+ =
b
b
a
a a
I
L
I
L
E
WL
Y
3
2
max
where:
W =radial load
L
b
=distance between centerlines of bearings
D
b
=nominal shaft diameter between bearings
E =elastic modulus of shaft
I
a
=moment of inertia for the shaft diameter between impeller centerline and radial
bearing
I
b
=moment of inertia for the shaft diameter between radial and thrust bearings
Table 1 shows four hypothetical pumps all designed to have an SFI of 76.2. Generally
users of SFI would find this value acceptable. However two of the pumps do not meet
the seal face deflection criteria of API 610. For this reason some pump manufacturers
oppose dealing with SFI and have offered the alternative of testing for deflection. A
pump bearing housing would be rigidly mounted in some fixture and a known
standard weight would be attached to the shaft end. The deflection would be measured
directly. This deflection value would be furnished in all proposals and would provide a
direct way of comparing shaft stiffness in all pumps offered in a particular case. The
down side of this method is standardization of the fixture and test/measurement method.
With hydraulic loads also varying among pump designs and vendors, this still doesnt
really give a direct comparison.
Table 1, Comparison of deflection in four hypothetical pumps with equal SFI
L
a
(in) 12.5 12.5 12.5 12.5
D
a
(in) 2.25 2.25 2.25 2.25
L
b
(in) 7.75 10 7.75 10
D
b
(in) 3 3 2.5 2.5
X (in) 8.5 8.5 8.5 8.5
W (lbs) 250 250 250 250
L
3
/D
4
76.2 76.2 76.2 76.2
Deflection at seal face (in) 0.0018 0.0019 0.0021 0.0023
Deflection at impeller center (in) 0.0057 0.006 0.0067 0.0073
% deflection 100 105 117 128
All of this material has been debated in the WGTF for more than two years. Everyone in
the WGTF recognizes the limitations of an SFI comparison. Further there is no real
advantage to using SFD for a comparison. In spite of the weakness of this sort of
analysis the WGTF received multiple comments on this issue and there is a faction in our
industry that strongly desires to have this calculation performed to compare various pump
offerings. It is clear that if one line of pumps has disparately high SFI numbers there is a
good chance they cannot meet the deflection and dry bending critical requirements of
API 610. One the other hand if SFI numbers are disparately low one might suspect the
manufacturer is using overly large and expensive seals or he is exercising his creativity in
the use of numbers.
The WGTF therefore took a look at values of SFI for a number of lines of modern (API
7
th
and 10
th
Edition Pumps) with the result shown in Figure 5. It was found that if SFI is
plotted as a function of HQ/N (this number is proportional to shaft torque) on log-log
scales, the result is a straight line.
Figure 5. SFI for typical modern pumps
Overhung Pumps - Rotor SFI
0.01
0.1
1
10
100
1000
1 10 100 1000 10000 100000 1000000 10000000
Kt = QH/N
S
F
I
=
L 3
/
D
4
W'ton HN
Pacific SVCN7
FLS HPX
Pump-Turbines
6,100KtE-0.76
BWIP SC7
The DIS2 will have a standard simple method of calculating SFI and will refer to a figure
based on Figure 5 for guidance. A significant inconsistency with the value in the chart
would be cause for the purchase to perform a more in depth investigation before
purchase.
BEARING SYSTEM LIFE:
The API 610 individual bearing life requirements have been unchanged since the Xth
Edition in 19xx. In the current 11
th
Edition draft these requirements are as follows:
6.10.1.7 Rolling-element bearing life (basic rating life, L
10h)
for each bearing or
bearing pair) shall be calculated in accordance with ISO 281 and be equivalent
to at least 25 000 hrs with continuous operation at rated conditions, and at least
16 000 hrs at maximum radial and axial loads and rated speed.
NOTE 1 ISO 281 defines basic rating life, L
10
, in units of millions of revolutions. Industry practice is to convert this to
hours and to refer to it as L
10h
. ISO 281 also defines the method required to calculate bearing system life from individual
bearing life.
NOTE 2 For the purpose of this provision, ABMA 9 is equivalent to ISO 281.
It is the experience of the authors and many other users that fatigue failure of bearings is
not a significant issue in most plants. Typical bearing failure numbers are 8-10% of all
pump failures for conventional lubrication and fewer than 1% for oilmisted pumps.
Bearing failures are almost entirely lubrication related. Either the oil is contaminated or
there isnt enough oil. As a result the WGTF has not paid much attention to paragraph
6.10.1.7, feeling that current experience proves it is adequate.
All API standards contain a paragraph establishing the minimum design life for the
equipment covered. In the current draft of 610, the paragraph reads as follows:
6.1.1 The equipment (including auxiliaries) covered by this International Standard shall be
designed and constructed for a minimum service life of 20 years (excluding normal-wear parts as
identified in Table 19) and at least 3 years of uninterrupted operation. It is recognized that these
requirements are design criteria and that service or duty severity, misoperation or improper
maintenance can result in a machine failing to meet these criteria.
Design is defined in the following paragraph:
3.10
design
manufacturers calculated parameter
NOTE Design is a term that may be used by the equipment manufacturer to describe various parameters such as,
design power, design pressure, design temperature, or design speed. This term should be used only by the equipment
manufacturer and not in the purchaser's specifications.
There is clearly an inconsistency between the bearing life requirement of 25,000 hrs
(approximately but just less than 3 years) and the general requirement for the pump to be
designed for 3 years. This inconsistency comes about as a result of the fact that for the
bearing system to be designed for 3 years, the individual bearings must be designed for
more that three years. For two equally loaded bearings the individual bearings would
have to be designed for 40,000 hours L
10h
. The bearing system life can be calculated by
combining the individual bearing lifes as follows:
L
10h
(System) =[(1/ L
10hA
)
3/2
+(1/ L
10hB
)
3/2
+ +(1/ L
10hN
)
3/2
]
2/3
where: L
10hA
= Basic rating life, L
10h
per ISO 281 for bearing A,
L
10hB
= Basic rating life, L
10h
per ISO 281 for bearing B,
etc.
So, why not simply require the bearing system to be designed for 25,000 hours or 50,000
hours or whatever? In a nutshell longer bearing life requires larger bearings. Over many
decades pump manufacturers have tried to increase bearing life by installing larger
bearings. They have consistently had trouble meeting the bearing temperature
requirements of 610 when bearings larger than a 7314 are used at 3600 rpm. The author
is only aware of one pump manufacturer that uses a 7315 bearing in his largest OH2
bearing frame. Pump manufacturers are very concerned that requiring a 25,000 hour
system life will force redesign of their bearing housings which costs money and worse
might result in overheating of the oil and ball skidding failures. Further manufacturers
and many users view bearing fatigue life as a non issue. Other users strenuously object to
the inconsistency and some other API standard taskforces are receptive to higher bearing
system life requirements than 610 for other types of equipment.
To understand why bearing life is not an issue in single stage overhung pumps one can
look at the form of the equation for L
10h
bearing life. Before looking at the equation lets
review how the life calculation is performed. First the actual radial and axial forces on
the impeller are determined (HI 1.3 and multiple other texts discuss methods of
determining these forces). Then coupling and seal forces are determined. The forces are
then summed and the equivalent radial load, Pr, is calculated for each bearing.
P
r
= XF
r
+Y F
a
Where:
Fr = Radial Load
Fa = Axial Load
and
X & Y are factors from a table in ISO 281, these factors vary depending on the
type of bearing and the relative magnitude of the radial and axial loads.
At this point we go to the bearing manufacturers catalog and select the smallest bearing
that will give us an acceptable L
10h
life. L10 life is calculated from the following
equation:
L
10
=(C
r
/P
r
)
3
Where:
C
r
=Catalog load rating
P
r
=Equivalent radial load as above
This result is in millions of revolutions so we convent to hours by dividing by the number
of revolutions per hour. Now lets look at the equation. For a system of two equally
loaded bearings to have a 25,000 hr life, the radial bearing and the thrust bearing (40
degree angular contact duplex pair must each have a 40,000 hr life. The ratio of 40 to 25
is 1.6. Rearranging our life equation we find that as long as C
r
/P
r
is 1.17 or greater the
life will be 40,000 hours or greater. So how likely is it that the ratio of C
r
/P
r
is 1.17 or
greater?
To begin with we will restrict our discussion to single stage overhung, OH2, pumps.
Most manufacturers will have either three or four bearing housing sizes for their OH2
pumps. They will line their sets of hydraulics up against these standard bearing housing
sizes based upon the equivalent radial loads the bearings must deal with. Each bearing
housing will have a single set of hydraulics that represents the highest possible loads at
maximum diameter impeller and some arbitrary suction pressure. For the manufacturer
from which the following example comes, that suction pressure appears to be 250 psig.
This seems to be a sensible number to the authors in that a 250 psig suction pressure
probably covers 98% or more of all refinery services. (It is also noted that the pump
manufacturer has some tricks in this bag for higher suction pressures, such as differential
wear ring sizing and plugging balance holes in the impeller.)
The equivalent radial load is made up of the weight of the rotor, the radial thrust on the
impeller, the axial thrust on the impeller, seal compression load and the coupling axial
thrust. API 610 requires that the coupling axial thrust be input as the thrust at the
maximum allowable axial misalignment allowed by the coupling. It should be rare that a
user will stretch or compress the coupling to the maximum as opposed to repositioning
the coupling hub somewhere close to the right distance between shaft ends or hubs.
Additionally API 610 disallows pump selections that do not allow at least a 5% increase
in head. This latter requirement means that except in those cases where revisions to
pump hydraulics encroach on the 5% margin thrust will be lessened by 5% from the
worst case plus a decrease in thrust load due to something approaching proper coupling
installation. Next we have the arbitrary suction pressure (of 250 psig; where individual
bearing life is about 25,000 hrs). The suction pressure for any given pump doesnt have
to be much less than the 250 psig assumed in bearing selection for the ratio of dynamic
load rating to equivalent radial load to be greater than 1.17. Figure 6 shows the L
10h
life
for a 4 x 6 x15 OH2 pump at maximum diameter impeller It is seen that L
10h
life
becomes a huge number for most suction pressures encountered.
Figure 6. L
10h
Life for a typical OH2 Pump
0.00
200000.00
400000.00
600000.00
800000.00
1000000.00
1200000.00
0 50 100 150 200 250 300 350
Suction Pressure (psig)
L
1
0
h
L
i
f
e
(
h
o
u
r
.
The last point in this discussion is that this pump is represented as having the heaviest
loads of any set of hydraulics used for this bearing housing. All smaller sets of
hydraulics will have even longer L
10h
lives. This agrees with most user experience that
fatigue failures of API pump bearings are exceedingly rate. What is puzzling is why ball
skidding failures are not more common? The WGTF does not have agreement as to what
words will be in the DIS2 but it is likely we will simply require a 40,000 hour bearing
system L
10h
life.
PERFORMANCE TESTING:
The performance test tolerances of API 610 may be the only requirements in the standard
that have remained unchanged since the 1
st
Edition in 1954. During this time Hydraulic
Institute Standard 1.6, Pump Tests has changed its tolerances and evolved but it has never
agreed with API 610. Additionally ISO has a standard ISO 9906, Rotodynamic pumps
Hydraulic performance acceptance testsGrades 1 and 2, 2000 does not agree with
either HI 1.6 or API 610. In spite of this API 610 currently references both HI and ISO
9906.
The purpose of the reference is to use the test methods and allowable measurement
uncertainties and basically everything except the performance test tolerances. The API
610 tolerances are shown in Table 2.
Table 2. API 610 Performance Test Tolerances
As the WGTF has worked through the 11
th
Edition drafts we have received multiple
complaints about the tolerances in Table 2 (14). If one studies the table one will note a
number of interesting things. First if one converts the allowable tolerance in head from a
percent to pressure in PSIG (which is what is measured) the tolerance is shown in Figure
7.
Figure 7. Performance Test Head Tolerance Band Width in PSIG
Tabl e 14 Performance tol erances (API)
Rated
point
Shutoff
% %
Rated differential head:
2 +10
+5 10 a
2 +8
+3 8 a
2 +5
+2 5 a
Rated power +4 b
Rated NPSH 0
NOTE Efficiency is not a rating value.
a If a rising head flow curve is specified (see 5.1.13), the negative
tolerance specifiedhere shall be allowedonly if the test curve still shows a
rising characteristic.
b Under any combination of the above (cumulative tolerances are not acceptable)
Condition
0 m to 150 m (0 ft to 500 ft)
151 m to 300 m (501 ft to 1 000 ft)
> 300 m (1 000 ft)
Tolerance Band Width
0
5
10
15
20
25
30
35
40
0 500 1000 1500 2000 2500
Head
P
S
I
f
o
r
T
e
s
t
o
n
W
a
t
e
r
Band Width
3-.003*Head BW
In the DIS which was balloted last winter the WGTF proposed equations that plotted at
the purple line. A huge number of people objected to this as too complicated.
Another issue is that Table 2 (14) does not mention flow. The flow tolerance for the 9
th
and 10
th
Editions is contained in paragraph 7.3.3.3 b). The tolerance is +/-5%. This is a
big number. The tolerances have now defined a rectangle shown in Figure 8.
Figure 8, Typical Performance Test Curve
Figure 9 is a close up look at the Allowable Test Point Region or block.
The point of Figure 9 is the allowable test region allows plenty of room for discussion
and it allows a stack up of tolerances that can result in actually efficiency being
considerable lower than quoted and yet the pump would be acceptable. For typical
refinery pumps this may not be very important but for large pumps and pumps that have
mostly frictional system curves such as pipeline pumps, this is a very big deal. It should
be noted that API 610 bases acceptance on power not efficiency. This is because
efficiency is a derived or calculated value whereas power is measured directly. If one
takes the uncertainties (allowable inaccuracies) in the values measured during the test one
can see that with acceptable uncertainties in other variables the uncertainty in efficiency
if very large. Figure 10 shows the Allowable uncertainties in various test parameters
from the current version of ISO 9906. Note that 9906 and HI 1.6 are pretty close on these
values.
Figure 10 Allowable Uncertainties in test parameters from ISO 9906
As it turns out Hydraulic Institute and the ISO 9906 Working Group have gotten together
and the next edition of 9906 will be adopted back at HI 1.6. So the two test standards
will at long last agree. Further the Convenor of ISO 9906 is a member of the 13709/610
WGTF and so there is a very good chance that the tolerances that appear in API 610 11
th
Edition will be adopted by ISO and HI and there will be consistency in the world at last.
Unfortunately the current ISO 9906 Committee Draft still persists in assessing whether a
pump is acceptable based upon a single test point. This means that the full values of both
random and systematic uncertainties apply to the point. The WGTF proposes to require
that the test points be fit using either a spline or a suitable polynomial and then rated flow
inserted into the resulting equation and the head and power tolerances applied to the
result. This eliminates the uncertainty in flow from consideration as well as averaging
out the random error.
It is expected that the following in paragraph will appear in the DIS2:
Quanti ty Symbol Grade 1
%
Grade
2%
Fl ow Rate e Q
2.1
3.5
Speed of
Rotati on
e n
0.5
2.0
Torque
e T
1.4
3.0
Pump Total Head
e H
1.5
5.5
8.3.1.2 Performance and (NPSH) tests shall be conducted using the methods and
uncertainty requirements of ISO 9906 grade 1, HI 1.6 (for centrifugal pumps) or HI 2.6 (for vertical
pumps). Performance tolerances shall be in accordance with Table 15. Evaluation of results
shall be in accordance with 8.3.3.3 b).
And
8.3.3.3 Unless otherwise specified, the performance test shall be conducted as specified in
below.
a) The vendor shall take test data, including head, flowrate, power and vibration at a
minimum of five points. These points will normally be
1) shutoff (no vibration data required),
2) minimum continuous stable flow (beginning of allowable operating region),
3) between 95% and 99% of rated flow,
4) between rated flow and 105% of rated flow,
5) approximately the best efficiency flow (if rated flow is not within 5% of best efficiency
flowrate)
6) end of allowable operating region.
b) The test data shall be fit to a spline or appropriate polynomial (typically third or fourth order)
for head and for power using a least squares method. The rated/guarantee flow shall be
inserted into the resulting equation and a value for head and power calculated. These
values shall be corrected for speed, viscosity and density (specific gravity). The corrected
values of head and power shall be within the tolerance bands allowed in Table 15.
In the case of high-energy pumps (see 6.1.18), integral-gear and multistage pumps, it
may not be feasible to test at shutoff. Some low specific-speed pumps cannot achieve
120 % of BEP flowrate for the rated impeller diameter.
c) Unless otherwise agreed, the test speed shall be within 3 % of the rated speed shown on the
pump data sheet (see example in Annex N). Test results shall be corrected to rated speed.
d) The vendor shall maintain a complete, detailed log of all final tests and shall prepare the
required number of copies, certified for correctness. Data shall include test curves and a
summary of test performance data compared to guarantee points (see 10.2.4, 10.3.2.2 and
example in Annex M).
e) If specified, in addition to formal submittal of final data in accordance with 10.3.2.2, curves
and test data (corrected for speed, specific gravity and viscosity) shall be submitted within
24 h after completion of performance testing for purchaser's engineering review and
acceptance prior to shipment.
Having said this, we have one more round of review and comments and the performance
testing section will almost certainly draw numerous comments.
CONCLUSION
The main focus of the 11
th
Edition of API 610 is to improve clarity and to deal with a
limited number of key issues. The three key issues, Performance Testing, Bearing
System Life and Shaft Flexibility Index have attracted sufficient interest and comment
that the Working Group/Taskforce has decided to submit the document for an additional
round of comment. In the ISO world this is a second Draft International Specification.
The second DIS is expected to be distributed by ISO and API in early 2007 with the
comment period and ballot closing approximately mid year. Comments will be resolved
in the third quarter. Comment resolutions will be presented to the API Subcommittee on
Mechanical Equipment at the Fall Refining Meeting in early November 2007. The plan
is for the Final Draft International Specification and API Ballot to take place in early
2008. Publication is planned for mid 2008.
References:
This tutorial contains unpublished work by a number of members of the API 610
Taskforce/ISO 13709 Working Group. Among these members are:
Mick Cropper, Sulzer Pumps
Fred Blumentrath, CPC
Terry McGuire, Flowserve
Charle Heald, Consultant
J im Harrison, Flowserve
Additionally the ideas in this tutorial have been affected/developed through inputs from
the entire ISO 13709 Working Group/API 610 Taskforce. The authors asknowledge and
thank all of them for their contributions.
Hydraulic Institute Standard 1.3
Appendix 1 BACKGROUND DATA COVERING THE HISTORY OF API 610
AND API 682 GASKET REQUIREMENTS
API 610 1
st
Edition (tentative) 1954: Totally silent on gaskets or gasket
materials.
API 610 1st Edition January 1955: Text is silent on gaskets and gasket materials.
Datasheet has a block for gasket materials and blocks for Confined or Flat beside the
word Gaskets.
API 610 2nd Edition January 1957: Text is silent on gaskets and gasket
materials. Datasheet references are identical to 1
st
edition.
API 610 3rd Edition January 1960: Text is silent on gaskets and gasket
materials. Datasheet references to gaskets have been removed. Thus 3
rd
Edition is totally
silent on gaskets.
API 610 4th Edition July 1965: Text and datasheet are silent on casing gaskets
but Seal Gland Plate gaskets are addressed in Section 24, item d.:
API 610 5th Edition March 1971: Addresses radially split casing gaskets in item
12. f.:
Seal End Plate gaskets are addressed in item 24. k.:
Casing Gaskets are addressed in the Materials Section in Table D-1:
Note that the table does not reference spiral wound gaskets at all and that the only choices
are variations of asbestos gaskets and Teflon. Teflon Casing Gaskets are addressed in the
General Notes, Note 10:
The datasheet has no reference to gaskets.
API 610 6
th
Edition January 1981:
Pressure casing gaskets are covered in 2.2.7:
Gland gaskets are covered in 2.7.1.17:
The datasheet has no reference to gaskets for either the casing or the seal gland plate.
The materials table is now E-1 but the requirements are identical to 5
th
Edition. The table
is now so large it is impractical to scan and insert into this record. Note 10 of the 5
th
Edition is now note 7.
API 610 7th Edition, February 1989:
In the 7
th
Edition there are two references in the text related to pressure casing gaskets.
These are 2.2.6 and 2.2.10:
The fact that 2.2.10 gives the requirements for o ring grooves implies that o-rings might
be acceptable gaskets. Seal gland gaskets are covered by 2.7.1.16. This latter claus
specifically classifies o-rings as controlled compression gaskets further implying that o-
rings can be used as gaskets on pressure casings.
Datasheet has no blanks or references to gaskets for the casing or seal gland plate.
After the text has created this ambiquity. Annex H completely contradicts the
acceptablility of o-rings by only calling out spiral wound gaskets on the pressure casing.
This applies to both casing and seal gland gaskets. Further note that this is the first time
spiral wound gaskets are mandated. No previous edition calls for them.
API 610 8
th
Edition, August 1995:
The pressure casing gasket is covered in 2.2.7 and the o-ring groove is covered in 2.2.10.
For the first time 2.2.2 is crystal clear, o-rings can be used as casing gaskets.
Seal gland gaskets are covered by 2.7.3.23. It is also crystal clear that o-rings are allowed
between the pump casing and the seal gland.
Table H-1 has been reduced in size to fit on a single page but the requirements for
gaskets are unchanged and only spiral wound gaskets are called out.Note that there is
another paragraph in the mechanical seal section 2.7.37 which says that seal gaskets and
hard faces shall be specified from the seal materials tables. This gets us a material for the
gland gasket but there is no reference directing materials for other pressure casing gaskets
if o-rings are used.
Note that there is another paragraph in the mechanical seal section 2.7.37 which says that
seal gaskets and hard faces shall be specified from the seal materials tables. This gets us
a material for the gland gasket but there is no reference directing materials for other
pressure casing gaskets if o-rings are used.
API 610 9
th
Edition, January 2003 and 10
th
Edition, October 2004
The pressure casing gaskets and o-ring groove requirements are covered by 5.3.10 and
5.3.12. They are unchanged from the 8
th
Edition.
Clause 5.8.3 for the first time refers to spiral wound gaskets being used on the seal
chamber joint. Note that this paragraph implies that bolting might have to be increased in
size to properly crush a spiral wound gasket. This could cause manufacturers to prefer to
use o-rings on this joint.
Seal chamber gaskets are covered by 5.8.11 and are unchanged from the 8
th
Edition.
Table H-1 continues to only call out spiral wound gaskets for the pressure casing.
API 682 3rd Edition
API 682 has three paragraphs and two annex sections relevant to o-rings. The first is the
most relevant. Paragraph 6.1.6.7.2 mandates o-rings on the joint between the seal
chamber and gland plate for services below 350 F.
The other two paragraphs, 6.2.1.2.2 and 6.2.2.2.2, are identical and are also identical to
the requirements in API 610:
API 682 also gives temperature limitations for o-ring materials and a tutorial on their
selection. These two sections follow.
Standard Paragraphs, Revision 23B, November 3, 2005
The standard paragraphs have only one paragraph relative to pressure casing joints. This
paragraph, 6.2.4, discourages the use of o-rings but is really aimed at compressors or
steam turbines.