You are on page 1of 41

Chapter 1

INTRODUCTION

1.1 General

Terrorism is a great concern for the world today. With the development of science and
technology the capability of the terrorists to destroy a structure has also been increased. At the
same time to cause maximum hazard, they aim at the vital structures which can breakdown the
bone marrow of the nation or the locality. So to protect our existence, the vital structures should
be resistant to a certain level of terroristic activity.

Water tanks play the most vital role in the existence of population. Because if a water tank is
destroyed, it affects a huge population. Similarly the petroleum tanks and chemical tanks play
very important role in the transportation, production, economy of a country. But if they fail due
to terroristic attack, then it may be catastrophic and at the same time paralyzing also. So these
structures should be equipped through proper research in order to prevent the destruction.

Though research has been carried out in the field of blast response of tanks, no research work
has been done in the base isolation of these tanks. It has already been proved that the isolators
reduce the response of structures due to dynamic loadings, so the isolation will definitely have
some effect on the response of tank. The research proceeds in the track line as:-study of
explosives, process of detonation, evaluation of blast pressure, response of single degree of
freedom system to impact and blast pressure, numerical analysis of response of isolated and non-
isolated tank using three lumped mass model system, finite element modeling of tank and liquid
mass and finding the response of fixed base and isolated tank and validating the results.

The dynamics of explosion on the basis of conservation of energy is studied by Henrych


(1979). Different blast wave front parameters have been studied by Rankine (1870) and Brode
(1955) independently. Baker (1983) and Kingery (1984) have represented the blast wave
parameters graphically. Blast wave effects on plates have been studied by Rajendran and Lee
(2009). Remennikov (2003) has studied blast pressure on buildings. United states army has
developed a manual TM-5-1300, now known as UCF 3-340-02, which gives details of blast
resistant design. Tawadros and Glockner (1972) carried out experiment to get dynamic response
data of shells under simulated blast wave. Ruiz et al. (1989) studied the elastic response of thin
walled cylindrical shell to blast loading; analyzed both rotationally symmetrical and sideways
loadings; the results obtained from both loadings were compared with those obtained
experimentally. Gefken et al. (1988) studied the response modes of thin cylindrical shells when
impulsively loaded with radial axisymmetrical loads; both internally pressurized and
unpressurised shells were analyzed. Xi et al (2000) gave theoretical and finite element solution
for an orthotropic thick cylinder under arbitrary impact loading. Jhung et al (2006) investigated
the dynamic response characteristics of a structure impacted by high speed projectile and
validated using LS-DYNA and ANSYS. Housner (1963) has given two-degree of freedom liquid
lumped mass model. Haroun (1983) has modeled the liquid as three-degree-of –freedom liquid
lumped mass model. Shrimali and Jangid (2003, 2004) studied the response of base isolated
liquid storage tanks to seismic excitation. Jaiswal et al. (2007) reviewed seismic codes on liquid
containing tanks.
1.2 Objectives
The aim of the present project is to study the blast response of fixed and isolated liquid tank
numerically and by rigorous finite element modelling and to compare the results. In specific,
objectives are:
1. To numerically find the response of isolated liquid filled tank under blast loading using
three-degree-of-freedom liquid lump mass model.

2. To compare the results with non-isolated liquid filled tank.

3. To find the response of tank with and without liquid.

4. Validating the results.

5. To model the liquid mass and the tank using rigorous finite element analysis

6. To do parametric study

1.3 Organization of report

The report consists of four chapters including this introductory chapter. The chapter 2
consists of the review the state of art of explosion process, evaluation of blast pressure and other
blast wave parameters, liquid mass modeling, blast response of non-isolated tank and earthquake
response of isolated tank. Chapter 3 consists of response of single degree of freedom system to
impact loading and blast loading. Chapter 4 consists of response of isolated and non-isolated
tank using three degree of freedom lumped mass model.
Chapter 2

LITERATURE REVIEW

2.1 General

The study of blast effect is a combination of physical and chemical phenomenon. When
an explosive is ignited, it decomposes violently with evolution of heat and gases. The rapid
expansion of these gases results in shock pressure on the surface of solid bodies. The detail
process of explosion, explosives used, and the evaluation of blast pressure are discussed in this
chapter. The state of art is studied under the headings of liquid mass modeling, blast effect of
tank without isolation and earthquake response of tanks with isolation.

2.2 Explosions

Explosions can be physical, nuclear or chemical. When explosion takes place due to
sudden failure of physical structure is called physical explosion e.g. catastrophic failure of a
cylinder of compressed gas. When explosion takes place due to formation of different atomic
nuclei by redistribution of the protons and neutrons within the interacting nuclei is called nuclear
explosion and when an explosion involves rapid combustion of different chemical compounds is
called chemical explosion. Usually in terroristic attacks, explosives used are chemical in nature.
So we will study the chemical explosion in more detail.

A chemical explosion involves the rapid oxidation of fuel elements composed of


hydrocarbon compounds. The oxygen required for combustion is contained within the
compound. Most explosives are condensed under high pressure. When explosives react, they
decompose violently with evolution of heat and gases. The rapid expansion of these gases results
in generations of shock pressure in any solid materials which are in near vicinity.

2.2.1Explosives

The explosives which are usually used in military and terroristic activities are organic by
origin. The elements which constitute the explosives are hydrogen, oxygen, carbon, nitrogen, and
sulphur. Some commonly used explosives are
1- Pentaerythritol tetranitrate (C5H8O12N4), commonly known as PETN is used as a
main charge in munitions and also in demolition detonators. The heat of explosion or
mass specific energy Q of this material is 5940kJ/kg and generates 0.79m3 of gas per
kg of the material.
2- 1,3,5-trinitro-1,3,5-triaza-cyclohexene (C3H6O6N6), commonly known as cyclonitrite
or RDX is used as a filling for military munitions and special applications such as
linear cutting charges.
3- 1,3,5,7-tetranitro-1,3,5,7-tetra-aza-cyclo-octane (C4H8O8N8), commonly known as
HMX, which is a white crystalline solid used as a filling for high performance
munitions.
4- N-methyl-N-nitro-2,4,6-trinitroaniline (C7H5O8N8), commonly known as tetryl or CE
which is used in military detonators.
5- 2,4,6-trinitrotoluene (C7H5O6N3), commonly known as TNT. Probably it is the mostly
widely used explosive. It is a pale yellow-brown crystalline solid having mass
specific energy 4520kJ/kg and produces 0.73m3 of gas per kg of TNT. It is used in
military munitions being mixed with other explosives such as RDX, HMX or
ammonium nitrate.
6- Nitroglycerine (C3H5N3O9), an oily liquid, commonly known as blasting oil. It
explodes with great violence producing 6700kJ/kg of heat and 0.74m3/kg of gas.
7- Nitroglycol (C2H4O6N2) explodes releasing very high energy i.e. 6730kJ/kg of
material and liberates 0.74m3 of gas.

Among all the materials TNT is the most widely used and very efficient explosive. So it
has been accepted as the basis of conversion for all other explosives. An equivalent conversion
factor for all the explosives has been decided according to their mass specific energy. The factors
are given in tabular form below. For calculations of different blast parameters the explosive
mass is converted to TNT mass by multiplying the TNT equivalent.
Table 2.1

Mass specific energy TNT equivalent


Explosive
Q(kJ/kg) Q/QTNT
Amatol 80/20(80% ammonium
2650 0.586
nitrate, 20%TNT)
Compound B(60%RDX, 40%
5190 1.148
TNT)
HMX 5680 1.256
RDX 5360 1.185
Lead azide 1540 0.340
Mercury fulminate 1790 0.390
Nitroglycerine 6700 1.481
Pentolite 50/50 (50% PETN, 50%
5110 1.129
TNT)
PETN 5800 1.282
Tetryl 4520 1.00
TNT 4520 1.00
Torpex (42% RDX, 40% TNT,
7540 1.667
18% aluminium)
Blasting gelatin(91%
nitroglycerine, 7.9%
4520 1.00
nitrocellulose, 0.9% antiacid,
0.2%water)
60% nitroglycerine dynamite 2710 0.600

2.2.2 Terminology

Combustion is the term used to describe oxidation process in which oxygen required may
be consumed from outside or that contained within.
Deflagration is the combustion process in which the rate of decomposition is much lower
than the speed of sound in the material. Deflagration is propagated by the liberated heat of
reaction. The flow direction of reaction products is in opposition to the direction of
decomposition.

Detonation is the form of reaction of an explosive which produces a high intensity shock
wave. The reaction rate is described by detonation velocity lies between 1500m/s to 9000m/s
which is much faster than thermal conduction and radiation.

Blast wave is the highly compressed layer of air which contains high pressure energy and
so it propagates or expands at a very high speed in order to balance its pressure difference it
possesses as compared to the atmosphere.

2.2.3Generation and propagation of blast wave

When a condensed high explosive is ignited, the explosives produce gases at a high rate
as a product of detonation which is can be at a pressure of 10-30Gpa and a temperature of 3000-
4000o C. A violent expansion of gases takes place and the surrounding air is forced out of the
volume it occupies. As a consequence the layer of air surrounding the gaseous products is
compressed, and this layer, which contains high pressure energy, is called the blast wave. As
disequilibrium is established between the blast wave and surrounding air, the blast wave travels
outwards from the centre of the explosion (expansion of blast wave). The pressure in the wave
front decreases as it moves away from the centre of explosion. Due to the momentum of the
gases, it over-expands and the pressure decreases below atmospheric pressure. This creates a
negative or suction phase, which thus applies a decelerative force to the surrounding gas
molecules, resulting in a backward flow of air towards the explosion centre.

The variation of blast pressure with time at a fixed point is shown in the fig. 2.1.ps is peak over
static pressure, p0 is ambient pressure or atmospheric pressure, Ts is the time of positive pressure,
ts- is time of negative pressure and ta is time of arrival. The negative pressure is much less than
positive pressure. The plot shows that after the time ta the structure exhibits the maximum
pressure ps and then the pressure decreases exponentially to a negative pressure and then attains
equilibrium. The positive pressure is of our concern.
p

ps

Peak
overpressure
p(t)

p0
Ambient
pressure

t
ta Ts ts-
Positive phase Negative phase
duration duration

Fig. 2.1 variation of pressure with time

2.2.4Blast wave parameters

During the detonation process, a detonation wave generates and propagates in the
explosive. The parameters that are used to assess the detonation performance of an explosive are
the Chapman-Jouguet (C-J) detonation pressure, the temperature of detonation and the
detonation velocity (Rajendran and Lee, 2009). Typically for an explosive(TNT) with a density
of 1650kg/m3, the Chapman-Jouguet (C-J) detonation pressure is 21,000MPa, the detonation
temperature is 3720k and the detonation velocity is 6950m/s. once the process of detonation is
completed, the interaction of product gases with the surrounding medium takes place. The
product gases with high pressure and temperature expand outward by generating pressure wave.
The gaseous products are assumed to be in viscid at high temperature and thus the viscous forces
are not considered for explosive modeling. In air explosion, the shock wave moves with the gas-
air interface. An equation of state (EOS) of the explosive relating energy, pressure and volume is
essential for the numerical modeling of the detonation process (Rajendran and Lee,2009).the
most commonly used EOS to describe the state of detonation products is Jones-Wilkins-Lee
(JWL),which is given as
    R1V     R2V 
p jwl (V ,U in )  A 1   e  B 1  e  E (2.1)
 RV 1   R2V  V
Where A , B , R1 , R2 and  are material constants, p jwl is the pressure, V is the relative volume

compared to the initial volume of the explosive and U in in is the internal energy per volume and
E is the initial energy per unit mass of explosive. The first in the JWL equation is known as the
high pressure term dominates first for close V to 1, the second term is influential for V close to
2 and last term corresponds to the expanded state.
The blast pressure time history is given by Friedlander’s equation (Smith and Hetherington,
1994)
 t  b  (t  ta ) 
p(t )  ps 1   exp   (2.2)
 Ts   Ts 
Where, ps is peak over static pressure

Ts is time of positive blast pressure

t a is time of arrival

b is decay coefficient
Peak over static pressure is the maximum pressure at a particular point which may occur for a
larger distance with a larger intensity of blast and can also occur for less distance and less
intensity of blast. So, scaled distance is used to evaluate the intensity of blast pressure at a point
considering the distance and weight of explosive used in terms of TNT (Smith and Hetherington,
1994).
R
Z= (2.3)
W1 3
Where, R is the radial distance from the center of explosion
W is the weight of TNT in kilogram
For chemical explosions, the peak overpressure is expressed as (Brode, 2009):

(a) Near field explosion condition,


6.7
ps   1 bar (ps> 10 bar) (2.4)
Z3

(b)Far field explosion condition,

0.975 1.455 5.85


ps   2  3  0.019 bar (0.1<ps<10 bar) (2.5)
Z Z Z
For chemical explosions, the peak overpressure is expressed as (Rajendran and Lee, 2009),

808 1   Z 4.5  
2
pS
   (2.6)
p0 1   Z 0.048  1   Z 0.32  1   Z 1.32 
2 2 2

The blast wave velocity U s , and the peak dynamic pressure, qs is given by (Smith and

Hetherington, 1994)

6 ps  7 p0
Us  a0 (2.7)
7 p0

5 ps 2
qs  (2.8)

2 ps  7 p 0 
Where, p 0 =ambient atmospheric pressure

a0 =speed of sound in air

The peak particle velocity, U p , peak wind velocity behind the shock front is ( Rajendran, 2009)

 
5 ps  a0 
Up  (2.9)

7 p0 1  6 p 7 p 2 
1
  s 0 

The air density, s , of the air behind the shock front is related to the ambient density, 0 as
(Smith and Hetherington, 1994)

6 ps  7 p0
s  0 (2.10)
ps  7 p0

The peak overpressure ps is related to the ratio of maximum positive overpressure (pso+) to the
maximum negative suction pressure (pso-) as (Lam et al., 2004),

 p 
log e  b so-   b  1  0 (2.11)
 pso+ 
Also, a commonly used simplified equation for the decay coefficient expressed with reference to
scaled distance, Z is expressed as (Lam et al., 2004),

b  Z 2  3.7 Z  4.2 (2.12)

The time of arrival of shock wave for a radial distance R from an explosive of radius re is given
as (Rajendran and Lee, 2009)

1  2
r
1
ta     dr (2.13)
a0 re 1   6 ps / 7 p0  

The time of positive pressure for chemical explosion in millisecond is given as (Rajendran and
Lee, 2009)

980 1   Z 0.54  
10
TS
   (2.14)
W 3 1   Z 0.02   1   Z 0.74   1   Z 6.9 
1 3 6 2

The impulse of the shock wave is given as (Smith and Hetherington,1994)

ta Ts

is   p t  dr
ta
(2.15)

2.3 Liquid mass modeling

To analyze a liquid containing tank numerically, it becomes inevitable to model the liquid
mass according to their responses to external forces so as to define the degree of freedom.
Housner (1963) developed a two-degrees-of-freedom theoretical lumped mass model of ground
supported liquid storage tank which are associated with sloshing mass and impulsive mass.
Rosenblueth and Newmark (1971) modified the expressions of Housner to estimate the sloshing
and impulsive masses to evaluate the seismic forces coming on to the tank. Haroun (1983)
developed a three-degrees-of-freedom lumped mass model of ground supported liquid storage
tank which are associated with sloshing mass, impulsive mass and rigid mass. The part of the
liquid which moves independent of tank movement is called sloshing mass and the other part
which moves in unison with the tank is called impulsive mass. When flexibility of tank wall is
considered then the part of the impulsive mass moves independent of tank movement and the
other part moves along with the tank called the rigid mass. Haroun has considered the flexibility
of tank wall where as Housner has ignored that. Haroun has also developed design charts for
estimating these masses with the assumptions that the liquid contained in the tank is
incompressible and has irrotational flow. Shrmali and Jangid (2003,2004) have used the three-
degrees-of-freedom lumped mass model to analyze the response of elevated liquid storage tank
to seismic excitation.

mc u
c

mi u
i
kc
cc ki
ci
mr ub
kb cb

(a) (b)

Fig. 2. Liquid mass models used by Shrimali and Jangid

Jaiswal et al. (2007) reviewed many seismic codes on liquid containing tanks across
world and said that as liquid storage tanks possess lower energy dissipating capacity than
conventional buildings and during external excitation, tanks are subjected to hydrodynamic
forces, they have been assigned with higher seismic forces compared to conventional buildings
by all the codes. The modeling in which two-degrees-of-freedom model is referred as rigid tank
wall model and those which use three-degrees-of-freedom model are referred as flexible wall
model. All codes except NZSEE use rigid tank wall modeling. At the same time they said that
those codes which use rigid tank wall modeling, considers the flexibility of wall in evaluation of
impulsive mode time period. So flexibility of wall is neglected in impulsive mass evaluation
(Jaiswal et al., 2007). ASCE 7 and Eurocode 8 use the absolute summation rule whereas ACI
350.3, D-110, D-115, D-100, API 650 and NZSEE use the SRSS rule to get the combined
response of the constituting masses (Jaiswal et al., 2007). The stresses in the tank wall depend on
the hydrodynamic pressure distribution along the height which may be curvilinear or linear
(Jaiswal et al., 2007). Expressions for hydrodynamic pressure on tank base is given in NZSEE
only, but the effect of hydrodynamic pressure on tank base in obtaining the overturning moment
is considered in all the codes(Jaiswal et al., 2007).

2.4 Sloshing wave height

When tank is excited by external force, the sloshing component of liquid mass undergoes
vertical displacement and so providing freeboard becomes necessary for preventing the liquid
from spilling off. Also it may damage the roof top of the tank. So different codes have given
different expressions for sloshing height (Jaiswal et al., 2007).

ACI 350.3 and D-110 (𝐶𝑠 )𝑐 𝑅0 Where, (𝐶𝑠 )𝑐 is the convective mode base
shear coefficient and 𝑅0 is the radius of the
tank.

Eurocode 8 and NZSEE 0.84(𝐶𝑠 )𝑐 𝑅0

ASCE 7 and API 650 (𝐶𝑠 )𝑐 𝑅0 𝑅𝑤𝑐 Where, 𝑅𝑤𝑐 is the response modification factor
for convective mode.

D-100 (𝐶𝑠 )𝑐 𝑅0 (1.4𝑅𝑤𝑐 )

2.5 Liquid element modeling using ANSYS

Dynamic non-linear fluid structure interaction analysis were performed for two large
steel tanks (Tank-1, with 92.3m diameter, 21.4m height and Tank-2, with 60.5m diameter, 19.8m
height) with floating roofs using finite element modeling (Asfura et al., 2003). The tank walls
were modeled with plate elements, the soils with equivalent non-linear spring elements, the fluid
with equivalent non-linear fluid elements (ANSYS element Type FLUID 80). The FLUID 80
element is highly incompressible and it is free to move vertically and tangentially relative to the
tank wall and also free to move horizontally relative to the tank roof. The elements at the
junction of the tank bottom and the foundation were modeled using a non-linear contact friction
element (ANSYS element Type CONTACT52). This element acts only in compression and
provides the sliding resisting forces between the tank bottom and foundation. The contact
elements between tank bottom and the soil and the soil itself in Tank-1 were modeled as a
combination of elements that allow uplift and friction (combination of ANSYS element Type
CONTACT52 and COMBIN40). The Tank -2 is supported on concrete pad on piles. The soil
elements below this pad were modeled by linear equivalent horizontal and vertical soil springs
(ANSYS element Type COMBIN40). These elements do not allow uplifting and the sliding of
the pad. The interface between tank bottom and the pad were modeled by non-linear contact
friction elements. The stiffness constants of these springs including the effect of the piles and the
soil, were estimated using soil pile interaction analysis using computer code SASSI. The gravity
analysis, modal analysis and dynamic analysis for seismic load were carried and the results were
matched with API and New Zealand recommendations.

2.6 Fluid structure interaction

During dynamic excitation to a fluid tank, interaction between fluid and tank wall takes
place, which is studied under fluid structure interaction. The seismic response of liquid tank was
tried to analyze using finite element modeling software, ANSYS (Austin and Rhee, 1999).
Though fluid structural element is available in ANSYS for considering the fluid structure
interaction problem, it can be used to perform unsymmetric or damped modal, harmonic and
non-linear transient analysis (Austin and Rhee, 1999). For complex structures performing a
nonlinear transient analysis is impractical (Austin and Rhee, 1999). If finite element linear
analysis method is adopted, it neglects the fluid structure interaction (FSI), so it results in more
conservative design (Austin and Rhee, 1999). Method developed by ABB and KOPEC can be
used to find the response using the response spectrum method. The component which is used to
develop this method was the KNGR (Korean Next Generation Reactor) Control Element
Assembly (CEA) Shroud Assembly (IBA). This is a large cylindrical shaped assembly which is
supported from and mounted inside of the Upper Guide Structure (UGS) barrel. The water in the
annulus between the IBA and the UGS produces the fluid structure interaction effect. The IBA
structure is modeled using plate (stiff63), solid (stiff45), beam (beam4) and lumped mass (mass
21) elements. The gap between the IBA and the UGS is kept 1.125 inches.
Method is available for fully coupled fluid-structure-soil interaction analysis of liquid
storage structures on/in horizontally layered half space (Yun and Chang, ). To simulate the fluid
structure interaction effect, the contained fluid is modeled using mixed finite elements with two
fields (displacement and pressure) and the structure-soil interaction effect is considered by using
finite elements and dynamic infinite elements.

2.6 Blast or impact loadings on similar structures

Non-linear dynamic response of PVC shells with different geometrical shapes under
simulated blast loading was observed (Tawadros and Glockner, 1972). Spherical and parabolic
shells with 32 different shapes were observed under blast loadings and also calibrated shots were
applied on dummy models and responses were observed by high speed television cameras. The
results were presented in tabular, graphical and photographical forms.

Rotationally symmetrical and sideways loading were applied on thin-wall cylindrical


shell and strength of the shells were calculated using numerical, analytical and experimental
techniques (Ruiz et al., 1988). The rotationally symmetrical loading case the analytical and
theoretical results were validated by experimental results. For sideways loading the analytical
result was validated by experiments using similar models loaded in shock tube apparatus. Finite
element package, ABAQUS, was considered to be the only practical tool to solve the sideways
loading case and deformation shapes has been given for the case (Ruiz et al., 1988).

Structural response modes of thin cylindrical shells to external radial impulsive loads
were observed for with and without internal pressure (Gefken et al., 1988) . For unpressurised
shells the response modes consisted of dynamic pulse buckling and followed by large inward
deflection and for pressurized shells the response modes consisted of inward deflection followed
by large outward deflection. In unpressurised shell, at critical impulse, the fracture occurs at the
mid height. In shells with internal pressure, at critical impulse, during outward movement, large
hoop strains are produced as the loaded surface rebounds past its original position, resulting in an
axial fracture which ruptures the shell.

A numerical model for non-linear dynamic analysis of blast loaded cylindrical shell
structure was presented by Jiang and Olson (1990). The model was based on a transversely
curved finite strip formulation and can be used to find the transient response of isotropic and
stringer-stiffened shells. The model had been validated by comparing results with other available
results or experimental results. Responses of simply supported stiffened plate panel, simply
supported shell roof structure, simply supported stiffened shell roof structures and cylindrical
shell under blast loading was presented.

The dynamic response of water storage tank under impact loading was studied by Jhung
et al. (2006). The impact of 300 kg projectile on a water storage tank was simulated by ANSYS
and LS-DYNA. An alternate impact analysis method was proposed which is equivalent to an
explicit dynamic analysis. The effect of fluid on the tank wall was also considered. When
frequency of water tank decreases by more than 50% by inclusion of fluid in to it (Jhung et al.,
2006) i.e. fluid is a very good absorber of impact load.

The theoretical solution for an orthotropic thick cylindrical shell under impact loading is
found by making use of finite Hankel transform and Laplace transform (Xi et al., 2000).
Dynamic formulas were derived for a sudden load and followed by an exponential decrease
phase. Finite element analysis for the same problem was also carried out using Algor (Super sap)
and it was found that both the theoretical and the finite element method yield same results.

Dynamic response of a short cylindrical shell which is made from a rigid, perfectly
plastic material under blast loading is investigated by Li and Jones, 1994.
Chapter 3

IMPACT AND BLAST

RESPONSE OF SINGLE DEGREE OF FREEDOM SYSTEM

3.1 Preview

In real life situations the dynamic excitations are neither harmonic nor periodic. So we
study the dynamic response of single degree of freedom system to excitations varying arbitrarily
with time. Blast loadings are usually idealized as impact loadings, as they act on structures for
few milliseconds. Here response of a SDF system is studied for different types of impact
loadings like rectangular, half sine wave and triangular and a general procedure is presented to
analyze SDF system subjected to force 𝑝(𝑡) varying with time. This result will enable analytical
evaluation of response to force varying with time.

Our aim is to find the solution of the differential equation of motion

𝑚𝑢̈ + 𝑐𝑢̇ + 𝑘𝑢 = 𝑝(𝑡) (3.1)

Subjected to initial conditions

𝑢(0) = 0 𝑢̇ (0) = 0

For finding the general solution, 𝑝(𝑡) is considered as a sequence of impulses of


infinitesimal duration, and the total response of the system to 𝑝(𝑡) is the sum of the responses of
individual impulses. The individual responses can be found from the response of the system to a
unit impulse.

3.2 Response to unit impulse

When a very high force is applied for a very short duration but with a time integral that is
finite is called as impulsive force. As shown in fig. 3.1 force 𝑝(𝑡) = 1/𝜀 with time duration
starting at 𝑡 = 𝜏. As 𝜀 → 0, force becomes infinite; however the magnitude of the impulse
,defined by the time integral of 𝑝(𝑡) remains equal to unity. Such a force in limiting case 𝜀 → 0
is called unit impulse.
𝑝

1/𝜀
𝑡
𝜀
𝜏

Fig. 3.1 Unit pulse

According to Newton’s second law of motion, if a force 𝑝 acts on a body of mass m, the
change in momentum of the body is equal to the applied force.

𝑑
(𝑚𝑢̇ ) = 𝑝 (3.2)
𝑑𝑡

For constant mass, this equation becomes

𝑝 = 𝑚𝑢̈ (3.3)

Integrating both sides with respect to 𝑡 gives

2𝑡
∫𝑡 𝑝 𝑑𝑡 = 𝑚(𝑢̇ 2 − 𝑢̇ 1 ) = 𝑚∆𝑢̇ (3.4)
1

In the above equation the left hand side is impulse and the right hand side is change in
momentum. So the magnitude of impulse is the change in momentum. The above equation
represents a SDF mass-spring-damper system, when the effect of spring and damper are not
there. Thus for a unit impulse at 𝑡 = 𝜏, imparts to the mass, m, the velocity

1
𝑢̇ (𝜏) = 𝑚 (3.5)

The initial displacement is

𝑢(𝜏) = 0 (3.6)

Following the solution of differential equation 𝑚𝑢̈ + 𝑘𝑢 = 0 and initial conditions,


𝑢̇ (0)
given as 𝑢(𝑡) = 𝑢(0)𝑐𝑜𝑠𝜔𝑛 𝑡 + 𝑠𝑖𝑛𝜔𝑛 𝑡 , the solution for undamped system is
𝜔𝑛

1
ℎ(𝑡 − 𝜏) = 𝑢(𝑡) = 𝑚𝜔 𝑠𝑖𝑛[𝜔𝑛 (𝑡 − 𝜏)] 𝑡≥𝜏 (3.7)
𝑛

The solution of damped system is with damping constant 𝜉

1
ℎ(𝑡 − 𝜏) = 𝑢(𝑡) = 𝑚𝜔 𝑒 −𝜉𝜔𝑛(𝑡−𝜏) 𝑠𝑖𝑛[𝜔𝑛 (𝑡 − 𝜏)] 𝑡≥𝜏 (3.8)
𝑛

ℎ(𝑡 − 𝜏) represents the unit impulse response function.

3.3 Response to arbitrary force

A force 𝑝(𝑡) varying arbitrarily with time can be represented as the summation of
number of impulse function. The response of one impulse at time 𝜏 of magnitude 𝑝(𝜏)𝑑𝜏, is this
magnitude times the unit impulse function.

𝑑𝑢(𝑡) = [𝑝(𝜏)𝑑𝜏]ℎ(𝑡 − 𝜏) 𝑡>𝜏 (3.9)

Now, response of system at time 𝑡 is the summation of the responses to all the impulses up to
time 𝑡. So the displacement function becomes,

𝑡
𝑢(𝑡) = ∫0 𝑝(𝑡)ℎ(𝑡 − 𝜏)𝑑𝜏 (3.10)

Substituting the values of unit impulse response function gives Duhamel’s integral:

1 𝑡
𝑢(𝑡) = 𝑚𝜔 ∫0 𝑝(𝜏)𝑒 −𝜉𝜔𝑛(𝑡−𝜏) 𝑠𝑖𝑛[𝜔𝐷 (𝑡 − 𝜏)]𝑑𝜏 (3.11)
𝐷

For undamped system the above equation becomes

1 𝑡
𝑢(𝑡) = 𝑚𝜔 ∫0 𝑝(𝜏) 𝑠𝑖𝑛[𝜔𝑛 (𝑡 − 𝜏)]𝑑𝜏 (3.12)
𝑛

Duhamel’s integral gives the solution for evaluating the response of a linear SDF system
to arbitrary force. As it is based on principle of superposition, it holds good for linear systems
only. For complicated forcing functions numerical methods are adopted to get the solution.
3.4 Response of SDF system to pulse excitation

3.4.1 Response to rectangular pulse

Let’s consider a simple SDF without damping and which is excited to rectangular pulse
described as follows.

 p0 t  td
mu  ku  p (t )   (3.13)
0 t  td

With at rest initial conditions: u (0) = u (0) = 0. It should be analyzed in two phase i.e. forced
vibration phase and free vibration phase.

1. Forced vibration phase: During this phase the system is subjected to step force
u (t ) 2 t
 1  cos nt  1  cos t  td (3.14)
 ust 0 Tn

2. Free vibration phase: After the force application duration,

u (t )  t    t 1 td  
  2sin d  sin  2    t  td (3.15)
 ust 0  Tn    Tn 2 Tn  

The response of SDF system to rectangular pulse is shown in fig. 3.2. It is observed from the
response that the response greatly depends on the td / Tn value. When force is applied to the

structure, it oscillates with static displacement  ust 0  p0 / k and time period Tn . As we are

discussing about undamped system, there will be no decay in force after the excitation ends. So,
the system retains the deformation it possesses at the end of pulse, and continues oscillating
about its center with that displacement as amplitude and time period Tn . So when the pulse ends

at n Tn (n=1,2,3…), the system comes to rest after excitation and when pulse ends at n Tn /2, the

system oscillates with maximum amplitude as  ust 0 .


2 2
td/Tn=1/8 td/Tn=1/4

1 1
u(t)/(ust)0

u(t)/(ust)0
0 0

-1 -1

-2 -2
0.00 0.05 0.10 0.15 0.20 0.25 0.0 0.1 0.2 0.3 0.4 0.5

2 2
td/Tn=1/2
td/Tn=1

1
1
u(t)/(ust)0

u(t)/(ust)0
0
0

-1

-1

-2
0.00 0.25 0.50 0.75 1.00
-2
0.0 0.5 1.0 1.5 2.0

2
td/Tn=1.25
2
td/Tn=1.5
1

1
u(t)/(ust)0

0
u(t)/(ust)0

-1
-1

-2
0 1 2 3 -2
0 1 2 3

2
2
td/Tn=1.75

td/Tn=2
1
1
u(t)/(ust)0

u(t)/(ust)0

0 0

-1 -1

-2 -2
0 1 2 3 4 0 1 2 3 4

t/Tn t/Tn

Fig. 3.2 Dynamic response of undamped SDF system to rectangular pulse force
3.4.2 Response to half sine wave pulse

Let’s now consider the following half sin wave step function to study the behavior of
undamped SDF system.

 p0 sin( t / td ) t  td
mu  ku  p(t )   (3.16)
0 t  td

It also follows the two phases, but here the behavior of the system becomes different when

td / Tn =1/2. So we will discuss this in another case.

Case 1: td / Tn  1/ 2

Forced vibration phase:

u(t ) 1   t  Tn  t 
 sin     sin  2   t  td (3.17)
 ust 0 1  Tn 2td 2   td  2td  Tn  

Free vibration phase:

u(t ) Tn / td  cos  td Tn    t 1 td  


  sin 2    t  td (3.18)
 ust 0 Tn / 2td   1
2
  Tn 2 Tn  

Case 2: td / Tn  1/ 2

Forced vibration phase:

u (t ) 1  2 t 2 t 2 t 
  sin  cos  t  td (3.19)
 ust 0 2  Tn Tn Tn 

Free vibration phase:

u (t )   t 1
 cos 2    t  td (3.20)
 ust 0 2  Tn 2 
2 2
td/Tn=1/4
td/Tn=1/8

1 1
u(t)/(ust)0

u(t)/(ust)0
0 0

-1 -1

-2 -2
0.00 0.05 0.10 0.15 0.20 0.25 0.0 0.1 0.2 0.3 0.4 0.5

2 2

td/Tn=1/2
td/Tn=1
1 1
u(t)/(ust)0

u(t)/(ust)0
0 0

-1 -1

-2 -2
0.00 0.25 0.50 0.75 1.00 0.0 0.5 1.0 1.5 2.0
A

2 2

td/Tn=1.5 td/Tn=2
1 1
u(t)/(ust)0

u(t)/(ust)0

0 0

-1 -1

-2 -2
0 1 2 3 0 1 2 3 4

2
2

td/Tn=2.5
1 td/Tn=3
1
u(t)/(ust)0

u(t)/(ust)0

0
0

-1
-1

-2 -2
0 1 2 3 4 5 0 1 2 3 4 5 6
t/Tn t/Tn

Fig. 3.3 Dynamic response of undamped SDF system to half sine wave pulse force
2 2
td/Tn=1/8 td/Tn=1/4

1 1

u(t)/(ust)0
u(t)/(ust)0

0 0

-1
-1

-2
-2
0.0 0.1 0.2 0.3 0.4 0.5
0.00 0.05 0.10 0.15 0.20 0.25

2
2
td/Tn=1/2
td/Tn=1

1
1
u(t)/(ust)0

u(t)/(ust)0
0
0

-1
-1

-2
0.00 0.25 0.50 0.75 1.00 -2
0.0 0.5 1.0 1.5 2.0

2
td/Tn=2
2
td/Tn=1.5
1
1
u(t)/(ust)0
u(t)/(ust)0

0
0

-1
-1

-2 -2
0 1 2 3 0 1 2 3 4

2
td/Tn=2.5 2
td/Tn=3

1
1
u(t)/(ust)0

u(t)/(ust)0

0
0

-1
-1

-2
0 1 2 3 4 5 6 -2
0 1 2 3 4 5 6
t/Tn
t/Tn

Fig. 3.3 Dynamic response of undamped SDF system to triangular pulse force
3.4.3 Response to triangular pulse

The function of triangular pulse is given by

 2 p0t
 t 0  t  td / 2
 d
mu  ku  p (t )  
 p  1  2  t  td / 2   td / 2  t  td
 0  td


t T 
2   n sin 2 
 td 2 td 
Chapter 4

NUMERICAL ANALYSIS OF LIQUID FILLED INDUSTRIAL TANK

USING LUMPED MASS MODEL

4.1 Introduction

In the literature we have discussed about the two-degrees-of freedom lumped mass model
(Housner, 1963) and three-degrees-of-freedom lumped mass model (Haroun, 1983). In this
research work three-degrees-of-freedom lumped mass model has been used for analysis as it
considers the flexibility of the tank wall.

When tank is excited by blast force, a part of liquid moves independent of the tank,
called the convective mass or sloshing mass (mc). This mass lies towards the upper part of the
tank. The second mass which also does not move in coordination with tank called as impulsive
mass (mi). This mass comes into picture when the flexibility of the tank wall is considered. This
mass lies around the central portion of the tank. The third mass which moves in coordination
with the tank called as rigid mass (mr). This mass lies just below the central portion of the tank.
The corresponding stiffness constants have been worked out according to the properties of the
tank wall and liquid mass.

These masses when applied to blast load, induces substantial hydrodynamic forces on the
tank wall which in turn generates design forces such as base shear and overturning moments. The
base shear is used for designing of the isolation system. The overturning moment develops high
stress on one side of the tank and may cause uplift of anchor on the other side. Due to
overturning moment buckling stress also develops on the tank wall. The designing becomes more
complex when the liquid-structure interaction is taken into consideration.

Here effort has been done to find the blast response of isolated tank numerically in order
to reduce the base shear and overturning moment which are the basic reasons for the failure of
the tank. Blast responses have been found out first for fixed base tank and then for isolated tank
and the responses are compared. Here responses found out for two types of tank i.e. broad tank
and slender tank.
uc
kc cc
mc

ki ci ui
mi
H
ub mc
mr mi
mr
kc
ki
cc
ci

(a) (b)

Fig. 4.1 (a) Liquid lumped mass model of non-isolated tank (b) stick model of non-isolated tank

kc cc uc
mc

ki ci ui
mi
H
ub
mc
mr mi
mr
kc ki mc

cc ci
Rubber isolator
mb

(a) (b)

Fig. 4.2 (a) liquid lumped mass model of isolated tank (b) stick model of isolated tank
4.2 Structural model of liquid storage tank

A structural model of ground supported liquid storage cylindrical tank with fixed base is
shown in fig. 4.1(a) and the stick model is shown in the fig. 4.1 (b). The base isolated tank model
and stick model are also shown in the fig. 4.2(a) & (b).

The liquid masses sloshing mass, impulsive mass and rigid mass are referred as mc, mi, mr
respectively. The stiffness constants of the springs which are associated with convective mass
and impulsive mass are kc and ki respectively. The damping constants which are associated with
convective mass and impulsive mass are cc, ci respectively.

The fixed base tank has two-degrees-of-freedom which are associated with sloshing mass
and impulsive mass where as in isolated tank there are three-degrees-of-freedom, the new degree
of freedom is associated with the rigid mass or tank base. uc, ui, ur are referred to as the absolute
displacement of these masses.

The liquid lumped masses can be evaluated by using the chart given by Haroun. The
chart values corresponds to the aspect ratio of tank, S (H/R) and th/R. (where H is the height of
liquid level and R is the radius of the tank)

For th/R =0.004 the values are presented.

1
𝑦𝑐 1.01327 −0.8757 0.35708 0.06692 0.00439 𝑆
𝑦𝑖 −0.15467 1.21716 −0.62839 0.14434 −0.0125] 𝑆 2
{𝑦 } = [
𝑟 −0.01599 0.86356 −0.30941 0.04083 0
𝑃 𝑆3
0.037085 0.084302 −0.05088 0.012523 −0.0012
{𝑆 4 }

Where, Yc , Yi , Yr are the mass ratios defined as

mc
Yc  (4.2)
m

mi
Yi  (4.3)
m

mr
Yr  (4.4)
m
m   R2 H w ( w is the mass density of liquid) (4.5)

The natural frequencies of sloshing mass, impulsive mass are expressed as

g
c  1.84   tanh 1.84S  (4.6)
R

P E
i  (4.7)
H s

Where, E and  s are modulus of elasticity and density of tank wall respectively and g is the
acceleration due to gravity.

The equivalent stiffness constants and damping coefficients of the convective and
impulsive mass are expressed as

kc  mcc 2
(4.8)

ki  mii 2
(4.9)

cc  2c mcc (4.10)

ci  2i mii (4.11)

Where c and  i are damping ratio of convective mass and impulsive mass respectively.

4.3 Governing equation of motion of non-isolated system

The governing equation of motion of non-isolated system subjected to blast force


expressed in matrix form is

mx  cx  k x  F (t ) (4.12)


Where  m  ,  c  ,  k  are the mass matrix, damping matrix, and stiffness matrix respectively. x

 x   xc , xi 
T
is the displacement vector. Here will give the absolute displacement of the

respective masses as, xc  uc and xi  ui . The matrixes are

m 0
 m   0c mi 

k 0
 k    0c ki 

c 0
c   0c ci 

F (t ) is the force vector which is multiplication of the pressure from blast and the area on which
blast pressure is acting. The blast force will be acting at the center of masses. So area to be
multiplied has to be considered accordingly. Now, by solving the differential equation the
displacement, velocity and acceleration are found out.

After calculating the, acceleration the base shear can be calculated as

Fb  mcuc  miui

4.4 Governing equation of motion of isolated system

The governing equation of motion of non-isolated system subjected to blast force


expressed in matrix form is

mx  cx  k x  F (t )

Where  m  ,  c  ,  k  are the mass matrix, damping matrix, and stiffness matrix respectively. x

 x   xc , xi , xr 
T
is the displacement vector. Here will give the absolute displacement of the

respective masses as, xc  uc  ub , xi  ui  ub and xr  ub . The matrixes are


 mc 0 mc 
 m   0 mi mi 
 mc mi M 

 kc 0 0
 k    0 ki 0 
 0 0 kb 

cc 0 0
c   0 ci 0 
 0 0 cb 

Where kb and cb are expressed as

cb  2b M b

2
 2 
kb    M
 Tb 

The expression of base shear is given as,

Fb  mcuc  miui  mr ub
12.00
10.00
8.00
6.00
4.00
2.00
0.00
-2.00 0 500 1000 1500 2000 2500

-4.00
-6.00
-8.00

Base displacement (in m)


0.03
0.02
Displacement (in m)

0.01
Series1
0.00
Series2
-0.01 0 500 1,000 1,500 2,000 2,500

-0.02
time (in millisec)

0.4

0.3

0.2

Series1
0.1
Series2

0
0 500 1000 1500 2000 2500
-0.1

-0.2
501 BROAD

Chart Title
6.00

4.00

2.00 Series1
0.00 Series2
0 500 1000 1500 2000 2500
-2.00

-4.00

0.05

0.04

0.03

0.02
Series2
0.01
Series1
0
0 500 1000 1500 2000 2500
-0.01

-0.02

-0.03

0.0040

0.0030

0.0020

0.0010 Series1
0.0000 Series2
0 500 1000 1500 2000 2500
-0.0010

-0.0020

-0.0030
502 SLENDER

1.5

0.5
Series1
Series2
0
0 500 1000 1500 2000 2500
-0.5

-1

0.05
0.04
0.03

0.02
Series1
0.01
Series2
0
0 500 1000 1500 2000 2500
-0.01

-0.02
-0.03

0.0015

0.001

0.0005 Series2

0 Series1
0 500 1000 1500 2000 2500
-0.0005

-0.001
502 BROAD

10

4
Series1
2
Series2
0
0 0.5 1 1.5 2 2.5
-2

-4

-6

1.5

0.5
Series1
0
0 0.5 1 1.5 2 2.5 Series2
-0.5

-1

-1.5

0.35
0.3
0.25
0.2
0.15
Series1
0.1
Series2
0.05
0
-0.05 0 0.5 1 1.5 2 2.5
-0.1
-0.15
401 SLENDER

10

4
Series1
2
Series2
0
0 0.5 1 1.5 2 2.5
-2

-4

-6

0.35
0.3
0.25
0.2
0.15 Series1
0.1
Series2
0.05
0
-0.05 0 0.5 1 1.5 2 2.5
-0.1
-0.15

0.015

0.01

0.005
Series1
0
0 0.5 1 1.5 2 2.5 Series2
-0.005

-0.01

-0.015
401 BROAD

2.5

1.5

1
Series1
0.5
Series2
0
0 500 1000 1500 2000 2500
-0.5

-1

-1.5

0.04
0.03
0.02
0.01
Series1
0
Series2
-0.01 0 500 1000 1500 2000 2500

-0.02
-0.03
-0.04

0.0005
0.0004
0.0003
0.0002
0.0001
Series1
0
Series2
-0.0001 0 500 1000 1500 2000 2500
-0.0002
-0.0003
-0.0004
-0.0005
101 BROAD

2 Series1

0 Series2
0 500 1000 1500 2000 2500
-2

-4

-6

0.04
0.03
0.02
0.01
Series1
0
0 500 1000 1500 2000 2500 Series2
-0.01
-0.02
-0.03
-0.04

0.002

0.0015

0.001
Series1
0.0005
Series2
0
0 500 1000 1500 2000 2500
-0.0005

-0.001

101 SLENDER
15

10

5
Series1
Series2
0
0 500 1000 1500 2000 2500

-5

-10

0.08
0.06
0.04
0.02
Series1
0
0 500 1000 1500 2000 2500
-0.02
-0.04
-0.06

You might also like