BALLOT AGENDA ITEM 653 - 150: 2

nd
ballot January, 2008

Background and Rationale
The current API 653 Appendix B methodology for evaluating the FFS of a storage
tank for outofplane settlement does not differentiate between types of tank roof
construction; covers only tanks whose settlement includes a plane of rigid tilt that
follows a cosine (sine) wave; permits only one goodness of fit (R
2
) methodology;
permits data to be ignored; and penalizes the use of more closely spaced of
settlement readings.

A new methodology is proposed that addresses these issues. It has been developed
based on a parametric FEA research study, documented in separate reports and
briefs. The proposed methodology can be used on API 650 carbon steel and
stainless steel tanks, but is not applicable to aluminum tanks. A separate document
includes several solved examples.
Economic Impact
The limitations of the current methodology in API 653 Appendix B, particularly the
dependence of permissible outofplane settlement on a settlement measurement
spacing limit, may result in tanks requiring expensive shell releveling or more
rigorous analyses. Additionally there are tanks that can not be evaluated with current
methods. The proposed FFS approach should reduce the number of tanks needing
expensive repairs or more rigorous analyses.
Proposed Revisions
Proposed revisions to API 653 Section 12 and Appendix B are given. Current API
653 text that has been relocated is given in italics, new text is bold, and deletions
are indicated with a strikethrough.


SECTION 12 - EXAMINATION AND TESTING
12.5.1.2
Tank settlement shall initially be surveyed with the tank empty using the an even
number of bottom plate projection elevation measurement points, N, uniformly
distributed around the circumference., The minimum number of elevation
measurement points shall be as indicated by the following formula:

D
N=
10


where:
N= minimum required number of settlement measurement points, but no
less than eight. All fractional values shall be rounded to the next higher
even whole number. The maximum spacing between settlement
measurement points shall be 32 ft
D= tank diameter, in ft.

BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008



APPENDIX B EVALUATION OF TANK BOTTOM SETTLEMENT
B.1 UNCHANGED
B.2.1 SETTLEMENT ELEVATION MEASUREMENTS
Measurements for outofplane of tank settlement must be taken carefully. These
measurements should be performed by personnel experienced in the types
of measurement procedures being preformed, and using equipment
designed for the intended purpose. Tank settlement measurements should
be taken with instruments capable of sufficient accuracy to be able to
distinguish settlement differences.

The principle types of tank settlement consist of settlements that relate to the tank
shell and bottom plate. These settlements can be recorded by taking elevation
measurements around the tank circumference and across the tank diameter. Figures
B1 and B2 show minimum recommended locations on a tank shell and bottom
plate for settlement measurements. Data obtained from such measurements should
be used to evaluate the tank structure. Additional settlement readings may be
required to better define local bottom depressions or edge settlements, to refine
shell settlement measurements in areas suspected to have local outof
plane settlements, or to otherwise improve bottom or shell settlement
evaluation. Settlement measurement locations should be re-used in any
future settlement surveys and evaluations.

In cases of distortion or corrosion of the tank bottom extending beyond the shell,
shell settlement measurements taken near lap welds in the tank bottom can result
in significant errors in measured elevation. Repaired or replaced bottom plates, or
new slottedin bottoms may not have been installed parallel to the bottom shell
course. In some cases, more consistent and accurate results may be obtained by
surveying the elevation of the weld between the first and second courses.

Measure bottom and edge settlement carefully, taking into account the following:
that measurements taken when the bottom is not in contact with the soil or
foundation under the tank can overestimate or underestimate edge or bottom
settlement significantly. If the measured settlement is near the maximum allowable
settlement, consider repeating the measurement with the bottom forced down to the
soil, e.g., standing on it, or take an additional set of measurements in the same
area, where the bottom is in firm contact with the soil.
B.2.2 THROUGH B.2.2.2 UNCHANGED
B.2.2.3
Due to the fact that a tank is a rather flexible structure, chances are great that the
tank will may settle in a nonplanar configuration, inducing additional stresses in the
tank shell. The outofplane settlements of the shell can at the bottom edge lead
to outofroundness a lack of circularity at the top of the shell tank, and
depending on in the case of a floating roof tank, the extent of the induced outof
roundness ovality may impede the proper functioning of the floating roof in such a
way that re-leveling is required. The out-of-roundness caused by settlement
may also affect internal roof support structures such as columns, rafters,
BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008

and girders. Also, such settlements may cause flat spots to develop in the tank
shell. This type of settlement could affect tank nozzles that have piping attached to
them.
B.2.2.4
While uniform settlement and rigid body tilt of a tank may cause problems as
described in B.2.2.1 and B.2.2.2 above, the outofplane settlement is the
important component to determine and evaluate in order to ensure the structural
integrity of the shell and bottom. Based on this principle, a common approach is to
determine the magnitudes of the uniform settlement and rigid body tilt component
(if any a rigid tilt plane exists or can be identified) for each data point on the
tank periphery. Once this is carried out, the If a plane of rigid tilt can be
distinguished, it becomes is then important as a datum from which to measure
the magnitudes of the outofplane settlements. When the out-of-plane
settlement pattern of a tank has an easily distinguishable plane of rigid tilt,
the methodology in this paragraph can be used to evaluate the acceptability
of the tanks out-of-plane settlement. If a rigid tilt plane can not be readily
determined, the methodology in B.2.2.5 can be used to evaluate the
acceptability of the tanks out-of-plane settlement.

A graphical representation illustrating the effect of tank shell settlement components
with a rigid tilt plane welldefined by a cosine curve fit is shown in Figure B3.
The construction of this settlement plot has been developed in accordance with the
following:

a. The actual edge settlement (in most cases an irregular curve) is plotted using
points around the tank circumference as the abscissa.
b. The vertical distance between the abscissa and the lowest point on this curve
(point 22) is the minimum settlement, and it is called the uniform settlement
component. A line through this point, parallel to the abscissa, provides a new base
or datum line for settlement measurements called adjusted settlements.
c. The plane of rigid tilt settlement, if well-defined, is represented by the optimum
cosine curve between the maximum and minimum values of actual edge settlements.
Several methods exist for determining the optimum cosine curve. The least accurate
method is by free hand drawing techniques, a kind of trial and error procedure to fit
the best cosine curve through the data. A better method is to use the mathematical
and graphical capabilities of a computer.
d. The vertical distances between the irregular curve and the cosine curve represent
the magnitudes of the outofplane settlements (U
i
at data point i).
e. The most A commonly used and accepted method is to use a computer to solve
for constants a, b, and c, to find the optimum cosine curve of the form:

( ) Elev =a+bcos +c
pred


Where Elev
pred
is the elevation predicted by the cosine curve at angle theta. A
typical starting point for a computer bestfit cosine curve is a leastsquare fit where
a, b, and c are chosen to minimize the sum of the square of differences between the
measured and predicted elevations. The optimum cosine curve may be is only
considered valid (i.e., accurately fits the measured data) if the value
2
R is greater
than or equal to 0.9.

BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008

( )
2
S - SSE
yy
R =
S
yy


where:
S
yy
= the sum of the squares of the differences between average measured
elevation and the measured elevations,
SSE=
Sum of the square of the differences between the measured and
predicted elevations.

Linear least square fitting and the
2
R goodness of fit parameter are basic
statistical tools. The use of a more rigorous statistical method to determine
the optimum cosine curve, such as nonlinear or iterative procedures, may
be used by those experienced in their use.

Obtaining a statistically valid cosine curve may require taking more measurements
than the minimums shown in Figure B1. In many cases, the outofplane
settlement may be concentrated in one or more areas. In such cases, and the
leastsquares fit approach may under predicts the local outofplane settlement
and is not conservative. In these cases,
2
R will typically be less than 0.9, and more
rigorous curvefitting procedures should be considered. Alternatively, the
settlement may not indicate a welldefined rigid tilt plane and the
procedure in B.2.2.5 should be considered. Appropriately selecting a, b, and c
will result in very little difference between predicted and measured elevation at all
but one or two measurement points, and
2
R will be greater than 0.9. One method of
selecting a, b, and c in these cases is to ignore one or two points that do not appear
to fit the initially calculated cosine curve, and recalculate the optimum cosine curve.
The remaining points will show a good estimate of the true outofplane settlement
at the worst point. Figure B4 shows an example where one point at 135 degrees is
well off the initial curve,
2
R is 0.87, and S is just less than the maximum allowed. By
ignoring the point at 135 degrees, and recalculating the optimum cosine curve,
2
R is
increased to 0.98 (very good) and the outofplane settlement at the point at 135
degrees is just above the maximum allowed.
f. The vertical distances between the irregular curve and the optimum curve
represent the magnitudes of the outofplane settlements (U
i
at data point i). S
i
is
the outofplane deflection at point i. Refer to Figure B3.

Note: When using determining the optimum cosine curve described in B.2.2.4e,
taking additional measurements around the shell will result in a more accurate cosine
curve fit. However, using all of the measurement points in the equation shown in
B.3.2.1 will result in very small allowable out-of-plane settlements,
max
S , since the
arc length L between measurement points is small. It is acceptable to use as many all
measurement points as needed to develop the optimum cosine curve, but only use a
subset of these points spaced no further than approximately 30 32 ft (8 minimum)
when calculating
i
S and
max
S . The points used must include the points furthest from
the optimum cosine curve. For example, if 8 points are required, but 16
measurements are taken, and the arc length between measurement points is only 15
ft, calculate the optimum cosine curve using all 16 points, but use only 8 points to
calculate
i
S . The equations in Figure B3 would be revised to read:

BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008

( )
( )
i i i-2 i+2
11 11 9 13
1 1
S =U - U + U
2 2
1 1
S =U - U + U
2 2


h. If outofmeasured outofplane settlement exceeds the limits described in
B.3.2.1 using the optimum cosine curve method, a more rigorous evaluation may be
performed in lieu of repairs. This evaluation must be done by an engineer
experienced in tank settlement analysis.
B.2.2.5
If a well-defined rigid tilt plane can not be determined or the maximum out-
of-plane settlement determined in accordance with B.3.2.1 is exceeded, the
procedures given in this paragraph may be used in lieu of more rigorous
analysis or repair.
B.2.2.5.1
For settlement profiles without a well-defined rigid tilt plane, the settlement
arc length,
arc
S , and outofplane settlement at the point under
consideration,
i
S ,must be determined from a plot of the measurement
data. Figure B4a illustrated the various measurement terms and
procedures for determining estimates of the settlement arc length and the
corresponding out-of-plane settlement, including the refinement of
measurements, when needed.
B2.2.5.2
If a valid cosine fit of the rigid tilt plane can be determined, but the
maximum out-of-plane settlement determined in accordance with B.3.2.1 is
exceeded, the procedure in B.3.2.2 may be used to evaluate the settlement.
In this case, refer to Figure B-4b for an illustration of the determination of
the settlement arc length and the corresponding out-of-plane settlement.
B2.2.5.3
If an examination of the measured settlement plot indicates a fold pattern
about a diameter of the tank, the maximum outofplane settlement should
be determined using a settlement arc length of 50% of the tanks
circumference.
B.2.3.1 UNCHANGED
B.2.3.2
The formula given in B.3.4 can be used to evaluate edge settlement. Alternatively, a
rigorous stress analysis can be carried out for the deformed profile. The
determination of the deformed profile should take into consideration:

a. Measurements taken when the bottom is not in contact with the soil or foundation
under the tank can overestimate or underestimate edge settlement significantly. If
the measured settlement is near the maximum allowable settlement, consider
repeating the measurement with the bottom forced down to the soil, e.g., standing
BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008

on it, or take an additional set of measurements in the same area, where the bottom
is in firm contact with the soil.

b. a. Locating the breakover point where the settled area begins requires some
judgment. Placing a straight edge on the unsettled bottom as shown in Figure B-5,
and observing where the bottom separates from the straight edge will help define
the breakover point.

c. b. If the tank bottom is cone up or cone down, the settlement B, should be
measured from a projection of the unsettled bottom, not from level. See Figure B-6.
B.2.3.3 THROUGH B.3.1 UNCHANGED
B.3.2 Permissible Out-of-Plane Settlement
From the measurements procedures described in B.2 B.2.2.4 and B.2.2.5,
determine the maximum outofplane deflection settlement. The magnitude
(absolute value) of the maximum settlement shall be compared to the
permissible values given in B.3.2.1 or B.3.2.2, as applicable. Use the following
formula to calculate the maximum permissible out-of-plane deflection (see note
below): The permissible out-of-plane settlement given in B.3.2.1 and B.3.2.2
do not take into consideration abrupt changes in shell elevation (ridges) or
discontinuities near the bottom of the tank in the settled region, such as low
nozzles. They are also not applicable to fold patterns in cone roof tanks
when the fold line is adjacent to or through a line of one or more roof
columns, or to patterns of settlement that include combined shell and edge
settlements. The permissible settlement criteria in B.3.2.2 are applicable to
API 650 carbon steel and stainless steel tanks and diameter ranges given in
B.3.2.2. Out-of-plane settlement that does not meet these limitations
should be further examined by a more rigorous engineering assessment in
lieu of repairs.
B.3.2.1
When using the procedure with an optimal cosine curve approach defined in
B.2.2.4 to determine outofplane settlement, the permissible outofplane
settlement is given by the following formula (see note below):

( )
( )

2
L Y 11
S
2 EH
=
max, ft


Where:
S
max, ft
=
permissible out-of-plane settlement deflection, in ft (out of plane
distortion),
L= arc length between measurement points, in ft,
Y= yield strength, in lbf/in.
2
,
E= Youngs Modulus, in lbf/in.
2
,
H= tank height, in ft.

Note: This formula is based on Criteria for Settlement of Tanks, W. Allen Marr, M.
ASCE, Jose A. Ramos, and T. William Lambe, F. ASCE, Journal of Geotechnical
Engineering Division, Proceedings of the American Society of Civil Engineers, Vol. 108,
August, 1982.
BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008

B.3.2.2
When using the procedure in B.2.2.5 to determine outofplane settlement,
the permissible outofplane settlement is given by the following formula
(see note below):





max,in
x x x
arc
D Y
S = min K S , 4.0
H E


Tank Diameter
K, Open ( external
floating
roof) Tanks
K, Cone ( internal
floating roof)
Tanks
D 50 ft 28.7 10.5
50 ft < D 80 ft
7.8 5.8
80 ft < D 120 ft
6.5 3.9
120 ft < D 180 ft
4.0 2.3
180 ft < D 240 ft
3.6 Not applicable
240 ft < D 300 ft
2.4 Not applicable
300 ft < D
Not applicable Not applicable



where:
max, in
S =
permissible out-of-plane settlement, in inches,
arc
S =
effective settlement arc, see B.2.2.5.1, in ft,
D=
tank diameter, in ft,
Y= yield strength, in lbf/in.
2
,
E=
Youngs Modulus, in lbf/in.
2
,
H=
tank height, in ft.

Note: This formula is based on Final Report on the Study of Out-of-Plane
Tank Settlement, J. Andreani, N. Carr, report to API SCAST, May, 2007.

B.3.2.3
Serviceability may also be a concern for tanks with significant out-of-plane
settlement. Outofroundness can impede floating roof operation and also
affect internal roof support structures. The outofroundness that a tank
experiences with out-of-plane settlement is fairly sensitive to the actual
pattern of settlement. The Owner may wish to specify additional inspection
or a more rigorous assessment of the tanks outofroundness.

B.3.2.4
If measured outofplane settlement exceeds the applicable limits
described in B.3.2.1 or B.3.2.2, a more rigorous evaluation may be
performed in lieu of repairs. This evaluation should be done by an engineer
experienced in tank settlement analysis.

BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008

CHANGES TO FIGURES
Figures B-1, B-2 UNCHANGED


Figure B-3Graphical Representation of Shell Settlement per B.2.2.4

(New Title, Figure B-3 itself UNCHANGED)
BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008



S
i,1
S
arci,1
POINT 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
S
unif
has been
eliminated from all
measurements
S
j,1
S
arcj,1
S
k,1
S
arck,1
S
i,2
S
arci,2
initial 16 measurements
additional measurements to better define settlement arc and maximum settlement
S
i,N
= maximum out-of-plane settlement measured from indicated plane, Nth estimate
S
arci,N
= settlement arc corresponding to S
i,N


Figure B-4aGraphical Representation of Shell Settlement per B.2.2.5
(Tilt Plane Not Described by an Optimal Cosine Curve)

Notes:

The settlement arc length shall be taken as the circumferential length
between points on either side of the maximum outofplane settlement point
under consideration.

An initial estimate of settlement arc length and maximum settlement should
be determined then, additional settlement measurement points may be taken
halfway between the points indicating a change in direction of the settlement
slope to further refine the settlement arc length and location and magnitude
of the maximum settlement.

Depending on location of the new settlement points relative to the initial
points, a further refinement of the measurement spacing may be needed.

The best estimate of the settlement arc length and maximum settlement shall
be considered in the procedure given in B.3.2.2.
BALLOT AGENDA ITEM 653 - 150: 2
nd
ballot January, 2008




S
i
S
j
S
arci
S
arcj
S
i
=maximim out-of-plane settlement measured from tilt plane for i
th
arc
S
arci
=settlement arc corresponding to S
i
S
unif
has been eliminated from all measurements
POINT 1 2 ... 16


Figure B-4bGraphical Representation of Shell Settlement per B.2.2.5
(Tilt Plane Described by Optimal Cosine Curve)



END OF REVISIONS
(ILLUSTRATIVE EXAMPLES IN SEPARATE DOCUMENT)