 (2014)
529 1–13.
530 [8] K.P. Dharmasena, H.N.G. Wadley, K. Williams, Z. Xue, J.W. Hutchinson, Response of metallic
531 pyramidal lattice core sandwich panels to high intensity impulsive loading in air, Int. J. Impact
532 Eng., 38 (2011) 275–289.
533 [9] K.P. Dharmasena, D.T. Queheillalt, H.N.G. Wadley, P. Dudt, Y. Chen, D. Knight, A.G. Evans,
534 V.S. Deshpande, Dynamic compression of metallic sandwich structures during planar
535 impulsive loading in water, Eur.J. Mech. A Solids, 29 (2010) 56–67.
536 [10] S. Avachat, M. Zhou, High-speed digital imaging and computational modeling of dynamic
537 failure in composite structures subjected to underwater impulsive loads, International Journal
538 of Impact Engineering, 77 (2015) 147-165.
539 [11] Y.P. Zhao, Suggestion of a new dimensionless number for dynamic plastic response of beams
540 and plates, Archive of Applied Mechanics, 68 (1998) 524-538.
541 [12] N. Jones, Structural Impact, Cambridge University Press,UK, Edition 2, ISBN:
542 9781107010963,, (2012) 1-604.
543 [13] N. G.N., M. J.B, Deformation of thin plates subjected to impulsive loading - a review. Part II:
544 Experimental Studies, Int J Impact Eng,, 8 (1989) 171-186.
545 [14] G.N.Nurick, J.B.Martin, Deformation of thin plates subjected to impulsive loading - a review.
546 Part I: Theoretical considerations, Int J Impact Eng,, 8 (1989) 159-169.
547 [15] H.D. Espinosa, S. Lee, N. Moldovan, A novel fluid structure interaction experiment to
548 investigate deformation of structural elements subjected to impulsive loading, Exp. Mech., 46
549 (2006) 805–824.
550 [16] G.J. McShane, V.S. Deshpande, N.A. Fleck, The Underwater Blast Resistance of Metallic
551 Sandwich Beams With Prismatic Lattice Cores, J. Appl. Mech. (ASME), 74 (2006) 352–364.
552 [17] A. Schiffer, V.L. Tagarielli, One-dimensional response of sandwich plates to underwater blast:
553 Fluid-structure interaction experiments and simulations, International Journal of Impact
554 Engineering 71 (2014) 34-49
555 [18] V.S. Deshpande, N.A. Fleck, Isotropic constitutive models for metallic foams, Journal of the
556 Mechanics and Physics of Solids, 48 (2000) 1253-1283.
557 [19] V.S. Deshpande, N.A. Fleck, One-dimensional response of sandwich plates to underwater
558 shock loading, J. Mech. Phys. Solids, 53 (2005) 2347–2383.
559 [20] Z. Xue, J.W. Hutchinson, Constitutive model for quasi-static deformation of metallic sandwich
560 cores, Int. J. Numer. Methods Eng. , 61 (2004) 2205–2238.
561 [21] N.A. Fleck, and Deshpande, V. S., The Resistance of Clamped Sandwich Beams to Shock
562 Loading, ASME J. Appl. Mech., 71 (2004) 386–401.
563 [22] N. Jones, Structural impact, Cambridge university press, 2011.
564 [23] G.N. Nurick, G.C. Shave, The deformation and tearing of thin square plates subjected to
565 impulsive loads- an experimental study, Int. J. Mech. Sci. , 18 (2000) 99–116.
23
Page 23 of 32
566 [24] Y.W. Lee, T. Wierzbicki, Fracture prediction of thin plates under localized impulsive loading.
567 part II: discing and petalling, Int. J. Impact Eng., 31 (2005b) 1277–1308.
568 [25] Y.W. Lee, T. Wierzbick, Fracture prediction of thin plates under localized impulsive
569 loading.part I: dishing, Int. J. Impact Eng. , 31 ( 2005a) 1253–1276.
570 [26] V.H. Balden, G.N. Nurick, Numerical simulation of the post-failure motion of steel plates
571 subjected to blast loading, Int. J. Impact Eng., 32 (2005) 14–34.
572 [27] S. Kazemahvazi, D. Radford, V.S. Deshpande, N.A. Fleck, Dynamic failure of clamped
573 circular plates subjected to an underwater shock Journal of the Mechanics of Materials and
574 Structures, 2 (2007) 2007–2023.
575 [28] J. Zamani, K.H. Safari, A.K. Ghamsari, A. Zamiri, Experimental analysis of clamped AA5010
576 and steel plates subjected to blast loading and underwater explosion, J. Strain Analysis, 46
577 (2010) 201-212.
578 [29] W. Huang, W. Zhang, P. Ren, Z.T. Guo, N. Ye, D.C. Li, Y.B. Gao, An Experimental
579 Investigation of Water-Filled Tank Subjected to Horizontal High Speed Impact, Experimental
580 Mechanics, 55 (2015) 1123–1138.
581 [30] Z. Xue, J.W. Hutchinson, A Comparative Study of ImpulseResistant Metallic Sandwich Plates,
582 Int. J. Impact Eng., 30 (2004) 1283–1305.
583 [31] Y. Liang, A.V. Spuskanyuk, S.E. Flores, D.R. Hayhurst, J.W. Hutchinson, R.M. McMeeking,
584 A.G. Evans, The response of metallic sandwich panels to water blast, J Appl Mech, 74 (2007)
585 81–99.
586 [32] G.I. Taylor, Scientific papers, III: Aerodynamics and the mechanics of projectiles and
587 explosions, edited by G. K., Batchelor, Cambridge University Press, (1963).
588 [33] A. Schiffer, V.L. Tagarielli, The response of circular composite plates to underwater blast:
589 Experiments and modelling, Journal of Fluids and StructuresVolume 52 ( 2015) 130–144.
590 [34] G.R. Johnson, W.H. Cook, Fracture characteristics of three metals subjected to various strains,
591 strain rates, temperatures and pressures, Eng. Fract. Mech., 21 (1985) 31–48.
592 [35] D.J. Korteweg, Uber die Fortpflanzungsgeschwindigkeit des Schalles in elastischen Rohren,
593 Annals of Physics 5(1878.) 525–542.
594 [36] J.B.Martin, P.S. Symmonds, Mode approximations for impulsively-loaded rigid–plastic
595 structures, J. Eng. Mech.: ASCE EM5, (1966) 43–66.
596
597
Pressure Pressure Pressure
transducer Simulator tube transducer transducer Piston
(a) Camera Specimen Projectile Gas
gun
Light 66mm
source
Water
Camera
500mm
Pressure Pressure Pressure
transducer Simulator tube transducer transducer Piston
Projectile
(b)
Water Water
150mm 500mm
Annular Ring
(c)
33mm 48mm 48mm
33mm
Impulse 80mm 80mm
Clamped
Bolt Edge clamped region
24
Page 24 of 32
599 Fig. 1.Sketch of the apparatus employed in this study to explore the response of (a) air-backed
600 condition, (b) water-backed condition, (c) side view of clamped panel: (d) specimen in air-backed
601 condition (e) specimen in central patch r=0.7 loading condition.
602
604 Fig. 2. Geometry of the smooth specimen and post-test photographs for uniaxial tensile test (a)
605 room and high temperature quasi-static tensile specimens and (b) dynamic tensile specimens
606
400 400
(a) (b) o
20 C
o
100 C
Engineering stress (MPa)
Engineering Stress (MPa)
o
300 300 150 C
o
200 C
o o
0 250 C
200 o
200
0
o
45
o
100 45 100
o
90
o
90
0 0
0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.16 0.32 0.48 0.64
500 (c)
Engineering stress (MPa)
400
300
-3
200 1.2510 /s
713.72/s
1493.85/s
100 1733.31/s
1968.35/s
fitting curve
0
0.00 0.03 0.06 0.09 0.12 0.15
Engineerig Strain
608
25
Page 25 of 32
120 250
Reservior pressure 0.6 MPa (a) Reservior pressure 2.0 MPa (b)
Projectile velocity 75.8m/s Projectile velocity 147.5m/s
200
Pressure (MPa) 90 I=0.13 s I=0.24 s
Pressure (MPa)
150
60 Experiment
Experiment 100 Prediction
Prediction
30 50
0
0
-50
300
90
Reservior pressure 0.5 MPa (c) Reservior pressure 3.0 MPa
(d)
Projectile velocity 64.6m/s 250
Projectile velocity 162.6m/s
I=0.17 s I=0.44 s
Pressure (MPa)
200
Pressure (MPa)
60
150
0
0
-50
0 100 200 300 400 500 0 100 200 300 400 500
Time (s) Time (s)
613
614 Fig. 4. Experimentally measured and predicted pressure histories for four different projectile
615 velocities,(a)(b) decay time was 35µs ,(c)(d) decay time was 58µs.
616
617
618 Fig. 5. Evolution of normal deformation for 0.5mm monolithic aluminum plate subjected to
619 normalized impulse 0.16.
620
26
Page 26 of 32
9.0 11.0
dynamic deformation 0.75ms
post-test result I=0.25
Normal deflection (mm)
7.5
1.5 2.2
0.09ms
0.0 (a) (b)
0.0
0 10 20 30 40 50 60 0.0 0.3 0.6 0.9 1.2
Cross-section profile (mm) Time (ms)
621
622 Fig. 6. (a) Deflection profile evolution of and post-tested profile for 0.5mm plate, I =0.16, θ=35µs,
623 and (b) normal deformation history of midpoints, 1.0mm plates.
624
0 us 120us 180us
628
27
Page 27 of 32
629 Fig. 8. Typical failure modes of solid metallic plates.
4.5 6
4.0
(a) (b)
5
3.5
3.0 4
( %)
2.5
( %)
3
2.0
1.5 2
1.0
1
0.5
0.0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Time(ms) Time(ms)
630
631 Fig. 9 Strain histories at different points on the back side of plates, (a) h=0.5mm, normalized
632 impulse 0.16 with patch size r=1.0, and (b) h=1.0mm, normalized impulse 0.39 with patch size
633 r=0.7.
634
0.35
9.0
(a) I=0.21
(b)
Normalized deflection (L)
0.30
7.5
Transverse deflection
I=0.21
0.25
6.0
I=0.11
0.20
4.5
0.15 I=0.12
3.0
0.5mm
0.10 1.0mm
1.5mm 1.5 1.0mm
fitting curve
0.05
1.5mm
0.0
0.05 0.10 0.15 0.20 0.25 0.30 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Normalized impulsive (I) Time (ms)
635
636 Fig. 10 (a)Transverse deflections versus applied impulses, and (b) a comparison of mid-point
637 deflection rate.
638
639 Fig. 11 Normalized deflection as a function of normalized impulses and panel thickness, (a) map of
640 load-structure-performance, and (b) 3D bar picture of experimental data and fitting surface.
641
28
Page 28 of 32
0.4
crack
Normalized deflection (L)
0.3
0.2
s,
s,
(a) fitting curve
0.1
0.1 0.2 0.3 0.4
Normalized Impulse (I)
642
643 Fig. 12 Normalized deflection as a function of normalized impulses and fluid-structure interaction
644 parameter, (a) linear relations of load-structure-performance, and (b) 3D bar picture of
645 experimental data and fitting surface.
646
h=1.0mm, I =0.39
20
18
phase I
Transverse deflection
16
14
12 phase IV
phase II phase III
10
Central loaded plate, r=0.7
8
1.0mm,I=0.45
6 1.0mm,I=0.39
4
2
(a)
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Entire loaded plate, r=1
Time (ms) (b)
647
29
Page 29 of 32
648 Fig. 13. (a)mid-points transvers deflection history of the central loaded plates, and (b) a comparison
649 of deflection profile between r=0.7 and r=1 loading conditions.
650
651 Fig. 14. Evolution of normal deformation for 1.0mm central loaded plate subjected to normalized
652 impulse 0.39.
653
0.7
0.5mm,r=1.0
0.5mm,r=0.7 crack,I=0.35
Normalized deflection (R)
0.6
1.0mm,r=0.7
0.5 fitting curve
0.4
crack,I=0.37
0.3
0.2
0.1 (a)
0.00 0.10 0.20 0.30 0.40
Normalized impulse (I)
654
655 Fig. 15 Normalized deflection as a function of normalized impulses and loading patch size r, (a)
656 linear relations of load-structure-performance, and (b) 3D bar picture of experimental data and
657 fitting surface.
658
0.45
Normalized deflection (L)
0.5mm,air-backed
0.5mm,water-backed
1.0mm,water-backed
0.36 fitting curve -
crack,I=0.20
0.27
0.18
0.09
(a)
0.00
0.18 0.28 0.37 0.46
Normalized Impulsive(I)
659
30
Page 30 of 32
660 Fig. 16. Normalized deflection as a function of normalized impulses and normalozed thickness
661 under water-backed condition, (a) a comparison of linear relations of load-structure-performance in
662 two different loading conditions, and (b) 3D bar picture of experimental data and fitting surface.
663
15
water-backed
0
Front side Back side -50 -40 -30 -20 -10 0 10 20 30 40 50
(a) (b) Cross-section profile (mm)
664
665 Fig. 17. The postmertem images of 0.5mm panel subjected to I =0.28 under water-backed
666 condition, (a) failure modes of the panel, and (b) a comparison of post-test deflection of 0.5mm
667 panels subjected to identical impulse under two loading conditions.
668
669 Table 1. Mechanical properties of 5A06 aluminum alloy
Material Young’s Density A B C n m
modulus (kg/m3) (MPa) (MPa)
(GPa)
5A06 74 2780 167.0 458.7 0.44 0.02 2.3
670
671 Table 2 Results of experiments for effect of thickness
Contact Normalized Decay Initial Normalized Normalized Failure
condition thickness time velocity impulse deflection mode
h (µs) (m/s) I
air-acked,r=1.0 0.015 35 68.4 0.11 0.19 I
35 96.8 0.16 0.23 I
35 137.6 0.22 0.31 I
35 101.6 0.16 0.23 I
air-acked,r=1.0 0.030 35 70.8 0.11 0.15 I
35 158.5 0.25 0.29 I
35 129.3 0.21 0.23 I
35 147.9 0.24 0.24 I
air-acked,r=1.0 0.045 35 58.7 0.07 0.05 I
35 136.8 0.22 0.17 I
35 128.9 0.21 0.21 I
35 76.2 0.12 0.10 I
672
673
674 Table 3 Results of experiments for effect of thickness fluid-structure interaction parameter
Contact Normalized Decay Initial Normalized Normalized Failure
condition thickness time velocity impulse deflection mode
31
Page 31 of 32
h (µs) (m/s) I
air-bcked,r=1.0 0.015 58 140.5 0.37 failed IIc
58 102.0 0.27 0.29 I
58 132.6 0.36 0.35 I
58 76.8 0.19 0.14 I
675
676 Table 4 Results of experiments for effect of patch size
Contact condition Normalized Decay Initial Normalized Normalized Failure
thickness time velocity impulse deflection mode
h (µs) (m/s) I
air-acked,r=0.7 0.015 58 75.2 0.20 0.41 I
58 30.9 0.04 0.10 I
58 107.8 0.29 0.57 I
58 127.5 0.35 failed II
air-acked,r=0.7 0.030 58 104.4 0.28 0.39 I
58 166.9 0.45 0.49 I
58 145.5 0.39 0.45 I
58 119.0 0.32 0.36 I
58 76.2 0.21 0.14 I
677
678 Table 5 Results of experiments for effect of water-backed configuration
Contact Normalized Decay Initial Normalized Normalized Failure
condition thickness time velocity impulse deflection mode
h (µs) (m/s) I
water-backed 0.015 58 104.6 0.28 0.16 I
58 116.6 0.31 0.20 I
58 77.2 0.20 0.04 I
58 105.8 0.28 0.15 I
0.030 58 102.0 0.27 0.09 I
58 146.5 0.39 0.15 I
58 168.1 0.44 0.21 I
58 80.5 0.21 0.03 I
679
680 Table 6 Summary of material-structure-property relationships for clamped aluminum plates.
Figure Contact condition Effect factor Structure-performance relations
Fig. 11 air-backed h ( / R ) a b 0 .2 ( h )
0 .4 2
(I )
0 .8 7
Fig. 12(b) Ψ ( / R ) a b 6 .0 5
0 .8 4
(I )
1 .0 1
Fig. 15(b) r 1 .9 9
( / R ) a b 0 .9 5 r
0 .9 9
(I )
Fig. 16(b) water-backed 1 .2 5
( / L ) w b 0 .0 2 ( h )
2 .6 4
h (I )
681
682
32
Page 32 of 32