University of Southampton Research Repository

ePrints Soton

Copyright and Moral Rights for this thesis are retained by the author and/or other
copyright owners. A copy can be downloaded for personal non-commercial
research or study, without prior permission or charge. This thesis cannot be
reproduced or quoted extensively from without first obtaining permission in writing
from the copyright holder/s. The content must not be changed in any way or sold
commercially in any format or medium without the formal permission of the
copyright holders.

When referring to this work, full bibliographic details including the author, title,
awarding institution and date of the thesis must be given e.g.

AUTHOR (year of submission) "Full thesis title", University of Southampton, name

of the University School or Department, PhD Thesis, pagination

http://eprints.soton.ac.uk
Al. Appendix A - Comparison of connection test results and
predictions using component method

A l l Introduction
This appendix presents a complete set of results for the component-based methods and the
respective experimental test data under both static and dynamic conditions. Some tests used a
single loading ram to provoke a rotation whilst others used dual loading rams to provide direct
axial loads and restrict rotation. The dimensions (Figure Al-1) were recorded prior to each test.

LDG2 LDGl

tLoad CeB B Load Cell A

Figure Al-1: Test dimensions

For dynamic tests the physical movement of the column section is recorded and a comparison
against the prediction using component-based methods shown. In addition the applied
force/moment is plotted against the displacement/rotation. For static tests only the force/moment
against displacement/rotadon is presented as there was no need to display the time history analysis.

The dynamic tests are characterised by the diaphragm arrangement used to release the pressure
within the system, where A is the weakest diaphragm arrangement and E the strongest. Hence a
DYN (E) test is likely to have experienced a greater load than a DYN (A) test. The true load was
recorded for each test.

Al
AppgndixA

A1.2 Fin plate connections

Test Loading
Test Type Failure
Number Type
Moment-
FINl Dyn(q Bolt in shear
rotation
Moment-
FIN2 Static Bolt in shear
rotation
Direct
FIN3 Static Bolt in shear
teKion
No failure during the test but large
Direct
FIN4 Dyn (E) bearing defamation of the beam web
Tension
was observed
Moraent-
FIN5 Dyn (A) Bolt in shear
rotmtion

Ajf J J jFw cofwecAww fesf 7

Dynamic test (1 thick and 1 thin diaphragm) w i single loading ram.

:g. 60
q 50
40

0.01 0.02 0.03 0.04

TinK (:)

(a) Rotation-time (b) Rotational velocity - time

5.3
Free icte<i -Predicted
5 - Experimental
Exp erim enta /
4.5
1
4 -
S 3.5 -

/
/
2 -
//
1.5 -

0.5 y
0
0.01 0.02 0.03 0.04 0.01 0.02 0.03 0.04
Time (s) Time(s)

A2
 (c) Rotational acceleration - time (d) Centre of mass displacement
300
Predicted
. 1 - Expmmaital

' ''M
A rft
/ /
- /'/v'
tJ
1 t
. V 1 \
1 1
ll 1 1
ll
0 J ' 1 **
0.01 0.02 0. 0.
-50 "4 Predicted
- - Experimental
100 0 0.0] 0.02 0.03 0.04
Time (s) Time(g)

(e) Load - rotation (f) Moment - rotation

60
Predicted Predicted
- - Dynamic (FPCl) - Experimmtal / *
50

1 2 3 4
Rotation (d%)
Rotation ( d ^

# Good correlation between predicted and experimental results for rotation of flying
column
# Centre of mass displacement was over predicted
# Stiffness of connection from moment-rotation curve was predicted with reasonable
accuracy
Overestimation of ultimate moment capacity

Al.2.2 Fin plate connection test! (FIN2)

Static test with single loading ram.

A3
 (a) Load - rotation (b) Mcment - rotation
70
Stmdc (FPC2)

60
- - Static {F P C 2 / Predicted
Predict ed /
^ 50
"9
1
/.
/
& 30
8 40 /! /
l-j j
T3 / 1'
30
/
/i
7

10

0 1 1 1 1 r
0 1 2 3 4 5 6
0 1 2 3 4
Rotadon ( d ^
Rotation (deg)

# Stadc prediction shows good correlation against the experimental test in the
Grst phase up to a rotadon of 3.2 degrees. Post yield stifkess is overestimated
for specified rotation possibly due to crushing of beam flange
# Sudden increase in stiEkess is a result of bearing against the column which
changes the pivot locadon and thus results in an increase in moment. In the
experimental test this occurs over a rotadon of approximately 1 degree
# Failure rotadon and moment capacity are reasonably accurate and within 5%
of experimental values

Al,2.3 Fin plate connection test 3 (FIN3)

Stadc test with dual loading rams for direct tension tesdng.

250 -

1
Predicted
(2
Experimental.

-T
4 6 8 10 12 14
Central displacement (mm)

Figure 1-2: Load-displacement comparison (FPC3)

Inidal stiAiess due to hicdon is not very accurate

Increased stiffness w h e n bolts begin to bear against the plate at 2 m m displacement is
captured

A4
Appendix A

# General shape is predicted well but final displacement is off by approximately 3 J mm

which is reasonable considering the length of the column section is 2m

Dynamic test (2 thick diaphragms) with dual loading rams for direct tension testing. The
connection did not 6ul during the initial dynamic loading phase.

(a) Cmtre of mass - time (b) Acceleration - time

6 1

5 - y 1
E
1
& 4 /
' /
if' // i
& ' / on 0.125 0.15
1/ HJ
' 1
_w
O 2

11
Predicted F*redicted
- - Experimental
0
J,i 1 1
- - Experimental
I 1
0 0.025 0.05 0.075 0.1 0.125 0.15
T i m e (s) Time (s)

(c) Load - centre of mass displacement (d) Bolt group forces in conqxment model
200 100000
;e" 90000
z
I
c
80000
70000

"o 60000
3 m
E 50000
w
(2 40000
2 30000
o
Predicted
20000 B o l l Group 1
- - Experimental - - Bolt Group 2
10000 B o l t Group 3
0
1 2 3 4 6 0 0.025 0.05 0.075 0.1 0.125 0.15
Displacement (mm) Time(s)

No failure was predicted in agreement with the experimental result

DifGcuit to tesl two rams dynamically because the pneumatic system meant slighty unequal
loads were applied through the two rams leading to rotation and oscillations
The unequal loads recorded in the test were input into the prediction model hence the
resulting oscillations
The experimenlal yield load appears to be approximately 60 kN compared to a predicted
value of 40kN with Gnal displacements of 5 and 4 mm respectively

A5
Al.2.5 Fin plate connection test 5 (FINS)
Dynamic test (2 thin diapragms) with a single loading ram.

(a) Rotation - time (b) Rotational velocity - time

I
i Predicted
V Predicted
- - ExpcmncmW - - Expehmental
T r
$ 0.01 0.02 0.b3 0.04 0.05
0.01 0.02 0.03 0.04 0.05
-1 J

Timc(s) Time (s)

(c) Rotational a c c e l o ^ o n - time (d) Caitre of mass displacement

300 18 1

250 -

150 -

A
U 0.01 0.02 0.03 M 0.05

1 ^Predicted
Predicted
- - Expehmental
150 - - Expenmental T r
0.01 0.()2 0.p3 0.04 0.05
-200
Timc(s) Time (s)

A6
 (e) Load - rotation (f) Moment-rotation
70 - 70 -

60 -

50 -

40 - 1 40 -

30 -
J 30 - o
20 - Predicted
- - Experimental
10 ^
Predicted
- - Experimental
0
2 3 4
2 3 4 5 -10

Rotation (d%) Rotati(m (deg)

General behaviour showed a good correlation with the experimental data

Over prediction of the post yield stiffness and moment capacity

A7
A13 8mm FLEXIBLE ENDPLATE CONNECTIONS

Peak Peak
Endplate Peak Failure
Test Loading Loading applied
Thickness Moment Rotation Failure
Number Type Time axial load
(mm) (kNm) (deg)
(ms) (kN)
Endplate
FEPl 8 Static 50.2 6.2 - 58.8
&acture
Endplate
FEP2 8 Dyn (E) 75 5.8 40 97.7
&acture
FEP3 8 Dyn (A) 60.1 6.3 39 523 No
Endplate
FEP4 8 Dyn (B) 63.4 5.8 40 65.7
fracture
Endplate
FEP5 8 Static 483 6.7 - 56^
&acturc
Endplate
FEP6 8 Dyn (E) 73.1 7.0 47 84.2
&acture
FEP9 8 Dyn (A) 54.8 40 47.4 No Failure
FEP9 Endplate
8 Static 45.7 7.4 - 54.2
(conO fracture
Endplate
FEP14 8 Dyn (A) 57.7 5.6 37 45.0
&acture
FEP16
(datalogge Yes
8 Dyn (A) - -
r failed to (endplate)
trigger)
Static Endplate
FEP18 8 - 260.6
Tension fracture
Dyn (E)
FEP19 8 48 218.3 No Failure
Tension
FEP19 Static Endplate
8 - 281.1
fcont) Tension fracture
Dyn (E)
FEP20 8 53.2 157.7 No Failure
Tension
FEP20 Static Endplate
8 - - 280.5
(cont) Tension Fracture

Al.3.1 Flexible end-plate test 1 (FEPl)

8min thick endplate

Static test with single loading ram

A8
 Predicted
- o - T e s t (FEPI)

Rotation (deg)

# Experimental elastic stiSness is higher than that predicted however the general trend shows
a good correlation
# The point at which the beam flange contacts the column appears to be out by about ys of a
degree where the test data shows an increase in stifkess &om about 4 degrees rotation
# The prediction model overestimates the failure rotation by nearly 3 degrees

A13.2 Flexible end-plate test! (FEP2)

Rmm thick endplate

Dynamic test with single loading ram

no
100
90

ig 60 -

-5 50

0.01 0.02 0.03 0.04 0.05 0.06

Time (s)

A9
 (a) Rotadoo-tiine (b) Rotational velocity-time
7
Predicted
Expenmental 6

is

n4
>a 3

I ^ Predicted
- - Experimental

0
0.0] O^Q 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Tune(:) 'nme()

(c) Rotadonal acceleradon-tiiiie (d) Centre of mass displacement

350
Predicted Predicted
- - Experimental - - Expenmental
W 250 .

200 -

0.01 0.02 0.03 0.04 0.

Tnne(s) 0.01 0.02 0.03 0.04 0.05

TimeCs)

(e) Load-rotadon (f) Moment-rotadon

80

70 -

- ! 1 - i^ /#
60

E50
s
40 -
//
30
yij
20
' V

Predicted 10 Prcdided
- - Experimental - --Expcnmental
0 1 1 1 1 i 1

0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7
Rotation (deg) Rotation (dcg)

AlO
A133 Flexible end-plate test 4 (FEP4)
8mm thick endplate

Dynamic test with single loading ram

^ 40

0.04 0.08
Time (s)

All
 (a) Rotation-dme (b) Rotational velocity-time
4.5

/
/

1 t
/
,
f/ f /
M / & 3
i/ ' /

T3 / /
// V /
' /

' /
//i
O '/ ' /

15

1, ^ 1/

2 1 _ /^
/

/l Predic ted A' Predict

1 -
- - Experimental f\ - - Experimental
^ 1
0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Time (s) Time (s)
(c) Rotational acceleration-time (d) Centre of mass displacement
30
Predicted
- - Experimental

0.02 ; 0.04 ; 0.06

Predicted {
% -f
- - Expenmenta] 0 0.02 0.04 0.06 0.08
Time(s) Time (s)
(e) Load-rotation (f) Moment-rotation
80
70

60

.50 - y' /

40 -
' /
1/
30 '/

20 fj
Predicted 10 Predicted
- - Expeninental - - Experimental
0
1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Rotation (d%) Rotation (deg)

A12
AL3.4 Flexible end-plate test 5 (FEP5)
8mm thick endpiate

Static test with single loading ram

(a) Moment-rotation

-o- Test (FEP5)

Predicted
r-
2 3 4 5
RotaMoa (deg)

A 1.3.5 Flexible end-plate test 6 (FEP6)

8mm thick endplate

Dynamic test with single loading ram

90

r ' . 60

9 50

0.01 0.02 0.03 0.04 0.05 0.06

Time (s)

A13
 (a) Rotation-dmc (b) Rotational velocity-time

Predi(^ed
7 -i
- - ExpeiimentE 1
/

6
1 i 1
t/ /Jj,
95
- /

y
J
4
s //
//

2
Predicted
1 -
- - Experimental
0 -i 1 1 r
0 0.01 0.02 0.03 0.04 0.05 0.06 0 0.01 0.02 0.03 0.04 0.05 0.06
Time (s) Time (s)
(c) Rotational acceleration-time (d) Centre of mass displacement
35
- Predicted
-530 - - Experimental
S
150 -

I 1 I I 'J
0.01 0.02 0.03 0.04' 0.05 0.
Predicted
- - Experimental

TUne(g) 0 0.01 0.02 0.03 0.04 0.05 0.06

Time (g)
(e) Load-rotation (f) Moment-rotation

A14
 (a) Rotadon-dme (b) Rotational velocity-time

Predicted


. - - Experimental
1

M)
-
,
1 "'T"
!
t
11
* /I
y l
if
AwWed i 0.05 I ok " 0.i5 0(2 0.
' - Ezqxrimental -0.5 - -
0 I' i ' i :
0 0.05 O.i 0.15 02 0J5 -1
nmef$) Time (s)
(c) Rotational acceleration-tinie (d) Centre of maRs di^lacement
250

Predicted
- - Expenmental

= 100 -

Predicled
-50 .
iq_ ^ ' ' : r - - Exp*imeo(ml
'(
-100 0.05 0.1 0.15 0.25
Thne (:) Time (s)
(e) Load-rotation (f) Moment-rotadon

Z 60 -

30.6 ^=20 -

Predicted
- - Experimental Predicted
- - Expenmental
2 3 4 5 6
Rotation (deg) 3 4 5 6 7
Rotation (deg)

Component-based method overestimates yield load at this loading rate

This results in a stifPer connecdon and reduced rotadon compared to the test result
Both the test and predicdon model show that the connection does not f ^ l

A16
Al.3.7 Flexible end-plate test 11 (FEPll)
10mm thick endplate

Dynamic test with single loading ram

110 -

0.01 0.02 0.03 0.04 0.05

Rotation (deg)

A17
 Rotation-time (b) Rotational velocity-time
8 7
Predicted | j
7 6 P r edicted
- - E ^)cnm mtal 1 1 / ; I !
1 ' / - - E : iperimental j y |
6
^ 5
n " / A /
1 ' / / i /
T3 "34
i/'/i o
g4 / / :
y c
/ /
o
1 / 1 i
!

y 1 : 1
/ I
y 1
1 1 1 0
0 0.01
0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Time (g) Time (s)
(c) Rotational acceleration-time (d) Centre of mmRx di^lacement
25U
Predicted
Predicted
200 - - - Expeiiniental 330 - - - E^qxaimental

@ 150

% 100 -

IL. _0^2_ JD.03_ 0 . 0 ^ 10

0 0.01 0.02 0.03 0.04 0.05

Thne (s)
T i m e (s)
(e) Load-rotation (f) Moment-rotation
120 120
110 Predicted
100
/> - - Experimental
-
' -
90
80
Z 70
A if
is" 60 - - - -

J 50 8 50 .
40
30 '

20 Predicted
10 - Experimental"
0 1
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Rotation (deg)
Rotation (deg)

A18
Al.3.8 Flexible end-plate test 14 (FEP14)
8mm thick endplate

Dynamic test with single loading ram

< 15

0.05 0.1 0.15 0.25

RotmOoa (d*g)

A19
 No fWlure recorded in the test or prediction model
" Prediction model overestimates yield load at this loading rate resulting in stiffer connection
and reduced rotation

Al.3,9 Flexible end-plate test 19 (FEP19)

8mm thick endplate

Dynamic test with dual loading rams for direct tension tests

(a) Displacement - time (b) Load-displacemeiit

175

Predicted Predkted
Experimemal Expcrimeotal

0 0.025 O.OS 0.075 0.1 0.125 0.15 1 2 3 4 5

Dhpkcfmcal (mm)

No failure recorded in the test or prediction model

Reasonable prediction of overall load-displacement behaviour but diKcult to draw any
deGnitive conclusions
The predicted displacement-time history shows secondary oscilladons of approximately
120Hz

A21
A1.4 10mm FLEXIBLE ENDPLATE CONNECTIONS
Peak
Endplate Peak Failure Peak
Test Loading applied
Thickness Moment Rotation Loading Failure
Number Type axial load
(mm) (kNm) (deg) Time (ms)
(kN)
FEP7 10 Dyn (B) 86J7 6.2 45 69^ No Failure
Fracture of
endplate,
beam web
FEP7 (cont) 10 Static 8Z2 9.6 92.8
crushing and
weld-endplate
inter&ce
Endplate
FEP8 10 Static 98 13.2 - 116.7
fracture
FEPIO 10 Dyn (B) 79.8 7.9 40 71.4 No
FEPIO Endplate
10 Static 76.6 11.5 - 89.5
fomO fracture, bolt
Endplate
fracture, bolt,
FEPll 10 Dyn (E) 113.5 6.2 39 113.0
weld-endplate
interface
Endplate
FEP12 10 Dyn(q 100.4 9.8 40 90.7
&acture, bolt
Weld-endplate
FEP13 10 S&dc 74.7 10.1 - 89.1
interface
Weld-endplate
FEF15 10 Stadc 81.5 9.6 - 97.4
inkrface
Endplate
&acture, weld-
FEP17 10 Dyn (A) 91.4 7.1 40 871
endplate
interface

A22
AlA.l Flexible endplate test 7 (FEP7)
10mm thick endplate

Dynamic test with single loading ram

90

80

70

0.05 0.1 0.15

Tlmc(s)

A23
 * Over prediction of yield moment and post yield stiffness

A14.2 Flexible endplate test 8 (FEP8)

10mm endplate

Static test with single loading ram.

100
.
90 - - -
"
--

80

70
>
/
60
% ? ,
50

o 40

30

Predicted
10 hxper imcnttil
0 1 i
0 1 2 3 4 5 6 7 8 9 10 11 12
Rotation (deg)

Al.4.3 Flexible endplate test 10 (FEPIO)

lOmm endplate

Dynamic test with single loading ram

80

70 -

60

0 0.025 0.05 0.075 0.1 0.125 0.15

Rotation (deg)

A25
AlAA Flexible endplate test 11 (FEPll)
10mm thick endplate

Dynamic test with single loading ram

"S 70 .

0.01 0.02 0.03 0.04 O.OS

Rotation (deg)

A27
AL4.5 Flexible endplate test 12 (FEP12)
Dynamic test with single loading ram

-o 60
u 50
d
40
A 30

0 0.01 0.02 0.03 0.04 0.05 0.06

Time ($)

(a) Rotation - time (b) Rotational v d o d ^ - time

FYedicfed Predicted
r
--Expenmimtali , - - ExpcnmenW
7 -J ~ 1" ' /
! ; f \

/ /
' i ' *
. .
/ /
/ /
f /
/ /
35 - V / /
/
.
% t /
/ /

1 , / f y/
1 ^ /
/
/
/
/ /
y . ,t /
/ /
//

2 4 t /

1 - / y*j
1

0 -* 1 1 1 1 r-'
O.Oi 0.02 o.o:* 0.04 0.05 0.06 O.Oi 0.02 0.03 0.04 0.0) 0.06
Time (*) Time fs)

A29
Al,4.6 Flexible endplate test 15 (FEP15)
Stadc single ram test

% 60 -

Predicted
-o- Experimental (FEP15)
: 1 1 1 i 1 1 i 1
0 1 2 3 4 5 6 7 8 9 10
Rotation (deg)

* Close prediction of yield moment, plastic stiffness and ultimate capacity

# Bearing rotation is over predicted by a;q)roxiniate]y degree

A1.4J Flexible endplate test 17 (FEP17)

Dynamic test with single load ram

100

3 70 -

= 40 -
<30 -

0.02 0.04 0.06

Rotation fdeg)

A31
Appendix B

Bl. Single Degree of Freedom (SDOF) Methods

Bl.l Introduction
An investigation into current methods of structural analysis relating to the jGeld of blast and
explosion loading is presented. In addition a review of blast phenomena is conducted to provide
background to the threat and expected loading conditions.

A common method for analysis of blast effects on structures is the single degree of 6eedom
method (SDOF); which simpliGes the behaviour of a structure or structural element to one point in
the element with an equivalent mass and sti&iess. The single displacement variable of this model
can be found and then used to assess the behaviour of the system. SDOF systems have been widely
used to predict the e^ect of blast on a structure. TM5-1300 (1990) - .Sfrucfweg fo r&gisf (Ae
presents methods of design for structures to resist blast loading based on
SDOF systems. The method is based upon early work by Biggs (1964). Mays and Smith (1995)
provide an explanation of the basis of equivalent SDOF systems and the application of these
techniques to the design of elements is demonstrated for steel and concrete.

The equivalent single degree of freedom method uses the conservation of energy, internal strain
energy and virtual work to create an equivalent lumped mass-spring system. The transformation
Actors used to determine the equivalent mass, resistance and loading are based upon the deflected
shape function of the system and the distribution of mass and loading over the original structure.
The dynamic reactions of the element can be calculated using dynamic equilibrium.

An example of an equivalent SDOF system is shown in Figure 1-1.

Dimturbing
force, Fm(t)
Disturbing force, F(t) Equivalent
mass, Me

Displacement,
Maw end stiffness, M _ . ,
andK Equivalent
stiffness, Ke

a) A c t u a l P r o b l e m b) E q u i v a l e n t P r o b l e m

Figure 1-1: S D O F system

The equation of motion for the equivalent system is shown in Equation B l . Note that it is general
practice to ignore damping due to the short load duration.

Wgii + kgu = Fg(f) Bl.

Where Mg, w and fgCf) have the same meanings shown in Figure 1-1. Further analysis of the
method is presented in Section B1.3 but first expected loads need to be characterised.
Bl
Appendix B

B1.2 Background to the threat

Bl.2.1 Mechanism of a blast or explosion
In order to design structures to resist blast loading, the blast itself must be quantised. An explosion
is essentially an extremely rapid release of stored potential energy. How this energy is stored
deGnes the classiGcation of the explosive. For example the rupture of a pressure vessel can result in
shock waves within the air in what is often referred to as a mechanical explosion. Of interest in this
thesis is the use of chemical e^glosives; where the energy is stored within the explosive material as
chemical bonds. In general there are two main types of chemical explosive; high and low.

A hig^ explosive material can be deGned by a shock wave travelling through the explosive material
faster than the speed of sound which is known as detonation [Tyas (2010)]. Once the initial
chemical bonds are broken and energy is released, the pressure rises in the explosive
instantaneously. This increase in pressure causes surrounding bonds to be broken thus starting a
chain reaction until the resistance of the unreacted explosive balances the driving force of the
reacting material. These steady state conditions deSne the detonation wave speed for the explosive.
This speed is often used to indicate the power of the explosive.

The release of energy in a hydrocarbon explosion is signiScantly slower than that of a high
explosive. Typically the oxygen required is not stored within the chemical compound and thus the
chemical reaction is usually driven by the rapid burning of fuel with oxygen from the surrounding
air. This process is called deflagration and the intensity of the explosion is dictated by the fuel-air
mixture. The ideal percentage mixture for each fuel is called the stoichiometric mixture.

In either case once the reaction reaches the charge or fuel edge there is no more reactant to
perpetuate the reaction and the wave enters the surrounding air. The resulting wave of pressure
within the air is called the air blast. For a high explosive this wave can typically propagate at a rate
of 8000 m/s whilst for a hydrocarbon explosive this Ggure is more often 2000 m/s.

The interaction of this blast wave with surrounding objects is of most interest to structural
engineers as this is what constitutes the blast loading.

Bl.2.2 Determining the design characteristics of a blast load

Two key aspects of a blast wave are the incident or side-on pressure and the reDected pressure. The
incident pressure ( f j is that associated with a blast wave travelling unimpeded in free air or
parallel to a s u r f ^ . The reflected pressure (f^) is that which develops upon contact with a rigid
targeL This includes the effect of the air particles in the blast wave being brought to rest and
conversion of their kinetic energy to an increase in air density. Thus the reflected pressure is
always higher than the incident pressure. The term overpressure is used to deGne the pressure over
and above normal atmospheric pressure (fg).

B2
Append tc B

1
CL \
2
a \
i \
5
o
\
V

Time (ms)

Figure 1-2: Typical pressure-time history

The time of arriva] (fg) is the time that the blast wave strikes the rigid surface and the length of time
this pressure acts is called the time of duration (f^). As can be seen from Figure 1-2 the pressure
then becomes negative and reverse loads the object. Although this can be i n ^ r t a n t for long
duration loads (i.e. hydrocarbon or gas explosions) it is often disregarded for high explosives where
the principal damage is assumed to be caused by the very h i ^ peak pressures. A final important
parameter of a blast wave is the specific impulse ( Q which can be found by integrating the pressure
over the loading duration. The actual total impulse can then be calculated by multiplying this by the
area of the impacted target.

The methods of predicting these parameters W1 into three categories:

Mathematical
Empirical
Numerical
The mathematical methods provide a closed form solution of the shock equations in air [Tyas
(2010)]. Once formulated these are quick to perform but have limited ^)plicability. Air is assumed
to behave as an ideal gas travelling with a shock speed perpendicular towards a target. This
assumption works well for &r-field explosions but breaks down for very high intensity shock
waves.

U
M=
a
Mach number Where o is the ambient wave speed in air B2.

is the shock front velocity

7(^2 - 1)
Incident pressure P, = P. B3.

B3
Appendix B

n* . 8W: + 4

UnforOmately even simple situations such as a real spherical shock cannot be calculated using this
method so alternatives have been developed.

Following World War n the USArmy (1946) produced a report entitled c / im/wcf o W
which analysed the work conducted on high explosives during the war and details the
estimation of pressures and impulses. This method is an empirical approach Grst noted by
Hopkinson in 1915 which uses what is referred to as cube-root scaling which Baker, Cox et al.
(1983) describe as:

wovgg org af WenZico/ J c a W wAen nvo oyZofive

cAargigs ggo/mgfTy o/kf game eap/bfzve ore defo/iafeff
w fAg awMg afmo^Agrg".

This allows e use of a scale distance parameter z to predict peak pressures as long as the
equivalent w e i ^ t of c h a r ^ in TNT is known. This system works because as the spherical blast
wave expands in three dimensions, the energy density associated with the wave decreases by the
same &ctor. The energy of the explosion is directly proportional to the charge mass therefore a
scaled distance can be used.

Scaled distance 25
f = standoS distance (m)
W = charge mass (kg)
A convenient way of presenting blast wave parameters is to use charts produced from experimental
data plotted against scaled distance. Suitable charts can be found in TM5-1300 (1990) and Mays
and Smith (1995) and programs such as CONWEP use these to produce solutions for problems
with simple geometry. These provided a quick way of gaining initial estimates of the blast
pressures, however a big drawback is the inability to include the interaction effects of other
structures or more complex geometry.

The solution is either experimental work or to numerically model the air blast. Air3d is a piece of
software developed at Cranfield University as part of the prcyect "A/i fo (Ag gvo/wafion
on /inifg f g / n f - i f ^ t f g Arwcmrgg" by Rose (2001). The program uses a variant of
the Advection Upstream Splitting Method as devised by Wada (1997) to solve the Euler equations
in three-dimensions. The software can be used to investigate the interaction of objects and shapes
within a blast area and provides pressure-time history over the period of the blast.

Once the peak pressure on a structural element is calculated it is converted to an equivalent force
by multiplying by the area over which it acts and thus a force-time relationship can be predicted.

B4
Appendix B

B1.3 The equivalent method and its uses

The Equivalent SDOF method will be reviewed and used to evaluate a structural system. The US
Army Coips of Engineers, USAGE (1957), outlines the foundation for this method as well as
providing a detailed explanation for the use of a shape function for the deflected shape. This is used
to predict the deflection at all points on the original structure from the single displacement in the
lunq)ed mass-spring system. Theoretically if the sh^)e function used to calculate the transformation
Aictors is correct then the response of the SDOF system will replicate the resulting displacement for
that reference point in the structure exactly. However as the shape function is found from the
deflected shape it is not possible to predict it exactly and thus assumptions have to be made. The
accuracy of the approach is thus dependent on the chosen deflected shape. The most common
approximation is to use the static deflected shape under the same uniformly distributed load from
the blast loading, however other functions can be used. The use of transformation factors can then
be used to simpli^ Equation B6:

Kj^Mu + Ksku = B6.

Where:

Xw = mass transformation factor

M = mass of structure

jQ, = load transformation factor

A j = stifkess transformation factor

A: = sti&iess

A:, = equivalent stiffness

F = loading

The calculation of these transformation factors is done by equating the work done (WD), internal
strain energy ([/) and kinetic energy (KE) of the two systems. The equivalent system will have the
same maximum displacement (wm) and initial velocity ( w ) as the actual problem. Reducing the
displacements of a multi degree of freedom system to a single displacement variable requires the
assumption of a deflected shape. As an example the fundamental mode of a simply supported beam
will be used to demonstrate kinematic equivalence between the distributed and equivalent systems.
The deformed shape for the fundamental mode is:

uW ^ B7.

B5
Appendix B

Evaluating WD omf KE for this deformed gives:

ft ^ 2
= -F(f)l4naz^ B8.
^0 ^

KE = ^ [ u(x)^ dx = 0.25 pAL B9.

2 Jo

The same energies are calculated for the equivalent system in Figure 1-1 (b) from simple spring
theory giving:

WD = FMumax Bll.

KE = 0.5 B12.

V= B13.

Equating the equations for the actual system to the equivalent ones and recognizing that /lAZ, = M
(total mass of the beam) and fZ, is e total beam static load provides:

Load Factor
f . y WD 2 2
Bern work = B14.
done

Mass Factor
2
from kinetic j/ _ ^ - 2KEu^ oz
energy *7miz

Equivalent

B16
from strain " 4 [3 ' - 2p
energy

Actual beam stiAiess for simply supported beam from TM5-1300 (1990)
table 3-8:

384F/
k 5L3

k, 7 384E7 5?^*
Factor 27384"'^^^

These reladonships have been developed based upon the maximum displacement and the kinematic
equivalence of e two systems. The &equency of the equivalent system will now be compared
B6
Appendix B

against the frequency for the fundamental mode of the actual problem. The exact solution is given
by Baker, Cox et al. (1983) as:

E/
W = B19.

For the equivalent system.

= =" W
This shows that assuming the fundamental mode for the deformed s h ^ gives a natural frequency
B20.

equal to the beam's fundamental frequency and that the equivalent system's behaviour is correct.

The same process as detailed above can be done for assuming the deformed shape of the statically
deflected shape given below:

S^tic deflected ^ ^ B21.

shape 5L*

Equating the energies of the two systems gives the following results.

Km = 0.50

Kt = 0.64

kg = 0.64A: therefore Ks = 0.64

The stiKness of the equivalent system is the internal force trying to restore it to its static position
therefore it can be shown that the stiffness factor should always equal the load ^ctor.

B22.

Using these modified stiffness and mass factors, the frequency of e equivalent system is now:

w = 9.889 B23.
MA3
\

Some difference is to be expected between the static deformed shape and the fundamental mode
&equency due to the inclusion of other modes other than the first however this value is still within
0.2 percenL

To reduce the number of coeSicients required for analysis, it is common to define a load-mass
factor, where:

B7
Appendix B

B24.

Using these new parameters, Eqn 3.6 can then be simpliGed as follows.

+ ku = F(t) B25.

The equivalent system equation of motion can therefore be found using a single transformation
factor and assumed deflected shape. Under the loading, the system may undergo a change of state
from elastic through to plastic and a similar method can be done to produce transformation factors
for both plastic and elastic deformations with a variety of support and loading conditions as shown
in Figure 1-3. It should be noted that although a coefficient is applied to the mass to give an
equivalent mass, the actual stiSness of the element is used throughout the analysis.

Ronge Lwd Mojts

of
Edge Conditions and Loodi'ng Diogmmg8eh9v Focfer Factor
for Foe (or Kw
Eloiiic 0*4 (k80 O.M
PtosJic (180 033 0.6#

SlQitic 10 Oi4$ 0.49

Wl I . U2 Pioitic LO a33 033

C]>9)tic OW 048 07e
Elosto*
Plutlc 0*4 090 0.7a
Piaitic 050 053 0.66
Elostk 1.0 043 043
eiosto- 1.0 0.49 0.49
L/2 r L/Z lA oza 0.%
Elostlc 0@3 04I 0.;,
Elodo 064 050 0.73
MemHe
MwMc oao 033 OW
Elostic LO 037 0.37
L/2 _l L/2 Ptashc 1.0 033 033

CIcs'ic 040 OM OM
Plastic 0.30 033 ! oee

Elastic 1.0 024 0.Z4

Plaatic 1.0 033 0.33

Eiaatic OS2 oeo

I L/3 I L/3 I L, ^ Ptostic 1.0 ON OB6

Figure 1-3: Transformation factors for one-way elements from Table 3-12 of TM5-1300 (1990)

Bl.3.1 Numerical solution or constant-velocity method

Numerical methods are used to allow the response of the system to be analysed at each time step as
the loading is applied. The process starts at time equals zero, where the initial conditions are
known, and solves the equation of motion step by step for discrete time intervals. The method
assumes tMt between discrete time periods the velocity is constant (see Figure 1-4) and hence has
become commonly known as the constant-velocity method. The displacements, w, and at time

B8
Appendix B

periods j and respectively are used to calculate the acceleration at time s using the equation of
motion. The next stage is to determine the displacement at time g+7 using extr^x)lation.

Hg+l = U; + B26.

Where is the average velocity between time periods s and g+7. The average velocity is simply:

We
+ iLAC B27.
At

This is found by assuming that is an average acceleration throughout the time period. This is

equivalent to approximating the acceleration time curve to a series of straight lines as shown in
Figure 1-4.

Substitution of Eqn 3.27 into Eqn 3.26 gives:

y-s+t ^ 2Us - B28.

This allows the displacement at the next time step to be calculated.

The accuracy of the calculation is limited by the time interval, At, being kept small in comparison
to the rate of change of acceleration. Biggs (1964) states that accurate results can be achieved by
keeping the time interval smaller than 1/10 of the natural period of vibration of the system.

Assumed constant-
vek)cAy time Wocks

Actual accekratlon-
time curve

Figure 1-4: Constant-velocity procedure

The displacements can then be found for each time step starting at time t = 0. However a problem
arises in the Srst time period in that the previous displacement is unknown. It is therefore assumed
the acceleration is constant during the first time interval. This makes the displacement, for the first
time period only:

B29.

The equation of motion is thus solved using a spreadsheet and for each successive time period the
displacement is calculated.
B9
Appendix B

The advantage of this method is that the displacement of the lumped-mass is calculated at each
discreet time step. When all of these displacements are plotted on the same graph, the overall
displacement-tiinG history is depicted.

The general properties of a single degree system are given by;

Natural angular frequency: = J

Natural time period: T ^ 271^ B30.

Natural &equency: / = ^ ^

B1.4 Bending moments and shear forces

For simply supported beams with a distributed load, TM5-1300 guidance includes the method to

calculate the mammum support shear using:

Maximum resistance*Length
IvfaauiimrD Slieaf ==

The iTiATimnm resistance is calculated from the ultimate resistance of the beam. Formula for
alternative support and loading conditions are shown in Figure 1-5.

Edge Conditicos ond Loading D(oyom# Syppcft R#ac,ioom,Vm

r#L
2

L/2 I L/2 Z
L. Reoctlon

R. Roeian

L. Reaction IIRy
16
, L/t R. Reacfion

r.L
r" 2

L/% J L/2

111I UL

Ru
L/ajL/ai 1/3 2

Figure 1-5: Calculating support shear for one-way elements from Table 3-9 of TM5-1300 (1990)

BIO
Appendix B

An alternative method to calculate the shear force is to use the dynamic reactions method where the
analysis includes the inertia forces of the beam upon deflection. The free body diagram of a beam
subjected to dynamic loading is given in Figure 1-6.

Figure 1-6: Free body diagram of dynamic problem

The distribution of the inertia force acting on the beam, is identical to the deflected shape due
to the motion being harmonic. The dynamic reactions acting at the supports, depend upon both
the load, f (f), and the inertia force.

A convenient way of calculating the dynamic reactions can be achieved by considering one half of
the beam as shown in Figure 1-7. From symmetry the shear force at midspan, S, equals zero.

0.5 Fd)

Figure 1-7: Determination of dynamic reactions

The centroid of the inertia force acts at a distance JCf from the left hand support. The total moment
of area equals the sum of all the individual moment of areas this can be rearranged to find the point
at which the moment area acts, where u is a function of % de6ned by the deflected shape.

4 r /2
I u- = I lur' B31.
^0 -/Q

jg ux dx
z, = B32.

The deflected shape assumed will thus alter the position about which the centroid of the inertia
force acts.

Once the point at which the inertia force acts has been found it is possible to take moments about
this point thus obtaining:

Bll
Appendix B

- Wmw - ^Zf--j-0 B33.

The bending moment at midspan (Mmw) is calculated from the defected shape using the engineers
bending formula:

S'^u
B34.

Rearranging Eqn 3 3 3 and substituting for M gives:

1 / F(OZ.\ F(f)
fir;

This is a general formula to calculate the dynamic support shear force for any assumed deflected
shape.

Bl.4.1 Elastic behaviour

The shape functions most often used in analysis are the fundamental mode or the equivalent static
deflected shape given below [Baker, Cox et al. (1983)].

B36.

The point where the inertia force acts, x;, is found using Eqn 3.32:

jV2 _ 2lx^ + l^x)] X dx

Solving this gives:

61L
B38.

From Eqn 3 3 6 it is calculated that:

5^u 16u-^a-^(12Lx + 12x^)

B39.

Substitution from Eqn 3.34, the bending moment at midspan (%=l/2) is therefore:

48uina% rr
Mmuf = B40.

Equating Equations B35, B36, B38 and B40 provides:

B12
Appendix B

30.216U,
yCt) = 0.1066F(f) + B41.
L3

This can be used to calculate the equivalent support reactions throughout the analysis procedure.

An alternative shape function is the Grst mode of vibration:

u(x) = UmaxSin ( y ) B4Z

The same method is used to calculate the location of the inertia forces and it can be shown that i:/ is

. Similarly the midspan moment is . Substituting these values into Equation B31
;r U
gives the dynamic reactions as:

Elir^u,
V(t) = + Fa)(o.5-0
L3
B43.
31.006U
= 0.1073F(t) + EI
L3

The dynamic reactions of a typical SDOF system were analysed using the two shape functions
outlined above. The results are plotted on the same axes and are shown in Figure 1-8. These results
show that the two shape functions produce similar dynamic reactions but that using a sine curve
predicts slightly higher peak values (less than 1% diC^erence).

- S M i c Deflected Shape
Sin CuMe Defected Shepe
4 0 0 0 0 0 . 0 0

3 0 0 0 0 0 0 0

S 2 0 0 0 0 0 . 0 0 -I

100000 00

(3 .100000.00 -
c>
.200(X)0.00 -

- 3 0 0 0 0 0 . 0 0

Time (seconds)

Figure 1-8: Comparison of dynamic reactions using sin curve and static deflected shape

A similar comparison has been conducted for the midspan bending moments and once again the
sine curve produces a slightly higher peak.

B13
Appendix B

It must be noted that neither of these functions may be suitable assumptions under blast loading.
This is especially the case for strong, short duration loading where high curvature may expected at
the supports producing much higher shear forces.

Bl.4.2 Plastic behaviour

Equations B35 and B37 were calculated for elastic behaviour. If yielding occurs before the
mmYimiim load, then the beam plastic moment, Mp, replaces the midspan moment in the equations.
For the Grst mode of vibration (Eqn B43) 7 becomes:

igZMm
V(t) = 0 . 1 0 7 3 F ( t ) + B44.

Further, the plastic moment can be expressed in terms of the mATimTim resistance, deGned as
the magnitude of the stadc load necessary to cause a plastic hinge at the center of the beam, Smith
and Hetherington (1994).

= B45.

Thus Eqn. B44 can be expressed in the form:

24
V(t) = 0.1073F(t) + O . l l F M + 0.39A^ B46.

Eqn. 3.46 is of the form given in Smith and Hetherington (1994) Table 10.2 and many other texts
on blast loading. A difference of less than 1% is obtained when the coefGcients are calculated &om
the static deflected shape.

BL5 Summary of SDOF methods

Single degree of freedom systems have been used for a long time to allow the analysis of structural
elements under dynamic conditions and have been shown to provide accurate results [Naito and
Wheaton (2006)] as long as certain criteria are met in the analysis. Limitations of the method
include:

* End condition assun^)tions must be made (rigid or sin^)le).

* Local deformations of model are not possible.
* Non-symmetrical loading conditions are not accounted for.
* Whilst the method was developed to predict the correct midspan deflection, the
corresponding bending moments and shear forces are calculated &om the assumed
deflected shape which is not always the same as the static deflected shape. This
can lead to gross underestimations of loading at the support conditions as

B14
Appendix B

Paramasivam (2008) found. Alternative deflected shapes can be used but these
must be verified experimentally.

Taking into account these limitations, SDOF systems are useful in providing a quick method of
analysis however a more detailed methodology which can improve upon these limitations is soughL

B15