Professional Documents
Culture Documents
of Storage Tanks
D. Whittaker & D. Saunders
Beca International Ltd, New Zealand
2008 NZSEE
Conference
Paper Number 04
multi-component spring/mass analogy. The analogy allows the complex dynamic behaviour of a tank
and its contents to be considered as simple modes of response including a short period impulsive
mode, with a period of around 0.5 seconds or less, and a longer period convective (sloshing) mode
with periods up to several seconds. For most tanks the impulsive mode dominates the loading on the
tank wall and the convective mode dominates the wave height and required freeboard. The mechanical
spring/mass analogy for the assumed modes of vibration is shown in Figure 1.
Figure 1. Mechanical spring/mass analogy for impulsive and convective modes of vibration.
Damping levels for tanks have generally been assumed to be of the order of 2-5% for the impulsive
modes and 0.5% for the convective modes. The additional effects of radiation damping (i.e. energy
lost into the foundation) in reducing earthquake response can be considerable, providing equivalent
viscous damping levels of as much as 20-30%.
3 1986 NZSEE PUBLICATION
The 1986 NZSEE recommendations provided a detailed approach to seismic analysis and design for a
range of types of storage tanks. Practical design examples were included and detailed methods for
analysing the seismic behaviour of anchored or unanchored tanks were included.
The seismic loadings included in the document were based on early seismic hazard assessments for
New Zealand available at that time, however these are now considered to be well out of date. The
design method for steel tanks is an ultimate limit state approach, but it assumed no yielding or ductility
of tank elements is permitted.
Experience with use of this document in designing steel storage tanks has indicated that the
requirements were often significantly more conservative compared with design to the API 650
approach.
4 RECENT NEW ZEALAND SEISMIC DESIGN STANDARDS NZS 4203 AND NZS 1170
Since publication of the 1986 document there have been two major revisions to the loading standard
for buildings NZS 4203. In 1992 seismic loads in NZS 4203 were revised based on hazardbased
seismicity maps of New Zealand. Since 2004 a joint Australian and New Zealand Standard AS/NZS
1170 Design Loads for Buildings, has been introduced and is a means of compliance with the
performance-based NZ Building Code. The Standard now supersedes NZS 4203. NZS 1170.5 is the
section of the Standard that deals with seismic loads.
The design spectra in NZS 1170 are based on a comprehensive seismic hazard analysis using New
Zealand seismicity models and earthquake shaking attenuation criteria. NZS 1170 is a limit-state code
with the ultimate limit state based on strength design and serviceability limit state dealing with
damage and displacement control. Although NZS 1170 is not directly applicable to tanks, it has been
used as the basis for setting seismic loads for design of storage tanks, with a correction factor being
applied to better account for ductility and damping levels in tanks.
5 API 650
API 650 Appendix E has been widely used around the world for seismic design of steel storage tanks.
It covers the United States and a wide range of other locations around the world. A simplified analysis
of seismic effects including impulsive and convective modes and a working stress approach is used in
the code for checking stress limits in the tank shell.
Until recently the design loads for the most seismic region of the US were based on an impulsive
mode acceleration of up to a maximum of 0.24 g in the most severe seismic zone in US, together with
a simple overturning stability limit for unanchored tanks.
The 2006 edition of API 650 Appendix E is significantly changed, and is based on the following:
Seismic hazard maps and design spectra for the whole of the US (short period spectral
coefficients of up to 2.0 g are specified for earthquake shaking with a return period of 2,500
years).
The methodology allows use for locations outside the US, based on seismic coefficients
obtained probabilistic design spectra or peak ground acceleration values.
Force reduction factors are used to determine design seismic loads (e.g. 3.5 for unanchored
tanks).
Stress limits are working stresses with maximum stresses of up to 0.9 times yield stress.
Seismic load for tanks is in accordance with NZS 1170.5:2004, the national loading standard
for buildings, including probabilistic hazard based seismic design spectra.
Adjustments are made for the effects of damping levels and ductility levels appropriate for
tanks.
Some limited ductility is permitted and is reflected in the force reduction of design loads.
The ultimate limit state design methodology is in alignment with other NZ structural design standards
and can also be interfaced with other international codes and design guides.
The horizontal seismic force acting on a tank, associated with a particular mode of response, is
calculated from the expression in Equations 1 to 3.
Vi = Cd (Ti ) mi g
(1)
(2)
(3)
and
Vi
mi
Cd(Ti)
C(Ti)
= ordinate of the elastic site hazard spectrum for horizontal loading for the site
subsoil type, and the relevant mode. Obtained from NZS 1170.5 for tank
importance level.
kf(,i ) = correction factor for NZS 1170.5 elastic site hazard spectrum to account for
ductility and level of damping, refer Table 2.
Sp
Ch (Ti)
= spectral shape factor for site subsoil type and relevant mode, from NZS 1170.5.
Ru
= return period factor for the ultimate limit state, from AS/NZS 1170.5 for tank
importance level.
= displacement ductility factor for horizontal impulsive modes. Taken as 1.0 for the
convective and vertical modes.
Ti
This approach replaces the design loading section of the 1986 NZSEE document with loads from the
NZS 1170:2004 design spectra, with an additional correction factor for ductility and damping
appropriate for tanks.
Displacement ductility factors of either 1.25 or 2.0 are permitted for seismic design of tanks, as shown
in Table 1. Although not strictly ductile behaviour, certain modes of inelastic behaviour, such as base
uplift and elephants foot buckling of steel tanks, are assumed to be an equivalent ductile response. The
use of ductility is a departure from the 1986 Red Book approach and results in lower design loads.
Table 1. Displacement ductility factors.
Timber tanks
An example chart giving recommended levels of damping for the impulsive modes of steel tanks is
given in Figure 2. The figure gives the total damping, made up of the tank-liquid system fixed-based
damping plus the foundation radiation damping. Damping for the convective (sloshing) mode is
assumed to be 0.5%.
The NZS 1170.5 design acceleration spectrum coefficient for an elastic 5% damped system is
modified by a spectrum correction factor kf(,i ), shown in Figure 3, to account for assumed ductility
and equivalent viscous damping levels. These acceleration spectrum coefficients then generally have a
similar basis as those in NZS 1170.5:2004.
0.5
30
Damping Factor, %
25
1.0
20
0.3
15
1.5
10
2.0
5
H/R = 3.0
0
200
250
300
350
400
450
500
550
600
Figure 2. Example damping for horizontal impulsive mode. Steel tanks t/R = 0.002.
1.8
1.6
1.4
ductility = 1.0
Correction Factor kf
ductility = 1.25
1.2
ductility = 2.0
1.0
0.8
0.6
0.4
0.2
0.0
0%
5%
10%
15%
20%
25%
30%
Damping
The seismic load derivation is based on the same approach used in NZS 1170 using Importance Levels
to reflect the significance of the facility and appropriate Return Period Factors. The Importance Level
is derived from considering the consequences of failure, based on separate consideration of life safety,
environmental risk, community significance and adjacent property value. Recommended
classifications of risk for each of these aspects can be found in the document. The Importance Level
used in determining the design load is then based on the worst of these considerations. Proposed
Importance Levels and Return Period Factors are given in Table 2.
Table 2. Importance Level and Return Period Factors to be used in application of AS/NZS 1170.
Consequence of
Failure
Importance Level
Slight
100
0.5
Moderate
500
1.0
Serious
1000
1.3
Extreme
2500
1.8
NZS 1170 incorporates a Structural Performance Factor Sp in the derivation of design seismic loads
for buildings, with values of between 0.7 and 1.0. The use of Sp = 1.0 is specified for tanks in
recognition of their limited ductility capability.
7 COMPARITIVE EXAMPLES
The seismic loads specified by the NZSEE 2008 method are compared with the requirements of the
1986 NZSEE and API 650 (2003 and 2006) methods for several example steel tanks. Comparison is
also made with seismic loads required by the existing water retaining structures code NZS 3106:1986.
Details of the example tanks are summarised in Table 3.
Table 3. Details of example tanks for comparison.
Tank
Material
Product
Stored
Anchorage
Design
Return
Period
(yrs)
Ground
Type
Case
Location
Wellington
Steel
Diesel
Unanchored
500
Wellington
Steel
Petrol
Unanchored
2,500
Auckland
Steel
Diesel
Unanchored
500
Auckland
Steel
Petrol
Unanchored
2500
Deep soil
Wellington
Steel
Diesel
Anchored
500
Deep soil
Wellington
Steel
Petrol
Anchored
2500
Deep soil
Wellington
Concrete
Water
Unanchored
1000
Rock
Auckland
Concrete
Water
Unanchored
1000
Rock
Shallow soil
Deep soil
Shallow soil
Figure 4 shows the impulsive mode base shear coefficients required by each design method for the
example tanks. The seismic design of tanks is generally governed by the impulsive mode base shear,
so comparison of the impulsive coefficients given by the various methods gives a good indication of
the overall seismic loads.
The API 650 methods are based on working stresses, so the coefficients shown in the figure include
scaling factors (2.0 for the 2003 method and 1.1 for the 2006 method) for the purposes of comparing it
with the NZSEE ultimate limit state approach. For reference, Wellington has similar seismicity levels
to the most active areas in California.
The 2008 NZSEE method generally gives lower impulsive mode base shear coefficients than the 1986
method. The method gives seismic coefficients and wall thickness of a similar order to the API 650
method, although greater in some cases and less in others.
The 2006 revision of API 650 Appendix E alters the seismic loads for many tanks compared to the
previous 2003 document. The geometry limits of tanks that are required to be anchored has been
tightened resulting in a greater number of tanks than before. The NZSEE 2008 document allows more
tanks to remain unanchored than API 650 permits.
Based on these comparisons the NZSEE 2008 method appears reasonable in relation to the other
methods available.
8 SUMMARY
The 2008 NZSEE publication entitled Seismic Design of Storage Tanks is summarised. The document
updates the original 1986 document to be consistent with the current New Zealand structural loading
and materials standards. Several example tanks have been considered to compare the seismic design
loads required against other available methods.
ACKNOWLEDGEMENTS
The authors wish to acknowledge the financial assistance of the following industry sponsor
companies: New Zealand Earthquake Engineering Society, Mobil Oil New Zealand Ltd, Shell Oil
New Zealand Ltd, Caltex Oil New Zealand Ltd, BP Oil New Zealand Ltd, NZ Refining Ltd., Shell
Todd Oil Services Ltd, New Zealand Insurance Ltd.
The NZSEE members that prepared the 2008 document were: David Whittaker (Convenor), Dean
Saunders, Rob Jury, John Wood, Barry Davidson, Graeme McVerry and John Mason.
2.0
Case 1
type: unanchored steel petrol tank
location: Wellington
risk: Moderate
1.5
1.0
0.74
0.5
0.48
0.47
0.36
0.0
NZNSEE
1986
API 650
2003
NZNSEE
1986
API 650
2003
API 650
2006 &
NZS1170
2.0
1.5
0.69
0.60
0.5
NZSEE
2008
Case 3
type: unanchored steel petrol tank
location: Auckland
risk: Moderate
1.0
0.40
0.5
0.24
0.15
Case 4
type: unanchored steel petrol tank
location: Auckland
risk: Extreme
1.5
0.97
1.0
0.5
0.35
0.30
0.22
0.12
0.0
0.0
NZSEE
2008
NZNSEE
1986
API 650
2003
NZSEE
2008
API 650
2006 &
NZS1170
NZNSEE
1986
API 650
2003
API 650
2006 &
NZS1170
4.0
4.0
Case 5
type: anchored steel petrol tank
location: Wellington
risk: Moderate
3.0
2.0
1.40
0.62
0.48
0.40
API 650
2003
API 650
2006 &
NZS1170
Case 6
type: anchored steel petrol tank
location: Wellington
risk: Extreme
3.36
1.08
1.0
API 650
2006 &
NZS1170
2.0
0.0
3.0
2.0
1.12
1.0
0.60
0.60
API 650
2003
API 650
2006 &
NZS1170
0.0
NZSEE
2008
NZNSEE
1986
NZSEE
2008
NZNSEE
1986
3.0
3.0
Case 7
type: prestressed concrete water reservoir
location: Wellington
risk: Intended Facility (Serious)
2.0
1.95
Case 8
type: prestressed concrete water reservoir
location: Auckland
risk: Intended Facility (Serious)
2.5
2.5
1.5
0.0
NZSEE
2008
1.0
Case 2
type: unanchored steel petrol tank
location: Wellington
risk: Extreme
1.76
2.0
1.73
1.5
1.0
0.66
2.0
1.5
1.0
0.86
0.5
0.5
0.35
0.23
0.0
0.0
NZSEE 2008
NZNSEE 1986
NZSEE 2008
NZS 3106
NZNSEE 1986
NZS 3106
Figure 4. Comparison of design impulsive mode bas shear coefficients for example tanks to various design
standards.
REFERENCES:
American Petroleum Institute. 2003. API 650 Welded Steel Tanks for Oil Storage.
American Petroleum Institute. 2006. API 650 Welded Steel Tanks for Oil Storage.
Environmental Risk Management Authority, Hazardous Substances (Dangerous Goods and Scheduled Toxic
Substances) Transfer Notice, (Amended) 08 August 2006.
Ministry of Works and Development, 1981. Recommendations for Seismic Design of Petrochemcial Plants.
New Zealand National Society for Earthquake Engineering, 1986. Recommendations for Seismic Design of
Storage Tanks.
New Zealand Society for Earthquake Engineering. 2008. Seismic Design of Storage Tanks (In Preparation).
Standards New Zealand. NZS 1170.5, 2004, Structural Design Actions, Part 5 Earthquake Actions.
Standards New Zealand. NZS 3106, 1986, Code of Practice for Concrete Structures for the Storage of Liquids.
Whittaker, D. and Jury, R.D. 2000. Seismic Design Loads for Storage Tanks. Proceedings 12th World
Conference on Earthquake Engineering, Paper Number 2376.