Sei sulla pagina 1di 6

Stress Analysis of Ammonia Storage Tank, Providing a basis for In-Service Inspection

Abstract
This paper presents a finite element based, stress analysis of an 23 year old ammonia storage
tank of Rashtriya Chemical and Fertilizers Ltd. Based on this analysis, past inspection report of
this tank and reported inspection results of such tanks elsewhere, an inspection plan is furnished.

1. Introduction

1.1 Purpose
Rashtriya Chemicals and Fertilizers Ltd., (RCF), uses a 15,000 tonne atmospheric storage tank
located at Pir Pau terminal, Chembur, Mumbai, for storage of ammonia. The tank was
commissioned in 1974 and inspected once in 1985. After 12 years of service the tank is to
undergo a second statutory inspection in 1997. In order to limit the extent of examination while
ensuring that all the important locations get examined, the plan of inspection (consisting of the
nature of examination, the locations to be inspected and the extent of inspection) needs to be put
on a sound footing. For this purpose RCF approached BARC. Following plan of action was
proposed:
• Finite element analysis of the outer tank (with the roof) and the inner cup.
• Piping analysis for evaluating the nozzle loads on the tank shells/roof.
• Evaluation of the local stresses in the shell/roof due to the nozzle loads.
These analyses will be for the most important loading, namely, internal pressure including
hydrostatic load and thermal stresses. Having the detailed picture of stresses at significant
locations will give the relative importance of different locations and will help in deciding the
locations to be inspected. It was also pointed out that the results of the detailed stress analysis
can be used for a fitness-for-purpose evaluation if such a need arises. The recommendations for
the inspection are arrived at based on this analysis and are supplemented on the basis of a
literature survey on the operating experience of similar equipment elsewhere.
The paper concludes with a final set of recommendations regarding zones to be inspected, the
extent of examination and method of examination

1.2 Structure Description


The tank in question is being used for the storage of liquid ammonia. The main storage tank
is a cup which is placed in an another tank. The tanks are of welded construction .The
thickness of the roof plates is 5 mm and that of the bottom plates is 10 mm. The shell
consists of several courses of plates of different thickness. The roof is constructed with
cantilevered support beams in the spherical segment and with connection between the roof
plate and the beams. Fig. shows the schematic diagram of the arrangement.
The outer tank and cup are anchored to reinforced concrete foundation with tie rods. It has
been designed in accordance with API 620 code along with Appendix R. [Ref.1]
INNER CUP
The diameter of the inner cup is 35 meters and its height is 23.8 meters. The cup contains
liquid ammonia stored at a temperature of -33 0C. The relative density of liquid ammonia is
0.683. The material used for construction is tough, fine grain steel of the specifications:
I. TT St E 36 (a German Specification equivalent to SA 516 Gr 70 )
Yield stress 360 MPa
U.T.S. 500 MPa
II. TT St E 29 (a German Specification equivalent to SA 516 Gr 55 )
Yield stress 290 MPa
U.T.S. 400 MPa
The plate thickness has been calculated with an allowable hoop stress of :
150 MPa for type I steels
120 MPa for type II steels
The cup has two stiffener rings. The upper stiffener ring is at a distance of 5 mm. from the top
edge of the cup.The lower one is at a distance of 1000 mm. from the top edge of the cup. It
is constructed on the inner side of the cup. This ring is supported by brackets which appear at
an interval of 7.5 0.
All plates on the cup wall are butt welded.
OUTER TANK
The outer tank shell is 36.6 meters in diameter and its height is 24.6 meters. Its purpose is to
act as a safety envelop in case the inner cup fails. The steel used for construction is same as
that used for cup.
The tank shell has two stiffener rings. Both are placed on the inner side of the wall. The upper
one is at a distance of 400 mm from the top edge of the shell. This stiffener ring is supported
by brackets at an interval of 7.50. The lower set is at a distance of 17900 mm. from the
bottom edge of the tank. This also is supported by brackets at an interval of 7.50.
The roof of the storage tank is supported by the outer tank. The roof rests on roof rafters.
There is an arrangement on outer tank shell to support these roof rafters. The rafters rest on
the console plates. These are at a distance of 150 mm from the top edge of the tank.
These are placed at an interval of 60, to support sixty roof rafters.
ROOF
The roof is designed as self-supporting with rafters in the form of a spherical segments of radius
36.6 meters, with composite construction between roof covering and roof rafters. There are thirty
pairs of roof rafters, making a total of sixty rafters placed at an interval of 60. The rafters of each
pair are connected by four polygonal bars. The roof rafter rests on console plate on one end and
on crown ring on the other end. The radius of crown ring is 1624 mm. It has an ‘I’ section. The
crown ring contains a pair of orthogonally placed cross-bars. The roof plates of thickness 5 mm
lie on the top of rafters, crown ring and cross-bars. The roof plates on the edge of the roof are 22
mm in thickness. These serve as compression ring.
TANK BOTTOM
The tank bottom consists of lap welded plates placed over insulating material. The whole
structure rests on a 0.61 meters thick concrete slab which is supported on a number of concrete
pillars. The length of each pillar is 1.016 mts
The tank contains a number of nozzles instrumnetation and fluid flow. Of these the following four
piping layouts have been considered for analysis.

1. NH3-Liquid inlet at Nozzle I (from jetty to tank)


2. NH3-Gas outlet at Nozzle II (from tank to suction of the compressor)
3. NH3-Liquid return at Nozzle III (from chiller to tank)
4. NH3-Liquid outlet at Nozzle IV (from pump to tank)

1.3 Loadings
The storage tank is designed for hydrostatic pressure, vapour pressure, wind and earthquake
loads. For deciding the locations of in-service inspection, the first two are the important ones and
analysis was performed for these loads. Analysis was also performed for thermal stresses which
were not considered during design.
DESIGN LOADING
The inner cup is designed for a 23.8 m high column of liquid ammonia. [Ref. 3]
Specific gravity of liquid ammonia = 0.683 Kg/m3
Equivalent pressure = 0.1626 Mpa (1 Mpa = 10 Kg/cm2 = 142 psi)
Pressure varying linearly from zero at the top to the maximum value at the bottom.
The outer tank is designed for a 22.02 m high column of liquid ammonia along with a vapour
pressure of 1050 mm of H2O (condition corresponding to spillage of liquid ammonia into the outer
tank)
Max. hydrostatic pressure = 0.1504 Mpa
Pressure varying linearly from zero at the top to the maximum value at the bottom.
Overpressure = 0.0105 Mpa
This is the only pressure acting on the roof, but for the tank walls, this is added to the pressure
due to hydrostatic pressure.
LOADING DURING NORMAL OPERATION
During normal operation, the inner cup is subjected to hydrostatic load corresponding to the actual
liquid level. Since this is just a scalar multiple of the design load, no separate analysis is required.
The stresses during normal operation can be obtained from the analysis for design loading by
multiplying with a suitable factor.
However, during normal operation the outer tank along with the roof is loaded only by the vapour
pressure of ammonia which is maintained between 400-800 mm of water column. There is no
hydrostatic load. Hence, a separate analysis was performed for the outer tank for an internal
pressure of 800 mm of water (0.008 Mpa).
For piping the main objective is to find out the forces and moments generated at the Nozzle-Shell
(outer wall of the tank) junction when the pipe line is subjected to a differential temperature of -63
deg. C ( from ambient temperature, taken as 30 deg. C to -33 deg. C that of the liquid Ammonia).

2. Finite Element Modelling


The analysis was done using a general purpose FEA software, COSMOS/M ver 1.70. The
modelling details are presented below.
MODEL OF THE INNER CUP
The tank cup has a 7.50 symmetry and could have been modeled using only a 7.50 sector.
But since the outer tank had to be modeled using a 900 sector, a similar 900 sector was
modeled for the inner tank also. The tank wall was meshed using four noded shell elements.
The stiffener rings and the supporting brackets were also modeled using four noded shell
elements. The bottom edge of the tank model was completely fixed. (For justification see the
para. on bottom plate of outer tank). Symmetric boundary conditions were applied on the
straight edges. The top edge was kept free. The finite element mesh is shown in Fig.The
model had 708 elements and 750 nodes. The total degrees of freedom were 4176 .
MODEL OF THE OUTER TANK AND ASSEMBLY
The second model was of outer tank shell, roof and tank bottom. The shell assembly is very
complex and some reasonable assumptions needed to be made. A 900 symmetry was
utilized for modeling. The tank shell was modeled using four noded shell elements. The
stiffener rings and the brackets were also modeled using four noded shell elements. 120
elements were used along the circumference to model the tank. The finite element mesh is
shown in Fig.
While modeling the console plates and their brackets following assumptions were made. The
console plate was omitted and the roof rafter web was connected directly to the tank wall.
This assumption was required as the size of the console plate is very small in comparison of
tank circumference. Breadth of the console plate lying on the plane of the tank wall is 150
mm. If number of elements in the circumferential direction were to increase such that a
compatible mesh is obtained with console plate, it would have resulted in 6480 elements on
the tank wall (compared to 2160 elements at present). Since compatibility of the mesh has to
be maintained while modelling roof and tank bottom also, the number of elements then would
have exceeded the elemental and nodal limitations of the software. The approximation in the
stresses at the junction of the rafter with the shell is later removed by performing a
substructure analysis of this zone alone. This is described below.
The discontinuity at the shell-rafter junction affects the results only in a small local region.
Hence the global results are not affected by the approximation. In order to get the correct
stress distribution near the discontinuity, a small region (substructure) surrounding the
discontinuity was isolated from the global structure. The finite element mesh in this
substructure was refined to model the junction details correctly. This refined mesh is shown
in Fig. The boundary of this substructure was given the displacements and rotations
calculated in the global model and the load on this region was applied. This model was then
analysed and the stresses at the shell-roof-rafter junction are taken from the results of this
analysis.
The Roof
The Cross Bar : Since a 90 degree sector is taken , half of a cross bar is modelled. The
cross bar has a ‘T’ section . The roof plates lie on the top of cross bar. To achieve mesh
compatibility , this ‘T’ section was replaced by an equivalent ‘ I ’ section of thickness 18.8
mm. The cross bar is attached to the crown ring as shown in the figure below for preserving
mesh compatibility.The finite element mesh is shown in Fig.
The Crown Ring :The crown ring was also modelled using 4 noded shell elements. No
approximations were made while modelling it. The finite element mesh is shown in Fig.
The Roof Rafters and The Polygonal Bars :The roof rafter and polygonal bars were
modelled using 4 noded shell elements. The rafters have an inverted ‘ T ‘ section and are
modelled similarly. The finite element mesh is shown in Fig.
The Roof Plates :The roof plates were modelled by using 3-noded shell elements.They
were modelled in two parts , both parts being sections of spheres. The first segment starts
from upper edge of the tank, ending at the outer edge of crown ring . The second section
starts from the outer-upper edge of crown ring and extends inwards. This is supported by the
cross-bar. The finite element mesh is shown in Fig.
Model of the Bottom Plate
Considering the thickness of the bottom concrete slab and the number of supporting columns
underneath, it was felt that the slab will not deform significantly and the bottom of the shell could
as well be assumed to be fixed without modelling the bottom plate. In order to check this, a model
consisting of the shell and the bottom plate was analyzed. The bottom plate was modelled as a
0.61 meters thick concrete slab supported on concrete columns of length 1.016 meters. The
diameter of the concrete slab is 37.6 meters. These dimensions were measured on site as no
drawing describing these parts was available. The arrangement of columns below the concrete
slab was observed and the distances were measured at the site. The arrangement is shown in
Fig. , and the finite element mesh for this model is shown in Fig. The concrete slab was modelled
using three noded shell elements. The value of Young’s modulus taken was 2.0 x 104 N/sq.mm.
The supporting columns were modelled using 3-D Beam elements. The base of the concrete
pedestals was assumed to be fixed. The model had 10608 elements and 7597 nodes. The total
degrees of freedom were 44737. Hydrostatic pressure corresponding to the testing condition (the
maximum possible load) along with the overpressure of 1050 mm of water was applied on the slab
as well as on the shell. The maximum deflection was observed to be only 2 mm. The deflected
shape is shown in Fig.
From the above analysis it is clear that the bottom slab is very rigid and need not be modeled in
the analysis. The tank can be assumed to be fixed at its bottom edge. Therefore, all further
analysis for both the inner cup and the outer tank was performed without the bottom slab.
A full model of the outer tank assembly is shown in Fig. The model had 9910 elements and
8834 nodes.
The elbows in the locations of expansion joints and turnings were modeled using the ELBOW
element and all other straight sections are modeled using the PIPE element of the general
purpose finite element code Cosmos/M.

Potrebbero piacerti anche