Sei sulla pagina 1di 28

5.

61
BIGGER & CHEAPER LNG TANKS?
OVERCOMING THE OBSTACLES CONFRONTING
FREESTANDING 9% NICKEL STEEL TANKS UP TO
AND BEYOND 200,000 m
3
DES RESERVOIRS DE GNL PLUS GRANDS
ET MOINS CHERS?
COMMENT SURMONTER LES OBSTACLES AUXQUELS DOIVENT
FAIR FACE LES RESERVOIRS AUTOSUPPORTANTS EN ACIER AU
NICKEL A 9% DUNE CAPACITE DE 200 000 m
3
ET PLUS
Bob Long
Technical Director
Whessoe LGA Gas Technology Ltd.
Brinkburn Road
Darlington DL3 6DS, U.K.
ABSTRACT
Much was made at LNG 11 of the high cost of the storage components of LNG
projects to the point that the cost of LNG tanks could have an adverse influence on the
economic viability of potential projects. This together with the steadily increasing unit
capacity of LNG tanks, driven by financial and effective use of space considerations,
make the constraints on tank capacity and cost imposed by the design codes more than an
inconvenience.
Both API 620 Appendix Q and BS 7777 : 1993 include limitations on allowable
design stress and maximum shell plate thickness and impose hydrostatic testing
requirements which limit the effective capacity of LNG tanks built to economic (height to
diameter) proportions. This is illustrated by means of examples.
The opportunity exists to rewrite some of these rules within the current deliberations
of the CEN Committee TC 265 as it seeks to produce a new Eurocode for low
temperature storage tanks. This paper presents the case for the use of the higher (BS
7777) stresses with the partial hydrotest requirements, which would allow tank capacities
up to, and beyond 200,000m
3
without excessive shell thickness. The conventional
wisdom relating to the benefits of a full hydrostatic test, with particular reference to 9%
Nickel steels and Nickel based alloy weld metals, will be challenged by considering
currently available material quality and crack initiation and/or arrest fracture control
philosophies, based on the wealth of material test data already available to the industry.
5.62
The influence of these proposals on tank capacity, cost and construction timescale
will be illustrated by way of examples.
RESUME
Pendant la LNG11, le cot lev des composants entrant dans la constitution des
systmes de stockage de GNL a fait lobjet dune grande attention. En effet, le prix des
rsevoirs pourrait avoir des rpercussions ngatives sur la viabilit conomique des
projets potentiels. Ceci ajout laugmentation croissante de la capacit des rservoirs,
pour des raisons financires et dutilisation optimale de lespace, rend encore plus
pnibles les contraintes lies la capacit et aux cots imposes par les normes de
construction.
Les normes API 620 Annexe Q et BS 7777 :1993 dfinissent des limites sur les
contraintes admissibles ainsi que lpasseur maximale des parois et imposent des essais
hydrostatiques qui limitent la capacit utile des rservoirs de GNL construits avec des
proportions conomiques (hauteur par rapport au diamtre). Cette question est illustre
par des exemples.
Lopportunit de modifier certaines de ces rglementations se prsente dans le cadre
des dlibrations actuelles du Comit CEN TC 265 charg de produire un nouveau
Eurocode pour les rservoirs de stockage basses tempratures. Ce texte prsente des
arguments en faveur de lemploi de contraintes plus leves (BS 7777) avec des essais
hydrostatiques partiels qui permettrait de porter la capacit des rservoirs
200 000 m
3
et plus sans augumenter de faon excessive lpaisseur des parois.
Lattitude conventionnelle relative aux avantages dessais hydrostatiques complets, avec
un rfrence particulire aux aciers au nickel 9% et aux alliages dapport base de
nickel, est remise en question par ltude de la qualit des matriaux actuellement
disponibles et des philosophies relatives au contrle de la fissuration, cest--dire
dclenchement et/ou arrt de la fissuration, en se basant sur le grand nombre de donnes
dessais disponibles dans lindustrie.
Linfluence de ces propositions sur la capacit des rservoirs, leur cot et les dlais de
construction est illustre au moyen dexemples.
5.63
BIGGER & CHEAPER LNG TANKS?
INTRODUCTION
A significant amount of attention has been focused in recent years on the need to
reduce the CAPEX in LNG projects from wellhead to end user. The storage component
of both export and import terminals represents a significant proportion of the total costs,
particularly in the latter instance. The cost of LNG Storage tanks was discussed at LNG
11 in Birmingham and will doubtless be subject to close attention at LNG 12.
At the same time there is a clear tendency towards LNG tanks of greater capacity.
The tanks constructed at Canvey Island in the 1960's were of 9000m capacity. During
the 1970's and early 1980's the British Gas peak shaving tanks were 50,000m. For a
while 100,000m seemed a popular choice and for Dabhol 160,000m tanks are being
considered.
The increase in tank capacity is driven by the desire for lower unit costs ($/m stored)
arising from scale effects together with savings associated with the reduction in fixed
costs (i.e., pumping systems, instrumentation, utilities, etc.) and economic use of land
area.
This paper examines the constraints on the size of above ground free standing 9%
Nickel Steel primary containment which arise from various design conditions, codes and
possible modifications to these codes. The paper also considers the methods for
demonstrating and achieving tank integrity and safety. We are confident that practical
and economic outer tanks of single, double and full containment types can be provided to
match the inner tank sizes discussed in this paper.
One obvious means of arriving at a lower unit storage cost is to choose single
containment rather than full containment. This however is a complex decision involving
local codes and regulations, the outcome of comprehensive and detailed hazard analyses,
safety studies and the owners preferred storage philosophy. Frequently the seemingly
apparent economic choice becomes less certain when the observers viewpoint choice is
broadened from the single subject of storage to the operating facility as a whole together
with its environs. This aspect is not considered in detail in this paper.
THE UNIT COST/SIZE RELATIONSHIP
It has been presumed in the introduction that increases in tank size brings economies
of scale and a reduction in unit cost.
This was demonstrated by the following exercise:
Tank Types
Single containment (Fig 1)
Full containment (Fig 2)
5.64
Figure 1
Figure 2
5.65
Tank Capacity
50, 000 m
100, 000 m
150, 000 m
200, 000 m
250, 000 m
Design Basis
Code API 620 Appendix Q (Ref 1)
Design stress Maximum allowed by code
NPSH heel 1.0 m
Ullage for slosh, etc. 1.0 m
Design SG 0.48
Design Temp - 165C
Foundation Ground bearing slab
(No piles required)
The dimensions used for the various tank capacities are listed in Table 1.
Table 1. Tank Dimensions For Unit Cost/ Size Calculations.
Capacity
(m
3
)
Di
(m)
Do
(m)
Hi
(m)
Ho
(m)
50,000 52 54.4 25.68 27.97
100,000 65 67.4 32.31 34.61
150,000 75 77.4 36.14 38.44
200,000 85 87.4 37.44 39.74
250,000 90 92.4 41.52 43.80
These examples have been costed on a UK supply and construction basis, each being
fitted with the following:
- 4 off intank pumps complete with pump columns, foot valves and handling
equipment.
- Tower stairway and roof access equipment.
- Full set of instrumentation including LTD measurement.
- Full electrical installation including base heating.
- The normal pipework to a termination point at grade.
The unit cost/tank capacity is shown in Figure 3.
5.66
Figure 3
This curve shows that the unit cost does decrease as capacity increases for both single
and full containment tanks. The rate of reduction decreases with increase in capacity.
Detailed examination of the estimates behind these curves suggests that for the very big
tanks the tonnage of materials, in particular 9% Nickel steel for the inner tank shell and
the carbon steel outer tank shell begins to increase in a disproportionate way. It is also
interesting to note that the cost difference between single and full containment is modest
and reduces from around 17% to 12% as tank capacity increases over the size range
covered. We have not included the costs of secondary impoundment for single
containment tanks. When these are included together with costs associated with
additional fire protection and economic use of site area, then it is quite probable that
many real cases will indicate full containment to be the most economic choice.
STORAGE COSTS AS A PROPORTION OF TOTAL COSTS
The costs of a number of typical LNG import terminals have been considered.
Clearly the cost of the storage component varies from one project to another depending
on many considerations but on average varies between 45% and 65% of the total facility
cost. This suggests only that storage costs are a significant cost of the total facility costs
and as such warrant a detailed review to ensure that their costs are minimised. It is also
usually the case that the use of large tanks maximises the use of available site area.
PARAMETERS WHICH INFLUENCE TANK SIZE
From the foregoing it is clear that greater tank capacity equates to reduced unit
storage cost.
Unit Cost Ratio vs Tank (Usable) Capacity
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
50 100 150 200 250
Capacity x 1000 CuM
R
e
l
a
t
i
v
e

c
o
s
t

/

m
3

s
t
o
r
e
d
Single
Full
5.67
The other parameters which will influence or limit the size of tank which can be
constructed are:
i) Design code or method
ii) Material properties
iii) Soil conditions
iv) Base insulation
v) Seismic design data
vi) Local planning rules (i.e., maximum roof height)
Each of these are discussed in subsequent paragraphs.
DESIGN CODE OR METHOD
The two most commonly used codes for the design of LNG tanks are API 620
Appendix Q and BS 7777:1993 (Ref 2).
API 620 Appendix Q
This standard was written for single containment tanks and has served the industry
well for many years. It requires a minimum hydrostatic test equivalent to 1.25 times the
maximum product load on the tank foundation and bases the allowable stresses for design
and test on the weld metal properties allowing maximum values of 400 N/mm Yield and
690 N/mm
2
UTS. This means that the shell thickness is based on the operation condition.
The maximum permitted shell thickness for the purposes of this exercise has been
assumed to be 38.1 mm (1 ).
BS 7777:1993
This standard has its origins in the EEMUA 147 document (Ref 3) which was in
many ways a reaction to the incident in 1977 at Umm Said in Qatar where the plant was
destroyed by the failure of a liquid propane tank in a brittle manner. The subsequent
technical and legal processes served to throw little light on the real causes of this event.
This was however a very influential event which through the EEMUA 147 document
pushed the industry towards secondary containment, full height hydrostatic testing and in
the case of 9% Nickel steels, increased fracture toughness. This has been discussed in
some detail by J.B. Denham (Ref 4).
BS 7777:1993 incorporated all of the major requirements outlined in EEMUA 147.
Those most significant to this paper are the increased hydrostatic testing level to the full
product height (for all products), the increased minimum thickness requirements for shell
and bottom plates and the increased allowable stress for the operating conditions.
CEN TC 265
This is the new Eurocode which will have a section devoted to low temperature
tankage. This document is in the process of being written, although the rate of progress
leaves something to be desired. The creation of a Eurocode means that all national codes
5.68
within the EEC are subject to standstill (including BS 7777) and consequently cannot
be amended.
If the industry wishes to promote any new ideas and design methods for low
temperature tanks, then the Eurocode is perhaps the only suitable vehicle.
Worked Examples
In an effort to quantify the influence of the design codes together with a variety of
possible amendments to those codes, we have carried out a number of designs of a 9%
Nickel steel inner tank with a useable capacity of 200,000 m . A number of different
height to diameter ratios have been examined. The basis of these calculations is the same
as stated above. The codes used and the modifications made are listed in Table 2. The
results of this exercise are shown in Figures 4 and 5 where the calculated total inner tank
weight (shell, shell stiffening, bottom and annular plates) has been plotted against tank
diameter and D/H ratio.
Table 2. Design basis assumed for designs of 200,000m
3
LNG Inner Tank shown in Figs
Design Stresses
Case Hydrotest
Sd
(N/mm
2
)
St
(N/mm
2
)
Modifications to
Basic Code
BS 7777 Full 260.0 340.0 None
API 620 Partial 229.9 340.0 None
BS 7777 (mod 1) Partial 260.0 340.0 Hydrotest
BS 7777 (mod 2) Partial 260.0 340.0 Hydrotest
Adopt API min thickness
BS 7777 (mod 3) Partial 275.0 340.0 Hydrotest
Adopt API min thickness
Increase Sd
5.69
Figure 4 Figure 5
Figure 6
Weight vs Diameter for various design codes
1500
1700
1900
2100
2300
2500
2700
2900
7
5
7
7
.
5
8
0
8
2
.
5
8
5
8
7
.
5
9
0
Diameter (m)
W
e
i
g
h
t

(
T
e
)
API
BS
BS(mod1)
BS(mod2)
BS(mod3)
Weight vs Ratio D/H for various design Codes
1500
1700
1900
2100
2300
2500
2700
2900
1.6 1.8 2 2.1 2.4 2.5 2.8
D/H
W
e
i
g
h
t

(
T
e
)
API
BS
BS(mod1)
BS(mod2)
BS(mod3)
1st Course Shell Thickness vs Tank Diameter
20
25
30
35
40
45
50
55
75 77.5 80 82.5 85 87.5 90
Diameter (m)
1
s
t

c
o
u
r
s
e

t

(
m
m
)
API
BS
BS(mod1)
BS(mod2)
BS(mod3)
5.610
From Figure 4 a number of issues become apparent. The first is that BS 7777
unamended produces a very uneconomic design. In fact the bottom course shell
thickness plotted in Figure 6 exceed the thicknesses allowed by the code and which
experience would suggest as sensible. The main difference between the unamended BS
7777 designs and the remaining cases considered is due to the requirement full/partial
hydrostatic testing which is discussed below.
The ratio of D/H has an influence on the total weight. The larger diameter designs
are lighter that their smaller taller alternatives by between 14% and 22%. Whilst this is
interesting it is probable that costs associated with the outer tanks, the insulation and the
land usage may consume all or part of these differences. These are also practical
considerations which may dictate D/H ratios associated with local soil conditions or
seismic design problems.
The influence of the minimum shell and bottom plate thickness requirements of the
BS and API codes are more than anticipated contributing a difference in total weight of
between 5.0% and 6.5%. Clearly there are differences unseen by this exercise associated
with the increased erection costs for the thinner shells and the increased complexity of the
shell stiffening. It would however seem that this is a worthwhile and uncontentious
saving which equals or exceeds the possible and more contentious savings associated
with increasing the allowable stress in the tank shell.
MAXIMUM SHELL THICKNESS
BS 7777 proposes a maximum thickness for types IV and V 9% Nickel Steel of 30
mm but leaves the door open for thicknesses in excess of this figure to be agreed between
the purchaser and the manufacturer.
Fabrique de Fer de Charleroi of Belgium who supplied the 9% Nickel plate for the
tanks in Greece and Trinidad, have provided plate and welding data which indicates that
the required properties can be met in the plate and weld metal in thickness up to 38.1 mm
(1). This is considered therefore a sensible and supportable maximum thickness for
the purpose of this exercise.
Hydrotest Test Level
The European Standard EN 1473: 1997 (Ref 5) requires the primary container to be
tested to 125% of the operating base loading and states quite bluntly that higher
hydrostatic test heights with the purpose of stress relieving the steel are pointless for
containers made from 9% nickel steel. There is a considerable support for this view,
particularly where the new enhanced toughness crack arresting grades of 9% nickel steel
are concerned. The development of enhanced grades of 9% nickel steels, the test work
and the development of crack arrest criteria for refrigerated gas storage tanks is
comprehensively covered by Denham (Ref 6).
API 620 Appendix Q was not influenced to change its test height criteria as a result of
the Qatar incident which it viewed as relevant only to LPG tanks and chose to amend
only Appendix R. Thus a considerable practical body of satisfactory experience relating
to the adequacy of the partial hydrotest exists for LNG tanks built to the API rules.
5.611
Inspection
The main problem on 9% Ni tanks is the density difference between the parent plate
and the high nickel weld metal which on the radiograph creates a marked difference in
photographic density. Thus in order to achieve the requirements for sensitivity and
density the parameters of the radiographic technique need careful adjustment between
film type, Kilovoltage and exposure time.
The resultant technique must also provide an overall exposure time which produces
an acceptable production rate as radiography is commonly a critical path item on the
programme. This is a particular problem when 100% radiography is specified for both
vertical and circumferential seams.
Production rate can be exacerbated by the size of the equipment, i.e., physical
handling problems within the tank. As plate thickness increases so does the size of the
equipment with typically 250-320 kV being required for 25-35 mm thicknesses and at
these power rates only metal ceramic X-ray heads/power sources are suitable for site
production. However, it still requires proper study to provide the correct handling and
support systems for the equipment in order to maximise production. As thickness
reduces then X-ray power requirement reduces which eases handling problems in terms
of weight but increases them in terms of height above grade (dependent on construction
method).
It is considered that more effort should be given to justifying the use of Gamma
radiography. This technique is acceptable to the major codes but not commonly accepted
by clients or consultants.
The Gamma ray technique has been shown to be capable of achieving density and
sensitivity requirements and by using implanted defect plates its ability to detect crack
like defects has been demonstrated.
In the future more detailed work on the use of ultrasonic inspection using time of
flight techniques for recording purposes in conjunction with longitudinal shear wave
probes could provide the basis for an alternative inspection system. Real time
radiography (i.e., pulsed X-ray with computer enhanced real time imaging) is capable of
application to 9% Ni tank on a mechanised basis and could dramatically improve
production rates. However, clients, consultants and codes would have to accept the
technique and potential equipment costs are high.
The commonly specified requirement for 100% radiography for both horizontal and
vertical seams is an expensive luxury. If we consider the basic requirements laid down in
API 620 Appendix Q, taking the 200,000 m
3
tank as an example, the length of seam
radiographed is some 35% of the total. This is a substantial saving in money and may
well take radiography off the critical path. There is a good case to be made for the basic
API level of radiography based on an ECA approach or for circumferential seams only to
combine radiography with some ultrasonic examination (which does not require separate
shift working), a combination which could be used to advantage.
5.612
MATERIAL PROPERTIES FOR LNG INNER TANKS
Materials Requirements
The integrity and safety of the 9% Ni Steel inner tank, defined as the avoidance of
catastrophic failure depends on three factors.
i) 9% Ni Steel plate and weld metal properties
ii) Fracture resistance properties as controlled by the principal stress, secondary
stress and residual stresses and the fracture behaviour under the influence of a
defect.
iii) Freedom of the welds (including the heat affected zone) from deleterious
flaws.
The proposals made in this paper are dependent on these three criteria and it can be
shown that these conditions can be satisfied.
9% Ni Steel Plates
The most widely accepted specification for 9% Ni steel plates is ASTM A553
(quenched and tempered). For this material API 620 Appendix Q requires a minimum
Charpy v-notch impact value for full size specimens tested at 196C of 20J (transverse).
In reality, the toughness achieved can be summarised as follows:-
10 mm thick: 95-237J (Whessoe LGA data)
22.3 mm thick: 151-225J (Whessoe LGA data)
38 mm thick: 165 205 J (>1.5 lat exp) (Fabrique de Fer data)
Hence, the quality of steel available exceeds that which is considered to be acceptable
for API 620 Appendix Q.
Weld Metal Properties
The most widely used weld metal is ASME Section 2c SFA SU EN CrMo-6. This
electrode provides a weld metal with undermatching strength, however the typical results
are:
YS ) 420 N/mm
2
)
UTS ) 690 N/mm
2
) Whessoe LGA data
CV (-196) ) 70J )
When the CTOD values are considered for both the 9% Ni Steel HAZ and the weld
metal, the results exceed 0.3 mm at 165C at thicknesses up to 38 mm.
Thus it can be seen that more than adequate properties can be attained for the welded
structures.
5.613
Freedom from Defects
Extensive welding experience has shown that the achievement of the workmanship
defect acceptance levels required by API 620 Appendix Q (i.e., ASME VIII standards)
can be achieved without excessive repairs or resort to extreme welding procedures.
When this is coupled with the allowable defect sizes incorporated into the PD 6493
based ECA, it can be concluded that the risk of unacceptable flaws remaining the welds is
acceptably low.
Fracture Behaviour
The results of wide plate tests (Ref 7) show that failure in the presence of a large
defect occurs at a factor of 2-3 times the design stress. The fracture behaviour
demonstrated is that of crack growth initiation. It is not possible to conclude whether the
crack will propagate or be arrested subsequently.
There is a view that the fracture control should incorporate crack arrest in the event
that crack growth initiation should occur. This may be an additional feature, but within
the existing design principles, crack arrest is not required.
SOIL CONDITIONS
Poor soil conditions may limit the proportions or the size of tank which can be
constructed due to limited ground bearing strengths. This may also be a complication to
be considered together with seismic design. This is outside the scope of this discussion.
INSULATION
The majority of LNG tanks are founded on cellular glass manufactured by Pittsburgh
Corning of the USA. For many years this material was manufactured with a minimum
average strength of 0.74 N/mm (107 psi) and a minimum single value of 0.54 N/mm (78
psi).
EEMUA 147 and BS 7777 give permitted factors of safety for cellular glass under
compression arising from operating and hydrostatic test loadings. In the case of BS 7777
which required a hydrostatic test to the full product level, the use of this material and the
given factors of safety will restrict the maximum product to 22.0 m. This is a severe
restriction which in the case of the 200,000 m useable capacity example would require
an inner tank of some 110 m in diameter, clearly an uneconomic prospect.
For API 620 Appendix Q which requires a partial hydrostatic test to a level providing
a minimum foundation overload of 25%, the use of this material is less restrictive
allowing a maximum product height of 36.7 m which in the case of 200,000 m example
would require an inner tank diameter of 85 m, not an unreasonable choice.
In recent years Pittsburgh Corning have produced a range of stronger grades of
Foamglas now known as HLB 800 through to HLB 1600. This latter material has a
minimum average compressive strength of 1.6 N/mm and a minimum single value of 1.1
N/mm. This would permit a tank with full hydrostatic test a maximum product full
5.614
height of 44.85 m. This should allow a tank unrestricted geometry at a modest cost
premium.
In addition to the stronger grades of Foamglas there are a number of other materials
which could be used for LNG tanks base insulation. One which Whessoe LGA have
been actively considering is Divinycell, a PVC expanded foam made by DIAB of
Sweden. This comes in a wide range of strength grades and should not impose any
restrictions in tank geometry.
SEISMIC DESIGN
In recent years a number of LNG terminals have been built in areas of the world
which are subject to significant seismic events. Above a certain level the seismic design
criteria begin to dominate the tank design. In Turkey at the Marmara Ereglisi LNG
Terminal the tank geometry was restricted (Ref 8) whilst at Revithoussa Island LNG
Terminal in Greece the extreme seismic parameters led to the use of seismic isolation of
the complete full containment LNG tanks (Ref 9).
The inclusion of seismic design criteria in this paper will confuse the basic issues
under discussion and have not been pursued further.
LOCAL PLANNING RULES
Local planning rules commonly give a maximum shell or dome roof height which
will obviously influence the shape and/or size of tanks which can be constructed on a
particular site. This is clearly outside the scope of this discussion.
CONSIDERATIONS OF INTEGRITY AND SAFETY OF LARGE LNG
INNER TANKS
The design and construction of very large tanks will inevitably cause some concern.
Although we are proposing the use of various design rules similar to or identical with
API 620 Appendix Q which has been the main guideline for the industry for many years,
the tanks we are proposing are big and will pioneer a number of new areas of territory in
terms of size alone.
We have considered two proposals aimed at providing reassurance:
Increase Hydro Test Height
An increase in the hydrostatic test fill height up to the point where the thickness of the
lowest course is the same for both operating and test design criteria can be achieved
without adding cost for the outer tank shell. The only costs relate to the stronger
foundation to withstand the additional loadings and a modest sum for the additional test
water.
This allows the API maximum overstress to increase from 25% to 48% and to push
the point where the overstress reduces to zero further up the tank shall as indicated in
Table 3.
5.615
For the BS 7777 mod 1 design the effect is less marked as can be seen in Table 4
where the overstress increases from 25% to 31%.
Table 3
Effect of increasing hydrostatic test height such that
course 1 operating and test thicknesses are equal.
(Based on 200,000m
3
, 80m dia., API 620 APPENDIX Q Design)
Course Width
(m)
Thickness
(mm)
API Test Fill
Ht
(overstress %)
Increased fill
Ht
(overstress %)
1 3.501 33.54 25.0 47.9
2 3.501 30.67 17.2 42.2
3 3.501 27.80 7.8 35.4
4 3.501 24.93 0 27.0
5 3.501 22.06 0 16.4
6 3.501 19.19 0 2.7
7 3.501 16.33 0 0
8 3.501 13.46 0 0
9 3.501 10.59 0 0
10 3.501 9.60 0 0
11 3.501 9.60 0 0
12 3.501 9.60 0 0
Adjust Operating Stress for Courses Which Do Not Experience a Significant Test
Overstress
There is a perception that parts of the tank receiving a small hydrostatic test
overstress are more at risk than those parts which get a higher overstress. A reaction to
this is to reduce the operating stress in certain parts of the tank shell. The details of this
adjustment are described in Table 4.
The penalties in terms of weight and consequently cost are modest.
5.616
Table 4
Hydrostatic Test Height increased such that course 1 operating and
test thicknesses are equal.
Effect of reducing operating stress by 10% for parts of the shell
receiving an overstress of <10%.
API 620 Appendix Q
Course Width
(m)
Test
Overstress
(%)
Original
Design
(mm)
Adj. Design
(mm)
1 3.501 47.9 33.54 33.54
2 3.501 42.2 30.67 30.67
3 3.501 35.4 27.80 27.80
4 3.501 27.0 24.93 24.93
5 3.501 16.4 22.06 22.06
6 3.501 2.7 19.19 21.11
7 3.501 0 16.33 17.96
8 3.501 0 13.46 14.81
9 3.501 0 10.59 11.65
10 3.501 0 9.60 9.60
11 3.501 0 9.60 9.60
12 3.501 0 9.60 9.60
Total Weight
(tonnes)
1600.5 1642.5
i.e., +2.6%)
BS 7777 MOD 1
Course Width
(m)
Test
Overstress
(%)
Original
Design
(mm)
Adj. Design
(mm)
1 3.501 30.7 29.41 29.41
2 3.501 23.4 26.87 2687
3 3.501 14.6 24.34 24.34
4 3.501 3.7 21.81 23.99
5 3.501 0 19.27 21.20
6 3.501 0 16.74 18.41
7 3.501 0 14.20 15.62
8 3.501 0 12.00 12.84
9 3.501 0 12.00 12.00
10 3.501 0 12.00 12.00
11 3.501 0 12.00 12.00
12 3.501 0 12.00 12.00
Total Weight
(tonnes)
1496.8 1553.4
(i.e., +3.8%)
5.617
ASSESSMENT OF TANK INTEGRITY
The proposals described above to provide reassurance of adequate or indeed
increased integrity require some form of confirmation that the desired effect is being
achieved.
To this end an Engineering Critical Assessment was made for one of the 200,000 m3
designs (plus a 50,000 m3 design for comparison purposes) considered in this paper.
The assessment was made by applying PD 6493 1991 (Ref 10) to the designs, to
determine whether brittle fracture would be likely. PD 6493 was selected because it is
recognised world-wide. There are other assessment methods and it would be expected
that these would provide a similar analysis.
Two postulated flaws were considered;
1) Through thickness flaw - 30mm long
2) Surface flaw - 5mm deep x 100mm long
These flaw sizes were chosen as they would be easily visible on a radiograph and
detected by surface flaw examination techniques, i.e. dye penetrant inspection.
The assessments carried out can only be used as a comparative measure.
The material properties used were based on the weld metal.
At room temperature
Yield Stress = 400 N/mm
2
UTS = 690 N/mm
2
At -165 C
Yield Stress = 550 N/mm
2
UTS = 1000 N/mm
2
CTOD value = 0.3mm
The primary stresses used in the assessments were based on the circumferential
stresses in the tank shell, i.e., those applied to the vertical seams.
The residual stress in the weld before hydrostatic testing was assumed to be equal to
the yield strength of the weld metal. When the tank was cooled down the increase in
residual stress, due to the differential contraction between the weld and parent materials,
was taken as 50 N/mm
2
(Ref 11).
The Level 1 approach within PD 6493 was adopted. This is the simplest assessment
level and incorporates in-built safety factors for fracture and plastic collapse, averaging
about 2.
The reduction in residual stress associated with hydrostatic testing is defined in PD
6493 as follows;
5.618
When a proof test has been performed for subsequent fracture analysis at lower
temperatures, the residual stress level after the proof test can be assumed to be the lower
of

Y
or 14 .
'


n
f
Y
where

n
is the net section stress under the proof load conditions

Y
is the appropriate material yield strength at the proof test temperature and

f
is the flow strength (assumed to be the average of the yield and tensile
strengths) at the proof test temperature.
The results of the assessment are presented in Figures 7 and 8, the Failure Assessment
Diagrams (FADs) for each course of the tank. The axes of the graphs are labelled
Fracture and Stress.
The bold lines marked on the FADs define the boundaries within which failure will
not occur. If the horizontal boundary is exceeded then failure will be by fracture and if
the vertical boundary is exceeded then failure will be by plastic collapse.
From the FADs it can be seen that even for a tank which does not receive a
hydrostatic test the postulated flaws, which are readily detectable, would not cause failure
of the tank by crack growth initiation. This would give a measure of confidence which
would be increased when a hydrostatic test is applied. Also from the FADs, the benefits
of the effects of hydrostatic testing on the residual stress can be seen. Obviously the
maximum benefit is derived on the bottom course with a decreasing effect moving up the
tank. At Course 5 any benefit from a 1.25 hydrotest is negligible, similarly at Course 7
the benefit of a 1.48 hydrotest is also negligible for an API 620 Appendix Q design.
In assessing an API design for a 50,000m
3
(1.25 hydrotest) tank the results for the
lower courses are no different than for the 200,000m
3
tank for a through thickness flaw.
This is because of the comparatively large radii involved with tanks as opposed to
pressure vessels. Therefore any change in radius of the tank has a minimal effect on the
flaw. In this respect there is no significant increase in risk from failure by fracture for an
80m dia. tank than for a 52m dia. tank. However for a surface flaw when the thickness of
the shell is more important, the 50000m
3
tank is apparently more at risk from failure.
There are many API 620 tanks of around 50,000 m
3
in service which should provide
comfort for the larger size of tanks currently being considered.
5.619
Figure 7a Surface Flaws
COURSE 1
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
COURSE 2
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
COURSE 3
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
5.620
Figure 7b Surface Flaws
COURSE 4
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
BS7777 -10%
COURSE 5
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
BS7777 -10%
COURSE 6
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
BS7777 -10%
API -10%
5.621
Figure 7c Surface Flaws
Figure 7c Surface Flaws
COURSE 7
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
BS7777 -10%
API -10%
COURSE 8
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31 HYDRO
BS7777 -10%
API -10%
COURSE 9
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31 HYDRO
API -10%
5.622
Figure 7d Surface Flaws
COURSE 10
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31 HYDRO
COURSE 11
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31 HYDRO
COURSE 12
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31 HYDRO
5.623
Figure 8a Through Thickness Flaws
COURSE 1
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
COURSE 2
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
COURSE 3
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
5.624
Figure 8b Through Thickness Flaws
COURSE 4
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
BS7777 -10%
COURSE 5
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
BS7777 -10%
COURSE 6
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
BS7777 -10%
API -10%
5.625
Figure 8c Through Thickness Flaws
COURSE 7
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
BS7777 -10%
API -10%
COURSE 8
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
1.25 HYDRO (50000)
BS7777 1.31HYDRO
BS7777 -10%
API -10%
COURSE 9
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31HYDRO
API -10%
5.626
Figure 8d Through Thickness Flaws
COURSE 10
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31HYDRO
COURSE 11
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31HYDRO
COURSE 12
0
0.7071
0 0.8
STRESS
F
R
A
C
T
U
R
E
NO HYDRO
1.25 HYDRO
1.48 HYDRO
BS7777 1.31 HYDRO
5.627
The results for the BS 7777 (modified) tank show a minimal increase in risk from
fracture for the lower courses, this is due to the higher allowable stresses in the shell
plates. The upper courses show a decrease due to the greater minimum thickness that BS
7777 imposes, 12mm compared to 9.6mm with API 620.
The principal benefit of hydrotest is the reduction in residual stress which occurs.
The reduction is calculated by PD 6493 as shown previously. The loss of this benefit can
be compensated for by reducing the operating stress such that the sum of the primary and
residual stresses is the same as it would be for a full hydrotest. This would result in some
courses being increased in thickness. It is necessary to decide at what level of overstress
the benefits of residual stress reduction are not significant. For the purpose of this
assessment the level was chosen as 10% overstress. Any course which did not receive a
10% overstress during hydrotest, except the minimum thickness courses, was thickened
such that the operating stress was reduced by 10%. This figure was chosen nominally in
order to demonstrate the effects of this proposal.
From the results shown in the FADs for the surface flaw, it can be seen that this
increase in thickness has a small beneficial effect for all courses (more so with the higher
courses) with respect to the fracture failure of the tank. From the FADs for the through
thickness flaws the benefit is less beneficial than for the surface defect.
The only stresses considered in the assessments were the primary membrane and
secondary residual stresses. Other secondary stresses which exist, such as vertical
bending in the bottom course have not been considered. A more detailed analysis would
be required on an individual basis to determine actual fracture assessments and confirm
the trends indicated in this paper. A hydrostatic test will always be carried out on storage
tanks whether it is a 1.25 overload or greater in order to test the foundations. Therefore
the annular to shell weld will always receive a measure of reduction in residual stress as it
has done in the past.
The value used for the CTOD has been validated within the industry. Wide plate tests
have been carried out with failures occurring at values two to three times greater than the
CTOD values would indicate. (Ref 7)
As mentioned, the circumferential seams are under little or no primary stress and
therefore could survive defects much larger than indicated in this paper for the vertical
seams. The circumferential seams would not benefit from a hydrotest with respect to a
reduction in residual weld stresses and their integrity has been proven over the years. The
trend towards 100% radiography of these welds in larger tanks, with respect to guarding
against fracture, seems to be unnecessary
CONCLUSIONS
Tanks of 200,000 m
3
capacity in accordance with the API 620 Appendix Q rules
appear to be a safe and economic proposition.
Modifications to the BS 7777 rules lead to further savings. These modified rule
could be taken up by CEN TC 265 if the committee so choose.
5.628
The Engineering Critical Assessment using the rules of PD 6493 suggest that these
large tanks are safe structures. The two comfort factors suggested (increase hydrotest
height and decrease applied stress) give small increases in integrity when judged
according to the PD 6493 criteria.
There is a good case to be made for using the API 620 level of radiography in
association with some ultrasonic inspection for circumferential seams to take inspection
activities off the inner tank critical path.
REFERENCES CITED
1. API 620, Design and Construction of Large, Welded, Low-Pressure Storage Tanks,
Ninth Edition February 1996
Addendum 1 December 1996
The American Petroleum Institute, Washington, DC, USA.
2. BS 7777: 1993, Flat bottomed, vertical, cylindrical storage tanks for low
temperature service, BSI London, UK.
3. EEMUA 147, Recommendations for the design and construction of refrigerated
liquefied gas storage tanks, Publication No 147, The Engineering Equipment and
Materials Users Association, London, UK.
4. Are storage tank standards holding back LNG import projects? J B Denham FEng,
FIMechE, FweldI, LNG Journal, November/December 1996.
5. BSEN 1473: 1997, Installation and equipment for liquefied natural gas Design of
onshore installations. BSI, London, UK.
6. Development of Crack Arrest Criteria for Liquefied Gas Storage Tanks, J B Denham
FEng, FIMechE, FweldI, TWI/HSE/UMIST Seminar, September 1995, The
Welding Institute, Abingdon, UK.
7. Review of Data on 9% Ni steels and weldments, with reference to cryogenic storage
tanks, A M Wood and A A Willoughby, The Welding Institute, Abingdon,
Cambridge, UK.
8. Seismic design of the LNG tanks at the Marmara Ereglisi, Turkey, LNG Terminal,
LV Scorsone and J Hoptay, Pittsburgh Des Moines Inc., Proceedings of Gastech
1993.
9. The seismic analysis and design of large cryogenic storage tanks with reference to
two ongoing projects, G N Trott and R O Long, Whessoe LGA Gas Technology
Ltd.
10. PD 6493: 1991, Guidance on methods for assessing the acceptability of flaws in
fusion welded structures, BSI, London, UK.
11. The use of 9% nickel steel for LNG applicatio, W P Carter and J D Harrison, The
Welding Institute Conference on Welding Low Temperature Containment Plant,
November 1973.

Potrebbero piacerti anche