Coe EEE
DYNAMIC ANALYSIS AND DESIGN OF
LARGE STEEL LIQUID RESERVOIRS
ABSTRACT
‘The structural response of liquid storage reservoirs or
tanks with flexible bottoms supported on the ground is
fundamentally different than the response of other
‘structures, including rigid bottom tanks. The different
‘factors that cause the response, including the effects of
‘the tank aspect ratios, foundation interaction and soils
characteristics are reviewed, along withthe history of
the applicable analytical techniques. An analysis using
linear and dynamic methods ts made and conclusions
‘and recommendations are given for their design.
INTRODUCTION
Liquid storage reservoirs, which have been in
construction over centuries, have become a major topic
in today’s seismic engineering world, One example of
this is the rupture of a 5 million-gallon concrete reservoir
in Westminster, California, which caused nearly $27
nillion in damages, on September 21, 1998, Other water
districts have been using this asa case example during
the design or retrofit oftheir reservoi
‘The performance of water, petroleum and chemical tanks
and reservoirs in an earthquake is critical to society. The
water supply is essential for controlling fires that usually
‘occur during an earthquake and which can cause more
damage and loss of life than the event itself. Broken
petroleum tanks can lead to large uncontrollable fires,
while they, and chemical spills, can result in enormous
‘environmental damage.
319
Richard L. Hess
‘Structural Engineer/President
oss Engineering Ine, Los Alamitos, CA
George Pytlik
Civilstructural Engineer
Hess Engineering nc, Los Alamitos, CA
Aswin Rangaswamy
Associate Engineer
Hess Engineering Inc., Los Alamitos, CA
‘The most common cause of tank ruptures is the failure
due to longitudinal compression and radial tension that
can burst a vertical seam and spill the entire contents. In
steel tanks, this takes the form of bulging or an
“elephant’s foot” a the base before actual rupture. Other
forms of damage include roof damage due to sloshing,
failure ofthe support systems or tearing out ofthe anchor
chairs. Excessive movement of the tank can break
connecting pipes that do not have sulficient flexibility
built into them, resulting in los of liquid contents.
Finally, failure of the foundation due to liquefaction or
lateral movement of the supporting soil can result in oss
of support and rupture ofthe vessel.
Over the last several decades, engineers have been using
the standard linear static procedure recommended by the
American Water Works Association and American
Petroleum Institute's publications, Prior to that,
theoretical solutions had been developed for se
analysis but most tanks were designed for static and wind
loads with rule-of-thumb safety factors developed by
petroleum and water supply companies.
Lately, there have been more requests to do linear
dynamic analyses using site-specific response spectra
However, there has been no standard procedure
published forthe dynamic analysis of such large tanks
for the practicing engineer that would cover all
conditions of differing aspect ratios, tank anchorage and,
base rigidity or soil interaction.
‘The intent ofthis paper isto present the many variables
that may be involved, and, as an example, the results of
the analysis of a 5 million-gallon steel reservoir using the
linear static and response spectra analysis. This willre
Fe
vat}
iy
FIG. 2. Site Specific Horizontal Response Spectra (0.58g)
Source: Soil Report by “Geolabs - Westlake Village", dated Sep. 4, 1998 (Westlake Village, CA),
220describe the various methods for obtaining the period of
the fank through analytical and computer programs.
THE BASIS FOR THE LATERAL TANK
ANALYSIS
Cylindrical tanks containing liquids, with flexible
bottoms resting directly on the ground or base mat,
constitute a unique category for structural design. ‘This is
because the entire weight ofthe tank contents contributes
to the lateral seismic force while only a small percentage
of that weight helps to resist overturning.
Further complicating the analysis isthe fact thatthe
lateral force consists of two components: impulsive
forces and convective forces. The fist isthe type of
force that structural engineers are familiar with, which
relates to the inertia ofa portion of the liquid along the
walls and the bottom which moves in unison wih the
tank as a rigidly atfached mass, The second (convective)
force is caused by the movement of the remaining fluid
inside the tank, which is the subject of fluid dyna
analysis.
‘The relative importance of these two forces depends on
the physical configuration of the tank. Because the
lateral component of the seismic forces is primary, the
larger width-to-height ratios allow the convective forces
‘to come more into play. Whereas when the height is
‘more than the width (H/D > 1), the impulsive forces are
predominant.
‘The natural period for these two components is also quite
different. When a dynamic response spectrum is used in
the analysis, the period for the impulsive force is
Spills a fraction of a second while the convective
is several seconds fong. The way that these
Uifferent components are handled varies with the analysis
‘methodology that is used.
‘The defining consideration in the analysis of the tank is
whether the overturning moment is large enough to result
in significant uplift of one side of the tank wall. If this
‘were allowed to occur, the longitudinal compression and
tangential tension on the other side would become
excessive and cause buckling and probable rupture.
Uplift ofthe tank shel is resisted by the weight of the
shell and supported roof plus a band of liquid adjacent ro
it, The width of this band of liquid depends on the
stiffness (or thickness) of the bottom plate inside the
321
shell, which is called the annular ring, ‘The designer can
thicken this ring but there are imitations because it
‘cannot be thicker than the shell. If this isnot sufficient,
additional restraint in the form of anchors must be
provided.
‘There is an intermediate zone in which some uplift will
occur but anchors are not required by AWWA D100 (16)
or API 650 (17). ‘This is because it has been found in the
past thatthe uplift was too small to create failure of these
tanks.
‘These two codes are based primarily on past experiences
of failure of large storage vessels. ‘The seminal work was
published by Wozniak and Mitchell in 1978 (4). Mr.
‘Wozniak had been with Chicago Bridge and Iron
Company, a major tank fabricator, and Mr. Mitchell, a
‘member of SEAOC, with Standard Oil Company of
California, They had access to a large database,
including the effects of the great Alaska earthquake of
1964, and were able to take the theoretical data that was
available and put it into a set of equations that can be
used by practicing engineers.
‘This appears to have been the basis of both the API 650
Appendix E and the seismic provisions of AWWA D100,
‘The latest edition of the later has incorporated
provisions that can employ response spectra in the
analysis. A 1998 edition of API 650 has been published,
‘but was not examined atthe time this paper was written
When anchorage for uplif is required, it can be provided
by piles or rock anchors. One approach, that used rock
anchors wit steel bolts that were designed to yield, was
described in a paper by John Shipp (12) atthe 1993
‘SEAOC convention and can be found in the Proceedings.
This approach may be especially useful in a retrofit
situation where the annular ring is not thick enough to
climinate uplift and the excess force can be taken by the
new anchors that yield at a predetermined level. A more
advanced study ofthis type of retrofit anchorage is
contained in reference (19).
Resistance to uplift ofthe shell by the bottom annular
ring cannot begin until an initial vertical displacement
‘occurs which permits the bottom to resist the
displacement in tension. Some installation specifications
call fora pad to be placed under the edge of the bottom
to give it an initial displacement, under no load, with the
bottom. This is called pre-uplift and is described in
reference (8).