Sunteți pe pagina 1din 13

Cent. Eur. J. Eng.

• 2(4) • 2012 • 496-508


DOI: 10.2478/s13531-012-0023-3

Central European Journal of Engineering

On pulse pressure loading of plates with holes


Research Article

Graham K. Schleyer1 , Nicholas. J. Underwood1 , Hyung Min Do2 , Jeom Kee Paik3∗ , Bong Ju Kim3

1 School of Engineering, Impact Research Centre, University of Liverpool, UK


2 Korean Register, South Korea
3 PNU-LR International Joint R&D Centre, Pusan National University, Pusan, South Korea

Received 02 April 2012; accepted 06 June 2012

Abstract: This paper reports the application of an energy solution to a complex problem involving large inelastic deformation
in thin, clamped ductile square plates with either square or circular holes under the action of transverse pulse
pressure loading. The work is part of a collaborative project to study blast loading of steel plates with penetrations
as used for deck plating or bulkheads that may be required to resist loading far in excess of their design limit due
to the effects of an accidental explosion. A novel differential pressure loading device was used to impart dynamic
loading to 1/8 scaled 0.5 m, 1.1 mm thick, clamped mild steel square plates with a central aperture up to 4% of the
plate area. This data was used to validate the energy approach that considers both plastic hinge formation and
extensional effects. Accounting for strain rate, scaling and dynamic effects in the tests gave more than acceptable
results when compared with final deflections in the tested plates. It is concluded that the energy approach together
with small-scale test validation paves the way for a versatile robust design methodology which can be used to
advantage for screening purposes and/or early stage conceptual design studies.
Keywords: Extensional mode • Energy methods • Thin ductile plates • Pulse pressure loading • Strain rate effects • Scaling
effects
© Versita sp. z o.o.

1. Introduction wider remit of this collaboration is to develop research


programmes in the area of response of structures under
blast and impact loading for the benefit of academia and
A project has been launched to study blast loading of steel industry concerned with oil and gas safety.
plates as part of a new collaboration between the PNU- The purpose of the study reported in the paper is to con-
LR International Joint R&D Centre at Pusan National sider the effect of penetrations in plated deck structures
University, Korea and the Impact Research Centre at the found on offshore topsides and ship bulkheads so as to
University of Liverpool, School of Engineering, UK. Both provide improved limit states assessment criteria to indus-
centres have an international reputation for conducting try on the design of deck structures under blast loading
world-class research over many years in structural impact [1–4].
and limit-state design of steel-plated structures. The The study utilizes a differential pressure device developed
at the University of Liverpool Impact Research Centre
capable of producing repeatable uniform transient pulse

E-mail: jeompaik@pusan.ac.kr pressure loading on 0.5 m by 0.5 m mild steel plate speci-

496
G.K. Schleyer, N.J. Underwood, H.M. Do, J.K. Paik, B.J. Kim

mens nominally 1.1 mm thick corresponding to a scale of


approximately 1:8. The facility is particularly efficient for
experimental studies of both a fundamental and applied
nature. The device can impart typically 200 kPa under
10 ms which, at this scale, can simulate representative rise
in overpressure produced by a semi-confined gas explosion.
Details of the experimental arrangement are described
later and in Figures 3 and 4. The test data generated in
this study will be used to validate numerical simulations
and analytical models.
It is well known that confined vapour gas explosions can
impart far greater levels of damage than that of un-confined
ones due to the higher overpressures generated. Offshore
structures are particularly vulnerable to this type of ex-
plosion, due to their exposure to hazardous, flammable
hydrocarbons and their complex geometry. Topside instal-
lations are typically compact units consisting of complex
piping, structural members and accommodation quarters. (a)
This leads to confinement and congestion which in turn
creates more turbulence and consequently higher over-
pressure. Typical overpressure from accidental explosions
on topside structures is between 0.3 bar and 2 bar, with
load durations generally between 50 ms and 200 ms [5].
However, research conducted by the Steel Construction In-
stitute proved that these overpressures can be significantly
higher [6].
Previous studies on square plates and profiled blast walls
[7–12] have shown that it is possible within the limitations
of the procedures to relate the dynamic, large inelastic
deformation of small-scale models to full-size structures
provided the laws of geometrically similar scaling are
followed. The plan is to extend previous work on mild steel
plates by considering the effect of a central circular hole
(φ50, 75 and 100 mm) and extended circular hole (φ50
by 75, φ75 by 100 and φ100 by 125 mm) on the loading,
and localized stress and deformation. Studies on plates
with openings have had very limited attention in the past.
Langdon et al. [13] recently investigated plates with small
(b)

openings; however, this focused on using perforations to


Figure 1. Plate-with-hole geometry and large-deformation mode
mitigate the blast loading rather than their effects on the shape for (a) rectangular hole and (b) rectangular hole
plate. Li et al. [14] studied the explosion resistance of with circular ends.
square plates with openings for venting dust explosions for
applications in the process industry, however very limited
details were provided. Jain [15] investigated the effect of the conditions when one is more appropriate than the other
hole aspect ratio of rectangular plates on the normal stress, and when more sophisticated numerical methods should be
shear stress and deflection in the transverse direction when used [18]. Equivalent non-linear single-degree-of-freedom
loaded statically. However, to the authors’ knowledge no (SDOF) methods can also be used for modelling plates with
experimental studies have considered dynamic transverse large inelastic deformations provided in-plane membrane
loading on plates with openings. effects are included [19]. Some preliminary static finite
Analytical studies have demonstrated the applicability element analysis using Nastran and an energy solution
of elastic-plastic and rigid-plastic analytical methods for are presented but it is proposed in future to utilize explicit
determining inelastic response of plated structures under non-linear finite element methods (e.g. ANSYS Autodyn
blast loading [16, 17]. It is important to consider the and LS-DYNA codes) for the simulation of the physical

497
On pulse pressure loading of plates with holes

tests. The use of approximate design methods also serves displacement at the centre of the plate, and x and y are
to provide a useful physical insight into the underlying the plate coordinates with their origin at the centre of
mechanics of the problem. the plate as shown in Figure 1a for a rectangular hole
of dimensions 2a and 2b, and Figure1b for a rectangular
hole with circular ends of dimension 2a and 2b where a is
2. Energy solution the radius of the circular end and 2(b-a) is the distance
between centres of the circular ends. For a clamped plate,
An energy solution to the problem of large deformations yield lines will form around the clamped edges but their
in thin, clamped square plates with a central rectangular contribution to the overall resistance of the plate in the
or circular hole subjected to transverse pressure loading large deformation range is small and therefore the deformed
is presented here. shape adopted here is justified. This shape function is
The approach adopted is to extend a previous analysis for differentiated to obtain strains and substituted into the
thin, clamped square plates with no hole [17]. As in the appropriate relationship for strain energy per unit volume.
previous analysis, membrane action governs the behaviour Integrating the strain energy per unit volume over the
of the plate for displacements greater than the plate thick- volume of the structural element gives the total strain
ness. Flexural resistance is considered negligible in this energy.
problem. Consequently, for thin plates an appropriate de-
formed shape of the plate under a uniform pressure load
is given as 2.1. Energy formulation for a rectangular
πx πy plate with a central rectangular hole
w = C cos cos (1)
2X 2Y
where 2X and 2Y are the dimensions of the plate, w is the The extensional elastic normal strain energy in the plate
transverse displacement at any point, C is the transverse (¼ geometry) with a central rectangular hole is given by

   
Ehπ 4 Y 9 3 b πb 3 2πb
1− sin −− sin C 4+
512(1 − ν 2 )X 3 16 4π Y Y 32π Y
  
Ehπ 4X 9 a 3 πa 3 2πa
1 − − sin − sin C 4+
512(1 − ν 2 )X 3 16 X 4π X 32π X
   (2)
Ehνπ 4 1 a 1 2πb
1 − + sin C 4+
512(1 − ν 2 )X Y 16 Y 32π Y
  
Ehνπ 4 1 a 1 2πa
1 − + sin C 4+
512(1 − ν 2 )X Y 16 X 32π X

where E is the elastic modulus, ν is Poisson’s ratio, h (¼ geometry), Wk , by the pressure load, p, given by
is the plate thickness and 2a and 2b are the dimensions
 
of the central rectangular hole. The extensional elastic 4pX Y πa πb
Wk = 1 − sin sin C (4)
shear strain energy in the square plate (¼ geometry) with π2 2X 2Y
a central rectangular hole is given by

If an impulsive solution is sought, then the total strain


   
Ehπ 4
1 b 1 2πb energy, Equations (2) and (3), is equated to the kinetic
1− + sin C 4+ energy (¼ geometry), Ek , given by
512(1 + ν)X Y 8 Y 16π Y
   (3)
Ehπ 4 1 a 1 2πa
1− + sin C 4
ρh
 
X πa

Y πb

512(1 + ν)X Y 8 X 16π X Ek = X Y − a + sin b + sin Ċ02
8 π X π Y
(5)
A quasi-static solution is obtained by equating the total where ρ is density and Ċ02 is the initial velocity imparted
strain energy, Equations (2) and (3), to the work done to the plate by an impulsive load.

498
G.K. Schleyer, N.J. Underwood, H.M. Do, J.K. Paik, B.J. Kim

but is used to obtain acceptable engineering solutions to


a complex problem. First, yield in membrane mode is esti-
mated from considering the stretching of an axially loaded
structural member made from elastic-plastic material. The
transverse displacement at first yield (extensional mode)
is given by
r
4X σ00
w1 = (X ≥ Y ) (7)
π E
After first yield in extensional mode, a weighted average
displacement (Kel wel + Kpl wpl ) is computed based on the
combination of the large-displacement elastic analysis and
a rigid-plastic analysis of the plate (Kel + Kpl = 1). The
von Mises yield theory is adopted for the rigid-plastic anal-
ysis in which the bi-axial normal stress equals a constant
σ00 and the shear stress is zero. Substituting into the rela-
tionship for strain energy per unit volume and integrating
over the volume of the structural element for a rigid-plastic
solution gives the extensional plastic normal strain energy
Figure 2. Mode shape of rigid-plastic plate with yield lines adopted in the square plate (¼ geometry) with a central square
for plate geometry having rectangular holes with or without
circular ends. hole as

σ00 hY π 2
 
b 1 πb
(1 − ) − sin C 2+
The above analysis holds for elastic extensional defor- 32X Y π Y
(8)
σ00 hX π 2
 
mation and may be used to predict maximum displace- a 1 πa
(1 − ) − sin C2
ments. A variation of the previous procedure [17] is adopted 32Y X π X
here for dealing with deformation beyond an elastic limit
and estimating permanent displacements. First, small- The relative weighting of each analysis will depend on the
displacement (flexural), yield-line theory is used to predict load and strain level. It is suggested that when the strain
a limit load for a rigid-plastic plate with a deformed shape level is above about 0.2% then Kel will be less than 1.
as shown in Figure 2 and rectangular holes with or without Some comparisons with experimental data are given later
circular ends. and show that adoption of Kel = 0.8 when the membrane
For a clamped rigid-plastic plate with a central rectangular strain level is around 1% or above gives acceptable results.
hole of dimensions 2a by 2b, the elastic limit load, p0 , The energy approach is appropriate for the estimation of
associated with the formation of yield lines along the global deformation. It cannot be used to directly predict
edges and diagonal yield lines as shown in Figure 2 is local strains around the hole. Stress concentration factors
given by can be applied in combination with the global strain to
give more realistic estimates of maximum strain but this is
3σ00 h2 (2X − a) (2Y − b)
 
not presented in this paper.
p0 = +
2 (Y − b) (X − a) The quasi-static energy balance approach does not account
 3  (6)
(X − a ) (Y − b3 )
3 3 for dynamic effects. A further stage would be to incorporate
+ − 6ab
(X − a) (Y − b) dynamics and solve the subsequent non-linear differential
equations using numerical methods. An independent SDOF
where σ00 is the dynamic flow stress. This limit load, which procedure is given in the paper whereby dynamic effects
is based on the energy dissipated in the yield lines, is can be evaluated.
then used in the large-displacement analysis to predict
an elastic limit displacement, w0 . Deformation beyond 2.2. Energy formulation for a rectangular
this displacement is considered permanent even when the plate with a central rectangular circular hole
membrane strain in the plate is elastic. with circular ends
Due to the progressive development of plastic membrane
strain in the plate, only an approximation is suggested here It is suggested that the energy formulation for a plate
to account for deformation beyond an elastic limit in the with a central rectangular hole can be adopted for the
membrane mode. This approach is not rigorously correct case of a plate with a rectangular hole with circular ends

499
On pulse pressure loading of plates with holes

considered later in the paper and are not insignificant in


Table 1. Maximum, wmax , and final, wf , displacements (mm) for 0.5 m
square plate, 1 mm thick with square hole loaded statically this type of problem even at low to moderate strain rates.
to 1 bar. It should also be noted that with the clamped arrangement,
Hole size, FE simulation Energy solution some stretching may occur around the boundary and there-
2a (mm) (static) (Kel = 1.0) fore a reduced size mesh in the FE analysis in future will
wmax wf wmax wf be required around the clamped edges as well as around
No hole 9.39 3.19 14.0 7.45 the hole in order to capture local stresses and strains.
50 9.86 5.21 14.9 8.0 For now, the hole reduces the load on the plate and as
75 10.6 6.99 15.3 8.15
100 11.1 7.69 15.5 8.24
such could lead to an optimal arrangement that would be
150 12.9 9.59 15.8 8.19 advantageous in terms of topside weight reduction and
200 14.5 10.5 15.7 7.78 design of explosion pressure relief. Stiffening around the
hole will also help to reduce local effects.

as shown in Figure 2 with minimal error of the order of


1%. The strain energy formulation would be the same 4. Comparison with experimental
but the work done for the plate having a rectangular hole
with circular ends would be marginally higher than for a
data
rectangular hole having the same overall dimensions due to
the difference in area. Consequently the overall area upon 4.1. Static and dynamic loading of square
which the pressure load acts would be marginally increased plates with no holes
and produce a fractional increase in displacement. Since
the area of the hole compared to the overall area of the The result of a static test on a clamped 0.5 m square
plate is small, the overall error is small. The effect of the plate [8] of nominal thickness 1 mm is compared with the
shape of the hole on stress concentrations particularly at energy solution. The average yield stress of 176 MPa was
discontinuities is of more significance than the effect of the determined from static tensile tests. For a static load of
area on the global displacements. 366 kPa, the maximum and final displacements are given as
32.7 and 30.2 mm, respectively. Using an elastic weighted
factor Kel of 0.8 (Kpl = 0.2), the energy solution gives
3. Numerical simulation maximum and final displacements of 34.9 and 28.7 mm,
respectively and corresponds to a maximum strain of 1.2%
A preliminary non-linear static analysis of a 0.5 m square (extensional mode).
plate, 1 mm thick was performed using the analysis program The maximum and final displacements for a clamped 0.5 m
NASTRAN. The material properties used in the analysis square plate of nominal thickness 1 mm subjected to a
were E = 205 GPa, ν = 0.3 and σ0 = 215 MPa. The mesh triangular pulse load of peak pressure 101 kPa and load
size was 25 by 25 elements and the edges of the plate duration 40 ms is given as 14.7 and 9.7 mm, respectively
were modelled as fixed. The objective of this analysis was [8]. This compares with 14.1 and 7.5 mm by the energy
simply to compare global displacements with the energy solution with σ00 = 215 MPa and Kel = 1.0, corresponding
solution. Consequently, an elastic, perfectly plastic von to a maximum strain of 0.2% (extensional mode).
Mises strength model (not including strain rate effects)
was used in the numerical analysis. The results are given 4.2. Dynamic loading of square plate with
in Table 1 compared alongside the energy solution for a square hole
static pressure load of 1 bar.
It is interesting to note that the FE analysis produces The explosion resistance of a square plate with a square
a wider spread of results than the energy solution over hole was studied [14]. A dynamic pressure load was im-
the range of hole size. It is clear that the FE analysis parted to a square clamped plate measuring 1.1 m by
is more able to simulate local effects than the simplified 1.1 m with a 0.44 m by 0.44 m central square hole. The
energy method with its single global mode of deformation. plate thickness was 1.63 mm and the flow stress was given
However, the energy approach gives results that are within as 190 MPa. Some plates had reinforcement around the
acceptable limits as an engineering solution in contrast edges of the square hole. Reinforcement around the edge
to a rigorous mathematical solution. Discounting local of the hole will strengthen the plate at this point but
effects, both analyses suggest no significant reduction in will not significantly reduce the global deformation of the
resistance due to the central hole, the load being reduced plate. Unfortunately, the paper gives no comparison in
in proportion to the hole area. Strain rate effects are terms of global displacements between the two types of

500
G.K. Schleyer, N.J. Underwood, H.M. Do, J.K. Paik, B.J. Kim

plate. At a pressure of 0.5 bar, the plate displacement at 4.3.2. Estimation of strain rate effects
the corner point of the square hole was around 25 mm
The deformed shape of the plate in Equation (1) is used to
and at the middle of the edge of the hole was just above
estimate the strain rate encountered in the pulse pressure
30 mm. No details are given of the load history. The
quasi-static energy solution for this case (with Kel = 1.0)
experiments. The average rise time recorded in the two sets
of experiments at nominal pressures of 25 psi and 50 psi
gives a maximum displacement at the corner of the hole of
were 0.006 sec and 0.012 sec, respectively. It is assumed
24.4 mm and at the middle point of the hole of 30.2 mm.
that the peak amplitude of the plate displacement occurred
At 1.14 bar, the corner displacement in the test is given as
around the same time as the peak magnitude of the loading
around 50 mm. This compares with 47 mm for the quasi-
(i.e. quasi-static response). The strain associated with
static energy solution (Kel = 0.8). The ultimate loading
stretching of the plate (i.e. membrane mode) is given by
½(w 0 )2 where w is the shape function, Equation (1), and w 0
pressure before rupture of the plate without reinforced hole
edges was given as 0.67 ∼ 1.14 bar and with reinforcing
as 1.19 ∼ 1.42 bar indicating considerable variability in
is the derivative of w with respect to the plate coordinate.
The energy method gives a maximum strain of C 2 π 2 /(16L2 )
the test conditions. Rupture of the plates initiated at the
where C is the peak displacement at the centre of the plate
hole corners. Without further details of the experiments, it
and L is the ½ span of the plate. This gives strain rates
of the order of 1 s−1 for the rates of loading encountered
is difficult to draw conclusions as to the effect of arching
around the hole, boundary conditions and load dynamics
in the experiments. The maximum strain by the energy
on the above results.
method is estimated to be of the order of 0.01. Using the
Cowper-Symonds relationship σ00 /σ0 = 1 + (ε̇/D)1/q , where
4.3. Dynamic loading of square plates, cen- σ00 is the dynamic flow stress and ε̇ is the strain rate, and
tral rectangular hole with circular ends the experimentally determined coefficients gives a ratio of
the dynamic flow stress to static flow stress of 1.31. Using
the average static 0.2% engineering proof stress in the
4.3.1. Material properties longitudinal direction of 200.5 MPa determined from the
material tests in the above Cowper-Symonds relationship
The material used in these experiments is Docol Form 01,
gives a dynamic flow stress of 262.7 MPa. This value will
a cold-reduced, mild steel used in a variety of applications
be used in the theoretical approach to compare the energy
for its formability, ideal where ductility is an important cri-
model predictions with the experimental data.
teria as in impact and blast protection. It comes in sheets
4.3.3. Estimation of inertia effects
of various widths and lengths with thickness ranging from
0.5 mm to 3.0 mm. Material characterization tests were
carried out using a servo-hydraulic tensile test machine SDOF methods can be used with due attention to their
at a strain rate of 0.00033 s−1 and gave average 0.2% limitations to estimate the dynamic loading effects on the
engineering proof stress of 200.5 MPa and 205.9 MPa square plates. It is well established that the rise time
in the longitudinal and transverse directions, respectively. of a triangular pressure pulse has the most significant
The ductility at failure was around 42%. The material influence on the dynamic response of the plate. The usual
was also tested for strain rate sensitivity using a split bi-linear resistance function is appropriate for structural
Hopkinson tensile bar over a range of strain rate from members with flexural resistance but the plate’s resistance
189 s−1 to 534 s−1 and gave 0.2% engineering proof stress in stretching mode is a function of C 3 and so highly non-
of 368 MPa to 412 MPa. Some tests were also carried linear. The elastic period of the plate, T , in stretching
out using a mini-bar at lower strain rates as only low to mode was estimated to be of the order of 0.005 sec based
moderate strain rates were expected in the experiments. on an averaged stiffness over the maximum plate response.
These gave a range of 277 MPa to 355 MPa 0.2% engi- This compares with the fundamental elastic period of the
neering proof stress over strain rates of 5 s−1 to 72 s−1 . It plate in flexural mode of around 0.05 sec. The load duration
is clear that the material is rate sensitive even at low to in the tests ranged from around 0.2 to 0.8 sec. while the
moderate strain rates. The Cowper-Symonds coefficients rise time ranged from around 0.005 to 0.013 sec. Using a
were determined from the above data to be D = 298 s−1 peak pressure load of 1 bar and load duration of 0.2 sec.,
and q = 4.89. The influence of thermal softening on the the effect of varying the rise time was evaluated using a
material has not been investigated in this study. With SDOF analysis. The SDOF parameters applied in the
the expected strains and strain rates in the experiments, analysis were 100 kN/m2 over an area of 0.25 m2 , total
adiabatic heating of the material due to plastic dissipation mass of 2.159 kg, stiffness of 1.76 MN/m, and load-mass
is not expected to have any significant influence on the KLM of 0.67 (Load factor KL = 0.46, Mass factor KM =
material behaviour. 0.31), where KLM = KM /KL . The load-mass factor converts

501
On pulse pressure loading of plates with holes

a fuse wire; this initiates rupture of the diaphragm and a


rapid blow-down of the chamber. The pressure difference
is created by the fact that the pressurized air evacuates
much quicker through the nozzle than through the holes
in the plate. The rig can be operated in static mode by
having end cover plates attached at both ends, as shown
in Figure 5(b).
The internal pressure is recorded by four pressure gauges
at a frequency of 25 kHz, two on each side of the support
plate, located at the top and at the side, as shown in
Figure 4. As a first approximation, the load on the plate
is simplified to be the pressure difference descr ibed by
the pressure transducers. This simplification is justified
by results from experimental characterization tests where
additional pressure transducers were used and spaced as
Figure 3. Load-deflection curve for quasi-static energy analysis show- far apart as practically possible on the centre of the test
ing linearized function over range of data.
plate and the support plate. Little or no variation between
the pressures recorded on the test plate and the support
plate were observed.
the distributed load and mass into an equivalent lumped
mass and concentrated load. The non-linear, elastic load- 4.3.5. Test plates
deflection curve1 is given in Figure 3 from which was
The geometry of the plates is given in Figure 6. Each plate
extracted a linearized function over the range of data.
was clamped all around the boundary with equi-spaced
This was subsequently used to obtain the linear stiffness
studs each torqued to the same value of 100 Nm. The
parameter applied in the SDOF analysis. For a tr /td = 0.5,
studs were 20 mm in diameter and the clamping holes
where tr is rise time and td is load duration, the maximum
22 mm in diameter. The geometries of the holes were
elastic response was 14.4 mm; 16.0 mm at tr /td = 0.05, and
chosen to represent idealized apertures in bulkheads on
16.2 mm at tr /td = 0.025. The maximum elastic response
ship structures. The outer dimensions for each plate are
jumps to 25.6 mm when a tr /td = 0.01. Therefore, it is
660 mm by 660 mm, while the area exposed to the pressure
concluded that dynamic effects are not negligible (∼10%)
load is 500 mm by 500 mm minus the area of the hole.
over the range of test conditions.
Each plate has a nominal thickness of 1.1 mm, so plastic
membrane stretching is assumed to be the primary mode
4.3.4. Test rig
of deformation.
A differential pressure device, Figure 4, was used to impart
a pulse pressure load to mild steel square plates with 4.4. Pressure-time curves
circular holes. The blast tests were carried out at the
University of Liverpool. This device is capable of producing Two nominal peak pressure loads of 172.4 kPa (25 psi) and
a repeatable and uniform transient pressure loading over an 344.8 kPa (50 psi) were applied across the range of plate
exposed area of 0.25 m2 . The test specimen is attached to specimens. This produced permanent deformation in the
the support plate with a clamping frame and 36 bolts. First, plates that could be measured while the plates were still
the chamber is sealed with a rigid cover plate covering in post-test situ. A summary of the test results for these
a 500 mm diameter nozzle at one end and a thin film plates are given in Tables 2 and 3. The complete load
diaphragm covering a 500 mm diameter nozzle at the other, histories are given in Figures 7 and 8. There is little or
as shown in Figure 5(a). The rig is then pressurized to no variation in the pressure recordings from the two sets
the desired pressure on each side of the support plate so of gauges located on the top and side of the support plate.
that no net load is applied to the test plate. At firing, the The rise to peak pressure appears almost instantaneous
diaphragm is melted along its circumference by energizing due to the scale used to capture the full pressure-time
curve but there is a finite rise as indicated in Tables 2
and 3. The decay part of the curve is much longer due to
1
This solution includes bending energy as well as mem- the small area of the hole in the test plate compared with
brane energy terms for a square plate with no hole. The the area of the exit nozzle. The air evacuation process
bending energy contribution over this large deflection varies with test pressure and is not linear. The air first
range accounts for about 1% of the resistance. has to push the diaphragm material away; normally this

502
G.K. Schleyer, N.J. Underwood, H.M. Do, J.K. Paik, B.J. Kim

Figure 4. Expanded view of pulse pressure loading rig (PPLR).

(a) (b)

Figure 5. PPLR assembled (a) with film diaphragm and clamping ring (dynamic mode) and (b) with end cover plates (static mode).

comes away as a whole piece if the fuse wire melts all the tion that pressures with duration less than 1 ms (scaled)
way round but still leaves a narrow piece of diaphragm would not affect the response of the plate significantly.
material around the edge of the nozzle. Any variation in An idealised triangular pressure pulse can be constructed
the width of this remaining piece could affect the pressure by extrapolating a line from the point of maximum pres-
profile. As the hole size increases, the decay to ambient sure through a point on the pressure curve at 10% of the
pressure and hence the load duration becomes shorter as maximum pressure as shown in Figure 9.
air from one side of the rig escapes through the hole in the
test plate more rapidly. The decay phase of the loading is
4.5. Maximum and final plate deflections
less significant than the rise phase of the pressure pulse
and SDOF analysis confirms this for the geometries, loads The permanent midpoint deflections of the tested plates
and times encountered in this series of tests. are summarized in Tables 2 and 3. It can be seen that
The pressure data was filtered removing any spike less there is no significant variation over the range of hole
than 0.125 ms (1 ms scaled) duration [6] on the assump- size for a given nominal test pressure; some variation may

503
On pulse pressure loading of plates with holes

Figure 6. Square plate specimens with circular holes.

Table 2. Test data for square plates with circular holes loaded to nominally 25 psi.

Hole geometry Hole area Hole area as % Rise time tr Peak pressure Impulse Final Load duration
(mm) as ratio wrt of plate area (msec) (psi) (psi-msec) centre disp. td (msec)
φ50 hole (mm)
φ50 1.00 0.78 5.3 26.8 9071.8 15.0 677
φ50 by 75 1.64 1.29 7.4 26.0 6019 12.4 463
φ75 2.25 1.75 5.5 26.3 4181.7 13.7 318
φ75 by 100 3.20 2.51 7.0 25.7 2942.7 12.4 229
φ100 4.00 3.14 6.0 25.6 2419.2 13.7 189
φ100 by 125 5.27 4.14 5.7 24.8 1736 14.6 140

be caused by the small differences in peak pressure and 0.2) was used given the large permanent deformations (and
rise time. The decrease in stiffness associated with the plastic strain) in the test plate. This is consistent with
increasing area of the holes is more than compensated analysis of other static and dynamic plate test data. The
for by the decrease in the load. Visual inspection of the main purpose of the tests is to validate analytical and
plates, Figure 10, identified plastic hinges at the edges, numerical methods which in turn can be used for further
but no plastic hinges within the plates. The primary load studies.
bearing mode is that of membrane stretching.
The energy methods described earlier were used to predict
maximum and final plate deflections and compare with the 5. Discussion and conclusion
permanent midpoint deflections of the tested plates. These
are given in Tables 4 and 5. The quasi-static formulation A study of plates with perforations will benefit the offshore
gives more than adequate correlations when the dynamic industry in terms of improved design guidance and optimi-
flow stress of 262.7 MPa is used to account for strain rate sation of deck plates, blast walls and blast relief panels.
material effects. An elastic weighted factor Kel = 0.8 (Kpl = New novel small-scale (1:8) dynamic pressure tests on

504
G.K. Schleyer, N.J. Underwood, H.M. Do, J.K. Paik, B.J. Kim

Table 3. Test data for square plates with circular holes loaded to nominally 50 psi.

Hole geometry Hole area as Hole area as % Rise time tr Peak pressure Impulse Final Load duration
(mm) ratio wrt φ50 of plate area (msec) (psi) (psi-msec) centre disp. td (msec)
hole (mm)
φ50 1.00 0.78 13.2 49.9 18388.2 23.3 737
φ50 by 75 1.64 1.29 12.6 50.2 15009.8 22.4 598
φ75 2.25 1.75 9.6 48.9 10684.7 22.8 437
φ75 by 100 3.20 2.51 10.2 48.0 7464 23.0 311
φ100 4.00 3.14 13.1 46.7 5674.1 21.3 243
φ100 by 125 5.27 4.14 13.0 45.3 3420.15 22.1 151

(a) (a)

(b) (b)

Figure 7. Pressure-time curves for plates with circular holes loaded Figure 8. Pressure-time curves for plates with extended circular holes
to (a) nominally 25 psi and (b) nominally 50 psi. loaded to (a) nominally 25 psi and (b) nominally 50 psi.

0.5 m square plates with central circular holes has been the test data. Future work will consider detailed analysis
conducted at nominal test pressure of 25 psi and 50 psi, of the test data using numerical finite element methods.
simulating an extreme over-pressure event. This data has The following gives both a qualitative and quantitative
been used to validate energy based analytical models with evaluation of the sources of error likely to be encountered
very good overall results. The energy approach that con- in the small-scale tests and in applying the simplified
siders large inelastic deformation and membrane effects analytical approach to these types of non-linear problems.
gave more than acceptable results when compared with Notwithstanding these sources of error, the simplified ap-

505
On pulse pressure loading of plates with holes

Table 4. Results of predicted maximum and final deflections for plates


at centre of plate loaded to nominally 25 psi by energy
method.

Experimental Energy solution (Kel = 0.8)


Hole geometry wf (mm) Peak wmax wf (mm)
(mm) pressure (mm)
(psi)
φ50 15.0 26.8 20.4 13.5
φ50 by 75 12.4 26.0 20.7 13.5
φ75 13.7 26.3 21.0 13.1
φ75 by 100 12.4 25.7 21.3 13.3
φ100 13.7 25.6 21.4 14.1
φ100 by 125 14.6 24.8 21.6 14.1

Table 5. Results of predicted maximum and final deflections for plates


Figure 9. Pressure-time history for plate with φ100 by 125 mm hole at centre of plate loaded to nominally 50 psi by energy
showing construction of idealised triangular pulse load method.
(pmax = 24.8 psi, tr = 5.7 msec, td = 140 msec.).
Experimental Energy solution (Kel = 0.8)
Hole geometry wf (mm) Peak wmax wf (mm)
(mm) pressure (mm)
(psi)
φ50 23.3 49.9 28.5 22.2
φ50 by 75 22.4 50.2 29.5 23.0
φ75 22.8 48.9 29.4 23.0
φ75 by 100 23.0 48.0 30.0 23.4
φ100 21.3 46.7 29.8 23.1
φ100 by 125 22.1 45.3 30.1 23.3

proach together with small-scale test validation paves the


way for a versatile robust design methodology which can
be used to advantage for screening purposes and/or early
stage conceptual design studies.

(a) 5.1. Limitations of approximate design meth-


ods
A simplified method which is based on global deformations
is not able to consider local effects and so this must be
factored into a design methodology that incorporates the
energy approach.

5.2. Loading
The rise time in the tests scales up to about 40–100 msec
which is representative of vented hydrocarbon explosions
in offshore structures [5].

5.3. Inertia effects


(b) It has been mentioned earlier that the load duration is
very long (longer than representative confined hydrocar-
Figure 10. Deformed 0.5 m square plates with (a) φ100 mm hole bon explosions) in these tests because the air was forced
and (b) φ100 by 125 mm hole showing plastic hinges at
the edges and ∼21–22 mm permanent central deflection.
to escape through one nozzle only. A shorter load du-
ration can be achieved when the air inside the vessel

506
G.K. Schleyer, N.J. Underwood, H.M. Do, J.K. Paik, B.J. Kim

exits both nozzles on either side of the test plate in a which scales down to 23.5 mm but in the small-scale plate
sequential manner thus producing a near equal rise time with φ100 mm hole, the final deflection is only 20.2 mm.
and decay time pulse load. However, the load rise time The strict requirements for geometrically similar scaling,
is more important in categorizing the loading event as which demands equality of the dynamic flow stresses at
being one of quasi-static, impulsive or dynamic. Inertia all scales, has been violated and leads to around 7% re-
effects are always significant when the loading is applied duction in permanent deflection in the small-scale plate.
almost instantaneously as in the case of shock loading. This effect is not insignificant and is more important at
The maximum displacements associated with a load pulse higher strain rates. Clearly, it is important to estimate the
having a finite rise time tr and a long duration thereafter influence of material strain rate sensitivity when conduct-
are within approximately 20% of the quasi-static value ing dynamic tests on small-scale models, but this can be
when tr /T > 0.9, approximately. The smallest rise time done fairly reliably. Thus, with this proviso, the laws of
in the experiments was around 0.005 sec and the longest geometrically similar scaling would be satisfied provided
around 0.012 sec, thus for an average elastic period in no rupture occurs.
the large deflection range of around 0.005 sec, tr /T > 0.9.

5.6. Boundary conditions


This explains why a quasi-static solution gives such good
results. When tr /T < 0.5 inertia effects are important and
a quasi-static analysis is less accurate.
The plates were assumed fixed by reason of the clamping
arrangements in the tests. Small axial movements at the
5.4. Strain rate effects boundaries can lead to appreciable lateral displacements.
An average value for ε̇ was used in the analysis based
The large number of bolts and a uniform tightening torque
of 100 Nm ensured minimal movement at the boundaries.
on an estimate of global strain rate in the tests. Strain
This can only be assessed more accurately through mod-
rate varies both temporally and spatially throughout the
elling the actual boundary conditions in a finite element
structure during loading.
analysis.

5.5. Scaling effects


The dimensions of the largest steel plate studied here Acknowledgments
are just over 2 times larger than the 0.5 m square plate
specimens. Nevertheless, the analytical methods give more The authors gratefully acknowledge the funding support
than acceptable predictions for the maximum permanent given by the UK EPSRC and the Korean government
transverse displacements of all specimens regardless of (Basic Science Research Program through the National
size. This agreement implies that there is no significant Research Foundation of Korea (NRF) funded by the Min-
departure from the laws of geometrically similar scaling. istry of Education, Science and Technology under Grant
It is well understood that the phenomenon of material No. K20903002030-11E0100-04610) in this project.
strain rate sensitivity introduces a known size effect, as
shown in §11.3.2 of reference [1], or

σ00 1 + (ε̇/βD)1/q References


0 = (9)
Σ0 1 + (ε̇/D)1/q
[1] Jones N., Structural Impact (2nd Edition), Cambridge
where σ00 and Σ00 are the dynamic flow stresses for a model University Press, Cambridge, UK, 2012
having a geometric scale factor, β, and a full-scale proto- [2] Paik J.K., Thayamballi A.K., Ultimate limit state de-
type, respectively. ε̇ is an average strain rate and D and sign of steel-plated structures, John Wiley & Sons,
q are the Cowper-Symonds coefficients introduced earlier. Chichester, UK, 2003
Now, if we take β = 1/8 = 0.125 for the 0.5 m test plates [3] ISO 18072-1, Ships and marine technology – Ship
and 4 m full-size plates, then using D = 298 sec−1 and Structures – Part 1: General requirements for their
q = 4.89, Equation (9) predicts that the dynamic flow limit state assessment, International Organization
stress for the larger full-size plate would be about 12% for Standardization, Geneva, Switzerland, November
smaller than the flow stress for the 1/8 scale plate. This 2007
will lead to larger permanent deflections in the full-size [4] Paik J.K., Thayamballi A.K., Ship-shaped offshore
plate e.g. for a quasi-static pressure of 3 bar, the final de- installations: Design, building, and operation, Cam-
flection in the full-size plate with φ800 mm hole is 188 mm bridge University Press, Cambridge, UK, 2008

507
On pulse pressure loading of plates with holes

[5] UKOOA & HSE, Fire and Explosion Guidance Part 1: [13] Langdon G.S., Rossiter I.B., Balden V.H., Nurick G.N.,
Avoidance and Mitigation of Explosions, UK Offshore Performance of mild steel perforated plates as a blast
Operators Association, 2003 wave mitigation technique: Experimental and numeri-
[6] The Steel Construction Institute, Blast and Fire En- cal investigation. International Journal of Impact En-
gineering for Topside Structures Phase 2, The Steel gineering Vol. 37, 2010, 1021–1036
Construction Institute, 1998 [14] Li G., Chen B.-Z., Deng X.-F., Eckhoff R.K., Explosion
[7] Schleyer G.K., Hsu S.S., White M.D., Blast loading of resistance of a square plate with a square hole, Journal
stiffened plates: experimental, analytical and numer- de Physique IV, Vol. 12, 2002, 121–124
ical investigations, In: Proceedings of ASME/JSME [15] Nitin K.J., Analysis of Stress Concentration and Deflec-
Joint Pressure Vessels and Piping Conf., Structures tion in Isotropic and Orthotropic Rectangular Plates
Under Extreme Loading Conditions, San Diego, PVP, with Central Circular Hole under Transverse Static
Vol. 361, 1998, 237–255 Loading. World Academy of Science, Engineering and
[8] Schleyer G.K., Hsu S.S., White M.D., Birch R.S., Pulse Technology, 2009
pressure loading of clamped mild steel plates, Interna- [16] Schleyer G.K., Hsu S.S., A modelling scheme for pre-
tional Journal of Impact Engineering, Vol. 28, Issue 2, dicting the response of elastic-plastic structures to
2003, 223–247 pulse pressure loading, International Journal of Impact
[9] Langdon G.S., Schleyer G.K., Inelastic deformation Engineering, Vol. 24, 2000, 759–777
and failure of clamped aluminium plates under pulse [17] Schleyer G.K., Jones N., On the simplified analysis
pressure loading, International Journal of Impact En- of square plates under explosion loading. Pressure
gineering, Vol. 28, Issue 10, 2003, 1107–1127 Equipment Technology: Theory and Practice, Eds:
[10] Schleyer G.K., Hsu S.S., White M.D., Scaling of pulse WM Banks and DH Nash, Professional Engineer-
loaded mild steel plates with different edge restraint. ing Publishing (Institution of Mechanical Engineers),
International Journal of Mechanical Sciences. Vol. 46, 2003, 225–233
Issue 9, 2004, 1267–1287 [18] Langdon G.S., Schleyer G.K., Inelastic deformation
[11] Langdon G.S., Schleyer G.K., Inelastic deformation and failure of ¼ scale profiled stainless steel blast wall
and failure of ¼ scale profiled stainless steel blast panels. Part II: analytical modelling considerations.
wall panels, Part I: experimental investigations, In- International Journal of Impact Engineering, Vol. 31,
ternational Journal of Impact Engineering, Vol. 31, Issue 4, 2005, 371–399
Issue 4, 2005, 341–369 [19] SBEDS v.4.1: Protective Design Centre, U.S. Corps
[12] Schleyer G.K., Lowak M.J., Polcyn M.A., Langdon G.S., of Engineers, 2009, https://pdc.usace.army.mil/
Experimental investigation of blast wall panels under software/sbeds/
shock pressure loading, International Journal of Impact
Engineering, Vol. 34, 2006, 1095–1118

508

S-ar putea să vă placă și