You are on page 1of 17

Seismisk analyse / dimensjonering

av beholdere/tank
Anton Gjrven
Thanh Ngan Nguyen

Background
The purpose with this presentation is
to demonstrate different calculation methods and design principles for
seismic response of "full containment" (stand-alone steel inner tank,
separated from outer concrete cylinder) LNG (liquefied natural gas) tanks
to show how an earthquake can impact the design of the steel inner tank.
The basic principles of anchoring/no anchoring of the steel inner tank is a
significant factor of the costs of an LNG tank.

NEED2012

Background
Relevant projects
Risavika: H = 20 m (Ht = 21 m), R = 22.5 m, H/R = 0.89
Lysekil: H = 37.5 m (Ht = 38.2 m), R = 16 m, H/R = 2.34
Lysekil: High H/R-ratio is a challenge when considering safe shutdown
earthquake (SSE - return period 4975 years. Operating basis earthquake
(OBE) - return period 475 years)

NEED2012

Background
Response calculations and design are normally based on hand
calculations using standards
Analyses, both explicit and implicit, have been executed to compare
and validate the hand calculations
Parameters of interest:
Base shear and overturning moment (foundation, stresses in bottom
insulation layers)
Compressive stress in tank wall ("elephant foot" buckling, EC8-4 A.10)
Uplift and anchorage of tank

Typical "elephant foot buckling"


NEED2012

Foto: Prof. J.M. Rotter, The University of Edinburgh

Basic design assumptions


General: Eurocode 2 (Concrete) and Eurocode 3 (Steel)
Tanks: Eurocode 1, Part 4 (Action on tanks) and Eurocode 3, Part 4-2
(Design of steel tanks)
Eurocode 8 (Earthquake), Part 1 (General)
Chapter 3: Ground conditions and seismic excitation

NEED2012

Ref.: NS-EN 1998-1:2004/NA:2008 Figure NA.3(903)

Ref.: NS-EN 1998-1:2004/NA:200


Figure NA.3(901)

Eurocode 8, Part 4 (Silos, tanks and pipelines)


Chapter 2: General
Chapter 4: Specific rules for tanks
Annex A: Seismic analysis procedures for tanks

Basic design assumptions


EN 1473 (Installation and equipment for LNG)
Structural parts vital for the plant safety shall remain operational after both
operating basis earthquake (OBE) and safe shutdown earthquake (SSE)
EN 14620 (Design and manifacture of steel tanks for storage of LNG)
Part 1 (General)

7.1.4 Earthquake design: "For full containment tanks, the primary liquid container shall be
designed to contain the liquid during an OBE and SSE action."
7.3.2.2.13: OBE earthquake
7.3.3.3: SSE earthquake
Annex C: Seismic analysis

Part 2 (Metallic components)

NEED2012

5.1.2.2: Requirements to allowable tensile stress in tank anchorage for OBE and SSE (NB:
Allowable stress theory, not limit state theory)
5.8.1: Other requirements to tank anchorage

Part 3 (Concrete components)


Part 4 (Insulation components)

6.3.2.2.1: Overall safety factor for brittle materials (insulation) for OBE and SSE (NB: Allowable
stress theory)
Annex C: Tank bottom insulation - Limit state theory

Part 5 (Testing, drying, purging and cool-down)

Design of LNG tanks

Foto: Norconsult. Steel roof


under construction.

NEED2012

Foto: Norconsult. An LNG tank is a complex structure. Here is the outer concrete
wall from one of our projects - picture is taken from below and upwards.

Design of LNG tanks


The steel inner tank is the part where the seismic conditions can have a
significant influence to basic design, for example the anchoring of the
inner tank.
The "slenderness" of the tank or the ratio H/R combined with the
ground condition govern the anchoring system to be used.

NEED2012

Explicit analysis - Abaqus

NEED2012

In an explicit analysis, the earthquake excitation is defined with a time


history
The model consists of the container a cylinder (open top) made by
steel filled with LNG
Two cases are studied without and with anchors (smeared
representation)
For each case, one earthquake definition based on Norwegian
Standards has been used
The FE model does not consider the outer concrete cylinder since the
obtained results (max. displacement during earthquake) indicate that
the interaction forces will be negligible
Model definition:
Steel tank: Shell elements
LNG: Continuum elements. Material defined by wave speed and dynamic
viscosity (Equation of state (EOS) material model - only available in
Abaqus/Explicit)

Explicit analysis - Abaqus


FE model, interactions:
Steel tank and LNG
Steel tank and bottom layers
(Foamglas, sand)
Bottom layers and rock
(analytical rigid surface)

NEED2012

Explicit analysis - Abaqus


Gravity load (LNG) is initially in equilibrium with a hydrostatic pressure
The initial stress state is obtained with a dynamic (explicit) approach
using time integration
This is done by increasing the gravity load with a smooth amplitude
curve in 10 seconds, thereafter continued 10 seconds further without
any load change in order to decrease oscillations of the unbalanced
solution

Stress state in LNG and tank wall is checked


NEED2012

Explicit analysis - Abaqus


The earthquake excitation is applied at the reference node for the
rock as prescribed acceleration in the 1- and 3-directions
The base acceleration has been multiplied by factors 1 respectively 0.3
for these directions and the earthquake is analyzed during 10 seconds

NEED2012

Explicit analysis - Abaqus


Without anchors

NEED2012

Explicit analysis - Abaqus


With anchors

NEED2012

Explicit analysis - Abaqus


Vertical section force lower part of wall (positive indicates risk of uplift)

without anchors

NEED2012

with anchors (higher


stress on
compression side,
reduced risk of uplift
on tension side)

Explicit analysis - Abaqus


Envelope of minimum contact pressure (negative indicates risk of
uplift)

Without anchors

NEED2012

With anchors

Explicit analysis - Abaqus


Envelope of maximum contact pressure

Without anchors

NEED2012

With anchors (higher


pressure!)

Explicit analysis - Abaqus


History plot of uplift at the corner of the tank

Vertical section force in anchors

NEED2012

EC8-4 A.9: Effect of uplift on stress in the wall (unachored)

EN 1998-4:2006 (E) Figure A.11:

NEED2012

EN 1998-4:2006 (E) A.9.2:

Explicit analysis - Abaqus


Rigid (dashed line) vs. elastic (solid line) bed, history plot of vertical
section force

Without anchors
(small differences)

NEED2012

With anchors (elastic


bed gives generally
higher compression
forces than rigid bed)

10

Explicit analysis - Abaqus


SUMMARY AND CONCLUSIONS

NEED2012

Material model for LNG


EOS material is used
Elastic material is also possible - gives more balanced initial state
Elastic material shows less damping during seismic excitation
Results when using anchors are very similar to the case without
anchors
Max. vertical displacement for the case without anchors is slightly
higher than with anchors
However, the uplift seems to have a rather small impact on the vertical
stresses in the tank wall
The case with anchors has actually increased section force compared to
the case without anchors
An explicit analysis is probably a very good approach to study the
dynamic behaviour for an LNG tank excited by seismic action

Simplified calcuation of overturing moment

= 0.8 40
= 1.0 0.8 0.5

2.5

2.5

1.0
= 0.667 2
2
1.5

OBE (475 years):


= = 1.5 30 000 3 490
NEED2012

0.667 2 18.75 = 276


3

SSE (4 975 years):


= 1.0, 3.0 = 276

3.0
= 552
1.5

11

Hand calculations - Malhotra


EC8-4 A.3.2.2
Simplified method for fixed base tanks
The response is splitted into impulsive and convective part, impulsive is
dominating for high tanks

EN 1998-4:2006 (E) Table A.2:

NEED2012

Hand calculations - Malhotra


Acceleration from elastic response spectrum, EC8-1
Return period other than TNCR = 475 years is taken into account with the
importance factor I (evaluated from EC8-1 2.1(4) Note)
Impulsive damping = 2 %, convective damping = 0.5 %
Impulsive and convective period EC8-4 Eq. (A.35) and (A.36)

NEED2012

Base shear and moment EC8-4 Eq. (A.37) and (A.38)

12

Hand calculations - Malhotra


Compressive stress =
Tension force =

Number of anchors = 2

Uplift?
Unachored EC8-4 A.9: Effect of uplift on stress in the wall

OBE

SSE

NEED2012

Ref.: EN 1998-4:2006 (E) Figure A.11

Response spectra for hand calculations

Malhotra/EC8-4:

" = 1.0" (elastic response spectrum)


=2%
Period 0.4 ?

NEED2012

Simplified calculations:
= 1.5
=5%
Max. value of spectrum

13

Partial factors for LNG tank design

OBE (475 yrs)

Limit state

Load
factor

Material
factor

Ordinary ULS

Yes

Yes

No (1.0)

No (1.0)

No (1.0)

Yes?

SSE (4975 yrs) Accidental ULS

NEED2012

Implicit analysis - Abaqus


EC8-4 A.2

Ref.: EN 1998-4:2006 (E) Figure A.1

NEED2012

Why not implicit analysis with response spectrum step?


Material model for the fluid?
Interaction between the fluid and the flexible steel wall?
Hydrodynamic pressure: Motion of the fluid due to seismic excitation is
preserved as "snapshot" of max. pressure ("pushover" analysis)

14

Implicit analysis - Abaqus


The rigid impulsive
(hydrodynamic) pressure is
applied on the tank wall for the
acceleration from the elastic
response spectrum
The LNG fluid is also applied as a
hydrostatic pressure
A contact algorithm is applied
between the tank bottom and an
analytical rigid surface, allowing
for separation
Base shear and moment
corresponds well with Malhotra's
simplified method
NEED2012

Implicit analysis - Abaqus


Stresses in tank wall and anchors, in addition to uplift may be studied

NEED2012

Results obtained from implicit analysis are in good agreement with


hand calculations
SSE: Uplift and stress in wall OK with anchors, not OK without. Unanchored
case: Highly increased stresses due to extensive deformations
OBE: OK with and without anchors. Unanchored case: Increased
compressive stress is moderate
An implicit analysis is more conservative than an explicit analysis. It is
in good agreement with hand calculations and may not give any new
information of the behaviour that can be found by simplified methods.

15

Summary
Several types of calculations/analyses - benefits and limitations
Simplified hand calculations
Simplified procedure - Malhotra
Implicit analysis
Explicit analysis
Which results are trustworthy?
Unachored tanks: Increased compressive stress when uplifted
(Eurocode, implicit model (moderate!))
Anchored tanks: Increased compressive stress due to tension in
anchors (explicit model)

NEED2012

Summary (continued)

Overturning moment
[MNm]
Hand calculations
(Malhotra)

Base shear [MN10]

Abaqus implicit
Tension in anchoring
[kN]

NEED2012

Abaqus explicit

Compression in tank
wall (with anchors)
[MPa 10]
0

500

1000

16

Conclusions
Calculation method may govern the decisions regarding the necessity
of anchoring the tank
Advanced FE methods (explicit analyses) tend to give reduced values of
the governing parameters (hand calculations are more conservative)
The complexity of explicit analyses is very high and need a lot of
engineering time

NEED2012

17

You might also like