You are on page 1of 54

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/271514531

CONSTRUCTION AND VALIDATION OF THE SHELLS OF THE AUTOMOTIVE


TOROIDAL LPG TANKS SAFE IN OPERATION

Chapter · June 2013


DOI: 10.13140/2.1.5004.8325

CITATIONS READS

0 632

1 author:

Marek Flekiewicz
Silesian University of Technology
44 PUBLICATIONS   39 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Low carbon fuel combustion View project

A device for mixing gaseous fuels on-board View project

All content following this page was uploaded by Marek Flekiewicz on 29 January 2015.

The user has requested enhancement of the downloaded file.


Marek Flekiewicz1, Tomasz Matyja1

CONSTRUCTION AND VALIDATION OF THE SHELLS OF THE


AUTOMOTIVE TOROIDAL LPG TANKS SAFE IN OPERATION

1. INTRODUCTION

Liquefied petroleum gas is commonly used as an alternative fuel for the internal
combustion engines in Europe. The LPG is stored and transported on road vehicles in
cylindrical or special shape containers.
The most popular toroidal oval-cross section LPG tanks are designed and manufactured by
several manufacturers in Poland, based on requirements of ECE Regulation 67.01 (The
United Nations Economic Commission for Europe – UNECE, Vehicle Regulation, LPG
vehicles) and specific standards in different countries of the world.
The LPG toroidal containers are low-pressure vessels since their work pressure is lower
than 3000 kPa. They are equipped with a suitable configuration of special hermetic valves and
are produced in large range of different heights and volumes. The range of height is from 180
to 400 mm and volume from 32 to 102 dm3. The main problem of the manufacturer during
design of toroidal container is to determine:
- Changes of stresses and strains in container walls in a function of pressure increase for
different materials and wall thickness.
- The burst pressures and place of its burst. According to requirements of ECE
Regulation 67.01 the measured bursting pressure (Pr) must not under any
circumstances be less than 6750 kPa 2.25 x test pressure (Pt =3000 kPa),
- The change in the volume of the container at the time of bursting, (8 % in the case of a
toroidal container).
Although toroidal shell is one of the lesser used shell components, a number of studies
have recently been published in literature, which highlights new applications. Galletly [1]
(1998) had studied buckling analysis of a complete toroidal shell having elliptical cross-
section using the shell buckling programs, BOSOR and INCA. Kaptan and Kisioglu [2]
(2007) studied and determined Pr and their locations for cylindrical LPG fuel tanks using both
experimental and finite-element analysis (FEA) techniques. Kisioglu et. al. [3] (2001) studied
and determined Pr and their failure locations for the refrigerant cylinders using both
experimental and FEA approaches. The same author Kisioglu et. al. [4] (2008) studied
optimum end closure design for the propane cylinders. Redekop et. al. [5] (1999) had studied
stability of fluid-containing toroidal shell and derived some numerical results using both FEA
and DQM methods. Wang et. al. [6] (2006) had studied theoretical modal analysis and mode
shapes for the thin and thick walled pipes and toroidal shells. Burst pressure P r and volume
expansions of toroidal containers had been studied by Kisioglu [7] (2009). Some aspects of
stresses and strains states in the toroidal container shells were presented by Velickovic [8]
(2007). Therefore, no similar body of knowledge appears to be available in the current

1
Faculty of Transport, Silesian University of Technology, Krasińskiego 8 Street, 40-019 Katowice, Poland

1
literature for Pr and volume expansion measurements of the toroidal oval-section LPG fuel
tanks.
The purpose of this work is to present a calculation method allowing for verification of the
toroidal container design. Method based on software Solid Works 2010/2011 provides to
designer a possibility of investigating the behavior of container shells and volume expansions
of the toroidal oval cross section LPG container. By this method one can predict the burst
pressure Pr and failure location using numerical approach, the actual shell and weld zone
material properties including thickness variations of the containers are investigated.
In this paper, the following aspects of the structural analysis of toroidal container are
discussed:
− the most important steps in preparing the input data (geometry, material characteristics,
load),
− basic problems connected with satisfying technical requirements, and
− the most important results of the final structural analysis.
The stress analyses of the nonlinear material and the geometrical properties are carried out
for a toroidal container with a given cross-section. The analyses are made using the finite
element method (FEM). The computation procedure is validated by comparing the FE results
with the analytical results.
The ratio of the wall thickness and other dimensions is such that this toroidal container can
be regarded as a shell and the shape of the container is the shell in the form of the surface of
revolution. The theoretical bases of the calculation of the stress state in the wall points of such
shells are given, for example, in [9] and [10], and the expressions for the calculation of the
stress state in the wall points of the circular torus submitted to the action of the internal
pressure can be found, for example, in [9],[11] and [12]. However, the initial shape of the
cross section of the toroidal container considerably differs from the circle (Fig. 1). Besides
that, it was estimated that the toroidal container fitting would be the source of the significant
stress concentration. Because of the cited reasons, it is not possible to calculate stress and
strain state components using analytical methods, but by the use of commercial software, the
structural analyses are performed.

2. TYPICAL SHAPES OF THE SHELLS AND THEIR MODIFICATION

Cylindrical Semi toroidal top and bottom


shell sphere

Fig.1. Main parts of toroidal container


Rys.1. Główne części zbiornika toroidalnego

2
In this paper under the name of toroidal tanks is considered the entire class of containers
with non-circular (eliptical or ovaloid) meridional cross-section.
Typical tank consists of three main components: the cup (inner pipe) and two semi toroidal
bottoms located top and bottom (Fig.1). Both internal pipe and two semitoroidal shells are
manufactured using the stamping process. After the spinning process, these parts are welded
circumferentially to form the toroidal LPG tanks. In addition, each tank is equipped with a
port for connecting the valve which, depending on the technical design solution, is welded to
the bottoms shells or to the cup. Some tanks have a special mountings shelf (Fig. 2), which
can additionally stiffen the cup. In order to maximize the volume of the container a closed cup
structure is applied (Fig.3). Overview of the structures used to store liquefied petroleum gas
can be found in ECE Regulation 67.01.

Fig.2. Container with an open cup and the mounting shelf, internal nozzle
Rys.2. Zbiornik z otwartym kielichem i półka montażową, króciec wewnętrzny

Fig.3. Samples of special toroidal containers with closed inner pipe (cylindrical shell)
Rys.3. Przykłady zbiorników specjalnych toroidalnych, pełnych z zamkniętym kielichem

3
The following items disturb the axial symmetry: the nozzle, any types of mountings
elements and connectors welded to the tank shells, any cut-out in the cup in the tanks full,
allowing the flow of gas. Nevertheless, strength calculations the shells of the toroidal tank are
usually carried out assuming axial symmetry.

3. LEGAL AND TECHNICAL REQUIREMENTS FOR THE CONSTRUCTION OF LPG


TANKS

The development process of the container can be represented as a "V-diagram" which


covers all phases from a pre-verification to the final product fulfilling requirements of all
standards and user expectations. The first step is to create and confirm the requirement for the
container. These requirements are derived from the container behaviors and properties as well
as its use for automotive market.
The next step is to define the requirements for vehicle with various powering systems,
followed by the design and configuration of these systems. The lowest level of the cascade
diagram consists of component design and the integration of individual components and
materials selection. The evaluation of the design and materials selection are followed by the
validation phase, which includes both the testing of containers without fire protection system
as and accessories fitted to it.
The process depicted in the V diagram can be repeated a number of times during development
of new container.
At the each iteration, the quality of design is improved and allows making the decision for
start the serial production.
Modern tools aiding the design process allow to significantly reduce the time to implement
new solutions of containers for LPG storage on vehicles. This important phase of the process
decreases not only the costs of preparing the prototype, but also significantly increases the
likelihood of receiving positive results of the verification tests. An overview of the container
development on the V model base is shown in Fig. 4.

Table 1. Specification of the main technical standards


Tablica 1. Główne wymagania prawne i normy techniczne
Legal requirements and technical standards Source
1 Regulation No. 67 UNECE
E/ECE/324/Rev.1/Add.66/Rev.3/Corr.1−
E/ECE/TRANS/505/Rev.1/Add.66/Rev.3/Corr.1
2 NFPA 58: Storage of LP Gas Cylinders NFPA USA
3 Directive 84/527/EC: Welded Unalloyed Steel Gas The Commission Of The European
Cylinders Communities
4 ISO 14245:2006 Gas cylinders -- Specifications and ISO International Standard
testing of LPG cylinder valves -- Self-closing
5 ISO 10464, Gas cylinders — Refillable welded steel ISO International Standard
cylinders for liquefied petroleum gas (LPG) —
Periodic inspection and testing
6 BS 5045 (or the equivalent EN Standard): Welded British Standard
Cylinders up to 130 Liters Water Capacity
7 UNE-EN 12805:2003 , Automotive LPG components - The Commission Of The European
Containers. Communities
8 Commission Directive 2007/37/EC The Commission Of The European
Communities

4
The process begins at the top of left-hand side of the V- diagram from requirements, the
target setting and proceeds down the left side of the V by the calculation and the FEM
simulation to produce the prototype. Then the process goes up the right side of the V by
necessary verification tests leading to decision of production start.

Fig.4. Scheme of the typical process of the validation of the tank structure
Rys.4. Schemat typowego procesu walidacji konstrukcji zbiornika

Table 2. Specification of the verification test


Tablica 2. Opis testów weryfikujących
Required test The aim of test
Checking the of welds quality and their dimensions.
Inspection of welds and dimensions of
1 Measurement of wall thickness, container diameter and
main parts of container
height.
Checking properties of parent material and the stress
2 Mechanical tests
resistance of containers shells
3 Microscopic examination Examination of a full traverse section of the weld
Checking the leakage and permanent distortion of the
4 Hydraulic test
container.
Checking the bursting pressure and specific change in the
5 Burst test under hydraulic pressure
volume of container
Checking that a container complete with the fire protection
6 Bonfire test system, specified in the design, will prevent the burst of the
container when tested under the specified fire conditions.
7 Cycling test Checking whether the container after 20.000 pressure cycles
is not failed and leaks.

5
4. PRELIMINARY ANALYTICAL CALCULATIONS

In literature there are known analytical solutions for thin-walled pressure vessels in shape
of ideal toroid. However, the real geometry of vessel’s shell usually differs from the shape of
ideal toroid. From that reason analytical calculations of vessels are approximate and on this
basis only preliminary assessment of stresses and displacement can be made. Initial thickness
of vessel walls can be also estimated.
𝐾
𝑂 𝜑 𝑟

𝑛
𝑟 = 𝑅2 𝑠𝑖𝑛𝜑
𝑃 𝑂1 𝑃 = 𝑅1
𝑑𝑧 𝑑𝑠1
𝑅2 𝑑𝑟

𝜑 𝑅1
𝑂2 𝑑𝜑 𝜏

𝑂1
𝐿

Fig.5. Meridional cross-section of rotational shell


Rys.5. Przekrój południkowy powłoki obrotowej

𝑂
𝜃
𝑑𝜃 𝑟
𝑟 𝑁𝜑
𝑁𝜑𝜃
𝜕𝑁𝜃𝜑 𝑟
𝑁𝜃𝜑 + 𝑑𝜃
𝜕𝜃 𝑁𝜃
𝑝𝜃
𝑁𝜃𝜑
𝜑 𝜕𝑁𝜃 𝑝𝜑 𝑝𝑟
𝑁𝜃 + 𝑑𝜃
𝜕𝜃 𝜕𝑁𝜑𝜃
𝑑𝜑 𝑁𝜑𝜃 + 𝑑𝜑
𝜕𝜑
𝜕𝑁𝜑
𝑁𝜑 + 𝑑𝜑
𝜕𝜑
𝑧

Fig.6. Equilibrium of shell element


Rys.6. Równowaga elementu powłoki

6
In the figure 5 presented is cross-section of the middle rotational shell, which is made by
rotation of meridional curve 𝐾𝐿 about the axis 𝑂𝑧. Geometry of shell is described by two
main radiuses of curvature: meridional 𝑅1 and latitudinal 𝑅2. Centers of curvature 𝑂1 and 𝑂2
are on the same straight line with direction normal to a shell surface in chosen point 𝑃. Center
𝑂2 additionally have to lay on the axis 𝑂𝑧. Position of normal to the shell in meridional cross-
section is described by angle 𝜑.
On the basis of figures 5 and 6 the following geometric relationships for rotational shell
can be stated:
𝑟 = 𝑅2 𝑠𝑖𝑛𝜑 ,
𝑑𝑠1 = 𝑅1 𝑑𝜑 ,
𝑑𝑠2 = 𝑟𝑑𝜃 ,
𝑑𝑟 = 𝑑𝑠1 𝑐𝑜𝑠𝜑 = 𝑅1 𝑐𝑜𝑠𝜑𝑑𝜑 ,
(1)
𝑑𝑧 = 𝑑𝑠1 𝑠𝑖𝑛𝜑 = 𝑅1 𝑠𝑖𝑛𝜑𝑑𝜑 ,
𝑅1 − 𝑅2
𝑑𝑅2 = 𝑐𝑜𝑠𝜑𝑑𝜑.
𝑅2 𝑠𝑖𝑛𝜑

Equilibrium conditions of shell element bounded by latitudes defined by angles {𝜑; 𝜑 +


𝑑𝜑} and meridians defined by angles {𝜃; 𝜃 + 𝑑𝜃} can be described by three equations [13] :

𝜕 𝜕𝑁𝜃𝜑
(𝑅2 𝑠𝑖𝑛𝜑𝑁𝜑 ) − 𝑁𝜃 𝑅1 𝑐𝑜𝑠𝜑 + 𝑅1 + 𝑝𝜑 𝑅1 𝑅2 𝑠𝑖𝑛𝜑 = 0 ,
𝜕𝜑 𝜕𝜃
𝜕 𝜕𝑁𝜃
(𝑅2 𝑠𝑖𝑛𝜑𝑁𝜑𝜃 ) + 𝑅1 𝑁𝜃𝜑 𝑐𝑜𝑠𝜑 + 𝑅1 + 𝑝𝜃 𝑅1 𝑅2 𝑠𝑖𝑛𝜑 = 0 , (2)
𝜕𝜑 𝜕𝜃
𝑁𝜑 𝑁𝜃
+ = 𝑝𝑟 ,
𝑅1 𝑅2

from which the first one presents equilibrium condition written in direction of tangent line to
the meridian, the second one in the direction of tangent line to the latitude and the third one in
the directions of normal to the shell (Laplace equation). Equations contain three unknown
transversal forces: longitudinal force (meridional) 𝑁𝜑 , circumferential force (latitudinal) 𝑁𝜃
and shear force 𝑁𝜑𝜃 .
In axisymmetric problems derivatives with respect to time are excluded and the second
equation contains only unknown tangent forces. Other equations are then independent from
tangent force. In case of uniform pressure load there is also:

𝑝𝜑 = 𝑝𝜃 = 0, 𝑁𝜑𝜃 = 0, 𝑝𝑟 = 𝑝 = 𝑐𝑜𝑛𝑠𝑡. (3)

Equilibrium conditions present now two equations:

𝜕
(𝑅 𝑠𝑖𝑛𝜑𝑁𝜑 ) − 𝑁𝜃 𝑅1 𝑐𝑜𝑠𝜑 = 0 ,
𝜕𝜑 2
𝑁𝜑 𝑁𝜃 (4)
+ = 𝑝𝑟 .
𝑅1 𝑅2

Calculating 𝑁𝜃 from the second equation and replacing it in the first equation it is obtained
after transformations:

7
𝜕
(𝑅 𝑁 sin2 𝜑) = 𝑅1 𝑅2 𝑝𝑠𝑖𝑛𝜑𝑐𝑜𝑠𝜑 . (5)
𝜕𝜑 2 𝜑

Assuming that the angle 𝜑 changes from initial value 𝜑0 (in case of shell closed from the
top 𝜑0 = 0) equation can be integrated:
𝜑
𝑝
𝑁𝜑 = ∫ 𝑅1 𝑅2 𝑠𝑖𝑛𝜑𝑐𝑜𝑠𝜑𝑑𝜑 . (6)
𝑅2 sin2 𝜑
𝜑0

Please note that curvature radiuses depend on angle 𝜑. Using geometric dependencies (1) it
can be stated (6) that:

𝑟
𝑝
𝑁𝜑 = ∫ 𝑟𝑑𝑟 . (7)
𝑅2 𝑠𝑖𝑛2 𝜑
𝑟0

𝑟
𝑟 𝐶
𝑃

𝐵 𝑂1 𝐴
𝑅1 = 𝑏

𝑅2 𝐶′

𝜑
𝑂2
𝑧

Fig.7. Basic geometric dependencies for toroid


Rys.7. Podstawowe zależności geometryczne dla toroidy

From the figure 7:


𝑟 𝑎 + 𝑏𝑠𝑖𝑛𝜑
𝑟 = 𝑎 + 𝑏𝑠𝑖𝑛𝜑 , 𝑅1 = 𝑏 , 𝑅2 = = , 𝑟0 = 𝑎 . (8)
𝑠𝑖𝑛𝜑 𝑠𝑖𝑛𝜑

On the basis of formula (7) it can be written that:

𝑎+𝑏𝑠𝑖𝑛𝜑
𝑝 𝑝 1
𝑁𝜑 = ∫ 𝑟𝑑𝑟 = [(𝑎 + 𝑏𝑠𝑖𝑛𝜑)2 − 𝑎2 ] . (9)
𝑅2 𝑠𝑖𝑛2 𝜑 2
𝑅2 𝑠𝑖𝑛 𝜑 2
𝑎

8
After transformations the formula for longitudinal (meridional) force in toroidal shell is
obtained:

1 2𝑎 + 𝑏𝑠𝑖𝑛𝜑 1 𝑎 + 𝑟
𝑁𝜑 = 𝑝𝑏 = 𝑝𝑏 . (10)
2 𝑎 + 𝑏𝑠𝑖𝑛𝜑 2 𝑟

Form Laplace equation, after transformations, circumferential force can be derived, which
is constant in all point of the shell:

1
𝑁𝜃 = 𝑝𝑏 . (11)
2

Form of the formula for meridional tension with use of radius 𝑟 is more convenient [14].
Directly from it stems the conclusion that meridional tensions are higher than circumferential.
In the shell there is flat state of stresses and principal stresses are respectively equal to:

𝑁𝜑 𝑁𝜃
𝜎1 = , 𝜎2 = , 𝜎3 = 0 , (12)
𝑡 𝑡

where 𝑡 – thickness of shell wall.


Reduced von Misses stress can be calculated from the formula:

√2 (13)
𝜎𝑟𝑒𝑑 = √(𝜎1 − 𝜎2 )2 + (𝜎2 − 𝜎3 )2 + (𝜎3 − 𝜎1 )2 .
2
𝑏
After replacing stresses (12) and introducing dimensionless factor 𝛽 = 𝑎 , 𝛽 < 1, it is
obtained:

𝑝𝑏 √3 + 3𝛽𝑠𝑖𝑛𝜑 + 𝛽 2 𝑠𝑖𝑛2 𝜑 𝑝𝑏
𝜎𝑟𝑒𝑑 = = Φ(𝛽, 𝜑) . (14)
2𝑡 1 + 𝛽𝑠𝑖𝑛𝜑 2𝑡

Function 𝛷(𝛽, 𝛼) is dimensionless reduced stress [15].


Because circumferential stress is constant, the reduced stress reach extremes in the same
points as meridional stress, which can be rewritten with use of 𝛽 coefficient:

1 2 + 𝛽𝑠𝑖𝑛𝜑
𝑁𝜑 = 𝑝𝑏 . (15)
2 1 + 𝛽𝑠𝑖𝑛𝜑

Derivative with respect to the angle 𝜑 equals:

𝑑𝑁𝜑 1 −𝛽𝑐𝑜𝑠𝜑
= 𝑝𝑏 . (16)
𝑑𝜑 2 (1 + 𝛽𝑠𝑖𝑛𝜑)2
𝜋 3𝜋
Extreme occurs in two points: 𝜑 = 2 and 𝜑 = 2 . Basing on the sign of derivative (minus
cosines function) in proximity of extremes, it can be determined that the first extreme
corresponds to the global minimum of the function and the second extreme corresponds to the
global maximum (respectively points 𝐴 and 𝐵 in cross-section of the vessel in the figure 7).

9
Polar graph of the function 𝛷 + 1, for two values of coefficient 𝛽 = {0.5; 0.7}, is shown in
the figure 8 (from the values presented in the graph one should distract 1 to obtain values of
dimensionless functions of reduced stress).

Polar pattern of the dimensionless equivalent stress function


3

1
b=1
 = 0.7  = 0.5
0

3
6 5 4 3 2 1 0 1 2 3

Fig. 8. Polar pattern of the dimensionless reduced stress function


Rys.8. Diagram biegunowy bezwymiarowych naprężeń redukowanych

Longitudinal and circumferential deformations of toroidal shell are calculated from the
generalized Hooke’s law:

1 1 𝑝𝑏 𝑎 + 𝑟 1 𝑝𝑏
𝜀𝜑 = [𝜎𝜑 − 𝜈𝜎𝜃 ] = [ − 𝜈] = [𝑎 + 𝑟(1 − 𝜈)] ,
𝐸 2 𝐸𝑡 𝑟 2 𝐸𝑡𝑟 (17)
1 1 𝑝𝑏 𝑎+𝑟 1 𝑝𝑏
𝜀𝜃 = [𝜎𝜃 − 𝜈𝜎𝜑 ] = [1 − 𝜈 ]= [𝑟 − 𝜈(𝑎 + 𝑟)] ,
𝐸 2 𝐸𝑡 𝑟 2 𝐸𝑡𝑟

where 𝐸, 𝑣 - respectively Young module and Poisson coefficient.


Change of circumferential length of an element:

𝑑𝑠2 = 𝑟𝑑𝜃, 𝑑𝑠2′ = 𝑑𝑠2 (1 + 𝜀𝜃 ) = 𝑟 ′ 𝑑𝜃 , (18)

is accompanied by change of radius length:

1 𝑝𝑏
Δ𝑟 = 𝑟 ′ − 𝑟 = 𝑟𝜀𝜃 = [𝑟 − 𝜈(𝑎 + 𝑟)] . (19)
2 𝐸𝑡

Relative change of radius length:

10
Δ𝑟 1 𝑝𝑏 𝑎 1 𝑝𝑏 2 + 𝛽𝑠𝑖𝑛𝜑
= 𝜀𝜃 = [1 − 𝜈 (1 + )] = [1 − 𝜈 ]
𝑟 2 𝐸𝑡 𝑟 2 𝐸𝑡 1 + 𝛽𝑠𝑖𝑛𝜑 (20)
1 𝑝𝑏
= Ψ(𝛽, 𝜑).
2 𝐸𝑡

Polar graph of the function 𝛹 + 1 for two values of 𝛽 coefficient and for 𝑣 = 0.3 is presented
in the figure 9. From the graph it steams that with increase of 𝛽 coefficient over the specified
value changes of radius length in proximity of point 𝐵 of the shell (fig.7) are negative. Critical
3𝜋
values of coefficient can be determined from the condition (assuming 𝜑 = 2 ) :

2 − 𝛽𝑘𝑟
1−𝜈 =0, (21)
1 − 𝛽𝑘𝑟
So:
1 − 2𝜈
𝛽𝑘𝑟 = . (22)
1−𝜈

4
For 𝑣 = 0.3 it is: 𝛽𝑘𝑟 = 7 ≈ 0.57. Theoretically such proportions of the toroidal vessel its 𝐵
point will not change the position.

polar pattern of the function  +1


1,5

1
 = 0.5

0,5

 = 0.7

0,5

1,5
1,5 1 0,5 0 0,5 1 1,5 2

Fig.9. Polar pattern of the function Ψ + 1


Rys.9. Wykres biegunowy funkcji Ψ + 1

As already mentioned, shapes of shells applied in practice differ from the ideal toroid. In
accordance with the recommendations of the toroidal tank for liquefied gas to be in the form
design, as set out in Annex 10 to the Regulations of the UN / ECE 67. Most, however consist
of three basic parts (Fig.10):

11
- the central part, which is an annular cylindrical tank,
- two identical ring bottoms hemispherical cross-section.

Fig.10. Toroidal cross-section of the tank


Rys.10. Przekrój zbiornika toroidalnego

𝐷
Toroidal tanks are thin-walled vessels because: 𝑡 < 20 , where: t – thickness of vessel
walls, D – outer diameter of the tank.

𝑟𝐴

𝑟𝐵

𝐵 𝐴
𝑁𝜑𝐵 𝑁𝜑𝐴
𝑁𝜑𝐵 𝑁𝜑𝐴

𝑁𝜑𝐵 𝑁𝜑𝐴

𝑁𝜑𝐵 𝑁𝜑𝐴

Fig.11. Division of the shell into parts


Rus.11. Podział powłoki na części

12
Vessel presented in the figure can be considered as composition of two halves of toroid
and two cylinders: outer one is exposed to the inner pressure, inner one is exposed to outer
pressure. If wall thickness is neglected then from the figure 10:

𝐷−𝑑 𝑑
𝑏= , 𝑎= + 𝑏. (23)
4 2

Each of cylindrical vessels is stretched by meridional forces (along the generatrix line of
the cylinder), which can be established basing on formula for toroidal shell in points 𝐴 and 𝐵:
1 𝑎 + 𝑟𝐴 1 2 + 𝛽
𝑁𝜑′ = 𝑁𝜑𝐴 = 𝑝𝑏 = 𝑝𝑏 ,
2 𝑟𝐴 2 1+𝛽
(24)
1 𝑎 + 𝑟𝐵 1 2−𝛽
𝑁𝜑′′ = 𝑁𝜑𝐵 = 𝑝𝑏 = 𝑝𝑏 ,
2 𝑟𝐵 2 1−𝛽

where 𝑟𝐴 = 𝑎 + 𝑏, 𝑟𝐵 = 𝑎 − 𝑏 .
For each of cylindrical shells it can be written by Laplace’s equations that:
𝑁𝜑′ 𝑁𝜃′ 𝑁𝜑′′ 𝑁𝜃′′
+ =𝑝 , + = −𝑝 . (25)
∞ 𝑟𝐴 ∞ 𝑟𝐵

Hence, circumferential forces in cylindrical shells:

1 2(𝑎 + 𝑏) 1 2(1 + 𝛽)
𝑁𝜃′ = 𝑝𝑟𝐴 = 𝑝(𝑎 + 𝑏) = 𝑝𝑏 = 𝑝𝑏 ,
2 𝑏 2 𝛽
(26)
1 2(𝑎 − 𝑏) 1 2(1 − 𝛽)
𝑁𝜃′′ = −𝑝𝑟𝐵 = −𝑝(𝑎 − 𝑏) = − 𝑝𝑏 = − 𝑝𝑏 .
2 𝑏 2 𝛽

The absolute value of circumferential force in the outer cylindrical shell is always greater
than the value of circumferential force in the inner cylindrical shell. It can be shown, using for
example symbolic calculation processor, that, the reduced stresses in cylindrical shells are
equal respectively to:


𝑝𝑏 √3𝛽 4 + 12𝛽 3 + 18𝛽 2 + 12𝛽 + 4
𝜎𝑟𝑒𝑑 = ,
2𝑡 𝛽(1 + 𝛽)
(27)
′′
𝑝𝑏 √3𝛽 4 − 12𝛽 3 + 18𝛽 2 − 12𝛽 + 4
𝜎𝑟𝑒𝑑 = .
2𝑡 𝛽(1 − 𝛽)

In the Figure 12 there are combined values of dimensionless elements of formulas for

reduced stresses. Differences occurring between stresses in points A (𝜎𝑟𝑒𝑑𝐴 𝑣𝑠. 𝜎𝑟𝑒𝑑 ) and B
′′
(𝜎𝑟𝑒𝑑𝐵 𝑣𝑠. 𝜎𝑟𝑒𝑑 ) of vessel’s parts junctions indicate disturbances of membrane state. The
reason is discontinuity of radiuses of curvature. Somewhat surprising may be the fact that
reduced stresses due to von Misses hypothesis in outer cylindrical shell are greater than
stresses in inner shell for 𝛽 coefficient less than circa 0.65, so in range frequently used in
existing constructions. In literature [16], there are recommendations for ideal toroidal vessels
to keep 𝑎 ≥ 2𝑏. Unfortunately, due to too rough problem presentation, there is no explanation
for the above relationship.

13
12


10 redA
dimensionless part of equivalent stress [-]

'

red

8 redB
''

red

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
 [-]

Fig.12. Dimensionless part of reduced stress (von Misses)


Rys.12. Bezwymiarowa część naprężeń redukowanych

Inner cylindrical shell is stretched in the direction of generatrix, but is circumferentially


compressed. That means occurrence of shearing stresses and the risk of buckling. For
comparison the reduction of stresses is made according to the hypothesis of de Saint-Venant
(Treska):
𝜎̂𝑟𝑒𝑑 = 𝜎𝑚𝑎𝑥 − 𝜎𝑚𝑖𝑛 . (28)

Comparison of forces in cylindrical shells leads to conclusions:

𝑁𝜃′ > 𝑁𝜑′ , 𝑁𝜃′′ < 𝑁𝜑′′ (0 < 𝛽 < 1). (29)

On that basis reduced stresses according to Treska’s hypothesis can be calculated:


𝑝𝑏 𝛽 2 + 2𝛽 + 2
𝜎̂𝑟𝑒𝑑 = ,
2𝑡 𝛽(1 + 𝛽)
′′
𝑝𝑏 𝛽 2 − 2𝛽 + 2
𝜎̂𝑟𝑒𝑑 = ,
2𝑡 𝛽(1 − 𝛽) (30)
𝑝𝑏 1
𝜎̂𝑟𝑒𝑑𝐴 = ,
2𝑡 1 + 𝛽
𝑝𝑏 1
𝜎̂𝑟𝑒𝑑𝐵 = .
2𝑡 1 − 𝛽

Dimensionless parts of formulas (30) are presented in the figure 13. In such a case the
greatest reduced stresses occur always in the inner cylindrical shell.
14
12

'

red
10 ''

red
Dimensionless part of equivalent stress [-]


redA

8 redB

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
 [-]

Fig.13. Dimensionless part of reduced stress (Treska)


Rys.13. Bezwymiarowa część naprężeń redukowanych (Treska)

Circumferential strains of cylindrical shells are equal to relative changes of its radiuses:

1 ′ 1 𝑝𝑏 2(1 + 𝛽) 2+𝛽 Δ𝑟𝐴 ′


𝜀𝜃′ ′
= [𝜎𝜃 − 𝜈𝜎𝜑 ] = [ −𝜈 ]=( ) ,
𝐸 2 𝐸𝑡 𝛽 1+𝛽 𝑟𝐴
(31)
1 1 𝑝𝑏 2(1 − 𝛽) 2−𝛽 Δ𝑟𝐵 ′′
𝜀𝜃′′ = [𝜎𝜃′′ − 𝜈𝜎𝜑′′ ] = [− −𝜈 ]=( ) .
𝐸 2 𝐸𝑡 𝛽 1−𝛽 𝑟𝐵

From the other hand relative changes of radiuses can be calculated from formulas for toroidal
shell:
Δ𝑟𝐴 1 𝑝𝑏 2+𝛽
= [1 − 𝜈 ],
𝑟𝐴 2 𝐸𝑡 1+𝛽
(32)
Δ𝑟𝐵 1 𝑝𝑏 2−𝛽
= [1 − 𝜈 ].
𝑟𝐵 2 𝐸𝑡 1−𝛽

Differences occur only in the first elements of the above formulas, what is obvious while
longitudinal stresses in cylindrical and toroidal shells junctions are identical. That can be
written:
Δ𝑟𝐴 ′ Δ𝑟𝐴 1 𝑝𝑏 2(1 + 𝛽) 1 𝑝𝑏 2 + 𝛽
( ) − = [ − 1] = ,
𝑟𝐴 𝑟𝐴 2 𝐸𝑡 𝛽 2 𝐸𝑡 𝛽
(33)
Δ𝑟𝐵 ′′ Δ𝑟𝐵 1 𝑝𝑏 2(1 − 𝛽) 1 𝑝𝑏 2 − 𝛽
( ) − = [− − 1] = − .
𝑟𝐵 𝑟𝐵 2 𝐸𝑡 𝛽 2 𝐸𝑡 𝛽

15
15
(2+)/
-(2-)/

10

vs -(2-)/ [-]
5

0
(2+)/

-5

-10
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
 [-]

Fig.14. Comparison of dimensionless parts of formulas (33)


Rys.14. Porównanie bezwymiarowych części wzorów (33)

In the figure above the comparison of dimensionless part of formulas is presented. Values
shown in the chart are significant. However, the effect of the scale cannot be forgotten. In fact
values should be multiplied by dimensional factor, which typically is of order 10−4. In
addition differences in displacements indicate that in junction zones of vessels bending
stresses occur.
From the above formulas results that radial displacements of points of inner cylindrical
shell are always negative. Compatibility of displacements condition in point B is met only in
constructions with sufficiently large 𝛽 coefficient. It can be easily verified that with 𝑣 =
0.3 ∧ 𝛽 = 0.7 the ratio of shell thickness ensuring displacements compatibility is technically

hard to met 𝑡 ⁄𝑡 ≈ 7.19.
Presented membrane theory for vessels does not produce the exact results because of few
reasons. In real constructions there occur stresses not equally spaced with the thickness, there
are also transverse forces, tangent forces and bending moments. Nonetheless, on the basis of
vessel analysis with membrane theory some conclusions can be formulated as to the method
of forming the vessel.
Inner side of toroidal shell (cup) should have thicker wall than the rest. Assume that toroid
√2
section is split to two parts in points 𝐶, 𝐶′. Let radius equal to 𝑟𝑐 = 𝑎 − 2 𝑏, what corresponds
to 60 degrees. Condition for stresses compensation in points A and B by change of the wall
thickness (t’- thickness of arc CC’) leads to the formula:

√2
𝑡′ 1 − 2 𝛽 √3 − 3𝛽 + 𝛽 2
= ,
𝑡 1−𝛽 (34)
√2 1 2
√(3 −
2 𝛽 + 2𝛽 )

which graph is presented in the figure 15.

16
2.6

2.4

2.2

1.8
t'/t [-]
1.6

1.4

1.2

0.8
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
 [-]
Fig.15. Graph of the dependencies (34)
Rys.15. Wykres zależności (34)

In case of modified vessel composed by toroidal and cylindrical parts bending forces are
inevitable. Points of both shells in junction (B) are moving in opposite directions, with the
values of coefficient 𝛽 < 𝛽𝑘𝑟 . Increasing the value of coefficient 𝛽 > 𝛽𝑘𝑟 ensures
compatibility of senses of radial displacements. While, achieving the same values of
displacement by change of walls thickness is technically unrealizable. In vessels of such a
type inner cylindrical shell cannot be unambiguously indicated as the most strenuous.
Vessel’s proportions and chosen method for stresses reduction are of the great importance.

5. EVALUATION OF THE STRESS AND STRAIN STATE USING FEM

5.1. Selection of the calculation model

Observations on the behavior of toroidal tanks during pressure tests allow determining
their structural characteristics:

1. Tanks increase their volume after a pressure test (3MPa), which indicates the
occurrence of permanent deformation. In some types of tanks plastic strain occurs even
at pressures below the test pressure. Filling up the tank several times causes the
stabilization of its volume and material of the shell strengthening.
2. Structural solution of connection between the cup and bottoms, and possibly between
the cover and the cup, has a very strong influence on the strength of the tank and burst
pressure. Important are the radii of curvature of the cup and bottoms in the connection
area of merger and arrangement of the joints. Joints stiffen the structure substantially in
the circumferential direction.
3. There is a close relationship between the cup stiffness and shell stiffness (it has an
impact on the shape and thickness of individual elements). In the linear elasticity range
the greatest stresses occur in the cup. However, during the burst test the pressure
increases and the plastic deformation appears, also the zone of the greatest stresses
moves towards shells connections area. If the cup is too stiff pressure will pull the
bottoms from the weld. Otherwise, the cup is damaged. Properly adjusted stiffness of
the cup and bottoms results in increase of the burst pressure value.

17
4. Research of burst vessels shows that permanent deformation of the cup are not axial
symmetric.
5. The elements of tank attachment to the vehicle, for example the shelf location to the
center of the cup, influence the deformations.

FEM model of toroidal container should take into account the above observation and
among other things provide:

1. Possibility of running a nonlinear analysis, taking into account large displacements and
nonlinear material (elastic-plastic material, the characteristics obtained from the tensile
test).
2. Possibility of simulating the strengthening of the container material. Implementation of
at least one cycle of filling and emptying (to test pressure) and transfer the results of
permanent deformation and stress for the proper simulation of plastic damage.
3. Accurate as possible mapping of cup and floors geometry: the radius of curvature, wall
thickness (the thickness can be varied by pressing the individual components), welds.
Providing boundary conditions that will not limit the deformations of container.
4. Determination of stresses and strains variation in vessel walls.

Finite elements method (FEM) models used to analyze toroidal tanks can be divided into
axially symmetric and spatial.
Toroidal tank has parts that disturb the axial symmetry, such as nozzle, barrier, and
fasteners. Nevertheless, the axially symmetric models (2D) are very often used in the
calculation because of the much smaller number of degrees of freedom in comparison with
the spatial models. The mostly applied finite elements are axisymmetric solid (2D) or surface
axisymmetric shell (1D).
Spatial models are usually shell models, rarely solid models. They allow modeling of
asymmetrical parts and asymmetrical boundary conditions. The simulation results can
therefore be closer to reality. The disadvantage of spatial models is the high cost of
computation due to the large number of elements needed for the discretization of the structure.
So often the axial symmetry is assumed and torus segments are analyzed only with additional
conditions (axisymmetric constrains) on the edge of segments.
In case of applying the elements of shell type, the inaccurate modeling of welds can be
expected. Elements of a solid type allow precise modeling of joints. However, at least four or
five layers of elements on the vessel’s wall thickness should be used in order to ensure
continuity of deformation and a satisfactory accuracy of calculations.
Bucki [17] has made a comparative analysis of four container models: axisymmetric shell,
axisymmetric solid, shell (1/4 container) and solid (1/4 tank). In addition, different densities
of the grids were analyzed. The most useful model in terms of accuracy and computing time
turned out to be axisymmetric solid, with an average of six elements on the wall thickness and
square shape functions. An additional advantage of axisymmetric solid models is a transparent
presentation of distributions of stresses in the walls of container.
To perform numeric calculations, one can use different software and different structural
models of the tank. It is good practice to perform the structural analysis of the theoretical
circular torus under uniform internal pressure at a first time in order to compare the obtained
results with the analytically solutions. When the accordance of the results for ideal toroidal
tank occurs, the numerical calculations of arbitrary shaped containers can be further
conducted.

18
5.2. Modeling of the material

Numerical calculations of stress and strain state require use of nonlinear material model. It
stems from the nature of vessel’s construction work under the pressure and from requirements
of regulations due to range of information from simulations. In case of steel constructions
undergoing high plastic deformations the most appropriate model is elastic-plastic with
isotropic hardening. Plasticity models including kinematic hardening are more suitable for
cases involving cyclic loading [18].
The method of defining of material depends on features of specific MES analysis system.
Usually elastic properties are defined separately: Young module 𝐸 and Poisson’s coefficient
𝜈. Next the tabular function of stress to plastic strain dependencies is given. The first point in
the table corresponding to the zero plastic strain defines also yield stress. Many MES systems
enable conducting coupled thermo-elastic-plastic analyses. In such a case a few tabular
functions of stress-strain for different values of temperature are defined. MES system
automatically interpolates data from the table and sets necessary values.
Data essential for defining properties of material are identified basing on the static tensile
test. Samples should rather come from shells of ready vessels than from the sheets of steel
being the material for production. Sheet of steel for vessel undergoes many technological
processes such as cutting, stamping, forming and welding and the whole vessel mechanical
restraining during the pressure test. All these operations can change internal structure of steel.

800

700

600

500
stress [MPa]

400

300 nominal stress - nominal strain


true stress - logarithmic strain
200 true stress - logarithmic plastic strain

100

0
-0.05 0 0.05 0.1 0.15 0.2
strain [-]
Fig.16. Stress- strain curves
Rys.16. Krzywe naprężenie-odkształcenie

The static tensile test provides information about curve nominal stress – nominal strain:

𝐹 𝑙 − 𝑙0 𝑙
𝜎𝑛𝑜𝑚 = , 𝜀𝑛𝑜𝑚 = = −1, (35)
𝐴0 𝑙0 𝑙0

19
where: 𝐹, 𝑙, 𝑙0 , 𝐴0 – appropriately tensile force in sample, actual sample length, initial length,
cross-section field of undeformed sample.

Due to the possibility of high deformations logarithmic deformation instead of nominal


should be used:

𝑙
𝑑𝑙 𝑑𝑙 𝑙
𝑑𝜀 = , 𝜀=∫ = ln ( ) = 𝜀𝑙𝑛 ,
𝑙 𝑙 𝑙0 (36)
𝑙0
𝜀𝑙𝑛 = ln(1 + 𝜀𝑛𝑜𝑚 ).

Using the definition of transverse deformations:

𝜀′
= −𝜈 , (37)
𝜀

actual sample volume in strain function can be estimated:

𝑉 = 𝐴 ∙ 𝑙 = 𝐴0 (1 − 𝜈𝜀)2 𝑙0 (1 + 𝜀) = 𝐴0 𝑙0 [1 + 𝜀(1 − 2𝜈) + 𝜀 2 (𝜈 2 − 2𝜈) + 𝜈 2 𝜀 3 ] ≈ 𝐴0 𝑙0 .

Coefficients containing 𝜀 are significantly smaller from one, what means that the material
of sample can be approximately treated as incompressible, so:

𝑙0
𝐴 = 𝐴0 . (38)
𝑙

Logarithmic deformation relates to actual stress defined as:

𝐹 𝐹 𝑙
𝜎𝑡𝑟𝑢𝑒 = = = 𝜎𝑛𝑜𝑚 (1 + 𝜀𝑛𝑜𝑚 ). (39)
𝐴 𝐴0 𝑙0

From the logarithmic deformation elastic and plastic parts can be extracted:

𝑒 𝑝𝑙
𝜀𝑙𝑛 = 𝜀𝑙𝑛 + 𝜀𝑙𝑛 . (40)

MES systems require setting the actual stress – logarithmic plastic strain curve, which can
be calculated from the formula:

𝜎𝑡𝑟𝑢𝑒 = 𝜎𝑛𝑜𝑚 (1 + 𝜀𝑛𝑜𝑚 ) ,


𝑝𝑙 𝜎𝑡𝑟𝑢𝑒 (41)
𝜀𝑙𝑛 = ln(1 + 𝜀𝑛𝑜𝑚 ) − .
𝐸

In the figure 16 three curves are given for example. The curves presenting dependencies:
nominal stresses to nominal strains, actual stresses to logarithmic strains, actual stresses to
logarithmic plastic strains.

20
5.3. The material hardening phenomenon

The problem of analysis of stress and strain states in toroidal vessels in nonlinear not only
due to nonlinearities in material. In the model large displacements should be allowed and the
fact that load coming from the pressure depends on the position should be taken into account.
All this results in MES calculations carried out by iterative-incremental method. The
parameter controlling the simulation is pressure in vessel. Usually the linear increment of
pressure during simulation can be assumed.
The aim of the simulation is determining the values of stress and displacement in a few
characteristic points of the pressure increment plot, usually: the operating pressure 2.4MPa,
test pressure 3MPa and minimal vessel’s resistance 6.75MPa (i.e. in cycle 0 − 2.4 − 3 −
6.75𝑀𝑃𝑎). The term minimal resistance means the state in which reduced stresses according
to Misses in vessel undergoing pressure 6.75MPa does not exceed maximal value 𝑅𝑚 (the
limit of resistance from the static stretching test).
Simulation can be conducted with constant or variable step of pressure increment
(automatically set by software), which changes in range from zero to minimum 6.75MPa.
From the authors’ experiences it comes that in case of vessels with irregular construction
simulation can finish for lower values of pressure. The reason is too high increment of
displacements in singular step of simulation (greater than assumed tolerance) and the lack of
possibility to reduce next step of pressure increment.
Despite the fact that regulations does not allow permanent deformations in vessel during
the pressure test, many contemporary vessels is constructed “on the verge” of these
requirements. In such cases the material hardening can be observed. Pressure test can be
treated as an additional technological procedure consisting in stresses equalization in effect
restraining of the material. Usually during the next filling the vessel to the test pressure
permanent deformations does not occur anymore.
Numerical simulations MES give results consistent with experimental researches if
material hardening phenomenon is taken into account. That means the simulation should be
conducted in pressure change cycle 0 − 3 − 0 − 2.4 − 3 − 6.75𝑀𝑃𝑎.
Figure 17 presents the phenomenon of material hardening on the example of reduced
stresses in two chosen integration points (Gauss) of elements in the proximity of top joint of
toroidal vessel.

5.4. An example of the structural analysis of toroidal container by FEM

5.4.1. Main dimensions of the container

The basis for preparing the FEM model was the geometry of the container which included
overall dimensions and the characteristic cross section shape. The main dimensions of the
containers are presented in table 3. We analyzed two types of the tanks: with open and closed
shape.
Table 3. Main dimensions of analyzed containers
Tabela 3. Główne wymiary analizowanych zbiorników
Parameters Container A Container B
Height -H [mm] 270 270
Outer diameter -D [mm] 680 720
Inner diameter –d [mm] 183,5 203
Thickness of shells -t [mm] 3,0 3,2
Thickness of inner pipe – t1 [mm] 4,0 3,6

21
Fig. 17. The strengthening of the material during the pressure test, calculated by FEM
Rys. 17. Wzmocnienie materiału podczas testu ciśnieniowego, obliczone MES

22
5.4.2. Main mechanical properties of the container material

All parts of the container, except accessories shelfs are made from the steel sheet. The
materials of steel sheet are P310 NB and P355 NB according to EN 10120. All parts are
connected by weld. The mechanical characteristics for the steel sheet material are following:

Table.4. Main mechanical properties of the containers material


Tabela 4. Główne mechaniczne właściwości materiału zbiorników
Material
No Mechanical parameters Comments
P355 NB P310 NB
1 Minimal value of the upper Container A
355 310
yield stress, Re [MPa] P355 NB for all
2 Minimal and maximal value parts
of the tensile strength, Rm 510-620 460-550 Container B
[MPa] P355 NB for inner
3 Elongation at reaching the pipe and P310 NB
24 28
tensile strength, A [%] for shells

Because for the structural analysis with nonlinear material properties the stress-strain curve
is necessary, mechanical properties of material were registered during tests.
Due to the possibility of plastic strain state occurrence, it is very important to know the
strain value corresponding to the yield stress.
On the basis of tensile test carried out on the material sample coming from the container,
the material model (the relationship between stress and strain) adapted to FEM software is
received, as shown in Figure 18. For steel P355 NB yield stress is about 395 MPa and the
limit of tensile strength 644 MPa and for steel P310 NB yield stress is about 320 MPa and the
limit of tensile strength 495 MPa.

Fig. 18. Stress – plastic strain curve for the P355 NB steel

23
Rys. 18. Krzywa naprężenie – odkształcenie plastyczne dla stali P355 MB

5.4.3. Analytically calculations

Typically, analytical calculations allow to roughly estimate the thickness of the shells. In
this case the analytical calculations will be used to estimate the coefficients of safety and
radial displacements the outer cylindrical shell. The following data are adapted: Young's
modulus = 2.05 ∙ 105 𝑀𝑃𝑎 , Poisson coefficient 𝜈 = 0.3, test pressure 𝑝 = 3𝑀𝑃𝑎 . The table
below shows the results of calculations (made on the basis of the equations in paragraph 4).

Table. 5. Analytically calculations results


Parameter: Container A Container B
radius 𝑎 [𝑚𝑚] 215.875 230.75
radius 𝑏 [𝑚𝑚] 124.125 129.25
shape coefficient 𝛽 = 𝑎/𝑏 [−] 0.575 0.560
inner cylindrical shell thickness 𝑡 ′′ [𝑚𝑚] 4.0 3.6
outer cylindrical shell thickness 𝑡 ′ [𝑚𝑚] 3.0 3.2
outer cylindrical shell radius 𝑟𝐴 = 𝑎 + 𝑏 [𝑚𝑚] 340 360
′′
reduced stress 𝜎𝑟𝑒𝑑 [𝑀𝑃𝑎] 224.87 260.87

reduced stress 𝜎𝑟𝑒𝑑 [𝑀𝑃𝑎] 238.32 233.26
minimum yield stress 𝑅𝑦′′ [𝑀𝑃𝑎] 355 355
minimum yield stress 𝑅𝑦′ [𝑀𝑃𝑎] 355 310
′′
safety coefficient 𝑋 ′′ =
𝜎𝑟𝑒𝑑
[−] 1.58 1.36
′′
𝑅𝑦

safety coefficient 𝑋 ′ =
𝜎𝑟𝑒𝑑
[−] 1.49 1.33

𝑅𝑦
outer cylindrical shell displacement [mm] 0.51 0.54
It is seen that both tanks are designed so that the shape ratio is close to the limit value. The
displacement in inner cup of the ideal toroidal tank would be close to zero. The selected
thicknesses of the shells provide sufficient safety.

5.4.4. Description of FEM calculation model

The analysis was made using the finite element method. Special attention in the analysis
was focused on elasto-plastic behavior of material of the container. The nonlinear finite
element method as a modeling tool of Solid Works 2010/2011 was used for simulation.
To determine the state of stress and strain the typical simulation of filling the tank
pressurized by the cycle 0 − 3 − 0 − 3 − 6.75𝑀𝑃𝑎 has been performed. The non-linear
displacement FEM version and von Misses plastic material model with isotropic hardening
were used.
Solid Works is a typical solid structures modeler (3D) and does not have in his library the
axisymmetric finite elements (2D). For this reason, the spatial finite elements model was
used. Due to the omission in the calculation of nozzle holes multivalve and mounting screws,
connector, and a ventilation hole in the cup portion of the tank was simulated, corresponding
to the 1/360 part, considering it as part of a symmetric. This assumption does not affect the
accuracy of the calculations, and significantly shortens their duration. It also enables to
generate accurate finite element mesh, providing a more accurate analysis of stresses and
displacements.

24
On the basis of technical documents structural models of toroidal containers, shown in
Figures 19 and 20, were developed. Axially symmetrical tasks were analyzed using finite
elements with 10 nodes and the full integration, taking into account large displacements. In
the model of container A were used 193383 elements and 299865 nodes while for container B
were used 264842 elements and 407355 nodes. The container model was loaded by internal
pressure and supported as shown in Figures 19 and 20.

Fig. 19. Structural model of container A


Rys. 19. Model strukturalny zbiornika A

25
Fig. 19a. Grid in shell conjunction: bottom of the tank and tube
Rys. 19a. Siatka w połączeniach powłok: dennicy i kielicha

Fig. 19b. Grid in shell conjunction: bottom of the tank and tube and shelf assembly
Rys. 19b. Siatka w połączeniach powłok: dennicy i kielicha oraz półki montażowej

26
Fig. 20. Structural model of container B
Rys. 20. Model strukturalny zbiornika B

Fig. 20a. Connection shells bottom with inner pipe


Rys.20a. Połączenia powłok dennicy z kielichem.

5.4.5. Results of numerical simulation

Below the simulation results obtained in the case of a container A are presented. The
results in the case of the container B were similar. Part of the results is also included in
Section 7 which describes the pressure test of the tank. Results are posted there for
comparison with experimental results (see Tables 9, 10, 11 and Figures 39 and 40).

27
Simulation test described in section 5.3 corresponds to a pressure test carried out in
accordance with the requirements of Rules 67.01 UNECE. During this test, the tank is
strengthening, the shells are slightly deformed.
During the first filling cycle the maximum stress equals to 496.3 MPa is located in a node
near the bottom weld, drawing the 21th. These stresses are the result of plasticity of the
material. The stresses, the value of which exceeds 395 MPa (yield stress), are present in an
zone containing only 0.20% of the volume of the tank shells (calculated by Solid Works). The
maximum displacement occurs in the lower lid and is equal to 3.452 mm (Fig.22).
The distribution of stress and permanent displacement of elements of the tank was
determined by reducing the pressure inside the tank from 3 MPa to 0.0 MPa. It was shown
that permanent stress and strain remain in the tank shells. Displacements in the selected
characteristic points of the tank are shown in Figure 23.
The technical requirements in regard to the strength of the container structure are presented
in Annex 10 of the ECE Regulation No. 67. These requirements specify that the container
should not be permanently deformed at a pressure of 3.0 MPa. This condition is satisfied
during the second cycle of filling the tank pressurized 3MPa.

Fig.21. Distribution of the von Misses stress (at design pressure, 3MPa for container A)
Rys. 21. Rozkład naprężeń von Missesa (przy ciśnieniu obliczeniowym, 3MPa dla zbiornika A)

28
Fig. 22. Distribution of displacements (at design pressure, 3MPa for container A)
Rys. 22. Rozkład przemieszczeń (przy ciśnieniu obliczeniowym, 3MPa dla zbiornika A)

Fig. 23. Distribution of displacements (at pressure 0 MPa after reducing from 3 MPa, container A)
Rys.23. Rozkład przemieszczeń (przy ciśnieniu 0 MPa po zredukowaniu z 3 MPa, zbiornik A)

29
Permanent displacements at points selected for comparison with experimental research (see
also Figure 39) are as follows:
- Midpoint of the upper shell - 0.197 mm
- Midpoint of the lower shell - 0.192 mm
- Circumferential weld joining the two bottoms - 0.003 mm,
- Circumferential weld joining the top of the cup bottoms - 0.133 mm
- Circumferential weld joining the bottom of the cup bottoms - 0.113 mm
- Measure cup - 0.001 mm
- Midpoint of the upper lid - 1.433 mm
- Midpoint of the bottom cover - 1.435 mm.
Location and value of the maximum deflection 1.546 mm are shown in Figure 23. First
pressure testing increases the volume of the reservoir with a value exceeding 1% of the
beginning volume.

Fig. 24. Distribution of the von Misses stress (at pressure 6.75MPa, for container A)
Rys. 24. Rozkład naprężeń von Missesa (przy ciśnieniu 6,75Mpa, dla zbiornika A)

Figure 24 shows the distribution of equivalent stresses at a pressure 6.75 MPa. Stresses
exceed the minimal value of the tensile strength 510 MPa. The zone exceeding the tensile
strength does not cover the entire thickness of the shell and includes only the 2% of the
volume of the material tank (Fig.25). The maximum deflection and a selected displacement
values are shown in Figures 26 and 27.

30
Fig. 25. The areas of exceedances the tensile strength
Rys. 25. Obszar przekroczenia wytrzymałości na rozciąganie

Fig. 26. Maximum deflection (6.75MPa)


Rys. 26. Maksymalne ugięcie (6.75MPa)

31
Fig. 27. Displacements (6.75MPa)
Rys. 27. Przemieszczenia (6.75MPa)

Fig.28 Distribution of the von Misses stress (at burst pressure 9.6MPa)
Rys. 28. Rozkład naprężeń von Missesa (przy ciśnieniu rozerwania 9.6MPa)

32
Fig.29. The areas of exceedances the tensile strength (pressure 9.6MPa)
Rys. 29 Obszary przekroczeń wytrzymałości na rozciąganie (ciśnienie 9.6MPa)

Fig.30. Maximum deflection (pressure 9.6MPa)


Rys.30 Maksymalne ugięcie (ciśnienie 9.6MPa)

33
Fig.31. Displacements (pressure 9.6MPa)
Rys. 31. Przemieszczenia (ciśnienie 9.6 MPa)

Drawings 28,29,30 and 31 respectively show: reduced stress, the zone of exceeding of
tensile strength (this time covering the entire thickness of the shells), the maximum deflection
and the displacement of the reservoir under pressure 9.6MPa. This pressure can be considered
as burst pressure.

6. OPTIMAL DESIGN OF TANK STRUCTURE

In the literature, there is currently no information on the development of effective methods


for the optimal design of toroidal shells. Typically the objective of the constructor is to get the
maximum internal volume of the tank (capacity) and a minimum of its own weight, in a given
space, limited dimensions of D and H. From this point of view it is a multi-criteria
optimization problem. At the same time, the conditions as to the value of the equivalent stress
are imposed. These conditions are not uniform, because they depend on the pressure in the
tank:

𝜎𝑟𝑒𝑑 < 𝑅𝑌 𝑖𝑓 𝑝 = 3𝑀𝑃𝑎


(42)
𝜎𝑟𝑒𝑑 < 𝑅𝑀 𝑖𝑓 𝑝 = 6.75𝑀𝑃𝑎

The stress state in the container can be determined only by means of FEM. This requires
time-consuming calculations for each analyzed. It should be noted, that if the local reduced
stress will exceed the 𝑅𝑀 limit, in a small area, in the points situated on the inner side of the
container, then it does not mean the destruction of the tank. This fact makes it difficult to
define the procedure for the automatic evaluation of the stress state (critical values), which is
necessary for optimization algorithm in each tank.
Another difficulty is the large number of parameters with which one can describe the
geometry of the tank shells. These parameters are often related to each other with a number of

34
geometric constraints. For example, two adjacent pieces of bottom plate may be different in
radii of curvature, but should have a common normal vector at the place of the connection.
This limitation does not apply to the shells that are connected by the welding. In this case,
other geometric conditions must be satisfied. The relative position of the connected parts must
ensure the assembly possibility. Shell geometry can be defined in a natural way, i.e. by using
successive arcs, with giving the specified radiuses of curvature and the angles of the arcs.
Alternatively, the geometry information can be described using the leading points and the
spline functions.
Taking into account all of the above difficulties, it appears that most conveniently the
optimization method of the container shells design can be to develop on basis of artificial
intelligence methods, for example genetic algorithms. Genetic algorithms allow formulating
the optimization problem in a less mathematically formal manner than for example the
gradients methods. This does not change the fact that such a task would be very complicated
and time-consuming to solve. The biggest challenge will be to develop an algorithm of
validation the geometric structures with randomly modified parameters.
An alternative to formal optimization methods are the methods of a case study. In this
method relatively simple is selecting the best design solution from an arbitrary set of proposed
solutions. Set of solutions can be prepared using CAD tools. Then there is no need for the
formulation of complicated geometric algorithms to validate each design solution.
In the paper of the authors [19], using the case study, the family of the containers was
analyzed. Geometric properties of the containers are described by a set of parameters shown
in the table 6 and Figure 32. The five pre-proposed design solutions were tested.

Fig. 32. Basic dimensions of the container


Rys. 32. Podstawowe parametry zbiornika

35
Table 6. Basic dimensions of containers tested
Container Height External Pipe Shelf Weld height R1 R2 R3 R4 R5
no. [mm] diameter diameter height [mm] [mm] [mm] [mm] [mm] [mm]
[mm] [mm] [mm]
cont.1 180 566 182 32 21 61 61 250 42 22.1
cont.2 225 580 182 60 24 61 61 177 61 27
cont.3 199 629 182 40 25.8 61 61 177 57 27
cont.4 200 600 182 40 21.2 61 61 177 57 27
cont.5 190 600 182 40 21.2 61 61 177 57 27

Chart analysis (Figure 33) shows that the best design solution is the case a reference
number 2. The greatest reduced stress, at the test pressure of 3 MPa, occurs in the bottom of
container in the weld area (the part of the container is further reinforced by the shelf).
However, they do not exceed the yield point 𝑅𝑌 = 355𝑀𝑃𝑎. With increasing the pressure the
zone of the greatest stress is moved to the area of the cup (Fig. 34 and 35). At this area, most
probably, the container will be destroyed 𝑅𝑀 ≈ 650𝑀𝑃𝑎. At a pressure of 6.75MPa tank is
still safe. Structures marked with numbers 4 and 5 demonstrate the worst behavior. Under the
test pressure the plasticity of the material can be observed. The destruction of these tanks
occurs near the bottom weld (Figure 36).

700

600 cont.1.cup
Von Misses stress [MPa]

cont.2.cup
500
cont.3.cup
400 cont.4.cup

300 cont.5.cup
cont.1.weld
200 cont.2.weld

100 cont.3.weld
cont.4.weld
0
cont.5.weld
0 2 4 6 8 10 12
Pressure inside conainer [MPa]

Fig. 33. The maximum reduced stresses as a function of pressure


Rys. 33. Maksymalne naprężenia redukowane w funkcji ciśnienia

36
Fig. 34. Reduce stress pattern in the cup (pressure 10MPa, tank 2)
Rys. 34. Mapa naprężeń redukowanych na kielichu (ciśnienie 10MPa, zbiornik 2)

Fig. 35. Reduce stress pattern near the weld (pressure 10MPa, tank 2)
Rys. 35. Mapa naprężeń redukowanych w okolicy spoiny (ciśnienie 10MPa, zbiornik 2)

37
Fig. 36. Reduce stress pattern near the weld (pressure 10MPa, tank 5)
Rys. 36. Mapa naprężeń redukowanych w okolicy spoiny (ciśnienie 10MPa, zbiornik 5)

Kukiełka et al. [20] has conducted the shape optimization of toroidal tank for LPG, using
ready-made tools: Goal Driven Optymalization in ANYS/WORKBENCH. Optimization
criterion was to minimize the stress and the total mass of the tank with given dimensional
restrictions. The calculations were carried out in the linear-elastic range. It should be noted
that the design solution that has the best features in linear elastic range, may not satisfy the
security requirements in the ranges of higher pressures (test pressure and /or minimum burst
pressure).

7. VERIFICATION FEM CALCULATIONS DURING PRESSURE TESTS

7.2. Measurements of displacements of the shells and the inner pipe (test pressure)

The answer to the question whether the possible occurrence of local deformation and
permanent hardening of the material construction of the reservoir under the pressure test was
obtained experimentally.
Tests were carried out as follows:
 checked the actual deformation of the finished tank shells at selected points, for which
numerical calculations were found to have the greatest stress,
 checked the size of permanent deformation (plasticity in) the tank after filling it with
liquid under pressure.

38
Fig. 37. Measuring points of the container shell and inner pipe displacements
Rys.37. Punkty pomiarowe przemieszczeń na powłoce i kielichu zbiornika

The tests were carried out in seven control points in accordance with the attached pictures
(Fig. 37). During tests at the designated points investigated the effect of evenly increasing
fluid pressure on the deflection of upper and lower shells and the inner pipe.
For the measurements used dial gauges measuring range 0-10 mm, the scale interval of
0.01 mm. Sample spacing of the sensors during tests of two types of containers is shown in
Figure 38. Results of measurements and the displacement obtained in the numerical
calculations (Fig. 39 and 40) for the pressure test equal to 3 MPa are shown in Table 7.

Container A Container B
Fig. 38. The test stand for measuring deformations of container
Rys. 38. Stanowisko pomiarowe przemieszczeń zbiornika

39
4

6
3

Fig. 39. Measuring points in the container A FEM model


Rys. 39. Punkty pomiarowe w modelu MES zbiornika A

3
8

Fig. 40. Measuring points in the container B FEM model


Rys. 40. Punkty pomiarowe w modelu MES zbiornika B

40
Table 7. Results of container A displacement measurements
The first test
Measured and / FEM calculated displacement
Pressure
No.
[MPa] Z1 Z2 Z3 Z4 Z5 Z6 Z7
[mm] [mm] [mm] [mm] [mm] [mm] [mm]
1 10 0,04 0.06 0,00 0,27 0,61 0,80 0,01
2 20 0,11 / 0,352 0,14 0,00 / 0,162 0,50 / 0,231 1,26 / 2,124 1,54 / 2,514 0,05 / 0,107
3 30 0,20 / 0,423 0,22 0,02 / 0,216 0,75 / 0,310 2,18 / 2,967 2,50 / 3,453 0,11 / 0,134
4 0 0,08 / 0.193 0,04 0,00 / 0,071 0,20 / 0,072 0,66 / 0,916 0,56 / 0.943 0,00 / 0,002

The second test


Measured and / FEM calculated displacement
Pressure
No.
[MPa] Z1 Z2 Z3 Z4 Z5 Z6 Z7
[mm] [mm] [mm] [mm] [mm] [mm] [mm]
1. 10 0,03 0,06 0,00 0,22 0,54 0,69 0,01
2. 20 0,07 / 0,411 0,12 0,00 / 0,186 0,39 / 0,268 1,06 / 2,540 1,35 / 3,150 0,05 / 0,105
3. 30 0,13 / 0,425 0,19 0,02 / 0,216 0,53 / 0,310 1,53 / 2,967 1,89 / 3,453 0,10 / 0,130
4. 0 0.01 0,00 0,00 0,00 0,05 0,03 0,00

Table 8. Results of container B displacement measurements


The first test

Measured and / FEM calculated displacement


Pressure
No.
[MPa]
Z1 Z2 Z3 Z4 Z5 Z6 Z7
[mm] [mm] [mm] [mm] [mm] [mm] [mm]
1 10 0,03 0,02 0,06 0,31 0,46 0,79 ------
2 20 0,13 / 0,111 0,06 3,41 / 0,256 0,89 / 0,196 1,23 / 1,315 1,98 / 1,385 ------
3 30 0,29 / 0,138 0,09 4,00 / 0,310 1,53 /0,427 2,10 / 1,707 3,27 / 1,780 ------
4 0 0,06 / 0,003 0,00 3,72 / 0,133 0.66 / 0,113 0,63 / 0,192 0,85 / 0,197 ------

The second test

Measured and / FEM calculated displacement


Pressure
No.
[MPa]
Z1 Z2 Z3 Z4 Z5 Z6 Z7
[mm] [mm] [mm] [mm] [mm] [mm] [mm]
1 10 0,07 0,04 0,01 0,33 0,52 0,92 ------
2 20 0,16 / 0,113 0,07 0,10 / 0,267 0,63 / 0,325 1,00 / 1,446 1,76 / 1,505 ------
3 30 0,24 / 0,138 0,09 0,34 / 0,311 0,92 / 0,383 1,44 / 1,708 2,52 / 1,783 ------
4 0 0,02 0,01 0,04 0,05 0,05 0,10 ------

Results obtained during the container A and B test indicate:


- the presence of plastic deformations in the range from 0.00 to 0.66 mm for container A
and from 0,00 to 4,00 for container B, (as a result of calculation, container A from 0,0
to 1,56 mm and container B from 0,06 to 0,197 mm);
- the existence of permanent deformations of the container A shells in the range from
0.00 to 0.66 mm and from 0,00 to 3,72 mm for container B shells, (as a result of
calculation, container A from 0,0 to 1,56 mm and container B from 0,06 to 0,197 mm).

41
The results obtained allow the conclusion that the local deformations of toroidal container
shells will always occur during any pressure testing, contributing to strengthening the material
and providing sufficient structural strength to the required toroidal container burst pressure.
The result of the analytical calculation is different from the actual deformation of the vessel
subjected to the influence of the test pressure. But both the FEM calculation and bench test
results show that the pressure test carried out on each tank, a strengthening of their structure
happen.
Table 9. Results of the FEM calculations
Point Displacement [mm]
Pressure Pressure Pressure Pressure Burst pressure
2,4 [MPa] 3,0 [MPa] 4,5 [MPa] 6,75 [MPa]
Container Container Container Container Container
A B A B A B A B A B
9,6 7,14
[MPa] [MPa]
1 2,540 1,446 2,967 1,707 5,436 3,536 8,771 7,364 16,311 8,111
σ<Re σ<Re σ<Re σ<Re σ<Rm σ<Rm σ<Rm σ<Rm σ<Rm σ<Rm
2 3,150 1,505 3,453 1,780 5,923 3,629 9,175 7,528 11,796 8,273
σ<Re σ<Re σ<Re σ<Re σ<Rm σ<Rm σ<Rm σ<Rm σ<Rm σ<Rm
3 0,410 0,113 0,425 0,138 0,506 0,203 1,651 0,436 7,093 0,430
σ<Re σ<Re σ<Re σ<Re σ<Rm σ<Re σ<Rm σ<Rm σ<Rm σ<Rm
4 0,186 0,325 0,216 0,310 0,429 1,089 0,846 2,381 2,689 3,000
σ<Re σ<Re σ<Re σ<Re σ<Re σ<Re σ<Re σ<Rm σ>Rm σ>Rm
5 0,268 0,267 0,310 0,427 0,591 1,191 1,067 2,561 2,561 2,326
σ<Re σ<Re σ<Re σ<Re σ<Re σ<Re σ<Rm σ<Rm σ<Rm σ<Rm
6 0,105 0,038 0,134 0,44 0,191 0,066 0,269 0,115 0,847 0,135
σ<Re σ<Re σ<Re σ<Re σ<Re σ<Re σ<Rm σ<Rm σ<Rm σ<Rm
7 2,193 2,376 27,734 35,502 34,489
σ<Re σ<Re σ<Rm σ<Rm σ<Rm
8 2,263 2,465 27,999 35,951 36,273
σ<Re σ<Re σ<Rm σ<Rm σ<Rm

The currently used methods of non-contact deformation measurements (laser instruments)


in the shells of the tanks, allow for automatic tracking of the correctness of each part of the
tank (control of dimensional tolerances), at every stage of production. From the point of view
of safety of tank operation, the most important are the measurements after the hydrostatic test
that can reveal any manufacturing defects. The Figure 41 shows an example of the
documentation from the measurements of deviations the nominal dimensions of the tank,
made in the STAKO Company.

42
Fig. 41. The documentation from the measurements of deviations the nominal dimensions of the tank
Rys. 41. Dokumentacja z pomiarów odchyłek od nominalnych wymiarów zbiornika.

7.2. Measurements of the volume increase and the burst pressure

During the pressure test can also be obtained information on the burst pressure of the tank
and the increase in its volume. Burst pressure depends on several factors encountered during
the production process of the tank and this parameter is usually the average of multiple tests.
Do not expect a good accordance with the results of FEM simulations. Of course, each tested
in this way specimen of the tank must meet the minimum requirements.

43
The Figure 42 shows a fairly typical case of toroidal tank bursts in the vicinity of the weld.
The corrugations of shells are visible (buckling). This effect is not possible to achieve with
use of simulation methods and axisymmetric FEM models.

Fig.42. The tank after the burst pressure into hydraulic test
Rys.42. Zbiornik po próbie rozerwania pod ciśnieniem hydraulicznym

Determination of volume changes on the basis of the displacements obtained by means of


the FEM is a difficult task. When using the most of FEM systems, it is necessary to develop
their own calculation procedures. However The Solid Works 1010/2011 is equipped with the
appropriate tools and supports the user in this regard.
Tables 10 and 11 show examples of experimental results of the burst pressure, compared
with results of numerical simulations tanks A and B with three different heights.

Table 10. Results of hydrostatic tests of containers type A


Tablica 10. Wyniki testu hydrostatycznego zbiornika A
No. Container
Results of burst
Burst volume
Container The calculated and pressure test/
pressure test increase during
height actual volume of Calculated burst
time test/ Calculated
container pressure
volume
[mm] [dm3]
[s] increase
[MPa]
[%]
1 250 82,0 / 82,8 480 8,5 / 8,7 9,1 / 7,1
2 250 82,0 / 82,8 510 8,7 / 8,7 8,8 / 7,1
3 270 91,0 / 91,0 490 10,0 / 9,6 13,8 / 11,25
4 270 91,0 / 91,2 540 9,45 / 9,6 14,4 / 11,25
5 300 102,0 / 102,8 540 9,45 / 9,55 9,6 / 9,2
6 300 102,0 / 102,8 570 9,80 / 9,55 10,0 / 9,2

44
Table 11. Results of hydrostatic tests of containers type B
No. Container
Results of burst volume
Burst
Container The calculated and pressure test / increase during
pressure test
height actual volume of Calculated burst test /
time
container pressure Calculated
[mm] [dm3] volume
[s]
[MPa] increased
[%]
1 250 84,0 / 84,4 480 10,6 / 7,0 9,51 / 9,2
2 250 84,0 / 84,6 420 10,3 / 7,0 9,10 / 9,2
3 270 92,0 / 92,4 480 10,6 / 7.14 9,46 / 9,6
4 270 92,0 / 92,4 580 10,3 / 7.14 9,84 / 9,6
5 300 103,0 / 103,4 540 10,7 / 8,2 9,13 / 9,9
6 300 103,0 / 103,4 660 11,4 / 8,2 10,15 / 9,9

8. THE INFLUENCE OF TEMPERATURE ON THE STRENGTH OF THE TANK (FIRE


TESTING)

8.1. Background of the material behavior in high temperature

Steel at high temperatures loses its strength. The yield strength and ultimate tensile
strength are significantly reduced. Figure 43 shows an example of stress-strain curves at high
temperatures. To experimentally obtain such charts it is required to the use of special
equipment and a large number of samples. Therefore it is quite expensive. Having no the
proper equipment one can use mathematical models describing the shape of stretching curves,
depending on the temperature. The paper of Kodur et al. [21] reviews high-temperature
constitutive relationships for steel structures currently available in American and European
standards, and highlights the variation between these relationships through comparison with
published experimental results.
Figure 44 shows the reduction factors at high temperatures, the basic parameters of carbon
steels such as the effective yield strength, proportional limit and the slope (Young's modulus)
obtained from Eurocode 3 [22].
According to Eurocode 3 standardized graphs are considered as the stress-strain curves as
shown in Figure 45. Adapted in the main part of a norm the standard elastic-perfectly plastic
model of the material does not give information on the reduction of the ultimate strength at
elevated temperature. For temperatures below 400 °C, the alternative strain-hardening option
described in Annex A of the Eurocode 3 may be used. The ultimate strength at elevated
temperature, allowing for strain hardening, should be determined as follows:

1.25 ∙ 𝑅𝑦 (𝑇) 𝑓𝑜𝑟 𝑇 < 300℃ ;


𝜎𝑢 (𝑇) = {𝑅𝑦 (𝑇) ∙ (2 − 0.0025 ∙ 𝑇) 𝑓𝑜𝑟 300℃ ≤ 𝑇 < 400℃ (40)
𝑅𝑦 (𝑇) 𝑓𝑜𝑟 𝑇 ≥ 400℃

45
Fig. 43. Elevated-temperature mechanical properties of high strength structural steel S460N [23]
Rys. 43. Właściwości mechaniczne stali konstrukcyjnej wysokiej wytrzymałości S460
w podwyższonej temperaturze [23]

Reduction factors for the stress-strain relationship of carbon steel at elevated temperatures

Effective yield strength


1 Propotional limit
Slope of linear elastic range

0.8
Reduction factor [-]

0.6

0.4

0.2

0
0 200 400 600 800 1000 1200
Temperature [°C]
Fig. 44. Reductions factors of carbon steel at elevated temperatures (Eurocode 3 model [22])
Rys. 44. Współczynniki redukcji dla stali węglowych w podwyższonej temperaturze (model Eurocode
3 [22])

46
Fig. 45. Standardized stress-strain diagrams of the Eurocode 3
Rys. 45. Standaryzowane wykresy naprężenie-odkształcenie według normy Eurocode 3

8.2. Fire test of pressurized container

In the case when the pressurised container is attacked by fire, its temperature rises and this
reduces the strength of container. This, combined with the pressure inside container, may lead
to its failure with catastrophic consequences.
In event of fire on vehicle fuelled by LPG, the primary protection system of the pressurised
container is PRD (pressure relief device). Functions of PRD may be realized by following
devices:
1. A fusible plug (temperature triggered), designed to open at a temperature of 120
±10 °C with flow capacity flow capacity of:

𝑄 ≥ 2.73 ∙ 𝐴 (41)

in which:
Q – flow of air in standard m3/min (100 kPa absolute and temperature of 15 °C);
A – exterior surface of the container in m2.
2. A pressure relief valve provided that its flow capacity complies the equation below:

𝑄 ≥ 10.66 ∙ 𝐴0.82 (42)

in which: Q and A have the same meaning as above.


3. A combination of the two above devices.
When the pressure relief valve is considered as a pressure relief device, the flow shall be at
least 17.7 standard m3/min.
The fire protection systems of LPG containers are currently designed according to
requirements of UN ECE Regulation No. 67.01, and containers with specified by
manufacturer system should to prevent their bursting during fire conditions. However it is
widely recognized some aspects of requirements are inadequate and inappropriate for

47
different cases of flame impact on container during the fire. Serious risks are in the following
cases:
1. Rapidly increasing container walls temperature (higher heat flux, for example Jet-Fire),
2. The flame impact of gas escaping from the PRD on a part of the surface of the container.
The second of these cases is shown in Figure 46, and the result of the flame impact on
container wall presents in Figure 47.

Fig.46. An impact of flame directly on of the container wall


Rys.46. Wpływ płomienia bezpośrednio na ściankę pojemnika

Fig. 47. The burst of container at a pressure p = 2 MPa, as a result of the impact a gas flame on its wall
Rys. 47. Rozerwanie pojemnika pod ciśnieniem p=2 MPa, jako wynik wpływu płomienia gazowego
na jego ściankę.

48
The theoretical aspects of the problem are illustrated in Figure 48 and 49 below, which
show idealised time histories for a pressurized container under fire attack. In Figure 48, the
container wall temperature increases slowly with time. The container strength initially
remains constant but starts to significantly loose its strength once a critical temperature is
reached (300-500oC, depending on the type of steel). The container is being depressurised and
applied stress reduces proportionally with the decreasing internal pressure. The container
strength always remains greater than the applied stress and thus the container is protected by
the well-functioning PRD. In Figure 49, the same container is exposed to more severe fire.
The higher heat flux causes the container temperature to rise more rapidly and the container
strength to decrease more rapidly. The container is being depressurised at the same rate as
before, however, the strength is dropping too rapidly and the container will fail (burst) once
its strength is less than the applied stress.

Fig. 48. The proper functioning of devices for Fig. 49. Too little efficiency of PRD, too big
container fire protection applied stress as a result of critical wall
Rys. 48. Właściwe funkcjonowanie urządzeń temperature – the destruction of the tank
zabezpieczających zbiornik przed ogniem Rys. 49. Za mała wydajność PRD, zbyt duże
zastosowane naprężenia jako wynik krytycznej
temperatury ścianki – zniszczenie zbiornika.

It should be emphasised that above example is greatly simplified. In reality the container
wall stresses will not only simply be the result of internal pressure, but non-uniform heating
by the fire will create additional thermal stresses. The impact of additional stresses explains
the case of Jet-Fire depicted in Figures 46 and 47, as one of the many possibilities.
Failure of a pressurized container subjected to fire attack is related to its strength at
elevated temperature. There are sufficient data available to fully describe the temperature-
dependent property of steel used in the manufacture of pressure containers used on vehicles
fuelled by LPG. According to requirements of Regulation 67.01 tests on LPG containers
under fire are performed for all new types of containers and on containers equipped with new
accessories, including PRD. The fire test should demonstrate that a container complete with
the fire protection system will prevent the burst of the container when tested under the
specified fire conditions. An example proper behavior of the container and its accessories
subjected to the fire attack is shown in Figure 50.

49
1000 5,0

900 4,5

800 4,0
Temperature,oC

Pressure, MPa
700 3,5

600 3,0

500 2,5

400 2,0

300 1,5

200 1,0

100 0,5

0 0,0
14:11 14:12 14:13 14:15 14:16 14:18 14:19 14:21

Time, min

płomień1/flame1 ściankaPod/wall beneath osprzęt/ accessory

płomień2/flame 2 ściankaNad/wall above ciśnienie/pressure

Fig.50. An example of changes in pressure inside container, temperatures of walls and fitted to it
accessories in the case of proper conduct of the bonfire test
Rys.50. Przykład zmian ciśnienia wewnątrz zbiornika, temperatury ścianek oraz zamontowanych w
nich akcesoriów w przypadku poprawnego przebiegu testu ogniowego

9. FATIGUE TESTS

Another important element of the validation of the pressure vessels structures are fatigue
tests. These will be particularly important in the case of tanks in which permanent
deformation during the pressure tests was observed.
On account of repeated filling and emptying of the tank its structure is under
nonsymmetrical cyclic loading. The pressurized containers, in which is only elastic
deformation below the proportional limit are completely safe (elastic shakedown). Fatigue
tests are particularly important in the case of tanks in which permanent deformation are
observed during the pressure test. There are two possible cases. In the first case plastic
deformation increases gradually, then growth slows down and plastic strain stabilizes (plastic
shakedown, fig. 51), the container structure is susceptible to low cycle fatigue. In the second
case plastic strain increase without bound in every load cycle (ratcheting, Figure 52). Plastic
strain accumulated until the container collapse.
Durability tests allow evaluating the strength of shell and the container accessories
subjected to cycles of increase and decrease in pressure. A change in pressure from minimum
to maximum value is realized with a certain frequency, characteristic of the device under test
(Figure 53). Toroidal tanks should be subject to a fatigue test of 20 thousand cycles, without
being damaged.

50
Fig. 51. The plastic shakedown behavior
Rys. 51. Cykliczne zmiany plastyczność

Fig. 52. The ratcheting behavior


Rys. 52. Kumulacja odkształceń plastycznych

Fig. 53. Stand for durability tests and changes in pressure chart in the cycle
Rys. 53. Stanowisko do badań trwałościowych oraz wykres zmiany ciśnienia w cyklu

51
10. CONCLUSIONS

Observed in the last decade rising steel prices and the widespreading desire to reduce
production costs mean that contemporary designs for LPG toroidal tanks are made on the
border of the minimum requirements. Safety of the tank operations depends largely on the
proper design verification.
This article presents a kind of road map, which is defining a set of actions that should
perform the constructor of the tank to verify the design with use of both computational and
experimental methods, in accordance with the law. Attempts were made to focus on various
aspects of constructing process, the preparation of the calculation model and the planning and
conduction of the simulation and experimental study. Very important are practical experience
of authors who had the opportunity to examine many of the structures during the works of
approval the tanks, carried out at the Department of Transport Silesian University of
Technology.
The study shows the process of verification of the selected container in two versions: cup
with a closed or open structure. FEM calculation results were compared with measurements
carried out during the pressure test. The very good accordance of results is obtained. the
examinations confirmed that tested tanks will be safe in operation.

REFERENCES

1. Galletly G.D.: Elastic buckling of complete toroidal shells of elliptical cross-section


subjected to uniform internal pressure, Thin-Walled Structures, 1998, 30, pp. 23-34.
2. Kaptan Y., Kisioglu Y. : Determination of burst pressures and failure locations of
vehicle LPG cylinders. International Journal of Pressure Vessels and Piping, Volume
84, Issue 7, July 2007, Pages 451–459.
3. Kisioglu Y., Brevick J.R., Kinzel G.L.: Determination of burst pressure and locations
of the DOT-39 refrigerant cylinders, Trans ASME J Pressure Vessel Technol, 123
(2001), pp. 240–247
4. Kisioglu Y., Brevick J.R., Kinzel G.L.: Minimum Material Design for Propane
Cylinder End Closures. Trans ASME Journal of Pressure Vessel Technology, vol.
130 (2008).pp.1-8
5. Redekop D., Xu B., Zhang Y.M.: Stability of a toroidal fluid-containing shell,
International Journal of Pressure Vessels and Piping, Volume 76, Issue 9, August
1999, Pages 575–581
6. X.H. Wang, B. Xu, D. Redekop: FEM free vibration and buckling analysis of stiffened
toroidal shells Thin-Walled StructuresVolume 44, Issue 1, January 2006, Pages 2–9
7. Kisioglu, Y., 2009, “Burst Tests and Volume Expansions of Vehicle Toroidal LPG
Fuel Tanks,” Turk. J. Eng. Environ. Sci., 33, pp. 1–9
8. Vladan Veličković: Stress and Strain States in the Material of the Stressed Toroidal
Container for Liquefied Petroleum Gas, Scientific Technical Review,Vol. LVII, No.3-
4,2007, pp. 94-105
9. Galletly G. D., 1998, “Elastic Buckling of Complete Toroidal Shells of Elliptical
Cross-Section Subjected to Uniform Internal Pressure,” Thin-Walled Struct., 30_1–4,
pp. 23–34.

52
10. Aksoley M. E., Ozcelik B., and Bican I., 2008, “Comparison of Bursting Pressure
Results of LPG Tank Using Experimental and Finite Element Method,” J. Hazard.
Mater., 151, pp. 699–709.
11. Xue L., Widera G. E. O., and Sang Z., 2010, “Parametric FEA Study of Burst
Pressure of Cylindrical Shell Intersections,” ASME J. Pressure Vessel Technol.,
132_3_, p. 031203.
12. Xue L., widera g. E. O., Zhifu Sang; Burst Analysis of Cylindrical Shells, Journal of
pressure vessel technology, 2008, vol. 130, no1, [Note(s): 014502.1-014502.5],
13. Nowacki W. Dźwigary powierzchniowe. PWN. Warszawa 1979
14. WARREN C. YOUNG, RICHARD G. BUDYNA, S.: Roark’s Formulas for Stress
and Strain. McGraw-Hill. 2002.
15. Magnucki K.: Wytrzymałość i optymalizacja zbiorników ciśnieniowych. Wyd. Nauk.
PWN. W-wa, Poznań, 1998.
16. Życzkowski M. red.: Wytrzymałość elementów konstrukcyjnych. PWN, Wa-wa,1988.
17. Bucki S.: Analiza możliwości dostosowania wózka widłowego napędzanego silnikiem
spalinowym do zasilania paliwem gazowym. Master's Thesis. Wydz. Transp. Pol.Śl,
Katowice 2007.
18. ABAQUS 6.10, Analysis User’s Manual, Volume III: Materials, Dassault Systèmes,
2010.
19. Flekiewicz M., Matyja T: Wpływ kształtu na stan naprężenia w zbiorniku
toroidalnym. Nowe technologie i materiały w metalurgii i inżynierii materiałowej.
Materiały IX seminarium naukowego, Katowice, 11 maja s. 409-412.
20. Kukiełka L., Kukiełka K., Geleta K., Cąkała Ł.: Optymalizacja zbiornika na gaz
płynny LPG, Autobusy, Technika, Eksploatacja, Systemy Transportowe, 5/2011,
p.222-226.
21. Kodur V., Dwaikat M., Fike R.: High-Temperature Properties of Steel for Fire
Resistance Modeling of Structures, Journal of Materials in Civil Engineering, Vol. 22,
No. 5, May 1, 2010.
22. European Committee for Standardization. _2005_. “General rules - Structural fire
design, EN1993-1-2.” Eurocode 3, Brussels.
23. Qiang X., Bijlaard K., Kolstein H.: Elevated-temperature mechanical properties of
high strength structural steel S460N: Experimental study and recommendations for
fire-resistance design, Fire Safety Journal 55, 2013, p.15–21.

53

View publication stats

You might also like