Académique Documents
Professionnel Documents
Culture Documents
steel pipe. A short time later a second,2 but more re- goods,' when revision of the physical-property require-
stricted, investigation mas made. Both of these.-.- - .-
investi- ~ne~!ts.of casing was being considered. The use of this
gations resulted in t h e development of formulas f o r method was estended to the derivation of lninimunl
the calculation of collapse strength t h a t were largely collapsing pressures f o r casing; h n d , a t t h e present
empirical. These formed the basis f o r the setting-depth time, the values thus obtained a r e the only ones avail-
tables published by t h e pipe manufacturers, and in 1934 able generally to the engineer f o r the design of casing
by the American Petroleuin Institute a s Supplement strings of seamless and continuous electric-weld casing.
No. 1 to +PI Code No. 5 : "API Settilly Depths for I n the application of the statistical method to this prob-
Casing. Dvill Pipe, axd T16bzng." Considerable variation lem, the minimum value was selected a s t h e average
o r spread in the results of these early collapse tests w a s collapse pressure minus three times the standard devia-
noticeable. It was recognized t h a t several factors con- tion froin the average value. Theoretically, 99.84 per
trlbuted to this effect, such a s variation in wall thick- cent of all results should be greater than this minimum
ness, eccentricity, out-of-roundness, coinposition of the value which, f o r casing, is only 65 per cent of the aver-
steel, physical properties of the steel, fabricating meth- age tollapse pressure. The use of this minilnum value
ods, working temperatures, etc. I t w a s t o be espected, is approsimately equivalent to a reduction of the safety
therefore, t h a t there \vould be a lack of close agreement factor fro111 1.5, according to present setting-depth
in the results of collapse tests. Even today, after many tables, to 1.0. I f this condition is a true representation
improvements and refinements have been made in t h e of the degree of uniformity of casing, then the processes
seamless process, a rather wide spread in test results is used in the nlanufacture of casing a r e f a r less capable
inevitable. The advancement has eliininated sensibly the of being controlled than has been realized by t h e user.
effects of the dimensional factors enumerated herein- The authors do not belleve t h a t this is a n accurate
before, b u t there still esist the variations arising from analysis of casing quality, and a r e of the opinion t h a t a n
the latitude required in the composition and properties error has been made in the manner i n which the statis-
of the steel and its processing to render i t feasible to tical method h a s been applied to this problem.
make pipe coinn~erciallyby any process. Greatest value attaches to the analysis of d a t a by
Undoubtedly, this wide spread i n measured collapse statistical methods, when the data have been obtained
strengths, together with the knowledge t h a t the pipe under controlled conditions 3-i.e., a set of observations
had much hlgher collapse strength than was required resulting fro111 measurements made under the sanie es-
for the comparatively sl~allowwells then being drilled, sential conditions and representing material produced
led to the adoption of the average collapse strength a s under the same essential conditions. The significance
the basis f o r determining the setting depths. A t t h a t of the interpretation of the observed results is influenced
time the setting depths arrived a t were safely con- to a marked degree by whether there is a foundation f o r
servative; because norlnally the physical properties of the belief t h a t the conditions were controlled. It is in
the steel were well above the minialum values f o r the this respect t h a t the error w a s made. The minimum
particular grades, and the setting depths were calcu- values were obtained from the results of collapse tests
lated f o r a safety factor of two. It is evident ; t h a t made, over a period of years, on seamless pipe repre-
there lnust have been: 1, failure to realize that; the senting the product of more t h a n one pipe mill. Cer-
selection of casing f o r a well actually is a n engineering tainly, changes in processing were made during this
problem; and, 2, lack of vision in foreseeing t h a t the period, a s well a s changes in the grade limits. I t is well-
deep wells of today cannot be cased either practically known t h a t not all pipe manufacturers work t o the saine
o r economically on the basis of a safety factor of hvo. limits of physical propert~es to produce plpe in ac-
Proof of the first of these two assertions is that i t would cordance with the A P I pipe specifications. This condi-
he estreinely difficult, if not entlrely in~possible,to find tion glves rise to a wider spread between masimum and
a n y other instance in which the engineered design of a minimum values and, consequently, a larger standard
structure is based on t h e average, instead of on t h e deviation from the average and lower minimum values.
ininl~nurn, physical properties of the structural ma- The composition of the steel varies from one inanufac-
terials. Even if such instances were known, they would t u r e r to anotller, a s well a s the detalls of the process
be condemned a s incompatible with sound engineering usecl in converting steel into pipe. There is a range in
practice. The t r u t h of the second assertion is coin~non physical properties of the steel f o r a given grade of
knowledge. casing, the estent of which varies between manufac-
As i t became necessary to set casing with reduced turers. These a r e variables which cannot be eliminated,
safety factors, t h e minimum collapse strength assumed but this fact should not be taken to imply t h a t all
inuch greater importance. It soon was realized, how- casing is so nearly alike t h a t i t can be looked upon a s
ever, t h a t the collapse-test data 011 current types and having beell produced under controlled conditions such
grades of casing were f a r too few in number to indi- a s a r e defined above. There a r e three other undesirable
cate the ininiinun~collapse strengths. To substitute f o r features i n the procedure used: 1, the assunlption t h a t
this lack of information, the pipe manufacturers advo- the spread in test results is the saine f o r all grades of
cated the application of the statistical method f o r t h e pipe; 2, the grouping of all sizes together on the basis of
analysis of data to the determination of t h e n ~ i n i n ~ u n ~ D
ratios; and, 3, t h e fact t h a t a n y iinproveinent in
values. The statistical metl~odfirst was applied to t h e
test data on physical properties of oil-country tubular product which would give less deviation would not be
reflected immediately, because the mass of accumulated COLLAPSE STRENGTH
data controls the resulting value. The several points
just enumerated a r e illustrated by the results of a Theory of Collapse
survey of the physical properties of casing steel,' which As has been brought out in the previous discussion,
was referred to previously. The data from this survey, the purpose of the ultimate specification, from t h e
showing the range in yield strength of the casing steel standpoint of the user, is to obtain interchailgeability
by grades, sizes, and n~anufacturers,are summarized and a product which will meet performance require-
in Table 1. Yield strength was chosen for this com- ments. I n the case when the design of casing to resist.
parison because i t is the physical property which has collapse is being considered, this performance require-
the greatest effect upon the essential characteristics of inent is nlet by setting a nlinimunl guaranteed collapse
casing, but similar comparisons could be made for pressure for the product; and, when this value is set,
any of the other physical properties. I t is unnecessary, the usel. then has all the information required to apply
although it would be informative, to discuss the results the product on a sound engineering basis. Also, when
of the survey in detail; but a study of the data in the specification is written on the basis of performance,
TABLE 1
Range in Tensile Yield Strength of API Casing Steel '
-
Yield Strength of Casing Steel (Pounds Per Square Inch)
48-111. to Over 78-In.
All Sizes 71-111. to 91-111. Over 91-11].
Manufacturer
, --
High Low High Low Low v- Low
GRADE "C"
GRADE "D"
' hlinimum specified yield s t r e n g t h : grade "C." 46,000 p s i ; gmde "D," 66,000 psi.
Table 1 will show a basis for each of the statements and the minimum collapse values a r e determined from
made herein with reference to controlled conditions. test, the user no longer needs 'to depend on theory f o r
I n this connection i t is of interest to note that f o r the estimating these minilnuin values. The theory, however,
average yield strengths of 58,920 psi and 67,222 psi will be discussed herein because i t has a bearing on
found for grade "C" and grade "D," respectively, the limiting values which can be expected.
minimum specified yield strengths permitted a factor The developnlent of collapse theory a s it applies t o
of slightly more than twice the standard deviation tubular goods has followed the accumulation of experi-
found for grade "C" casing (5,697 psi), and slightly mental data. Stewart,' in his early work, developed a
less than twice the standard deviation found for grade theory which fitted his experimental work. His em-
"D" casing (6,794 psi). This is due primarily to high pirical equations and data covered the product which
results, and not to low results, although a few results for many years met all of the user's requirements.
below the specified minimum yield strengths were in- When producing depths increased, higher-strength steel
cluded. As the principal requirement that the data be was used, and i t was necessary to correct Stewart's
obtained under controlled conditions is not satisfied , work to fit the new requirement. As depths cont~nued
fully, the authors believe t h a t the minimum values to increase, thicker tubes were used a s well a s higher-
should be determined by a more direct method, and strength steels, and the older experimental data no
that the average values for collapse strength, pullout longer covered the conditions encountered. There was,
strength, and bursting strength have no place in the therefore, a need for a more complete theory which
design of casing strings for deep wells. would cover the general range of conditions. However,
the development of such a theory could not ,be made As the agreement is so close, the authors favor the
until further experimental data were available. The retentioh of equation ( I ) , a s given in Blclleti?~5-C-2,
next major contributioa to the developnlent of a rational so a s not to involve further a n already complicated
and practical theory to cover the field of casing collapse discussion.
was given in a paper by Jasper and Sullivan?
The work of Sturin" was used a s a basis for more
accurately deternlining the collapse of long tubes in
the so-called elastic range. When failure is purely
elastic, only the elastic constants and the tube dimen- I t is believed t h a t this equation actually is no better
sions enter into deterlnining the pressure a t which than one o r more of t h e several which have been pro-
collapse will take place. Yield strength o r other physi- posed. It is a s good from the standpoint of practical
cal properties of the steel do not influence the failure. purposes a s any other equation which has been proposed
After a study of all the esperimental d a t a available- and, consequently, will be used in a n y of the following
including that of S t e w a r t 1 and Carlnen," together with discussions which refers to tlie envelope curve. This
information fro111 A. 0. Smith Corporation's esperi- curve, a s will be noted from the equation, i s independent
n~ental results-Jasper and Sullivan showed t h a t the of the yield point o r other strength properties of the
yield point of the material influenced the behavior in material, and-except a s the casing is influenced by
the region to the left of the envelope curve. From the the lnanufacturing processes-the values should follow
information which was compiled, a series of curves this curve.
was drawn. These curves were based entirely on es-
perimental results. By interpolation and estrapolation Yield-Point Curve
they were extended to the regions where esperinlental
data were not available. Subsequent papers have veri- As h a s been pointed out previously, the development
fied these curves 6s 7s12813 , and recent papers have been of the collapse curves f o r the conditions not covered
written to explain why these curves have this form, by the envelope curve originally was based on esperi-
and to develop a general theory which would cover all nleiltal results. The theoretical studies applied to this
~ o n d i t i o n s . ~ ~ ~The paper by Jasper and Sullivan
149 subject have attempted to show how i t might be possible
further applies the detailed investigations of Sturin, to approsiinate the esperimental curves by calculation.
Carmen, and Stewart to the considerat~onsof the effect As has been brought out in t h e theoretical papers, t h e
of length, end effects, out-of-roundness, and many other shape of the collapse curve depends on the stress-strain
special factors which enter into this involved problem. curve of t h e steel. This stress-strain curve is influenced
Many of the papers recently offered. give refinements in by the coinposition and treatment which the steel h a s
theory. I n many cases these so-called corrections, o r receivCd, and in no way is a fixed nor nearly fixed
modifications, have so little effect from the standpoint value such a s the elastic constants (modulus of elas-
of application t h a t they a r e of interest only from a n ticity and Poisson's ratio). I n t h e so-called plastic
academic viewpoint. region t h e e ~ p e ~ i l n e n t adata
l show t h a t the variation
It is felt by the authors t h a t the entire collapse ques- within practical limits is about proportional to the
tion a t times h a s been confused unnecessarily by t h e yield point of the material; and all fornlulas which
discussion of theoretical problems-which, from the have been proposed f o r this region a r e linear with
standpoint of the user, have but little o r no bearing respect to yield point, and directly dependent upon the
on the application of casing. Furthennore, the ques- yield point f o r a given fabricating process. The exact
tions of interpretation of data (which will be discussed values f o r collapse pressure in this region have been
later) ancl application of theory likewise have been shown, by the published experiinental data, to be de-
confused by discussion of factors which a r e within t h e pendent upon nlany factors. F o r instance, in the case
limits of esperiinental error, o r which a r e so small of seanlless pipe t h e results a r e influenced by the
t h a t they have no influence in the selection of a proper presence and magnitude of internal stresses," and in
factor of safety. The following discussions of special the case of special processed pipe '?-when directional
phases of collapse theory a r e iiicluded because of their properties a r e present in the steel-the collapse value
bearing on the discussion and application of data in depends on the compressive yield strength, and h a s no
the following sections. relation to the tensile yield strength. Even though
much h a s been written with respect to the theory of
collapse in the region covered by the yield-point curves,
Envelope Curve there is a t the present time no theory which gives t h e
The theory of collapse of long tubes h a s been studied basis f o r calculating the actual values in t h a t region,
by inany writers. The theories of S t u r m l " and Timo- and a n y curves o r equations which have been developed
shenko '' agree reasonably well. The equation given in to cover these conditions have been based on experi-
Bzilletirl 5-C-2"ives slightly higher values than obtained mental data. The fact t h a t such a theory o r practical
from Sturm's ecluatioi~. The equation given by Cline- equation is not available sl~ouldnot in a n y way compli-
dinst" gives still higher values, but the spread between . cate the present application of casing. The inore im-
all of these values is small and depends on the elastic portant problem, with respect to t h e yield-point region,
constants which a r e used in obtaining the equations. is t h a t of interpretation and analysis of data.
5
ZIALS
I11 thls paper prime importance is attached to esperi- and include casing made for both the old and new API
mental data and practical experience. I t is not intended grades. I11 Appendix 1 and elsewhere in the discussion
to imply that theoretical studies should not be made, the casing has been classified according to the present
nor that they have no place in the study of casing API grades-i.e., it has been classed a s grade H-40 if
problems. The general purpose of a theory is to develop the tensile yield strength was in the range 40,000 to
a basis for formulating, with greater precision, ideas 54,900 psi; a s grade J-55 if the tensile yield strength
and data which have been collected esperiinentally was in the range 55,000 to 79,900 psi; and a s grade
and by esperience, and to extend the range of applica- N-80 if the tensile yield strength was 80,000 psi or
tion of these data and esperience. If a theoretical more. I t would have been preferable to have data
investigation is complete, it usually is so coniplicatecl from tests of casing made only according to the present
lnathematically a s to be difficult or impossible of solu- API grades, but sufficient time has not elapsed since
tion and not readily applicable. If the theoretical de- their adoption to make the number of tests required
velopinent is simplified by making many assumptions, for this purpose. The error involved in using the test
the range of application may be so narrow as to be of results from casing thus reclassified probably is small,
little use. In any case, the theory must be checked a s the nliaiinum tensile yield strengths of grade H-40
by espe~imentaldata or experience to be certain that ancl grade 5-55 are identical, respectively, with the
it accurately represents the facts. Once the theory is old grade "B" and grade "D." I t is possible that soine
verified, it is very useful as the basis for extension of the casing reclassed a s grade H-40 was actually
to new developn~ents, and to the analysis of new old grade "C"; but, if this is the case, the collapse-test
problems 011 which esperiinental data or esperience is results do not reflect i t definitely. There is reason to
not available. In the case of the envelope curve, the believe that the deviation is greater than it would have
theory has been shown to represent the lnaxinlum which been if the same number of tests had been made on
can be obtained, and is usable only over a certain casing intended to meet the present grade requirements.
range-that range being determined from esperience As a matter of pertinent interest, the averages of
and esperiinental data. For the region where the all collapse-test results on API casing sizes were calcu-
envelope curve does not fit esperience and data, esperi- lated and coinpared with the corresponding average
mental .curves have been drawn; and, although the values in B~illetin5-C-8. The results of this comparison
theory esplains the general shape and reason for the are given in Appendix 2. The averages of the tests
lack of agreement, i t is not applicable to esact calcu- from the four sources which provided the greater por-
lations. tion of the data were calculated separately, and the
general average also was calculated for those sizes for
Collapse-Test Data which results were available from more than one source.
The con~parisonis facilitated by reference to the ratio
I t was stated hereinbefore that the authors do not of the general average to the API average. This com-
believe that a lninimunl collapse pressure only 65 per parison included 1,044 out of the total of 1,371 collapse
cent of the average collapse'pressure is representative tests. The sources of the data were designated a s
of casing now being used, and that such a low value is follows :
not suitable for use in the design of casing programs. A. Data used by Manufacturers' Subcolnmittee in the
Consequently, all of the pertinent available collapse- preparation of Bz~lle.ti)t5-C-2. All tests in this
test data have been analyzed for the purpose of arriv- group were on seamless pipe.
ing a t new, safe, and usable minimum values. These B. Data from Crawford's investigation of electric-weld
data, from 1,371 collapse tests, are given for reference casing.
in Appendix 1. They include the results of the tests C. Data from Frame's investigation of seamless casing.
made on seamless pipe which were used by the Manu- D. Data from Frame's investigation of seamless casiilg
facturers' Subcommittee in the preparation of the with low residual internal stress (these data in-
curves, formulas, and tables given in Bztlletin 5-C-2, cluded in Appendix 1, but not in Appendix 2).
the results of investigations by Frame" and Crawford,? E. Data from recent tests on one make of sea~nless
and results of soine recent additional tests on seamless casing.
casing. The casing tested was made by the seamless API. Average values in Bztlletin 5-C-2.
and continuous electric-weld processes, and both the
closed-end and open-end nlethods of collapse testing The same designations a r e used throughout. Specific
were used. The results of closed-end tests were cor- attention is called to the fact that the values given in
rected to opeh-end values, in accordance with the con- Bzilletin 5-C-8-which are designated herein as API
clusion reached by Frame: after an investigation of average values-are not recoinn~endedby the American
the two methods of collapse testing, which is discussed Petroleum Institute, but were published only a s infor-
more fully in connection with the proposed standard mation on casing.
collapse test. No collapse test was considered unless Examination of the ratios of the averages of tests
the longitudinal tensile yield strength of the steel also to the averages given in Bulletin 5-C-2 discloses soine
had been determined. It should be pointed out that ~ facts not brought out previously-which, however, need
the data supplied by the Manufacturers' Subcommittee not be discussed in detail, a s they did not enter into
represent tests made over a period of several years, the determination of the nlinimuin collapse pressures.
SETTINGDEPTHS FOR CASING 131
The averages and ratios provide additional confirn~ation considerations of stress distribution-resulting, in the
of the conclusion reached with 'egard to controlled open-end test, from the pressure of the packing glands
conditions. in the stuffing boses and, in the closed-end test, from
the end support provided by the end plates-are colnpli-
Con~putation of Minimum Collapse Pressure cated, and no useful purpose would result from their
inclusion. I t is the general opinion t h a t the closed-end
I n order to determine reasonable and reliable tem- test gives slightly higher results than the open-end test.
porary n ~ i n i m u n collapse
~ pressures, the test results The most conv~ncing data in confinnation of this
on A P I casing sizes a r e summarized in Table 2. The opinion a r e those obtained by F r a m e B from 49 sets of
hi,gh and low results for each size a r e listed in order tests in which one open-end and one closed-end speci-
to i n d ~ c a t ethe variation, and the ratio of the lowest men were cut from the same piece of casing, practically
result to the API average result (Bulletin 5-C-2) is free of residual internal stresses. F r a m e found t h a t
calculated. The ininimu~nvalues proposed a r e based on the difference between .the two collapse pressures w a s
these ratios, which represent 1,044 collapse tests on
greater a t low D values,
- and t h a t the average differ-
both seamless and continuous electric-weld casing. The t
only correction applied t o t h e test results w a s f o r t h e ence varied from 0 a t a n ratio
- of 28 to 7.5 per cent
conversion of closed-end values to open-end values, a s t
,
described previously. This correction was in the direc- at a D ratio
- of 16. This relation was expressed by
tion of lower values and, hence, added safety. The t
the following equation:
lowest ratio f o r grade H-40 and grade N-80 casing was
0.76, and f o r grade J-55 the lowest ratio was 0.71.
Only 3 tests on grade J-55 casing gave results below
the ratio 0.75, and, only 15 tests of the entire group
gave results lower' than 80 per cent of the average Pc=collapse pressure for closed-end specimen, psi.
values in Bztlletzn 5-C-2. Of these 15 tests, 12 were in Po=collapse pressure-for open-end specimen, psi.
the group of data f o r seamless casing used f o r the Because of the increased emphasis being placed upon
preparation of Bz~lletin5 4 - 2 . As already pointed out, the accumulation of collapse-test data, i t is desirable
it IS believed t h a t the reasons f o r the low results in t h a t standard conditions f o r making.collapse tests be
this group a r e t h a t the tests covered a perlod of several ' adopted a s soon a s possible. After a rather extended
years; and much of the pipe was made to the old API consideration of this question, the Manufacturers' Sub-
grades, and probably is not truly representative of committee reconunended t h e adoption of t h e open-end
the presept casing. test, prior to the annual meeting in 1939. No action
In the light of these facts i t is proposed t h a t the has been taken a s yet on this recommendation; but,
minimum collapse pressures be fixed temporarily a t evidently anticipating i t s adoption, several pipe manu-
values t h a t a r e 75 per cent of the average collapse pres- ,, facturers recently have installed open-end test equip-
sures given in B~illeti?~ 5-C-2. The authors believe t h a t ment. The actual differences between the results of
safety factors based on these n~inimums (which a r e open-end and closed-end tests a r e of no practical im-
given in Table 8) will be reliable, ' a n d t h a t only a portance, escept f o r the heavier weights of the smaller
slight hazard .would accompany the use of minimum sizes of casing. I n view of the fact that, with only one
values a s high a s 80 per cent of the averages. When o r two esceptioas, all collapse-testing equipment is f o r
the practice of working to minilnun1 values rather 'than open-end test specimens, and because the slight differ-
average values is established firmly, a s i t must be if
ence in results is in the direction of increased safety,
future problems in deep drilling a r e to be met, there
the authors propose t h a t t h e open-end collapse test
mould seem to be no reason why the minimum values
be adopted a s standard. Each pipe ~nanufacturerh a s
should not be 86 per cent or more of the average values.
designed and constructed collapse-testing equipment in
This is not too much to expect of modern pipe-manufac-
turing processes. accord with individual ideas, so t h a t there i s 110 uni-
formity in details of design o r dimensions. This is not
necessary a s long a s each test ecluipment fulfills certain
Standard Collapse Test essential requirements and data a r e obtained in a uni-
form manner. Adoption of the following rules will
One of the more controversial phases of the collapse
insure co~npliancewith the essential requirements.
problem has been the method by which collapse pressure
is determined. Two methods of testing have been, and
still a r e being, used. The names by which these two PROPOSED METHOD
methods a r e designated indicate the principal difference
between them. I n t h e open-end test t h e ends of t h e STANDARD COLLAPSE TEST FOR CASING
section of casing being tested extend, through stuffing 1. Scope
boses, beyond the pressure chamber; whereas in the
closed-end test the ends of the test section a r e closed, The collapse pressure of casing made in accordance
usually by melding on head plates, and all of the result- with API pipe specifications (API Standards No. 5-A)
i n g capsule is in the pressure chamber. The theoretical shall be determined by the inethod outlined herein.
TABLE 2
Sunin~arpo f Collapse-Test Data for API Standard Casing
Nominal Dimensions
Average A P I Ratio:
Actual Collapse
Wall D Collapse Actual LOW
- Pressure ( P S I )
Size Weight Thickness t Number of -> Pressure Average
(Inches) (Lb P e r Ft) (Inches) Ratio Tests High Low (PSI) API Source
GRADE H-40
3 3,900
10 5,240
14 5,800
6 2,570
9 4,260
3 4,475
1 ....
43 3,470
3 3,050
10 3,700
31 2,500
GRADE J-55
7 9,950
20 4,650
39 6,400
12 7,210
12 6,600
2 7,120
16 5,130
8 7,620
17 6,670
3 7,300
3 6,980
11 7,180
1 .. . .
10 7,925
2 8,990
3 2,610
3 3,250
6 2,660
18 3,960
14 4,200
17 5,550
18 5,000
1 ....
15 6,100
2 6,430
17 6,525
14 7,110
60 6,700
14 7,800
49 7,800
TABLE 2-Conti~lued
Size
Nominal Dimensions
Weight
h
Wall
Thickness
-D
t Number of
Actual Collause
Pressure (PSI)
Average A P I Ratio :
Collapse Actual Low
Pressure Average
(Inches) (Lb P e r F t ) (Inches) Ratio Tests High LOW' (PSI) API Source
GRADE 5-55-
16.55 10
13
2
14
20.33 3
10
15
GRADE N-SO
18.03 3 9,150 7,420
24 9,200 6,900
20.83 6 5,650 5,010
TABLE 2-Continued
Nominal Dimensions
A Actual Collapse Average A P I Ratio:
Wall D . Collapse Actual Low
Size Weight Thickness
-
t Nuinber of
Pressure ( P S I )
Pressure Average
(Inches) (Lb P e r F t ) (Inches) Ratio Tests Bigh
Low (PSI) API Source
GRADE N-80-Continued
18.52 4 7,910
located 18 in. on either side of the center of the collapse- There is a possibility t h a t operators a t soine time
test specimen. The outside diameter shall be the aver- may desire t o make collapse tests in t h e field. F o r
age of these eight readings, and this value shall be this purpose i t probably would be satisfactory to weld
D a cyl~nder-having the requisite strength and connec-
used to compute the -i- ratio.
tions f o r the pump and pressure gage-around t h e test
7-B. IYnll thickrtess: After the specimen h a s col- length of casing. Field collapse testing should comply
lapsed in the test, it shall be cut through the center with the standard test procedure a s closely a s possible.
of the section in which greatest flattening occurred.
Wall-thickness ineasureinents shall be made with a
ball-point microineter a t this section, a t 16 points
Effect of Tension oil Collapse
equally spaced around the periphery of the specimen. When a casing string is of uniform weight and grade,
The wall thickness shall be the average of these 16 its design is limited to checking its resistance to collapse
readings, and this value shall be used to compute the a t the bottom and to pullout strength a t the top. A t
D ratio.
- all intermediate points this casing string would be ade-
t cluately strong. Wit11 the advent of graduated strings
D D of either different w e ~ g h t so r grades, the problem be-
7-C. -t
ratio: Compute the 7 ratio by dividing
comes more complicated; because, in these cases, the
the average outside diameter by the average wall
casing is subjected to both tension and collapse stresses
thickness. a t the critical point, which is a t the change in grade
o r weight. This combination of collapse stress and
8. Longitudinal Tensile Yield Stretigtl~ tension reduces the pressure a t which the pipe will
Three longitudinal-strip tensile-test specimens, spaced collapse. The anlouilt by which t h e collapse pressure
120 deg apart, shall be prepared from the second section is reduced will depend on the resultant stress caused
cut from the piece of casing adjacent to the collapse- by the two components of stress.
test specimen. The test specimens shall be prepared Before t h e effect of the combined stresses can be
in accordance with Par. 27 of A P I Standards No. 5-A, determined. i t must be known how the material will
10th edition. The yield strength shall be t h a t stress fail under a coinbined stress. Four principal theories
which causes a total elongation of 1 per cent of t h e have been advanced to cover t h e type of problem t h a t
gage length, in accordance with Par. 1 3 of A P I Stand- is involved, a s follows:
a r d s No. 5-A, 10th edition. This total elongation shall 1. Mazil)~za>~-stless theory: This theory proposes t h a t
be measured by a suitable extensometer mounted on the elastic limit will be reached when the inaxi~num
the test specimen. The average of these three tests principal stress equals the elastic liniit under simple
shall be the longitudinal tensile yield strength of the tension o r compression.
steel. 2. Maxi?)tz~?)~-stra:i1~ theovy: This theory states t h a t
yielding will occur when the maximum unit strain
9. Recorded Data becomes equal to the unit strain a t which yielding
' occurs 111 simple t e n s ~ o no r conipression.
The following data shall be reported f o r each collapse 3. Maxintzint-shear t l ~ e o r y :This theory assumes t h a t
test : yielding s t a r t s when the nlaximuin shear stress becomes
Make and type of casing. equal to the maximum shear stress a t the yield point
A P I grade. in a simple tensile test.
A P I nominal size. 4 . Mazintzi?)~-strc~in-energy theo?y: This theory as-
A P I nominal weight. sumes t h a t the energy which may be stored in a
Collapse pressure. material attains a nlasimuin value independent of the
Average outside diameter. value of t h e applied stress.
Average- wall thickness. These theories have been discussed in detail in most
Average - D ratio. of the standard books on elasticity and strength of
t
materials. The two theories which might be considered
~ o n ~ i t u d i n " atensile
1 yield strength.
a s being applicable to this problem a r e those of
inaxirnum shear and maximum strain energy. The
10. Deformation before Collapse masin~um-shear theory is-for t h e conditions being
It may be desirable in soine collapse tests to measure considered-conservative, a s compared with the strain-
the deformation of the test specimen before collapse energy theory. I t has been used in the -4SME Code for
occurs. This may be done by filling t h e inside of the Design of Trnns?,tission S l ~ a f t i n g f, o r combining bend-
test specimen with fluid, and measuring the volume ing and tension stresses. If i t were applied to graduated
displaced from the inside a s the pressure is increased casing strings, i t would require heavier pipe than t h a t
b y t h e rise of the fluid in a calibrated capillary tube called for by the maximuni-strain-energy theory, and
connected to t h e inside of the specimen. When this, practically would eliminate all of the advantages which
is done, provision shall be made so that there can be inight be obtained from the use of graduated strings.
110 pressure buildup by the fluid on the inside of the . The maximum-strain-energy theory appears to be
specin~ento resist collapse. the most logical f o r this problem. It h a s been investi-
Maximum-Strain-Energy Curve.
FIG. 1
SETTINGDEP.T.HSFOR CASING 137
gated in detail by those interested i n materials; and the joint strength became the most important considera-
extensive tests, which have been made by Edwards and tion was when ~t became necessary to pull a casing
MillerZ4 on t h e casing problem, indicate t h a t i t i s string. Now, however, in the design of a long graduated
applicable. Sufficient data a r e not available to show casing string, the joint strength must be accorded the
clearly and conclusively what the final relation, based same careful attention a s the collapse strength because
on tests, wi!l be. Investigations in the field of strength of the effective weight of the casing, the effect of tem-
of materials have shown t h a t t h e following theoretical perature, and the residual load remaining a f t e r ce-
equation applies reasonably well: menting. Failure to allow properly for any one of these
factors in a deep well easlly lnay result in a parted
S"S,?+StZ+2 /.ISt S,. ........... (3) casing string. Although the setting depths in tension
TVl~ere: that were used for many years were arrired a t after
p=Poisson's ratio.
St=tensile or longitudinal stress.
S, =compressive or collapse stress.
S=yield stress in simple tension.
Using a value of p=0.26, this equation reduces to:
S'=S,'+Sta+0.52 St S, ......... - (4)
Fig. 1 shows a plot of the per cent of normal collapse
pressure against the ratio of longitudinal tensile stress
to minimuin yield stress in tension.
The longitudinal stress, a s caused by the weight of
the pipe, can be obtalned by the equation:
WL
A ................. (5)
St= -
Where:
W=welght of pipe, Ib per ft.
L=length of pipe, ft.
A = a r e a of cross-section of pipe, sq in.
As a n apl~rosimation, neglecting the effect of the
e s t r a coupllng weight :
W=12 6A
WILB?.~
:
6=density of steel (0.2833 lb per cu in)
and
A P I cooperative investigation,' and on electric-weld cas- the agreeinent was quite good, and indicates t h a t the
ing by Crawford.' Certain relevant comparisons may average values f o r joint strength in Bulletilz 5-C-2 f o r
be made from these investigations, and certain facts casing sizes up to and including 133-in. a r e suitable
were established which were used in the calculation of for the computation of the temporary i n i n i ~ n u njoint
~
minimual joint strengths. I n Table 3 a comparison is strength.
made between the average joint strengths found in the One purpose of both investigations was to determine
two investigations and the average joint strengths f o r how casing with the round-thread form (new A P I
corresponding sizes in Bzilletin 5-C-2, which con~prises standard thread) conipared in joint strength with cas-
the most recent information compiled by the Manufac- ing having the truncated sharp thread (old A P I
turers' Subcommittee.6 Only joints with short threads standard thread). This comparison is suininarized in
TABLE 3
Con~parisonof Average Pullout-Test Results with APl Average Joint Strengtl~
(Short Threads and Coupling)
Noininal Dimensions
Weight
P e r Foot
Average Joint Strength (Pounds)
(Threads and Wall
Size Coupling) Thickness API Seamless * E1ectric;Weld t
(Inches) (Pounds) (Inches) Grade API Casing Casing
59 14.00 0.244 H-40 174,000 214,500 178,300
54 17.00 0.304 H-40 220,000 ...... 241,500
and coupling were considered in this comparison. Tests Table 4, and is expressed a s the ratio of average joint
were made on 8 groups of seamless casing; and only strengths corrected to the specified n ~ i n i n ~ u i yield
n
one group (7-in. 20-lb-per-ft grade J-55) had a n aver- strength. The ratios were all very close t o unity, a n d
age below t h a t given in Biclletix 5-C-2. Tests were inade the d~stributionwas almost perfectly uniform-thereby
on 14 groups of electric-weld casing, and the averages justifying the conclusion that there is no detectable
of 8 of these groups were lower than the values given effect of thread form on joint strength. Therefore, in
in Bttlleti?~5-C-2. The differences were negligible, es- analyzing the pullout-test data, the two thread forms
cept f o r the 3 larger sizes-which gave averages of mere considered identical.
approximately 93 per cent of the figures published in Both investigations made a limited number of tests
Btilletin 5-C-2. All of the low values were for grade to develop the effect of long threads and coupling on
J-55 casing, and only two groups of this grade-61-in. joint strength. A t the present time grade 5-55 A P I
26-lb-per-ft and 8P-in. 36-lb-per-ft-gave results t h a t casing is made either with short o r long threads and
equalled o r exceeded the published values. I n general coupling, but grade N-80 A P I casing is inade only with
long threads and coupling. I n Table 5 average results based on the results of 111 tests in addition to the 48
from the two investigations a r e compared with the tests included in the A P I cooperative investigation.
values taken from Bulletin 5-C-2. There were two These equations a r e included here for convenient
groups of tests on seamless casing, and three groups of reference :
tests on electric-weld casing. The two groups of seam- Grade H-40, short threads and coupling:
less casing gave averages higher than the published
values, but all three groups of electric-weld casing P=72.5 (33.71-d) (-t-0.071 +24.45)a1. . . . . (8)
gave averages considerably below the published values. Grade 5-55, short threads and coupling:
Apparently, the average values f o r joint strength of
grade 5-55 casing with long threads and coupling a r e
not representative f o r electric-weld casing. The Bulletin
~=96.5(33.71-d) (& +24.45)a1. . . .. ( 9 )
5-C-2 averages, however, probably may be used to Grade 5-55, long threads and coupling:
compute minimum joint strengths for both types of
casing, a s Crawford' showed t h a t the standard devia- P=l59(25.58-d) (& +24.45
TABLE 4
Comparison of Strength of Joints Threaded with Sharp Threads and Round Threads
(Short Threads and Coupling)
Nominal Dimensions
A
/ Ratio of Corrected Average
Weight Joint Strength : Sharp
P e r Foot Threads to Round Threads
(Threads and Wall
Size Coupling) Thickness API Seamless * Electric-Weld t.
(Inches) (Pounds) (Inches) Grade Casing Casing
5& 14.00 0.244 H-40 1.09 .. .
58 17.00 0.304 H-40 ... 1.01
A P I cooperntivc inrrstigation.'
? Crawford's investigation.'
tion of pullout tests-based on values corrected to the Grade N-80, long threads and coupling:
minimum yield strength-is much less f o r the electric-
weld casing than i t is for the seamless casing. ~ = 1 8 E i(25.58-d) ( t-:.071
--- +24.45)a,. ..
Table 6 summarizes the data from the two investiga-
tions so a s to show the ratio of strength of joints with In these equations :
long threads and coupling to t h a t of joints with short P=load a t parting of joint, lb.
threads and coupling. Only one size of casing-7-in. d = actual outside diameter of casing, in.
30-lb-per-ft-was common to both investigations, and t =wall thickness of casing, in.
f o r this size the agreement was close. Regardless of a , = n e t area of metal a t gage point, sq in.
this, the conclusion reached by Crawford,' t h a t "the
use of long threads and coupling hardly appears war- ~~~~~~~~i~~ of ~ i joint strength
~ i
ranted on any but the heaviest weights of 5-55 casing,"
is not in agreement with the previous recommendation Crawford ' proposed that the minimum joint strengths
of the Manufacturers' Subcommittee '-which recom- be determined by the statistical analysis of the test
mendation proposed formulas f o r calculation of the data in a manner quite similar to t h a t used for the
strength of joints f o r round-thread casing t h a t were collapse-test data. When 3 times the standard devia-
DEPTHS FOR CASING
SETTING 141
tion w a s used, the minlinum values ranged approsi- the statistical inethod is applied t o data obtainecl on a
mately from 70 to 75 per cent of the average joint- few tests from each of a l a r g e number of groups. I n
strength values in B z ~ l l e t i )5-C-2.
~ I n a discussion of such cases t h e requisite controlled conditions do not
Crawford's paper, Clinedinst stated that, when a more prevail.
direct inetl~odof analysis w a s applied to the test data The desirability of arriving a t inininluin values froill
for seamless pipe, the minilnuin values amounted to 72.8 a large number of actual tests results, which h a s been
per cent of the averages. The statistical method is, emphasized in the discussion of collapse pressure, caa-
however, subject to the same criticisms and objections, not be contested. Therefore, this inethod w a s chosen
when used f o r this purpose, a s previously mere cited also for the selection of the proposed. minimum joint
against i t when applied to the determination of inini- strengths which a r e given in Tahle 8. The test re-
inuin collapse pressures. The following is a n esample sults on A P I casing sizes from t,he A P I cooperative
TABLE 5
Con~parisonof Average Pullout-Test Results with API Average Joint Strength
(Long Threads and Coupling)
Nominal Din~ensions
Weight
P e r Foot Average Joint Strength (Pounds)
(Threads and Wall
Size Coupling) Thickness API Seamless * Electric-Weld .i.
(Inches) (Pounds) (Inches) Grade API Casing Casing
GI 26.00 0.385 5-55
7 24.00 0.332 J-55
7 30.00 0.423 5-55
* A P I cooperative investigation
t Cmwfor~l'si n ~ ~ s t i g a t i u7 n
TABLE 6
Con~parisonof Corrected Average Joint Strengths for Short and Long Threads and Coupling
Nominal Dimensions
Ratio of Corrected Average
Weight Joint Strength: Long
P e r Foot Threads to Short Threads
(Threads and Wall
Size Coupling) Thickness API Seamless Elect~lc-Weld
(Inches) (Pounds) (Inches) Grade Casing Casing
61 26.00 0.385 J-55 ... 1.02
of the artificially low ininimunl values t h a t may result investigation and Crawford's investigation were used
from the grouping together of a few tests on nlany sizes for this purpose. The total number of tests w a s 145.
of casing, when the method suggested by Cramford is I n line ~ v i t hthe preceding discussion, no distinction was
applied. Crawford's data include a group of 18 tests made between round and truncated sharp threads, nor
on 55-in. 17-lb-per-ft grade 5-55 casing, and a group of between seamless and electric-weld casing: all were
1 8 tests on 7-in. 24-lb-per-ft grade J-55 casing. When considered a s equivalent. The ratio of t h e lowest in-
the ininilnuill was computed by taking 3 times the staad- divlclual test result f o r each size to the average value
a r d deviation based on the actual pullout loads, the in Bt~lleti?~ 5-C-2 was determined. These values a r e
value was found to be 91 per cent of t h e average value given in Tahle 7. The lowest ratio found w a s 91 per
in Bztlleti?~5-C-2 f o r the 51-in. casing, and 89 per cent cent of the average. I11 view of the f a c t t h a t compara-
f o r the 7-in. casing-n~iniinum values f o r both sizes tively few results representative of present practices
of casing f a r higher than the approsimate 75 per cent - a r e available, i t was considered t h a t 85 per cent of t h e
found by Crawford for all of the tests. It seems average joint-strength values in Bzilleti~z 5-C-2 consti-
quite clear t h a t a n unduly large penalty results when tuted a safe ancl conservative ratio f o r the selection of
the temporary miniinurn values. I t is thought quite the setting depth for such large casing is inuch lower
probable that, when a sufficient nunlber of test results in collapse strength than in joint strength.
have been obtained, it will be found that this ratio Minimum joint strengths for grade N-80 casing with
may be increased safely to 90 per cent. I t was neces- short threads and coupling were included also in
sary to inodify this procedure for the sizes of casing Table 8, a s i t is anticipated that there will be a need for
larger than 131-in., a s the values in B l t l l e t i ~ 5-C-2
~ this casing a t the bottom of combination strings, where
were obviously f a r too high. The averages for these the tension load is low. These minimums were com-
large sizes were calculated for grade 5-55 casing ac- puted from the corresponding grade 5-55 values by mul-
cording to equation (9), which constitutes an estra- tiplying the latter by the ratio of the constants in
polation, a s no tests have been made on the large casing
sizes. These values then were used to calculate the equations (10) and (11) (g), which represents the
corresponding average values for the grade F-25 casing difference between the t ~ o ' ~ r a d e s .
TABLE 7
Ratio of Measured Mi~~imonl-Joi~~t-Strength
to API Average-Joint-Strength Values
Nominal Dimensions Ratio :
API Miniinuln
-
Observed
Weight Average Observed Minilnuin
Per Foot Joint Joint to API
(Threads and Strength Strength Average
Size Coupling) Number of (Thousands of Pounds) Joint
(Inches) (Pounds) Grade Tests Strength
H-40 11
H-40 6
H-40 8
H-40 14
H-40 3
by first correcting them to grade H-40 values, accord- STANDARD JOINT-STRENGTH TEST
ing to the ratio of constants in ecluations (8) and (9)
(g), and then making a further correction to re-
There is a definite need for results of many more
joint-strength tests, particularly on the sizes of casing
used for oil strings in deep wells. The minimum joint-
duce these averages from grade H-40 to grade F-25.
In the absence of physical test data, the average yield strength values given in Table 9 were based on too
strength for grade F-25 was assuined to be 37,000 psi; few tests, and it is probable that higher nliniinunl
so that the average joint-strength values for grade values could have been proposed if there had been more
F-25 were 74 per .cent of the computed grade H-40 data on which to base them. Study of the figures either
. values, for which the steel has an average yield strength in Table 9 or in B t ~ l l e t i ,5-C-2
~ will show improbable
.of 50,000 psi. Then, as above, the ininiinuin values were values, and these should be corrected. This is true par-
taken a s 85 per cent of the average values. No doubt ticularly when a comparison is made of joint strengths
.need enter into the use of these nlinimuin values, a s for short threads and coupling with those for long
:S FOR CASING , 143
machine shall not esceed 3 in. per min. Longitudinal tensile yield strength of steel. i .
BURSTING STRENGTH I thereby aid progress in drilling to the pay zone; 2,
to pre;ent contai,~inationof fresh water in upper sands,
The bursting strength of casing had no practical and exclude water from t h e producing formations; 3,
impo.rtance until recently, when squeeze-cementing prac- to confine production of both oil and g a s to the well
tices were introduced, abnormal' formation pressures bore, and prevent waste and migration into
were encountered, and the use of graduated casing rocks overlying the productive horizon; 4, to provide
strings-which sometimes u t i l ~ z especial high-strength a means f o r control of pressure; and, 5, to facilitate
joints-became common practice. running tools when cleaning out o r reworking the well
Until recently t h e bursting strength of casing was and installing sub-surface ecluipment, when production
calculated according to Barlow's formula: is obtained by artificial-lift methods.
To perform these functions satisfactorily, t h e joints
with which the individual lengths of casing a r e con-
Wlze,re: nected together must be of a type t h a t will resist
P=bursting pressure, psi. leakage of oil, water, and g a s ; and the casing must
S=tensile strength of the steel, psi. have sufficient strength to resist the applied forces.
t=wall thickness, in. Also, the bore of t h e hole, o r diameter of the casing
D =outside diameter, in. w ~ t hwhich the well is completed, inust be of the proper
size.
I t is not necessary t h a t the pipe rupture or p a r t to
cause failure in a casing string. When t h e stress im- Hole or Casing Size
posed exceeds the yield strength of the steel, the re-
Several years ago drilling and production practices
sulting permanent deformation should be regarded a s
were such a s to justify completing a well with large-
co~lstitutinga failure of t h e casing. Therefore, work-
size casing; so that, when i t no longer was possible
ing stresses never should be allowed to exceed the yield
to obtain production in paying quantities, the well
strength of the steel. With this reasoning in mind,
could be "worked over" and drilled to a deeper horizon.
when ~t became desirable to include bursting strength
The finding of, and drilling for, oil a t g r e a t depth h a s
in the setting-depth tables, t h e values given in Bzclletin
increased production costs considerably, and every effort
5-C-2 were computed by substituting the lninin~unlyield
must be made to keep individual well-completion costs
strength f o r the minimum tensile strength in Barlow's
a t a minimum. This, together with the present re-
formula. These values were based on nominal wall
strictions regarding field allomables and the consequent
thickness and, consequently, do not make allowance f o r
longer payout and productive life of the field, no longer
the permissible variation of 121 per cent in wall thick-
makes i t desirable, from either a n econon~icalor practi-
ness. As i t recently was dec~dedt h a t i t is impracticable
cal viewpoint, to complete a well with larger size casing
to reduce this tolerance on wall thickness, for absolute
than is ~nechanicallynecessary to recover the reserves
safety i t is necessary to calculate the bursting strength of the surrounding area. I n many fields in which pro-
f o r rnininlum wall thickness. This basis has been used
duction is found a t two or more widely separated hori-
to arrive a t the bursting yield-strength values given
zons, i t is now the practice to drill individual wells into
in Table 8, which a r e 878 per cent of the corresponding
each of the several pay zones a t a minimum cost, rather
values in Bztlleti?~5-C-2. As the minin~urnwall thick-
than to delay drilling to a lower pay until t h e wells
ness occurs very rarely, i t was not considered neces-
drllled into a n upper zone have been abandoned.
s a r y to make any further correction to allow f o r
This change in 011-field development methods h a s re-
deviation t h a t naturally is to be expected. These values
sulted in greater attention being given to casing pro-
probably a r e too conservative, because they ]nay be
grams and the size of t h e hole with which a well is
exceeded approxiniately by the ratio of the tensile
to be completed. The trend is to finish a well with t h e
strength to the yield 'strength of the.stee1 before t h e
smallest size casing t h a t is practical, a f t e r all factors
casing will rupture. Ultimately, when test data a r e
involved have been analyzed carefully."
available, i t will be,determined how much these values
The size of the hole o r casing with which to coinplete
may be increased; b u t now there is no basis on which
a well is dependent on the nature and extent of the
to arrive a t safe higher mimmum values.
remedial work t h a t will be necessary in completing o r
I t is not feasible a t this time to propose a standard
working over the well, and the size and type of sub-
nlethod f o r b u r s t ~ n gtests; but, a s the test data a r e
surface equipment to be installed when production is
accumulated, the testing procedure should be such t h a t
obtained by artificial-lift methods. The several bottom-
it will be possible to determine the pressure a t the hole well-completion methods now employed will result
yield strength, a s well a s t h e pressure to cause rupture.
in greater o r less working clearance inside the oil string
I n details common to both collapse and burst testing,. o r liner? depending on t h e nlethod used; and, conse-
t h e burst-test procedure should agree with t h a t adopted
quently, the completion nlethod must be considered when
f o r the standard collapse test. selecting the proper hole o r casing size.
* External upset.
TABLE O C o t ~ t i l ~ u e d
3 9 10 11 12 13 16 17 18
Grade 5-66 Grade N-80
I
h'onilnal D~lnenslons Grade F-25 Grade H-40 Joint Strength Jolrlt Strength
h (1,000 Lb) (1,000 Lb)
n
0
bD ta
m 9
2~ to-
.-
- .-
a -a
Lm 0 a
zg U S
'
0 a
h U1
z2 90-2 -0 -
v 3 a ffl S
* External upset.
of each string with which the well is to be cased, also previously has been pointed out in this paper, this
the minimum allowable clearance between the casing practice of using average, rather than minimum, prop-
and open hole t h a t the operator's experience has proved erties to establish safe setting depths is not sound
to him to be satisfactory, the first step to be taken i n engineering practice.
the design of a casing program is the selection of It is not the purpose of this paper to reconln~endt h e
suitable factors of safety, based on the minimum factor of safety t h a t should be used; because this i s
properties of the casing. a matter t h a t must be decided by the individual user,
When 3,000 f t to 4,000 f t was the approximate limit in t h a t i t is related to the particular practice followed,
of depth reached with the drill, the size and grade of and is a responsib~lity t h a t the user must assume.
casing used were such t h a t factors of safety in collapse, However, in using the nlinilnum values in Table 8 a s
tenslon, and burst were high in comparison to those the basis f o r design, factors of safety t h a t can be
now in effect; and, consequently, the user did not heed accepted a s conforming to good practice, and which in
the fact t h a t published setting depths f o r casing repre- the past have been based 011 the average properties of
sented average properties of the pipe. casing ( a s shown in Bulletin 5 4 - 2 ) can be lowered,
I n drilling deeper wells, and wlth greater attention without changing the casing programs now used, by
being given to the econonlics of casing design, so-called inultiplying the factor of safety on average properties
factors of safety based 011 average properties have by the proper correction factor-which, f o r collapse,
been lowered unduly in some instances; and, a s a result, is 0.75; for joint strength is 0.85; and f o r bursting
loads have been imposed on casing in excess of i t s yield strength is 0.875.
TABLE 9
Oil S~ringsin Some Gulf Coast Fields
Factor of Safety *
P
Collapse
Depth Sizes Weight Average Ultinlate
(Feet) (Inches) (Lb P e r F t ) Grade Average -
; Strength of Joint
* Compr~ted on t h e basis of t h e published values In B ~ r l l e t i ~5l - G ? A pressure gradient of 50 psi per 100 f t uf depth w a s
assumed In calculating f a c t o r of safety i n collapse.
strength limitations, and have caused failure. This h a s External Pressure or Collapse
been noticeably true in the case of external pressures
sometimes applied to casing during the operation of After suitable factors of safety have been chosen,
squeezing cement into a forn~ation." the next consideration is the selection of the proper
Table 9 shows the depth to which oil strings of grade and weight of casing to resist external pressure.
various sizes have been r u n in some fields in the Gulf When the casing string is landed and cemented, i t
Coast, and the unusually low factors of safety in is subjected to balanced external and ~ n t e r n a l pres-
collapse t h a t have been used-particularly if it is sures equal to those of a column of drilling fluid of
assumed t h a t the strings a t some time will be void of the density required to counteract the maximum forma-
internal pressure. tion pressure that is encountered. The pressure exerted
The reason, probably, f o r the use of such low factors by a particular weight of drilling mud a t a given depth
of safety is t h a t i t has not been recognized generally, can be obtained directly from Fig. 4 to 6, inclusive.
nor understood until recently, t h a t there can be a n Fig. 4 is prepared in accordance with the mud-weight
appreciable difference between the average and inini- unit of pounds per gallon, commonly used in the Mid
mum collapse and joint strength of casing, and there Continent and Southwestern Districts; Fig. 5 with the
h a s been no acceptable information available on which mud-welght unit of pounds per cubic foot, co~ninonly
to establish a factor of safety on minimun~values. used i n the California District; and Fig. 6 with a unit
Factors of 'safety now coininonly used in designing of pounds pressure per 100 f t of depth, a s proposed
casing strings evolved a s the result of experience on f o r general use in A P I code 29 on drilling mud."
the p a r t of the user in making a large number of Surface and intermediate strings a r e generally not
satisfactory casing installations. Average properties subject to collapse failure, because pressure outside
of the casing, and present accepted factors of safety, the casing is opposed by a n intellla1 pressure equal t o
were the basis f o r design calculations. However, as t h a t of the inside colun~nof mud fluid sealed off by the
Pressure Resulting from Mud Column.
FIG. 4
Pressure Resulting from Mud Column.
FIG. 5
cenlent t h a t is placed around the shoe of the nest- suspended in air, in salt water, and in drilling muds
smaller-size string. However, if a n lnternlediate string of various weights. The equation used f o r Fig. 8 was:
a t some time may be used a s a n oil string, or there is e=
a possibility of the inside fluid level being lowered as .
[ 8 - 2 ~ ( 1 - ~ ) ] . . . . . . . . (14)
the result of lost circulation, then collapse becomes a Whwe :
factor in design, a n d must be considered. e=stretch, in.
Cement placed behind the casing will afford some L=length of string, ft.
protection against collapse; however, i t is not certain E=modulus of elasticity 130 (10') 1.
just how completely the pipe will be encased, nor can d=density of steel (0.2833 lb per cu in.).
t h e additional strength which is afforded t h e casing w=density of flotant, lb pey cu in. (equivalent to
be determined. Therefore, i t is custoinary with most specific gravity times 0.03613).
operators t o disregard any support t h a t is given the ~ = P o i s s o n ' s ratio (0.28).
pipe from this source except, perhaps, in the case of
Joint Strength
short strings t h a t a r e ce~nentedto the surface.
I n selecting a grade and weight of casing t o resist
If the casing is not cemented properly above the shoe,
possible failure by pulling out a t t h e joint, i t is t h e
i t will be subjected to a collapse pressure in the event
usual practice to disregard the buoyant effect of the
a cement retainer is used f o r squeezing cement into
drilling mud, and to assume in design calculations t h a t
and against the pay forniation under extremely high
the casing string is suspended in air. This provides
pressure f o r the purpose of shutting off water o r g a s ; f o r a n additional load t h a t will be imposed on the pipe
the squeeze pressure will be transmitted to the outside joints a s a result of frlction between the outside surface
of the casing a t a point above the retainer where the of t h e pipe and t h e wall of the'hole, ~f t h e string is
external pressure inay be critical. This condition, with111 raised a t a n y t ~ m e ;e.g., t h e weight of a casing string
the li~nitationof the bursting strength of the casing, t h a t is suspended in 10-lb-per-gal drilling mud is ap-
can be offset by applying pump pressure inside the prosinlately 0.85 of i t s weight 111 a i r ; and, therefore,
casing. a load equal to 1 5 per cent of t h e weight of t h e string
During the life of the well a decline in bottom-hole in air--or approsimately 1 8 per cent of its weight in
pressure will cause a correspond~ngreduction of pres- mud-could be applied in overconling friction before
sure ~ n s i d ethe oil string, and in the later life the inside the load f o r which t,he strength of the joint was calcu-
fluid level may be lowered even to the bottolll of t h e lated is exceeded. To reduce the load on the pipe joints
string. A condition of collapse, therefore, will be and r i g ecluipment when running casing, i t is custon~ary
present a t some stage a s a result of a greater pressure to install a float collar a t t h e bottom of the string,
gradient o u t s ~ d e the casing. To insure against a and float the pipe into t h e hole; however, a f t e r t h e
possible failure, therefore, the m a x i m ~ u npressure dif- casing point is reached and circulation is established,
ference t h a t inay be imposed a t all points of the string the load on the joints will be t h a t of the weight of
inust be determinecl; a n d a grade and weight of casing, t h e suspended string c o r ~ e c t e df o r buoyancy.
having a n ~ i n ~ m u icollal~se
n strength greater than this
pressure by a n amount t h a t will provide f o r t h e factor Release of Tetlsion
of safety used, may be selected from Table 8. Some operators follow t h e practice of hanging the
casing a t the casinghead before cementing, and not
Stretch Due to Tension releasing thereafter a n y of the initial tension imposed
on the pipe a s a result of i t s weight. Other operators,
When casing is r u n in a well to the required depth, however, release p a r t of t h e tension-although the
i t is subjected to longitudinal o r axial tension because amount t h a t is released varies considerably, and the
of its weight, the stress i n the wall of the pipe causing inethod used apparently does not conslder fully the
a corresponding elongation o r stretch. If the pipe is possibility of the casing being frozen p a r t of t h e way
assumed to be hanging freely in air, i t will stretch down the string. Unless this point is detern!ined, the
a n amount which is expressed by the following relation: tension left in the string cannot be calculated. A method
by which the length of f r e e pipe can be estimated was
proposed by Hayward." It is based o?l the measurement
of the change in stretch t h a t is obtained f o r a pre-
Wlt.ere:
determined amount of tension release. The formula
e z s t r e t c h , in.
used to calculate the length of f r e e pipe, when deter-
81densit.y of the pipe material (0.2833 lb per cu
mined by this method, is:
in. for steel).
L=length of string, ft.
E=modulus of elasticity [30 (lo8)]. ~1'11ere :
Because of the buoyant effect of a drilling fluid, the L=length of free pipe, f t .
a s i a l stress and, therefore, the stretch of the pipe will e ~ s t r e t c h in.
,
not be a s great when immersed in the mud fluid a s w=weight per foot of pipe, lb.
when suspencled In air. Fig. 7 and 8 give the stretch P = amount of tension released, Ib.
of steel pipe of u n i f o r ~ n cross-section when freely E =modulus of elasticity [30 (10') 1.
SETTINGDEPTHS FOR CASING
leeway f o r a possible increase in tension in the pipe ecluivalent to approsin~ately800 f t of 7-in. 28-lb-per-ft
after cementing, a s a result of t h e t e l l d e n c ~for the action. One corn-
casing f o r the pipe to fail by coluim~~
pipe to contract when circulating cold drilling fluid, pany, a t least, has adopted a practice of first determin-
or if there should be escessive espansion of gas in the
use of a bottom-hole choke, and to relieve a nart of the ing the poillt a t which the casing is free, and then
load from the casinghead connections. releasing o r increasing the tension in the pipe, so t h a t
If too much tension is released, the lon~erp a r t of the there will be a predetermined tension stress in the
free portion of the string ]nay be left in compression, pipe a t the bottom of the free portion.
Stretch Due to Tenlperature Change Other Factors
Immediately a f t e r casing h a s been r u n into t h k well h i a d d i t i o n to the forces already mentioned, casing
and the cement h a s set, the mean temperature of the will be subjected to column loading if the pipe is
casing ordinarily will be but slightly less than the released a t the surface prior to nippling up, or when
average of surface-formation and bottom-hole tempera- setting liners. If t h e hole is crooked to a n appreciable
tures; and, because of the increase in temperature of extent, o r the casing i s subjected to lateral earth
the pipe, i t will expand o r stretch a n amount t h a t is pressure-as in the case of shifting formations o r
expressed by t h e following : sudden heaving of unconsolidated sands-a bending
force will be applied which will add to the tension stress
e = 1 2 L a t ................( 16) already imposed 011 the pipe and joints. Shooting of
Where: wells with explosives 23 will induce stresses in casing-
e = stretch due to temperature change, in. which, a s i n the case of bending stresses and those
resulting from column loading, would be difficult, if not
L=length of pipe a t atmospheric temperature, ft.
inlpossible, to evaluate.
a=coefficient of expansion (0.0000069). Wear of casing by drill pipe o r tubing, and also
t = temperature change, deg F. corrosion, a r e factors t h a t sometimes a r e given con-
Fig. 9 shows the stretch in pipe f o r various tempera- sideration and lead to a preference f o r heavier wall
t u r e differences, in degrees fahrenheit. pipe than otherwise would be necessary.
When the well is p u t on production, a n increase in
the temperature of the pipe will result because of Graduated Strings
producing fluid a t a higher temperature than t h a t of
the casing a t the time t h a t i t was cemented. Thls It was the practice until two or three years ago to
increase in temperature, therefore, will cause a decrease use one weight and grade of pipe f o r each of the several
in the tension t h a t is already in the pipe, and con- strings of casing with which a well was oompleted.
ceivably might result in column loading-particularly ~f I n following this practice no advantage is taken of the
savings to be effected In the cost of casing by virtue
too much initial tension in the string was released a t the
',
of the fact t h a t the principal forces applied to casing-
time the casing was landed. which a r e used a s a basis f o r design-change uniformly
along the length of the casing string, and of the further
Bursting Strength fact t h a t over the central portion of the string the
magnitude of these forces is such t h a t pipe of less
As previously pointed out in t h e paper, the bursting weight o r lower strength than is used f o r the end sec-
strength of casing h a s become a n increasingly im- tions can be substituted in nlany cases without lowering
portant factor in the selection of casing strings; be- the overall efficiency of the casing installation.
cause, in drilling to greater depths, abnormal pressures The principal forces with which we have to deal in
approaching, and in some cases equal to, the e a r t h designing a string of casing a r e : 1, external pressure
pressure" a r e encountered. Fig. 10 is a plot of the tending to collapse or crush the pipe; 2, longitudinal
"normal" pressures t h a t occur in t h e Gulf Coastal o r axial tension tending to pull the pipe a p a r t ; and,
region, and of the "abnormal" pressures t h a t have been 3, internal pressure tending to burst t h e pipe.
observed in some of the fields in this area. External pressure is due primarily to the head of
The practice of squeeze cemelltillg may result also fluid sealed off outside of the casing by cementing
in extremely high pressures inside of the casing. For around the casing shoe, and is, therefore, a maxiinum
example, in a casinghead squeeze job high pump pres- a t the bottom and of zero value a t the top of the fluid
colunln. 1t' becomes a factor in the design of casing
sure, together with the weight of a column of cement,
only ~f the difference between the external and internal
may be necessary to force the cement Into the formation,
pressures t h a t may be applled to the casing a t a n y
and the resulting pressure a t the top of the oil string time should become critical from the standpoint of
will be so high t h a t the burstlng strength of the pipe collapse.
well can be the controlling factor in the selection of Axial tension, which is due principally to the weight
t h e , g r a d e and weight of casing to be used. of the suspended casing, is a rnaxinlum a t t h e top of
After cement has been placed around t h e shoe of the the string; and if no tension is released a f t e r cement-
casing string and the well closed in a t the surface, ing, the tension in t h e pipe decreases uniforn~lyuntil
pressure inside the casing may become critical if expan- a t the bottom there is no tension.
sion of the fluid in the casing7as a result of a n increase Internal pressure is the result of forillation pressure
in temperature caused by well tenlperature and t h e or a head of fluid to counteract the formation pressure;
heat t h a t is developed during the setting of the cement also, i t may be due to the weight of a column of drilling
-is restrained. This condition, however, can be con- fluid, together with pump pressure applied a t the
trolled, and a pressure t h a t is within the limitations surface in a n attempt wholly or partly to balance a n
of the casing be maintained by releasing fluid a t t h e external pressure unintentionally applied to casing
surface. when sclueezing cement into and against the formation;
or, a s previously mentioned, i t may be caused by t h e if i t is assumed t h a t a t some tiine during the life of
weight of a coluii~nof cement, together with a pump the well the string will be void of internal pressure,
pressure necessary to force the cement into the forma- a casing having a collapse strength sufficient t o resist
tion. I n all cases the internal pressure is a masiinuin a n external pressure equal to the inasilnum eserted by
a t the bottom of the string, but may be critical a t the column of fluid behind the casing is chosen f o r t h e
various points t h a t nlust be determined by taking lower section of the string. This section of casing
account of the difference between the internal and is extended upward to a point in the hole where the
esternal pressures. esternal pressure will perinit the use of casing of less
To take advantage of t h e unequal distribution of wall thickness o r lower grade and, therefore, of a lower
these forces over the length of the casing string by collapse strength than t h a t of t h e first section.
using what may be called a "graduated string," and A t the p o ~ n tof change, however, i t is not sufficient
to consider only the collapse strength of the pipe on This method of selecting sections of casing on the
the basis of external pressure alone being applied; basis of the abihty of the pipe to withstand esternal
because, in addition to a n external pressure, t h e casing presS'ui-e is continued until a point in the string is
is subjected to asial tension due principally to t h e reached where a s i a l tension, rather than collapse o r
weight of the sect1011 of casing below the point of external pressure, becomes the predominant factor in
change. This introduces a condition of biaxial stress the design of the string. F r o ~ l l this point on the
which, if the casing is of such proportions that plastic strength of the casing in tension, and particularly the
cpllapse \ y u l d be the mode of failure? will result in strength of the joints with which the individual lengths
the pipe Paving a lower collapse strength than if a r e connected together, may impose a restriction re:
external pressure only were applied. Therefore, i t is garding the use of pipe of a lower grade o r less wall
necessary to deternline from Fig. 1 w h a t the effective thickness. I n the case of long and heavy strings, i t
collapse pressure f o r the casing will be under the par- ]nay necessitate the use of a better grade o r thicker
ticular condition of stress, before the point of change wall plpe, or perhaps a change in the type of joint,
can be established. Having established the point of safely to resist the tension load a t the top of the string.
change, the second section of casing then is estendecl I n the previous discussion, only the effect of external
upward until a point in the hole is reached where the pressure and tension was considered. It has been men-
esternal pressure is such a s again will permit the use tioned t h a t internal pressure can be a controlling factor
of a still weaker section of pipe from the standpoint in the design of casing, and, therefore, i t is necessary
of collapse strength-taking into account, a s before, t h a t the bursting strength of the pipe also be considered
the effect of a s i a l tension. in the final selection of casing size.
Table 10 is a compilation of d a t a showing the relative ties deviating less from average properties-as only in
costs of three possible casing strings f o r a 10,000-ft this way is i t possible to obtain higher minilnun1 values
well. A saving of $1,177--or approxinlately 13.5 per without increase of average values. Any advance t h a t
cent-in the mill cost of casing i s effected by using a can be made in this direction.wil1 be very inuch worth-
graduated string, a s in case 3, in preference to a while.
string of only one weight and grade, a s in case I . The The method t h a t is preferred by the authors, and i s '
graduated string is also 15,000 Ib-or 9 per cent- used here f o r the determination of the mininlum per-
lighter in weight, and results in a further saving when formance properties, brings out very clearly the Im-
freight charges a r e considered. portance attached to results of tests and the,need f o r
The string shown a s case 2 is of the same weight many more test data. The practices now used by most
per foot throughout its length, and was r u n In a well of the pipe n~anufacturersand by several of, the large
i n one of the Louisiana fields. This string is $904 lower purchasers of casing insure t h a t a large number of
in cost than t h a t of case 1. results of collapse tests on present grades of casing
TABLE 1 0
Co~nparisonof Cost of 10,000-Ft Strings of 5%-In.-OD Casing
3lin1rnuni
Collapse Pressure
(PSI) If ~ n ~ m u m Factor of Safety
& Aflnirnum Bursting ,
We~ght(Lb Per F t ) Corrected Joint Yield Bursting
Length API No for Strength Strength. Joint T~eld Mill Price Cost of
(Feet) Nomlnal Calculated Grade Tens~on Tens~on (1,000 Lb) (PSI) Collapse Strength Strength (Per 100 F t ) Strlng
CASE 1
10,000 17.00 17.25 N-80 5,890 .... 340 7,740 1.18 1.97 1.98 $88.12 $8,812.00
CASE 2
CASE 3
* Long t h r e a d s a n d coupling.
Note: F a c t o r of safety on b u r s t ~ n gyield strengtll based on a srlrfnce pressure of 3.90(1 psi, from Fig. 9, curve No. 4.
E x t e r n a l pressure calculated on the b a s ~ sof a pressure gradient of 50 lb per 100 f t of depth.
2. The mini~numperforinallce properties should be normal pressures; and K. Lindley, of the Gulf 011
determined directly from the results of tests. Corporation, f o r preparation of p a r t of the material.
3. The A P I pipe specifications f o r casing s h k d be To P. D. Foote, of Gulf Research and Development
revised so t h a t grade classification is based on minimum Company; J. R. Suman, of Huinble Oil and Refining
limits specified f o r the performance properties, collapse' Company; and P. H. Bohart, of Gulf Oil Corporation,
pressure, joint strength, and bursting yield strength. f o r permissioti to prepare and publish the paper.
Additional minilnun1 requirements to insure adequate
ductility also should be included. BIBLIOGRAPHY
4. Temporary minimuin performance properties a r e
R T S t e w a r t . "Collapsing Pressures of Besscmer-Steel Lap-
proposed a s follows: ~reliled-Steel Tubes," Trtr~rs d11i. Soc dlccll E ~ r g r s 27, 750
(1'30G).
Collapse 'pressure-75 per cent of the "average - R T Stewart. "Col1al)siug Pressures of Laln~elded-Steel
Tubes." T r a ~ i s d l t t Soc Ueclr Lzt~grs 29, 1 2 3 (1907)
collapsing pressures," in Bzilletin 5 - c - 2 ; 3d.STLlI dla1~1ta1OIL I'iesctltatro~r of D a t a , Amerlcan Soclety
f o r Testing Alaterlals P h ~ l a d e l p h l a , Pa. (1933)
Joint strength-85 per cent of the average "Report of t h e ~ A n i m i t t e e on t h e S t n n d a r d i z a t ~ o n of Oil-
"ultimate-strength-of-joint" values, in Bz~llet.in Country T u b u l a r Goods," Proc. A P I 19 [IV] (Productiotr Bull. '
NO 228) 1 3 0 (1935).
54-2; "PI Li~uisro~rof Protlrtctroti Btrll Xo. 5-C-0. "Itrforlllation
o n C o l l a p s ~ ~ zPresslircs
g a ~ i dS c t t i n g Depths f o r Ctrsrng," 2nd
Bursting yield strength-87.5 per cent of ithe values edn , Mar ( 1 9 4 0 ) .
0 \Y A1 lei-ame, "Casiag Setti?g Depths a r e n o t Bssured by
f o r "internal pressure a t ininimuin yield Pllyslcal P r o l ~ e r t i e aof t h e Steel. Drr1lr11.gn~rtlProtlrtctto~lPruc-
strength," in Bulletin 5-C-2. ,tree, $ 2 2 A n i e r ~ c a n Petru1t.11111 Institute. New York (193S)
T. J. 'Cmwford, "Joint and Collapse S t r e n g t h of AIJI Grades
H-40 luld 5-55 Electrlci~lly-~velded Casing." Uvrl111ig a ~ t t l P r o -
5. Safety factors calculated from t h e proposed mini- t l r ~ c t i u ~I'rcrctlce,
r 43'7 (1930)
Inurn performance properties may be used with confi- "lieport of t b e Conim~ttve on t h e Standardization of 011-
Country T u b u l a r Goods," P r o c d P 1 201 [IVJ (I'roti1tctiurr Bull.
dence. No 2 3 ) lli (1930)
T. AIcLenn .Jasper a n d J. W nr. S u l l l m n . "The Collapsing
6. Additional test data on the performance properties Strength of T h ~ nTubes," Trans. Ain. Soc Jfecli. Engrs. ( d p -
pltctl LlIcclro~i!cs) 53, l i b (1931)
of casing a r e essential, and should be obtained a s soon I0R. G S t u r m , doctorate thesls submitted t o D e l ~ a r t m e n tof
a s possible, so a s to allow the establishment of perma- Theoretical a n d Applied Alechan~cs, U ~ r r u .Illtnots Uoll. Eng.
(192s)
nent minimum performance properties required f o r t h e U A 1'. Carman "The Coll?l8se of S h o r t T h i n Tubes," Unrv.
IlllllOlS Etig Exp. '&'tit. Uul!: Ao 99 ( 1 L J l i ) .
revision of the A P I pipe specifications. T. h1cLenn Jasper, I ~ n p u r t a n c e of Compress~on Yield
Strength In Establishing S e t t ~ n gDepths for Casing," U r i l l r ~ i g
aitd Procltlctio~rPractlcc, 250 (1939).
ACKNOWLEDGMENT l3 J L I - I o l n l c ~ ~a~ni d~ tA. Nadal ".4 Tlieoretical a n d Experi-
mental d l ) l ~ r o a c l ~to t h e robl lei^ of Collnl~se of L~eep-\\~ell
Casllig," 1Jrrllilrg alld P r o d n c t i o ~ rPracttce, 89:: (19.39).
The authors wish to express their, appreciation f o r I4\V 0. Cllnedinst "-4 Ratlonal Exlwrsslon f o r t h e Critical
Collaysilig Pressure a;f P i l ~ eunder E x t e r n a l l'ressure," Drilling
assistance in t h e preparation of this paper: a n d l'rodrrctio~r Practrce, 383 ( 1 0 3 9 ) .
lS \T7alter C Alain "Combining Bending nnd Hoop Stresses t o
To the Manufacturers' Subcommittee, f o r much of Determine Collal~sin'gPressure of Oil-Country Tubular (;oods,"
the test d a t a ; and in particular i n this regard, to Urillrlrg ant1 Protluctto~rPractrce, 431 (1339).
' B T ~ ~ ~ ~ o s h ea nu dk oh s s e l l s , d p p l l e d E l a s t i c ~ t y , Westinghouse
J. J a y Dunn and W. 0.' Clinedinst, of the National Kiglit School l'ress. E a s t Pittsburgh, P a ( )
l7 E r n e s t K. Parks, "Effect of Varying Well-Bore Diameters
Tube Company; W. M.,Frame, of The National Supply In C ; l l ~ t o r u ~Reservoirs,
a D r i l l ~ ~ ragn d Prodilctro~rPractrce, 259
Company; and E. J. Protin, of the Pittsburgh Steel (1939).
IS I W. A l c o r l ~a n d J U. Tengue, "Bottom-Nole Well-Comple-
Company. tion Aletl~ods I U t h e Gulf Coast," D r t l l i t ~ g (IILIZ Prodrtctiot~
Pmctrce, 30 ( 1 9 3 i l
To P. A. Mills, of the Moody Engineering Company; '9lJau1 IJ Torray, "Selective Esclusion of F l u i d s from IVells,"
D r l l l t ~ r gulrrl P r o d r c c t ~ u ~Pructlcc,
r 206 (10391.
E. F. Kindsvater, of the Phillips Petroleu~nCompany; 'Odl'1 Coflc A'o ?,? " I c ~ c o n t t ~ ~ o r t l cJ'ract~cc
d on Gta(?,dord
T. McL. Jasper and R. F u r r e r , of the A. 0. Smith Frclrl I'roccd~rrc f o r T c s t r ~ r gD r r l l r ~ l gFltltds ( t o r t n t t o c l , 1 s t
eiln , Oct. (19.3SI
Corporation; and J. E. Stillwell, of the American .T T l ~ a y m a r d ,"hletliods f o r Determining How AIuch nf a
Iprozen or Ceruented Culunin of Pipe 1s Free," Drlllb,rg a,td
Petroleun~ It?stitute, f o r collstructive criticism and P r ~ t l ~ c t i oI'ractrcr,
~i 1 6 (1!1:35).
suggestions. -- \V E 1'. d b m h n n ~ . "l;eological Aspects of Deep-Dr~llrng
I'roblen~s." J . 111sl Pt7trolclr~~rl'cclr (London) 23, 164-383
,*a,.,-,
To C. N. Bowers of the Gulf Research and Develop- {J.V.,,,
23 13anl l? Len-is, "Oil-Field Esplos~ves-Tlicir Character-
ment Company, for asselnbly and correlation of the istics a n d Use." UrrIlr~rg a~rcl Prodrcctto~t Prnctlcr, i:3 ( 1 0 3 4 ) .
24 S H E d w a r d s a n d (7 P. Aliller. "Discussion on t h e Effect
test d a t a ; Burton Atkinson, of the Humble Oil and of C o n ~ b ~ n e L
d o n g ~ t u J ~ n : \L
l p a d ~ n ga n d E s t e r n n l I'rvssnre on
the Strcngth 011-Well CRSIII~."Di11111rg a ~ i d P I O ~ I I G ~ I O I ~
Refining Company, for preparation of the data on ah- I'rnct~cc, 483 ( 1 9 3 0 ) .
APPENDIX 1
COLLAPSE-TEST RESULTS
Note: T h e r a l u e s in tlie c o l u h u hendcd "Corrected Collal~se Pressure" arc clusctl-end collnlrse-test results culrected to alien-end col-
1:il~se-test results, a s e s p l a ~ n ~ iinl tlie text
Nonilnal Dimensions
APPENDIX 1-Continued
-
h
- - sP
0 %
Bz
,
Nominal nimpnsions
- -
A
-
0
2
- s
"2
-"Z
b-
=U -
ir;
OP,
A
m
m
s
45
.-
m
.-0
-""a
2
CI
i
B
F
2"
h
L.
w $2
"-
.*,0. 2;- C
.,zc. r2r YL
* =2
a"
0
'Il
a-
- 2%I
;*
n
g p" F-
h - 2
4
. Y ~ C
.za -'6 Ebc
i.F.8 ;
I
5s
A
-
L-
2
'C
-
V "2
n m
5%
L. 2" 5%
0 .?: =;? z q(, oa
2 & ti- ;B a\- L. cm
U
C
2 g- U v
Grade H-40-Continued Grade H-40-Continued
SETTING DEPTHSFOR CASING 163
APPENDIX 1-Continued
Grade H-40-Continued
Grade H-40
Grade 5-55
6 17.00 0.275 21.82 79,500 5,790 5,570
79,200 5,950 5,720
6 20.00 0.324 18.52 76,550 7,510 7,080
79,800 7,550 7,120
61 24.00 0.352 18.82 56,800 4,100 ....
59,100 4,690 . .. .
61,100 4,545 . ...
APPENDIX l-Continued
Nominal Dimensions
APPENDIX 1-Continued
0
L
IL-
O-
=Cm
fi
Nonilnnl Diniens~oris
n
A -m
-+fi
-
A
5
U
2 m Vd
-
a
2 u- c*, i
fi
ii
r.
=-
zz
-;2 222
7
,;- E
-~270 ;-
w.
-n
=z a
c j,
0 D
? fi w
N
< -
:;z 6-
Grade J-55-Continued I Grade J-55-Continued
APPENDIX 1-Continued
----
Nominal Dlrnens~ons Norn~nalDt~nellsions
A w
2
-c
w
n
m
A
- n
w
U
z
ow
-
m
c n, e
m
n
,
=mon,m
U-
w
v w 5 k v w
92.-, 2" 5YI
om 2E
$i
:
&& U
FRAME'S
INVESTIGATION OF SEAMLESSCASINGWITH LOW
RESIDUALINTERNAL STRESS
Grade J-55
5 15.00 0.296 16.89 68,500 7,510 ....
66,700 7,320 ....
69,000 7,310 ....
6% 22.00 0.318 20.83 65,500 5,900 ....
62,600 5,850 ....
64,000 5,700 ....
65,700 5,870 ....
68 29.00 0.432 15.34 67,200 8,990 ....
70,000 8,550 ....
7 17.00 0.231 30.30 74,800 2,560 ....
74,800 2,350 ....
78,500 1,960 ....
75,500 2,000 ....
78,500 2,220 ....
7 20.00 0.272 25.73 67,200 3,770 ....
67,500 3,960 ....
65,800 3,600 ....
67,900 3,680 ....
69,200 3,720 ....
68,400 3,590 ....
71,300 3,150 ....
72,400 3,300 ....
71,900 3,650 ....
74,100 3,180 ....
74,100 3,450 ....
74,200 3,900 ....
73,100 3,840 ....
72,800 3,775 ....
78,000 3,100 ....
75,600 3,600 ....
75,600 3,900 ....
76,500 3,430 ....
7 22.00 0.301 23.25 63,700 5,050 ....
64,900 4,840 ....
63,150 4,950 ....
65,200 4,890 ....
67,200 4,530 ....
65,200 4,650 ....
68,400 4,140 ....
SETTING
DEPTHSFOR CASING 169
APPENDIX 1-Continued
I Noniinnl D~rnens~ons
Nominal D~menmons
z
m %
P h
-
Oc
F
="?
0
0
U
%E
-
a
f=: z2
nrn
=L z:
0-
i3
$0
U
68,300 68,000
67,600 71,200
68,300 70,500
67,600 71,100
68,900 70,200
71,000 71,600
65,600 77,500
66,000 63,700
66,100 70,500
64,600 71,100
68,300 65,200
67,900 63,800
67,400 64,800
67,600 30.00 0.423 16.55 60,600
67,600 61,000
APPENDIX 1- -Continued
N o n ~ ~ n nDimensions
l Nominal D~rnensions
APPENDIX 1-Continued
Grade N-80
6 17.00 0.275 21.82 83,550 5,950 5,720
' 83,050 5,620 5,410
81,800 6,190 5,950
APPENDIX 1-Continued
Nornlnal Dimensions Nominal Dlmensions
Grade N-80
DEPTHSFOR CASING
SETTING
APPENDIX 1-Continued
Nominal Dimensions
-i: 4:-
A
-
, ,
a
I
O w
A
m
C, L
2"
m
-2
-"Z
2.2-
22G
E
01
Zrn
O h
0
.cIw
-
6 2" : B * .-
d ~ 2&
I,
>-
dm
f
=.*;"
m"-
m .$$
5 qJy
zh
g- tE
P'
FRAFVIE'S
INVESTIGATION
OF SEAMLESS
CASING
WITH LOW
Grade J-55-Continued
RESIDUAL
INTERNAL
STRESS 29.70 0.375 20.33 60,200
Grade N-80 60,700
61,000
7 17.00 0.231 30.30 84,700, 2,250 .... 60,500
82,200 2,020 .. . . 58,800
83,200 2,150 . ... 57,400
7 22.00 0.301 23.25 83,500 4,920 .... 57,000
83,500 5,450 5,290 57,800
57,600
7 28.00 0.393 17.81 83,100 8,350 . . .. 57,000
80,200 8,220 ... . 62,400
7 30.00 0.423 16.55 82,400 8,940 .... 61,200
82,500 8,630 .... 62,000
81,300 9,200 . .. . 61,800
81,700 8,410 . .. . 59,600
80,400 8,000 . ...
86,500 9,200 ... . Grade N-80
85,000 9,200 .... 17.00 0.304 18.09 81,600
81,900 8,610 .... 81,600
80,500 8,610 ... . 81,600
85,500 9,400 .... 90,200
85,500 9,500 .... 82,200
83,600
86,500
81,000
Grade J-55 93,750
93,750
92,600
92,600
93,900
93,900
92,400
92,400
92,900
92,900
90,500
90,500
93,800
93,800
APPENDIX 1-Contiknecl
\Val1 .
D Ratlo:
S~ze IVelght Thickness t General I:cnernl Average
(Inches) (Lb Ppr F t ) (Inclres) Ratlo A B C E .i r e r a g e API' .iPI Average
GRADE H-40
GRADE J-55
44 16.00 0.334 14.22 .... .... 9, 290
59 i4.00 0.244 22.54 .... 4. loo ....
52 17.00 0.304 18.09 .... 5. 720 6. 830
6 20.00 0.324 18.52 7. 100 .... ....
6g 24.00 0.352 18.82 4. 460 6. 670 ....
68 26.00 0.385 17.21 6. 050 6. 870 6. 930
68 28.00 0.417 15.83 6.480 .... 8. 370
GW 29.00 0.432 15.34 7. 280 .... ....
7 20.00 0.272 25.73 2. 490 3.050 2. 600
7 24.00 0.332 21.08 .... 4. 690 5. 130
7 26.00 0.362 19.34 5. 880 5. 400 5. 760
7 28.00 0.393 17.81 .... .... 7. 020
7 30.00 0.423 16.55 6.810 6.990 7. 270
78 29.70 0.375 20.33 .... .... 5. 610
78 33.70 0.430 17.73 .... .... 6.810
81 32.00 0.352 24.50 3. 160 3. 150 4. 170
81 36.00 0.400 21.56 3. 600 4. 500 4. 940
88 43.00 0.487 17.71 5. 520 .... ....
88 44.00 0.500 17.25 5. 910 .... ....
9 40.00 0.425 21.18 4. 560 .... ....
3 45.00 0.484 18.60 .... .... 6. 440
APPENDIX 2-Continued
> '
Nominal Dimens~ons Average ~olfapse,Pressure(PSI)
h
Slze Weight
Wall
Thickness
-
D
t General
Ratlo:
Grnernl Average
(Inches) (Lh P e r F t ) t Inches) Ratio A B E Average API API Average
GRADE J-55-Continued
GRADE N-80
.... .... 8,510
5,400 .... ....
7,770 .... ....
.... .... 8,650
.... .... 9,410
4,520 .... ....
5,830 .... 7,120
6,770 .... 7,420
7,820 .... ....
.... .... 8,G90
.... .... ....
.... i .... 6,460
. .,.. .... 7,620
4,610 .... ....
5,380 .... 5,920
6,760 .... 7,120
4,680 .... ....
5,190 .... ....
APPENDIX 3
JOINT-STRENGTH-TEST RESULTS
Nominal Dimensions
Weight
P e r Foot
(Threads Longitudinal
and Wall Tensile Yield Pullout
Size Coupling) Thickness D Strength Load
(Inches) (Pounds) (Inches) -Ratio (Pounds)
t Grade Thread (PSI)
ELECTRIC-WELD CASING
21.95 J-55 Sharp-Short 60,957
59,730
21.95 H-40 Sharp-Short 49,063
22.54 I H-40 Sharp-Short 52,336
53,250
53,696
18.09 H-40 Sharp-Short 52,870
50,700
54,950
APPENDIX 3-Continued
Nominal Dimensions
Weight
P e r Foot
(Threads Longitudinal
and Wall Tensile Yield Pullout
Size Coupling) Thickness D Strength Load
(Inches) (Pounds) (Inches) -Ratio (PSI) (Pounds)
t Grade Thread
ELECTRIC-WELD CASING-Continued
18.09 H-40 Round-Short
APPENDIX 3 C o l l t i n u e d
Nominal Dimensions
Weight
Pe? Foot
(Threads Longitudinal
and Wall Tensile Yield , Pullout
Size Coupling) Thickness D Strength Load
(Inches) (Pounds) (Inches) -Ratio
t Grade Thread (PSI) (Pounds)
ELECTRIC-WELD CASING-Continued
7 24.00 0.332 21.08 H-40 Round-Short
Sharp-Short
Round-Short
Sharp-Short
Round-Short
Sharp-Short
Sharp-Long
Round-Long
Round-Short
5-55 Round-Long
H-40 Round-Short
H-40 Round-Short
J-55 Round-Short
J-55 Round-Short
H-40 Round-Short
APPENDIX 3-Continued
Nominal Di~nensions
Weight
P e r Foot
(Threads Longitudinal
and Wall Tensile Yield Pullout
Size Coupling) Thickness D Strength Load
(Inches) (Pounds) (~nches) t
-Ratio Grade Thread (PSI) (Pounds)
ELECTRIC-WELD CASING-Continued
0.430 31.10 J-55 Round-Short
SEAMLESS CASING
0.244 22.54 B Sharp-Short
B Round-Short
APPENDIX 3-Continued
Nominal Dimensions
Weight
Per Foot
(Threads Longitudinal
and Wall Tensile Yield Pullout
Size Coupling) Thickness D Strength Load
(Inches) (Pounds) (Inches) Ratio Grade Thread (PSI) (Pounds)
SEAMLESS CASING-Continued
25.73 D Sharp-Short
16.55 D Sharp-Short
25.73 D Round-Short
16.55 D Round-Short
16.55 D Sharp-Long
16.55 D Round-Long
27.34 B Sharp-Short
27.34 B Round-Short
25.66 D Sharp-Short
20.40 D Sharp-Short
25.66 D Round-Short
SETTINGDEPTHSFOR CASING 181
APPENDIX 3-Continued
Nominal Dimensions
Weight
P e r Foot
(Threads Longitudinal
and Wall . . Tensile Yield Pullout
Size Coupling) Thickness D Strength Load
(Inches) (Pounds) (Inches) t Grade Thread (PSI) (Pounds)
SEAMLESS CASING-Continued
91 47.60 0.472 20.40 D Round-Short 60,000
60,200
61,100
\ 60,300
91 47.60 0.472 20.40 D Sharp-Long 59,900
58,300
57,000
60,800
9B 47.60 0.472 20.40 D Round-Long 59,600
59,100
58,600
59,200
3. A P I casing specifications should be revised so t h a t Some operators sensed t h a t the actual physical prop-
grade class~fication is based on ininlmum a l j n ~ ~ t s ertles of casing would permit a material increase in the
specified for performance properties : collapse then generally accepted setting depths and, therefore,
strength, joint strength, bursting yield strength. proceeclecl to reduce their relatively h ~ g hsafety factors
Ductility r n ~ m a ~ urequirements
m should be added. through experimentation, which resulted in considerable
4. T e i ~ l p o r a ~m
y ~ m i n u m performance properties a r e advantage. Nevertheless, the usual result has been t h a t
recommended a s follows : the lack of known values and definite knowledge coacern-
ing the properties of casing has caused the use of
a. Collapse strength=75 per cent of average in
heavier walls o r hig!ler grades than actually were re-
Bltlletl)~5-C-2.
cluired, with resultant loss. Consiclering the oil industry
b. Joint strength = 85 per cent of average in Bttlle-
a s a whole, this loss has been tremendous, and is in
fix 5-C-8.
itself the best example of the necessity and importance
c. Bursting yield strength = 87.5 per cent of aver-
of the information given in this paper.
age i n Bttllctzr~5-C-2.
The casing of oil wells is not now and cannot hecoine
5. Safety factors celculated from the proposed mini- a n esact science. I t can be, and is being, brought to a
muni performance properties n ~ a ybe used with basis where definite knowledge of controllable factors
confidence. is available.
6. Additional test data on performance properties of Therefore, i t seems certain t h a t minimum perform-
casing a r e essential. ance values will be generally acceptable after they have
The authors should be co~nplimentecl011 their splendid been agreed upon finally, a s they will provide this illore
paper. definite basis f o r calculation ancl deslgn. The principal
result of establishing these minimums probably is found
in the better control of safety factors in comparison
H. W. Ladd (Stanolind Oil and Gas Company, Tulsa, wlth those now used with average values.
Okla.) (written) : Discussion of the paper under con- The A P I minimum collapse pressures h a r e been found
sideration first calls f o r the statement t h a t the paper by many to be too conservative and impractical; and
col~tainsa coa~pletesuinmation of the very excellent the average values, tllel-efore, have been used to provide
work on physical properties of casing completed during a inore economical basis of application. This practice
the past few years in which the authors individually has been generally successful, but always uncertain-
have taken a n important part. This h a s provided in- demanding the use of safety factors considered high #
formation and conclusions which give a more definite enough to avoid undue risk or a ganlble 011 the unknown
basis for cleterinining a safe and economical applica- probability of failure. The importance of the chances
tion of casing to its use in oil wells. taken increased with depths of the wells cased. The
On this basis, the authors propose to advance further use of minimum values a s proposed in this paper con-
the knowledge through application of aclclitional d a t a templates t h a t a certain quantity of product will fall
which they have cleveloped and data which they recom- below the proposed mimmum, which is 75 per cent under
mend be procured in the future. This is in keeping with the present A P I averages. This undoubtedly is t r u e of
the goal whlch long has been sought by all users of any percentage that might he selected.
casing-which is, briefly, to be able safely to protect I t mould be of great importance to know just what
their well investment in the most econonlical manner. the ratio of the number of such specinlens is to the total
The application of such a principle definitely must number of tests made a s f u t u r e collapse data a r e de-
be governed by a sound foundation f o r dec~slon,and veloped. Such information would be of value 111 calcu-
certainly this best can be expressed by minimum values lating the expectancy of failure and in balancing the re-
when assurance IS available t h a t such minimum values sult against the safety factor selected by the user.
a r e not in excess of the reasonable linlits of control In This would tend f u r t h e r to guarantee performance.
n~anufacturingpractice.
The proposed method of stanclard collapse test for
Not so long ago the selection of a n oil-well casing caslng seems satisfactory. I t is suggested, however,
string was nlade with little or no consideration a s to that all such data accuinulntecl be available to the Aineri-
how much the safety factors might be above those which
can Petroleum Institute f o r correlation.
were needed to protect the well properly and safely
I t is undoubtedly desirable that all work dealing with
from collapse pressure and tensional stress created by
the expected setting depths. the effect of tension on collapse in combination or
The empirical setting-depth tables, which provided in graduated strings continue, and t h a t the maxi-
the only information then available, were used with a mum strain-energy theory be considered applicable
factor of safety which now is considered higher than until further investigation adjusts or disproves i t s
required. The results of past esperlence also played application.
a considerable p a r t in the selections; however, in inany It has been mentioned previously by several t h a t 85
instances changing conditions failed to duphcate past per cent of the A P I average joint-strength values is too
experience which, therefore, was lessened in value a s high a ratio to set f o r a mininluin, in view of present
both product and application diffe~ed. data available, a s this appears now to be insufficient for
such a conclusion. The accumulation of further data, verse thereby defined a r e sufficient to design casing
of course, is indicated and desirable. programs with a n y collapse-failure risk desired.
I n cons~deringthe strength of joints, it is apparent Th'e writer has clevotecl considerable time during the
that, when the yield point has been exceeded, failure past year to this problem, and had hoped to have some-
already h a s occurred. Therefore, i t would be of possible thing concerning the problem in shape for presentation
value to determine the yleld point of the joint, when a t this session. An interesting fact revealed in the
making pullout tests, if this could be done with reason- invest~gationto date 1s t h a t the risk of collapse failure
able accuracy, a s the correlation of such information increases directly wit11 the number of sections in gradu-
m ~ g h become
t advantageous in the determmation of the ated-welght strings.
m i n ~ m u mjoint-strength value and the better selection It is well to point out t h a t this method f o r the eco-
of a suitable safety factor. nomic design of casing progralns In no way 1s limited
Burstillg strength based on the substitutio~lof mini- by the use of m i n ~ m u mstrength values. I t is only neces-
inuin yield f o r mininlum tensile strength, with a 12;-per- s a r y to know the statistics of the strength universe in
cent allowance for wall-thickness variation, is considered terins of the specified ininiinuin instead of the average.
to be logical and safe.
Too much stress cannot be placed on the ~inportance H. J. Hand (National Tube Company, Pittsburgh,
of properly designing casing strings, a s such design Pa.) (written) : The writer wishes to congratulate the
guarantees both safety and econon~y. The full advan- authors for their very enlightening analysis of this
tages of this have not yet been taken by the industry. subject-both a s to forces to which the string is subject
The paper under discussion provides aclecluate informa- during the life of a well, and to properties t h e casing
tion f o r the proper consideration of all factors which material must possess in order successfully to resist
a r e recluired in such design, either a s a straight string these forces. P~*obablyno logical argument can be ad-
of similar pipe or a combinat~onstring of different vanced t h a t essential properties should not be based on
weights, grades, and joints. ~nin~inum specifications f o r a nlaterial rather than on
I w ~ s hagain to state t h a t many pipe users a r e penal- average values. After a consicleratioi~of data included
izlng themselves by not making the most econoinical in the' appendices, however, the writer 1s not in accord
s a f e application of casing to their wells. The facility with the values of the minima suggested and, further-
for doing so is provided by this paper in a complete and more, does not share the authors' ideas relative to the
sufficient manner. The capable engineering work sum- application of statistical methods in the setting of speci-
marized and supplemented herein by the authors pro- fication requ~reinents.
v ~ d e sdefinite results which require no further effort It 1s agreed that, based on data included in Ap-
on the p a r t of the user, other than its appl~cationto pendis 1, i t is possible to set minimum values f o r col-
his needs on a rational o r common-sense basis. lapse t h a t a r e 75 per cent of the average, without
The authors a r e to be congratulated on this most ex- encountering escessive rejections in manufacture or
cellent preparation, and f o r their foresight in clearing necessitating a raise in t h e grade of steel. It is believed
the way for future inlprovenlent. t h a t more confidence could be placed in the latter,
however, if a value of 70 per cent of the average were
adopted f o r the limlt. It is not proved t h a t minimum
W. 0. Clinedinst (National Tube Company, Pitts-
values f o r pullout could be set at 85 per cent of the
burgh, Pa.) (written) : The authors a r e to be com- average \vithout raising present grades. I t is believed
mended highly on their efforts to provide a nleans f o r further t h a t the 87.5-per-cent figure f o r b u r s t ~ n g
designing casing prograllls on a minimum-strength strength cannot be substantiated, a s i t is not based
b a s ~ s . There is a rather p r o n ~ i s ~ nmethod
g f o r the eco- on experimental evidence. I n what follows, reasons f o r
nolnlc design of casing programs which h a s not been taking such a position will be discussed in detail, and
~nentlonedin the paper. The econon~iespossible appear arguments w ~ l lbe presented to show not only why the
'
to justify its consideration. statistical method should not be disregarded, but why
Absolute assurance t h a t a well will not be lost by i t is the only sound, scientific way to attack this general .
castng f a ~ l u r eis not necessarily t h e most econonlic con- problen~ of establishing tolerances. I t is urged that
dition. F o r the operator drilling a single well o r a statistleal methods be utilized when analyzing future
sinall group of wells, i t probably is. F o r a large group data on collapse and pullout tests.
of wells, the savings in casing cost by the use of lower-
strength material m a y ' f a r exceed the cost of wells lost
Collapse Tests
by casing failures. This is a problem which is handled
readily by probability mathematics, provided the statis- The treatment of the collapse test data is questioned
tics of the strength universe of casing a r e kno\vn. It in several respects.
was ivlth this thought 111 mind that statistical methods 1. Tl'h~l jctere ~ e s t t l t so f 365 o f the 1.371 tests ( i u ( ~ i 1 -
were a l ~ l ~ l i eind the detenliination of the n~inimunl-col- crble ilot inclzcderl in t l t ~uncclysis? Although it IS true
lapse curves presented in B1illeti)l 5-C-2. These minima t h a t sizes itlvolved f o r these tests were not shown in
curves were expressed a s 3 times standard deviation Tables 2 01%3 of Blslletin 5-(3-2, t h e ratio of actual to
minima, having a collapse-failure espectancy of 1 out average collapse (needed f o r t h e analysis) could have
.of 740 pieces of pipe. The statistics of the collapse uni- been obtalned froin the curves of Fig. 1of t h a t bulletin.
SETTINGDEPTHS FOR CASING 185
When these results were included, 7 tests were found yield encountered in this group. This may be taken a s
under 75 per cent ,of the average, and 37 under 80 I)er a change in the grade. If these tests and those of
cent-in contrast with the 3 and 15 found below sthe group. "A" be eliminated, only 474 tests remain on
respective inlnima when results of these tests were which to base a reliable min~muincollapse value. With
eliminated. To assume that the test pieces were ma- the exception of one test in group "C," all this material
terial not made to A P I specifications is not sufficient. meets the 75 per cent average satisfactorily, whereas
2. T l ~ ec1.7~thors'
nretllotl of Izancllt~tgdata from sozwce 7 of the 474 (1.5 per cent) fail to meet 80 per cent.
"A": Here results of tests made to old API-grade speci- Attention is called, however, to the f a c t t h a t this ma-
fications were classed a s grades H-40 and 5-55 on the terial (groups "B," "C," and "E") exceeds t h e average
basis of yield point. Distorted distributions of yield curves given in Bulletin 5-C-2 by 5 to 10 per cent.
Further, if the tests did not come from a "controlled
system of causes," 474 tests a r e insufficient a s a basis
to set u p reliable tolerance limits. (This point will be
developed in detail, subsequently).
I n view of these comments, i t is believed t h a t 7 5
per cent is the highest figure t h a t should be considered
when establishing nlinirnuln collapse data. And it is
quite possible t h a t additional data would indicate that-
10 -
-
1.5s
SVNTMETIC J-5sMRDE UP FROM
U LP z C O O P C R ~ T I V E PROCRMM
z
15-
3
r)
0 hCTUML DISTRIBUTION O N
S T E E L +?*DL TO J 5 9
g lo-
SPECIPIC~TIONS
t
Cu,n~ulativeDistributions of Per Cent of Norninal Pull-
5-
out Strength from Data Included in Appendix 3.
FIG. 2 (HAND)
I I I I I I I I h
50.000 70.000 a value a s low a s 70 per cent might be necessary, if
minimum collapse specifications a r e to be met wlthout
Yield-Point Distributions on "Synthetic" H-40 and generally raising steel grades.
5-55, Made by Selection from Old API Cooperative
Data Con~paretlwith Those Obtained in Production when Pullout Tests
Making Steel to H-40 and 5-55 Specifications.
Exception llkewise is taken to many "techniques"
FIG. 1 (HAND)
util~zedi n the handling of pullout data:
1. Elinti?la.tion of tests: Of 243 pullout tests, 211 of
points result from such synthetically madeup grades, which were from pipe produced under normal operating
quite unlike those occurring in practice [as shown in conditions, the authors included results of only 145 in
Fig. 1 (Hand)]. With such a difference in yield-point their analysis. Working with results of the 211 normal
distributions, collapse tests on material encountered tests, cumulative distributions were plotted separately
in practice could not be expected to approach those in for the two groups of tests involved in the study, and
group "A"-accounting for 583 tests of the 1,371. also f o r both thrown together. These a r e shown in
3. F ~ ~ r t h eisr ,it advisable to i?7clzccle the 314 tests on Fig. 2 (Hand), and the results a r e quite different from
gra(Ie J-55 of g r o ~ p"D"? These a r e froin a group of those of the authors'. Whereas the authors found no
material produced purposely with low internal stress, values under 91 per cent of the average, the present
and were of abnorn~allyhigh yield point f o r the grade. study indicates t h a t 011 all tests 1.8 per cent were under
5-55 h a s a minimum yield specification of 55,000 psi, but 85 per cent of the average, and 0.9 per cent under
a study of these data revealed 60,000 psi was the lowest 80 per cent.
2. A "co~tvol-cl~trt" study was made on these 211 controlled conditions cannot be reasoned out-they must
tests, and "non-controlled" conditions were found to be tested f o r by s t a t i s t ~ c a lcalculations. It might seen1
exist. Under such condit~ons,i t is folly to base toler- logical to assume that, because a product is received
ances to be used In specifications on so few tests. As from various manufacturers, control does not exist;
will be pointed out later, "non-controlled" conditions however, this is no criterion. What is meant by
mean t h a t non-constant actlng causes (ones which enter statistical control is t h a t variation between averages, or
and leave the system with time, or those t h a t a r e of variabillttes of "rational" sub-groups of- data, is
peculiar to certain sizes, grades, manufacturers, etc.) within limits of chance variation. The latter arises from
a r e present, and one might expect large differences be- residual variation within the rattonal groups, and be-
tween groups of tests. Such is actually the case with conles w ~ d e rwtth greater residual variation. If factors
materlal represented in the two test trials shown in affecting results vary to such a ~)roportional estent
Fig. 2 (Hand). One set of tests showed no pullouts \vithin a rational sub-group (e.g., a given outside dl-
under 85 per cent of the average, whereas the other had ameter, weight., o r grade of casing, made by a given
3.4 per cent under 85 per cent and 1.7 per cent under manufacturer) t h a t the variations between groups a r e
80 per cent. Such a condition is typical of test results all wlthtn ltinlts of chance variability, we have statisti-
obtained under "non-controlled" conditions, and should
' be interpreted a s arlsing out of the action of assignable
;;AVERRGE U R W C OF SERVICE L l r E PER MRNUFRCTYRCR 3 LOO0
causes, rather than representing fundamental differ- L I M I T S %MOWN 149C 30 L l W I T 3 , PUOBhBILITV OF BEING E X C C C O L O DUE T O
CIIRNCC e L l N G I I N 3 7 0
ences between the two groups of tests. Upon accumula-
tion of equal numbers of additional tests, the reverse
situation relatlve to test groups "A" and "B" reason-
ably might be anticipated. F u r t h e r accun-rulation of
data in units of 75 to 150 tests might show even wlder
fluctuation, a s nluch a s 10 t o 15 per cent under 85 per
cent of the average not being unreasonable. Under
such conditions, a large number of tests should be a v a ~ l -
able for the establishment of reliable tolerances; a
m~niinumof 1,000 would be desirable.
I n vlew of the foregoing, if no change in present
grade characteristics i s to be made, i t is not posslble
on the basis of the data presented to coinply with a
minimum pullout tolerance a t 85 per cent of the average.
lilnlts to insure efficient use of engineering materials, asslgnable causes a r e a p t to creep into the system-
all t h a t i s needed is to secure d a t a on 100 or more and to do so to a n unl~redictable extent. No reliable
instances (11. 63, Shewhart, loc. cit.) and set t h 6 ' l i n ~ l t estii%ate2,0f the degree to which these unl~redictable
where deslred. Under such a condition, if dealing Gith causes a r e a p t to enter can be made. Certainly, however,
averages of 4 or 5 indlv~dualtests, ilorlnal law proba- the best method f o r estimating i s to obtain d a t a on a
s esactly-here there is 1 chance in 740 of
b ~ l ~ t i ehold large number of tests-at least 1,000 would be desirable.
exceeding a minimum limit of average minus 3 stand-
a r d deviations; 1 in 40 of esceedlng a nii~llmumof aver- Additional Remarks Concerniiig the Applicatiot~ o f
age minus 2 standard dev~ations; etc. If statistical Statistical Methods in Setting Tolerances
control does not exlst, then the question of tolerance
is partly a s t a t i s t ~ c a lproblem and partly one of coin- Even when controlled conditions do not exist, use of
lnon sense or engineering judgment. Even here, how- statistical lnethods 1s advisable in setting up tolerances,
ever, the problem can be defined more clearly, and t h u s because t h e problem can be defined more clearly. The
better evaluated, ~f statistical methods a r e employed. following taken f r o m Dodge (loc. cit.) illustrates this
If control does not esist, i t must be stressed again t h a t point: When settlng u p a specification f o r the tensile
a large number of salnples-at least 1,000-is needed, strength of a zinc-base die-casting alloy, d a t a mkre
before a n y tolerance can be set t h a t is a t all rel~able. obtalned from each of five producers. Control charts f o r
Under these conditions, if tolerances a r e set u p based averages and standard deviations were set up. All
on relatively few tests, unexpectedly high proportions variations were due to chance (controlled conditions)
of rejections might result in p r a c t ~ c e ,due to the action except those of one producer with a low average and
of unpredictable "assignable" .or "non-constant-acting" one with a high average. All were satisfactory f o r
causes. standard deviation (which, incidentally, IS usually t h e
The authors' remarks on 1). 141-2, when discussing case). From a statistical viewpoint, the producer with
pullout test results, particularly t h a t p a r t reading, "it the low average had something wrong in his process-
seems qulte clear t h a t a n unduly large penalty results soine asslgnable reas011 why his product had a low
when statistical methods a r e applied to d a t a obtained average, and one which might be found. The limit
..
. . . uylien t h e requlsitc controllecl conditions do not finally decided upon should be set in the best interest of
apply," should have cons~derationa t f a r g r e a t e ~length all concerned, compromis~ng between t h a t which i s
t h a n is here possible. feasible to ~ r o d u c eand use economically-balancing the
Examples a r e clted from Crawford's data, t h a t on chances of a rejected product against failure in service.
Fi$-in. x 17-lb J-55, the 5 x SU lllnlt i s 9 1 per cent of With the aid of the control chart, such compromises
the averages given in Btclletirt 5-C-2; on 7-in. x 24-lb would not need to be made un\vlttingly. In t h ~ case s they
J-55, 89 per cent; whereas f o r all slzes i t is only 75 could be made with the knowledge that, if t h e Iiniit
per cent. This is due to the f a c t t h a t these groups were r a ~ s e dt o meet servlce needs, this action \vould
a r e both "out 'of control" OJI the h ~ g hside (actually requlde either 110 adjustment, or a n adjustment of a
checked, using control calculations). Some assignable definite kind in the process of a given producer. I n the
cause prevailed ill both these groups of d a t a which example under consideration m i n ~ m u i ntolerances could
gave unexpectedly h ~ g hvalues. The authors make no be set on the basis of 3 standard deviations below aver-
mention, however, of 7-in. x 17-lb J-55. Here, three tests age (probability of being exceeded d u e *to chance of
gave a n average pullout of 78 per cent of the average, 1 to 740), this "average" being based on only: 1, t h e
the lowest of the three being 75.2 per cent. This ma- highest; or, 2, the lowest; or, 3, the 3 inteiQiiiediatepro-
t e r ~ a lwas siniilarlp "out of control," this tinie on the ducers (which were all i n control). Tolerances thus
low side-again due to a n assignable cause-only in this could be deterlnined with a n exact knowledge of w h a t
case it acted adversely and not beneficially a s In the changes the various producers could o r would have to
first case. I t i s evident t h a t a penalty 1s necessary inake when one of the 3 finally w a s adopted.
f o r lack of control; but t h ~ s1s due to the s t a t e of lack Considering collapse or pullout values, the control-
of control, a n d not to a n y statistical system of ~ n t e r p r e - chart nlethod could be applied to determine a s s ~ g n a b l e
t i n g the data. If careful analysis of the d a t a indi- factors ~iifluencing differences between actual test re-
cates t h a t assignable (findable) causes a r e acting, t h e sults and average vnlues according to Bubletilz 5-C-2.
problem becolnes defined clearly. Until technology Such assignable factors might be associated with pro-
enables a n e h m i n a t ~ o nof these causes, we must expect t o ducers, class of product, or sizes. Based on results of
get values t h a t a r e undesirably low-and ones which this type of analysis, specifications could be set up with
must be taken rnto account when setting tolerance a sound knowledge of penalties inflicted if a given toler-
limits. Application of the statistical method In manu- ance limit were adopted-wlilch producer, class of
facture, if diligently applied, should lead quickly to t h e product o r sizes, if any, mould be penalized if rhyme or
elimination of such conditions-and, when this h a s beell reason existed with respect to groups found "out of
accomplished, the establishment of closer tolerances can control." If no regularity existed with respect t o pro-
be made without undue penalty. ducer, sizes, etc., f o r d a t a groups "out of control," such
So long a s such conditions creep into the cause sys- a study might indicate t h a t i t mould be economically
tem, economic tolerances cannot be set based on only a desirable to base tolerances on a consideration of t h e
few tests. 1V1th the present s t a t e of the a r t , such general cumulative distribution f o r all d a t a luinped to-
gether, selecting a ininiinun~point to exclude 0.5 per American Petroleun~ Institute, the purchaser is pro-
cent, 1 per cent, 2 per cent, o r any other per cent of the tected and is given the opportunity of balancing one
lowest values encountered. On the other hand, if all against the other in the engineering of each casing
data were found to come from a controlled condition, string.
except t h a t of one assignable low producer o r of a few The writer visualizes the performance A P I specifica-
sizes assignably low, i t might be well to have minimum tions somewhat a s follows:
tolerances based on data excluding the particular pro-
I. Dm~ensionsand tolerances:
ducer or sizes concerned, using the cumulative distri-
bution or average minus three standard deviations a s a a. Because of interchangeability the thread form,
basis ( t h e thought being t h a t the particular producer taper, lead, and length tolerances should be stand-
or slzes involved were low, due to a findable cause ardized so f a r a s the normal joint is concerned. This
which, in turn, should be eliminated f o r the general involves weight and thickness tolerances a s well. This
good of all concerned). Then the advantage of the is already in the 5-A specification.
statistical method becomes apparent; i t 1s clearly b. The sizes and thicknesses of pipe which most
possible to establish a n y adjustments necessary to set nearly fit the requirements f o r casing the wells in
up'tolerances higher than those determined by merely various fields should be standardized. This is already in
escludlng a certain low percentage of all values en- the 5-A specification.
countered. Wlth such a method of analysis available, c. Provision should be made f o r registering the vari-
establishment of a specific tolerance then becomes one ous makes of special joints together with their per-
of economic decision. formance record, thls performance record to be estab-
In conclusion, i t is urged t h a t statistical methods be lished thoroughly before registration is permitted. This
utilized in the final establishment of minimum limits f o r would tend to eliminate claims which vendors of par-
the very important casing properties discussed in the ticular joints might make which cannot be established
present report.. It firmly is believed t h a t only through upon thorough investigation. It should be of much
these methods .can reliable and economic tolerances be assistance to the casing user in selecting good joints
established to the best interest of all. It is suggested for service.
further that, when collecting f u t u r e data on this prob- 11. Collapse pressure
lem, the order in which tests a r e made by various
a. If a company will guarantee a minimum collapse
nlanufacturers be shown. Quite frequently the time ele-
pressure for each grade, size, and thickness of caslng,
ment aids in locating assignable causes of variability.
then if i t can supply routine laboratory tests which
show t h a t these tests a r e never below the guaranteed
T. McLean Jasper ( A . 0. Smith Corporation, Mil- value, this should be sufficient to allow t h a t conlpany to
waukee, Wis.) (written) : This paper represents a register its pipe and be given the privilege of stamping
large amount of laborious work, and the results a r e t h e pipe with "API" on the basis of this quality.
worthy of our very best attention. The authors of the b. I n addition to the above specifications, a method
paper well may feel gratified by the very large amount f o r field and laboratory testing f o r collapse should be
of apprecjation which the writer feels will be bestowed outlined. I t is believed t h a t a simple method f o r field
for the effort they have made. The help which the paper t e s t ~ n gcan be devised which will be practicable and
gives in clarifying the discussions which have occurred relatively inexpensive. This will tend to insure policing
durlng the past five o r six years also will be appreciated of the minimum collapse pressures registered f o r A P I
by the American Petroleum Institute and by the pipe casing.
user and the manufacturer, because ~t tends to elimi-
nate unnecessary discussion of highly technical o r 111. Pullout load
theoretical considerat~onst h a t have tended to confuse a. I f a company will guarantee a minimum pull-apart
the main issue. The real problem we have is to develop value, using any threads a s under "I" f o r each grade,
a logical and simply applied pipe specification f o r the size, and thickness of casing, and will supply tests to
casing of oil wells. substantiate the minimum values, then such pipe should
The writer visualizes a very distinct principle point- be registered a s "API" pipe and be given the privilege
ing toward a simple performance specification for cas- of stamping "API" upon i t on the basis of this quality.
ing which will not throttle progress in manufacturing b. Pullout tests require equipment capable of apply-
processes, and which will allow full advantage to be ing a large quantity of force. The satisfactory method
taken of improvements of whatever nature. It should f o r making such tests should be outlined.
not compel all A P I pipe to wear the identical pattern
(nor stoop to the stature of t h e least able method of IV. Ductility
manufacture) in order to meet the A P I requirement, a s a. If a company will guarantee some form of ductility
represented by its code, f o r a general class or grade quality f o r each grade, size, and thickness of pipe which
of pipe. i t makes, and will supply tests which substantiate i t s
I t is my belief t h a t one type of casing may vary from nlinimuln values, then t h a t company should be allowed
some other type in one or more of its properties. It may t o register a s supplying API pipe on this basis.
have superior pullout ppoperties but inferior collapse b. Outline the ductility test required, such a s ring-
properties. If this difference is so registered with the flattening test o r per cent elongation. The ductility test
DEPTHSFOR CASING
SETTING 189
TABLE 1 (FRAME)
Ratio of
M~nimum
Minlmum Corrected
Joint Joint
Nominal Strength Strength
Dimensions Average Minimum Corrected to Average
API Observed Observed to Minimum API
' Size Weight Joint Joint Yield Yield Joint
(Inches) (Pounds) Grade ,Thread Strength Strength Strength Strength Strength
55 14 H-40 Short 174 174 53,200 131 0.75
59 17 H-40 Short 220 229 50,700 181 0.82
7 20 H-40 Short 239 230 53,700 171 0.72
7 24 H-40 Short 296 274 48,100 228 , 0.77
55 14 J-55 Short 232 215 62,300 190 0.82
5: 17 5-55 Short 293 272 59,400 253 0.86
5: 20 5-55 Short 350 390 62,900 341 0.97
7 20 J-55 Short 318 288 64,800 . 244 0.77
7 24 5-55 Short 394 362 60,200 331 0.84
7 30 J-55 Short 506 480 63,300 417 0.82
7 24 6-55 Long 452 383 59,000 357 0.79
7w 30 5-55 Long 580 529 59,600 488 0.84
7s 30 5-55 Long 580 540 63,800 465 0.80
specifications. The only thing t h a t I would change The test specimens used in these investigations were
would be a s shown above in order to produce a flexible made from steels of inore or less average properties.
performance specification which represents operating No attempt was made to provide steels with minimum
service demands, and which can be more directly used physical propert~es. Yet the A P I specifications desig-
by the field engineer. nate the miniinuln properties of the steel, and we know
Field practices occasionally entail the use of h ~ g hin- that pipe is manufactured in some instances from steels
ternal pressures in casing. The authors have brought very close to the n~inlinum values. The "Report 011
this u p in connection with deep drilling, and have out- Physical Properties of Pipe," A P I Bztllett?z 223, shows
lined a method for caring f o r this. It is suggested that, that, f o r grade "D" steel, nearly 4 per cent of the tests
In thls connection, the experience of field operators reported were very close to the minimum specified yield
should be brought to bear on such requiren~ents, and strength of 55,000 psi.
t h a t the casing manufacturer become acquainted with I n Table 7 of the paper the minimum observed values
these requirements through the Engineers' Subconi- of joint strength were not corrected to the minimum
mittee. If such pressures a r e due to improper handling physical properties specified f o r the pipe. It is be-
of casing, then field-practice recommendations may be lieved t h a t this correction should have been made, and
able to eliminate any trouble from this source. If, how- in a manner similar to t h a t used by Crawford.
ever, certain casing programs call for high operating I n Table 1 (Frame) the corrected observed lninin~uln
pressures, then these properties should be included in -
the perforlnance specification. Prodltctto~i Birll. Xo 020, Exhibit "I::;' p 107-75.
t LJr1lll~lg c111c1I'rodrtctlo~~.Prctctice, -lo--S.! (1:1::01
RIALS
joint strengths taken from Crawford's Table 2 a r e The fit of the Edwards and Miller d a t a to equation
compared with the minimum observed values and the ( 2 ) is shown in Flg. 1 and 2 ( H o l m q u ~ s t ) .I n order to
A P I average values. The observed yield strengths a r e permit the asse~nblyinto one plot of d a t a on materials
shown also. The ratio of the miniinun~corrected joint
strength to the A P I average joint strength 1s shown
to be cons~derablylower t h a n t h e ratios of Table 7 of
the paper. Only 2 of tlie 13 ~ t e l n shave ratios of 85
per cent or better, and 6 of t h e 13 items have ratios
80 per cent o r under.
Consequently, ~twould a p p e a r t h a t , ~f t r u e minimum
values a r e desired, they should be based on a ratio of
not more than 75 per cent f o r t h e lighter sizes, and not
more than 80 per cent f o r t h e h e a v ~ e rsizes-instead of
85 per cent f o r all sizes.
tudinal tension load a n d external pressure which failed t h e IVescott, Dunlop, and Kelnler paper. In this case
by collapse. t h e collapse value f o r 5 5 i n . s 15-lb 5-55 casing is cor-
I n Flg. 1 a considerable scatter of the d a t a a1;b;'ut t h e rected f o r tension, t h e correction factor being 0.86 a s
curve 1s evident; however, f r o m previous esperlence, determined froin the curve of equation (1). If the curve
we a r e prel~areclto acceptjthls a s characteristic of col- of equation ( 2 ) were used, t h e correction factor would
lapse test data. It would be difficult, however, to d r a w be 0.79. E q u a t ~ o n( 1 ) gives only two-thirds of the cor-
a morelsatisfactory trend line through the d a t a t h a n rection o b t a ~ n e dby equation ( 2 ) In this ~nstance.
the curye shown, which represents the biasial-stress Ecl~irards and Mlller also concluded t h a t equation
case of the Hencky-von Mises theory. (2) was inore applicable to their d a t a t h a n equation
Fig. 2 sho\vs t h e fit t o the curve representing eclua- (1). They also plotted their d a t a in terms of the
tion ( 2 ) of those tests made under combined longitndi- variables: ratio of hoop compression stress to yield
nal tension load ancl external pressure which failecl by stress and ratio of longitud~naltension stress to yield
"stretching." stress. Fig. 6 of their paper shows t h e plot of their
I n Fig. 2 the scatter of the d a t a about the curve repre- d a t a which may be compared with Fig. 3. I t will be
senting equation ( 2 ) is much less than in the case of
Fig. 1. Thls is due to the f a c t 'that, in t h e stretch
Positive=tension.
.Negative =colnpression.
* F o r collapse failure, the simple collapse stress i s
used instead of simple yield.
+=stretch failures-groups 3 t o 12, inclusive.
"=collapse failures-group
n
6 ; -=16.5; yield stress
Miller and Edwards Data-Collapse ant1 Stretch Failures. t
=40,600.
FIG. 3 (HOLMQUIST)
=collal,se failures-group 7 ;
t
2
~ 2 1 . 7 yield
; stress
failures, t h e factor of instability does not intrude on =81,300.
t h e results a s in the case of t h e collapse failures. A s FIG. 4 (HOLMQUIST)
in the case of Fig. 1, i t would be difficult to d r a w a more
satisfactory trend line through the test d a t a t h a n the noted t h a t t h e pbints representing tube group 7, which
curve shown, mhlch again represents t h e biaxial-stress failed by collapse, occupy a different position in t h e
case of the Hencky-von Mises theory. plot froin a n y points i n Flg. 3. This i s due t o t h e f a c t
I n Fig. 3 (Hohncluist) the test d a t a f o r both t h e col- t h a t f o r these points Edwards ancl Miller used a s t h e
lapse ancl stretch failures a r e plotted. The curves yield strength t h e average stress computed froln Bar-
r e p r e s e n t ~ n gt h e Hencky-ron Mises theory a n d the maxi- low's f o n n u l a froin t h e collapse pressures f o r t h e open-
mum strain-energy theory a r e shown also. It is evi- end tests i n t h e group. F o r all other points t h e yield
dent t h a t t h e l a t t e r curve does not fit t h e d a t a a s well strength used w a s t.he valGe found i n t h e tenslon test.
a s t h e former. Therefore, i t is concluclecl t h a t t h e cor- All the plotted points in Fig. 3 used the tension yield
rection f o r t h e effect of tension on t h e collapse pressure strength.
more properly is made on the basis of equation (2) t h a n The use of a yield-strength value computed f r o m
equation ( 1 ) . Barlow's formula increases the scatter of t h e data, a s
I t should be noted t h a t there is a considerable differ- a comparison of the two figures shows. A discussion of
ence in t h e results obtained by t h e two equations. Con- the reasons f o r this would involve a discussion of t h e
sider case No. 3 of the example shown in Table 10 of effect of t h e shape of the stress-strain curve on collapse
resistance. As been discussed previously in publica- P. 143, item 6, the inethod of gripping the speci-
tions of the society,' i t is not desired further to discuss men siinply states "so t h a t the loading is truly axial."
the matter here. If it is assunled t h a t plugs and wedge grips a r e to be
\ used a s specified in ASTM E 8-36, Par. 19, should not
Bibliography this reference be given or suitable description of appa-
r a t u s needed?
' H(11niqulsta n d Naclai. "A Theoretical and Esprril~~riltal A11
13roacli ~ I the I I'r~vhlruiof Collal>st.of Dee{)-Well C'RIIIII~." LJr111-
I I I O n ~ r dP r r ~ r l r t c t t o ~Prctctrcc,
r 302 11?$9)
' .A N I I ~ I R I ." T ~ I C U of~ ~Strengtll.
PS Trcor8 AIII A'oc TrRt i l r g
M n t t : r ~ r r l n 55, APAI 55-15',(143.3). C. D. Miller (Barnsdall Oil Company, Tulsa, Okla.)
Ed\vards i ~ n t lAllllrr. D~scnssio~i on tllr Efrct of l:oinl~ii~rd
Longituilinal Lo?dlng a n d Esterl~alFrrsxur+, on tlie Strengtli (written) : +
of I ~ l i l - \ V r l l Casing." Dvillr,ig n,id I'rod~rctio~rProctrrc, 4%: Stcn~tnnry of Replies to cc Qltestionnaire on Proper
I 19:<9I
Aletlrod of Setting Casing: F o r some time nleinbers
of the Drilling Practice Co~ninitteehave given consider-
W. L. Schwalbe (University of Illinois, Urbana, Ill.) able thought to the subject of the proper method of
(written) : * The authors state t h a t design of deep- setting casing. A casual survey indicated t h a t there
well casings is a n engineering problenl. The modern was a great difference of opinion esisting ainong the
a r t of structural and machine design is based on the operators. As the longer strings of pipe came to be used
scientific inethod of experimentation and analysis, es- inore generally, the question of the proper method of
periinentation f o r the determination of the behavior setting the pipe became more acute. With the idea of
properties of the materials and members involve'd, analy- getting this question into the open, i t was decided to
sis for the determination of the relations w h ~ c hexist circularize a s many of the operators a s possible with a
between the test specimen o r the complete structure questionnaire, setting out the pertinent steps in the
and the environment in which behavior takes place. setting of casing in a n effort to determine what the
Present knowledge of materials and the behavior of various practices are.
long tubes under pressure is estensive enough now so The questionnaire sent out was rather long and com-
that a convincing case f o r the use of higher perforin- pl~cated,and was submitted .as a working model rather
ance values, whether average or m i n ~ n l u n can
~ , be made. than a s a n actual questionnaire; however, we believe
The evaluation of the available knowledge can be such the response to have been sufficient to indicate the prac-
t h a t the engineer may design with confidence. Confi- tices throughout the entire country.
dence in design must be supplemented with confidence In my opinion the meat of the subject appears to be
In the materials and menlbers t h a t realize the design. whether p a r t of the tension is slacked off or whether
Such confidence inay be obtained only on the basis of the entire string is held in tension and, if the string
tests the engineer himself inakes o r on the basis of is slacked off, the method used in determining the
specificatiol~su ~ r i t t e nby engineers in cooperation with amount to slack off.
manufacturers. A sulninary of the replies to the various questions is
I n regard to collapse pressures, specifications should a s follows.:
distinguish between the elastic collapse, sudden o r Q ~ t e s t i o ~No.
z 1. Do you leave the full anlount of ten-
gradual, and the inelastic o r plastic collapse. Specifica- sion In the string when nippling u p a f t e r the cement
tions f o r the latter should include limiting shapes f o r h a s set, or do you slack off p a r t of this tension?
the stress-strain curve of the material, and upper liinits Answer. Seventy-eight (787;)) per cent said they
f o r out-of-roundness and eccentricity of the tube. slacked off and twenty-two (22%) per cent said they
Not only a r e cooperative test lwograms on the effect held the pipe in full tension.
of aslal stress on collapse strength advisable, a s sug- Qttestiot~No. 2. Is the same procedure used for all
gested by the authors, but i t is des~rablet h a t some- strings, i.e., surface pipe and oil strlng?
what longer-range progralns, lnore in the nature of .-lnsuier. Fifty (50%)) per cent of the companies use
"academic" research, be proposed, to determine the re- the same procedure f o r all strings except when the sur-
sisting behavior of materials under biaslal and triasial face pipe is cemented to the surface.
stress and non-unifornl stress distributions in' general. Qttestion No. 3. If there is a n y difference in the
W. J. Putnain, professor of theoretical and applied method used in setting the two casing strings, what is
mechanics a t the University. of Illinois, who has tested the dlfference?
about 150 joints ranging from 51 in. to 102 in. in di- -4nswev. I t appears t h a t the general practice is to
ameter, for the A. 0. Smith Corporation, Milwaukee, cement the shorter strings of surface pipe to the sur-
Wis., during the past gear, offers the following face, and treat the intermediate strings the same a s the
comments : 011 string.
P. 143, iten1 9-A, "gage point" f o r measurement of Q~testzo~z No. 4. If solne of the weight o r tension is
inside dianleter is not defined. If this is a t t h e same slacked off before settlng in the casinghead, do you
position a s t h a t for ineasuring outside diameter, i.e., usually just slack off so many inches regardless of
"adjacent to the shoulder of the threads," i t should be depth?
so stated.
t r:r:~il by title This rllscrlssion was pr~srntedby Alr. AIiller
11s a rrpurt at the mretinp ot the Topical Coll~mittrr011 r)rillilr::
* Head by title. l'mcticr. Elnckatolie H u t 4 , Fort Worth, Texas, ,\la< 27, 1!)4lI.
THS FOR CASING 193
A?tszue~. Practically every operator considered the pipe and then keep all of the pipe above the freezing
depth in calculating the amount to be slacked .off... One point in tension.
operator, however, slacked off half the tension in the With respect to this question, one operator states:
pipe regardless of the depth. "Due to the buoyant effect of the mud and/or cement
Q ~ t e s t l o ?No.
~ 5 . O r do you attempt to slacli off a there is a neutral point in each string which is neither
certain distance per unit of depth, such a s a certain under tension nor compression. Determination of free
number of inches per thousand feet of plpe set? length of casing a f t e r the cement has set indicates t h a t
A ? ~ s ? u eA ~ . few operators use a certain distance per caslng usually freezes just below the shoe of the last
unit of depth, and in each case this unit w a s one ( 1 ) previous string and above the neutral point; hence the
inch per one thousand (1,000) feet of pipe above the free length of casing is usually in tension-at least
cement. such is indicated by the few tests t h a t have been made.
Qziestion No. 6 . If neither question 4 nor 5 is your The pipe below the neutral point is either cemented o r
procedure, do you slack off a certaln weight a s deter- frozen and, even though i t may be in slight compres-
mined by polnts on the weight indlcator? and sion, i t is not free to wobble and, therefore, causes no
Qliesttox NO. 7. IS this certain welght slacked off difficulties." This operator flanges up hls casing with
varied a s to depth of well? the submerged welght, and does not attempt to slack off.
Avts7uel.. Fifty-six (56%) per cent of the companies Another operator, on this same question, says: "Ten-
use the weight indicator in determining the amount to sion, if any, is usually slacked off when casing,,js set
be slacked off, and in almost every Instance the weight in casinghead a f t e r cement has set which is forty-eight
slacked off varled, of course, according to the depth of (48) hours o r longer after casmg is cemented. Our
the well. practice has been to test the amount of tension, which
Qlcestio)~No. 8 . Is the height filled by the cement cal- will vary from well to well, with identical length strings,
culated or estimated, and is this fact taken into con- due to temperature and tendency of formation to stick
sideration when tension In the casing strlng is slacked upper p a r t of string. After this is determined, we usu-
off? and ally slack off about one-third of the total tension."
Qziestion No. 9. O r is the amount of pipe t h a t is Another operator says: "Our practice has been
above the cement determined by a n y sort of test such changed. We used to drop about G in. slack in flanging
a s moving the pipe up and down or by a temperature up or nlpplmg up after cement had set. Now there is
test? and very little, if any, slack given to the pipe. I n no event
Qttestlott No. 10. If one of the methods in question 9 is additional tension taken."
is used, which inethod do you use? Still another operator says: "Casing is hung on hook
Arzsrue~.Seventy-two (72%) per cent answered ' i y e ~ " for several hours until caslng reaches hole tempera-
and twenty-eight (28%) per cent answered "no." A ture-r unt.11 weight or indlcator ceases to drop, and
small percentage of the operators determines the posi- then landed without slacliing off. I t is felt t h a t this
tlon of the cement behind the pipe by temperature tests, lengthening due to temperature change is a sufficient
and a small percentage determines the freezing polnt amount of slack off."
of the pipe by movement. Another operator uses a rather unique method,
Qziestion No. , I ] . If a certain amount of casing is whlch is a s follows: "On strlngs set below 7,000 ft, a
determined to be above the cement, how much, or what pressure gage is placed on cement head and pressure
per cent of weight or total above the cement is slacked is bled to predetermined amount, o r released colnpletely
off? if float valve holds, and p a r t of tenslon on casing slacked
Alzszuer. The percentage of weight above cement off each hour untll cement is set." This operator slacks
slacked off vaned from a minimum of 25 per cent to a off about twenty-five (25) to thirty-five (35) per cent
maximum of 80 per cent. of the total weight of the pipe above the cement indi-
Question No. 12.- Do you attempt to keep all pipe i n cated by a temperature test.
tension? Another operator slacks off approximately four ( 4 % )
per cent. H e states: "For example: A 9,000-ft string
Qz~estiolzNo. 13. Or do you attempt to keep the upper
of 81-in. casing is landed In the following manner:
string only in tension?
Qzsestion No. 14. Or do you attempt to keep the full 1. Weight of string a t time plugs bump is approsi-
s t r ~ n gin tension, tapering down to the last joint above nlately 75 points.
the cement, which would be neither in tension nor 2. About 4 hours later, casing is landed in slips under
con~pression? 72 points of tension.
ATzszuel-. Forty-two (42%) per cent of the operators By this method, the string is in tension by its own
attempt to keep the upper p a r t of the string in full weight a s modified by the effects of buoyancy, cementa-
tension, tapering down to the last joint above the cement, tion, and formational temperature changes, and the
which would be neither i n tens1011 nor compression. A increments of t e n s ~ o nin said string when landed theo-
number of the operators questlon this, indicating t h a t retically a r e distributed from top to bottom in effective
in most cases when tests have been made the pipe had relation to their distribution a t the time the plugs
become frozen some place above the cement, in which bumped."
event they tried to determine the freezing point of t h e Another operator says: "We prefer to slack off ac-
7
cording to amount of pipe above cement rather than exceeded the yield point of the weakest section, which
according to the weight, due to difficulty in keeping is generally in a joint. Such loads should never be im-
weight indicators in constant adjustment. All pipe kept posed except in case of emergency, when no other al-
in tension, but tension above cement relieved by slack- ternative exists. Altogether too little is known about
ing off." the effect of tensile loading upon joint tightness, and
There seems to be considerable difference of opinion this lack of information is a source of concern to the
a s to whether tension should be slacked off. I t is my operators.
belief t h a t a number of the operators who indicated. Mr. Clinedinst discusses more fully a point mentioned
t h a t they kept the pipe in full tension meant t h a t they by Mr. Ladd. I t is t h a t the saving possible by design-
kept the amount of pipe above the cement in tension ing casing strings with very low factors of safety may
rather than t h a t they kept tension on the full string. be greater than the cost of a well lost occasionally a s
I believe there a r e relatively few strings of pipe set a result of this practice. I t is true t h a t no direct men-
without letting off some of the tension. I n discussing tion of this was made in the paper. This omission
this problem with several of the pipe-company engi- was deliberate, however, because the authors believe
neers, they indicate t h a t their idea of the proper method t h a t this is a question of policy for each individual
of setting the pipe would be to determine the amount of operator. I t was expressed indirectly by the statement
cement behind the pipe or the point a t which the pipe t h a t "the responsibility of the user is to decide what
has been frozen, and then let off enough slack to have factor of safety is required, and to accept the conse-
the first joint above the freezing point in neither tension quences if failure occurs." However, these two costs a r e
nor compression. not always the only costs involved. Quite frequently,
I t is difficult to arrive a t any definite conclusions, but deep wells a r e wildcat wells drilled to explore struc-
it would be interesting to know the amount of trouble tures located by geophysical surveys, or to satisfy lease
t h a t h a s occurred while using the various methods of obligations in which time is a n important factor. I n
setting the pipe. If i t were possible to tabulate the such cases much more is involved than the cost of a
pipe failures which indicate t h a t the pipe h a s parted single well. It is also possible that the strict economic
and then determine the method t h a t was used in the consideration will be lost sight of in case of the loss of
setting of each of the strings, some definite conclusion a n expensive well. All of these factors a r e individual
might be reached. problems and, therefore, i t would be unwise to recom-
mend a general policy based on the considerations ad-
Authors' Closure vanced by Mr. Clinedinst. His remarks do bring t h e
question to the attention of individuals concerned, and
The authors wish to express their appreciation f o r the paper which he anticipates will undoubtedly be of
the discussions, which provide much useful and valua- very material assistance to operators who choose to risk
ble information. This is particularly t r u e a s regards occasional failures to obtain maximum economy.
the attitude towards the adoption of minimum values The discussion by Mr. Hand is very detailed, and the
instead of average values, and the readiness to accept authors welcome a n opportunity to reply in the same
the principle of performance specifications.' detail. The points raised by Mr. Hand will be answered
Mr. Carson has pointed out clearly the necessity f o r in the order used in his discussion.
designing casing strings on accurately known minimum Mr. Hand states his belief t h a t "the 87.5-per-cent
setting-depth values, and has emphasized the desira- figure for bursting strength cannot be substantiated,
bility of closer control over these properties during a s it is not based on experimental evidence." The au-
manufacture of the casing. The authors a r e in com- thors a r e a t a loss to understand the reason for this
plete agreement with these opinions as well a s his other objection-even when, a s pointed out in the paper and
remarks, including the various factors which enter by Mr. Hand, no bursting-test data a r e available. Pos-
into the design of casing strings f o r deep wells and sibly Mr. Hand overlooked the statement t h a t the
the need f o r additional test data on casing of recent bursting strengths were calculated for minimum yield
manufacture. Although no data yet have been presented strength, and not minimum tensile strength a s is com-
which would influence the authors to deviate from a n y mon in the use of Barlow's formula. This practice was
of the conclusions in the paper, they freely admit the adopted by the Manufacturers' Subcommittee for use
possibility t h a t additional new test data might give in Bulletin 5-C-2. I n addition, the authors applied a
rise to some slight modification of the minimum limits further correction of 121 per cent to allow for minimum
suggested. wall thickness. When the spread between yield strength
Mr. Ladd, in his discussion, points out the possibility and tensile strength is taken into consideration, to-
of reducing the investment in wells by taking advantage gether with the additional allowance f o r minimum wall,
of engineering principles in the design of casing strings. it appears t h a t this limit is t h e most conservative sug-
The reduction in cost is not insignificant to a n y indi- gested by a wide margin.
vidual operating company, and the total becomes im-
pressive when estimated f o r the entire industry. Mr. Collapse Tests
Ladd also calls attention to the need for more test data
on joint strength; and there can be no disagreement 1. The results of tests of A P I casing sizes only were
with this opinion, nor with the statement t h a t failure included, a s mentioned in the paper, because the only
h a s occurred when the load imposed on a casing string concern was with pipe which bears the A P I monogram.
SETTINGDEPTHS FOR CASING 195
I
The elim~nationof all other test data was decided upon tions" by only two manufacturers over a comparatively
before the test data were analyzed, and the authors short penod of time, i t would seem t h a t there- would
do not belleve any other data should have been ~ncluded, be small likelihood t h a t controlled cond~tions would
in view of the purpose for which this paper was written. exist for tests made of the products of all pipe manu-
2. Objection was made to the reclassification of data facturers over longer perlods of time. Mr. Hand's re-
from tests on old A P I grades C and D into grades H-40 marks lead to this conclusion, and in t h a t respect the
and 5-55, respect~vely. The minimum yield strength of authors a r e in agreement with h ~ m . I n view of the
grade C was 45,000 psi, a s compared to 40,000 1x1 for fact that non-controlled conditions were found to ap-
grade H-40. There was no difference between the yield ply to the group of all pullout tests available, does ~t
strengths of grade D and grade 5-55. Grade B casing seen1 probable. t h a t controlled conditions exlsted for
required a minimum yleld strength of 40,000 psi, identi- the collapse tests which were made by several differ-
cal with grade H-40, but t h ~ sgrade was practically ent manufacturers over a perlod of several years? Yet
unused. D~fference in distr~butlon might be expected these data were used statistically by the Manufacturers'
between grade C and grade H-40, but thls difference, Subcommittee to arrive a t the nnnimunl values glven
a s shown in Fig. 1 of the discussion, 1s quite small and in Bztlletnz 5-C-2. It would be ~ n t e r e s t i n g to know
much less than the difference between grade D and whether these d a t a were tested f o r controlled condi-
grade 5-55. Therefore, Mr. Hand's criticism does not tions. I t naturally would be assumed, in view of Mr.
seem to be supported adequately by his explanation. Hand's remarks, a s regards the reclassification used
I t is suggested t h a t the answer n l ~ g h tbe that changes by the authors, t h a t these data could not possibly satisfy
were found desirable in manufacturing operations be- controlled conditions-in which case the desirable
cause inspection of the collapse test results (Appendix n l i n ~ m u nof~ 1,000 results is f a r from satisfied even f o r
2 in the paper) shows t h a t pipe fro111 source "A" was collapse tests. I t is difficult to reconcile Mr. Hand's
quite consistently lower in collapse strength than pipe conclusion that two groups of tests reported by Craw-
from other sources. ford were "out of control" with Crawford's suggestion
3. The statement made by Mr. Hand in regard to the that the ininllnuin values be determined statistically
pipe with low Internal stress is belleved to be in error. from these same data. Other groups of collapse tests
The pipe represented run-of-mill production, a s f a r a s also were "out of control on the hlgh side"; and yet,
the authors were able to determine. in every ~nstance, the pipe was run-of-mill product.
I n spite of the objections just discussed, Mr. Hand Therefore, the authors consider 'these results to be
found no senous object~on to the minlinum collapse t y p ~ c a lof casing installed in wells. However, if the
limit suggested in the paper. minun~umresults a r e obtained by using three times the
standard devlation, the values a r e not hlgh enough for
Pullout Tests econonlical usage. As pointed out in the early p a r t of
the paper, minimum values for physical properties cal-
1. Mr. Hand is correct in stating t h a t some pullout culated on the same basis a r e f a r below the minimum
test data were not included. The authors believe the values allowed by the specifications. Such a position
number to be 111 instead of' 98-which, in his discus- is not tenable, and is not acceptable a s a basis for
sion, Mr. Hand infers were not considered. These drawing up performance specifications.
omitted results were used by the Manufacturers' Sub-
committee in preparing Bulletin 5-C-8.* Therefore, the
authors bel~evedthat they would be suitable also f o r Use of Statistical Methods
use in preparation of t h e paper. Requests were made F o r some reason, Mr. Hand h a s chosen to interpret
for these data when the paper was being written, and the authors' remarks to mean a basic objection to the
have been made subsequently. The Manufacturers' Sub- use of statistical methods for the analysis of data. T h ~ s
colnnllttee since has advised t h a t the data a r e not is not a t all true. The objection is t h a t the authors
suitable f o r this purpose, and f o r t h a t reason have do not believe t h a t the essential requirement of con-
not been subnntted to the authors.? The authors do trolled conditions has been satisfied in test data availa-
not understand clearly why data sultable for d e t e r m ~ n - ble to the pipe committee. If this 1s true, a s admitted
m g average properties a r e not suitable for determining by Mr. Hand, f o r the group of 111 pullout tests and
ininin~umproperties. Evidently these data were availa- the 2 groups of tests in Crawford's invest~gation in
ble to Mr. Hand, and the authors a r e pleased to note which unusual precautions were esercised, does ~t seem
the relation between the results of all tests and the likely t h a t control can esist when so many sources of
average values given in Bttlletin 5-C-2. product a r e involved? The authors cannot be conv~nced
2. It is of interest to learn t h a t non-controlled condi- t h a t i t is possible, especially in view of Mr. Hand's
tions were found to exist In the pullout test data not statement: "controlled conditions do exist a t times in
available to the authors a s well a s In the tests t h a t industry, although this is the esception rather than
were available to the authors. As the 211 tests were the rule." No satisfactory proof has been offered t h a t
from pipe "produced under normal operating condi- controlled conditions apply to all casing manufactured
f o r the oil industry. Mr. Hand should not lose sight of
* P~rodtrctioir llrrll A'o. ..?!2J.,item +a, 11 124, a n d Esliibit "J,"
p 245-55 the fact t h a t the operators a r e confronted with t h e
, I<eport of e t h e Sta~irl:irdlzation of Oil-Cocintry
C o m ~ i ~ i t t ron necessity f o r setting casing in wells every day: As a
T r ~ l ~ u l n(.:o~~ds.
r I'rotlrfctro~r Brrll Xo 226
guide in this work, they have only the test data made sympathy with the plan of deliberately selecting steel
available to them by the casing manufacturers. If the of lowest permissible physical properties or of correct-
tests made under non-controlled conditions a s "tested ing the results to allow for such possibilities. I t was
for by statistical calculations" a r e eliminated, a n in- with the possibility suggested by Mr. F r a m e in mind
sufficient number a r e left to satisfy the minimum num- t h a t the minimum value of 85 per cent was suggested
ber of 1,000 given by Mr. Hand. The operators cannot when the lowest result was 91 per cent. Until additional
wait several years for these data, because the problem test data show t h a t a further reduction is needed, there
is urgent already-and, therefore, the policy of waiting seems to be no reason to drop the values a s low a s sug-
until enough data a r e available for the application of gested by Mr. Frame.
statistical methods which he advocates will find little Mr. Holmquist proposes t h a t the effect of tension
favor among operating engineers. If statistically de- loading on collapse strength be calculated by the
termined minimum values a r e used either on all results Hencky-von-Mises equation rather than the maximum
o r on the remaining results, the minimum values a r e strain-energy equation. A t the present time, when so
artificially low and uneconomical. As there can be no few test data a r e available, i t is rather futile to at-
alternative to the use of minimum values, no logical tempt to determine which equation best fits the prob-
procedure is left except t h a t used by the authors. lem, especially a s the difference between them is not
The authors a r e particularly appreciative of the re- large. These considerations a r e in the formative stage,
marks made by Dr. Jasper. He clearly outlined the and the test data a r e seriously lacking in consistency.
ultimate form desirable in the performance specifica- I t is virtually impossible a t this time to judge even the
tions. Simply stated, if the user knows the minimum magnitude of the effect, a s there a r e some data indi-
guaranteed performance properties, he is able t o use cating t h a t the effect is relatively small. Therefore, i t
the casing available to best advantage. I n addition to would seem t h a t the equation used a t present may be
providing the user with the information of greatest largely a question of personal judgment. I n the tests
value to him, a performance specification has a n added made by Miller and Edwards,* two types of failure
virtue in t h a t it presents no obstacle to continuous were noted-stretch failures and collapse failures. We
progress. The authors a r e aware that Dr. Jasper has do not know which type of failure is most common in
visualized this objective f o r a long time. The coopera- service, because there is no opportunity f o r observa-
tive efforts of manufacturers and users of casing will tion. As regards the fit of the data of Miller and Ed-
cause i t to materialize. wards, it is evident from reference to Fig. 6 of Holm-
Mr. F r a m e brought out a question t h a t well may be quist's discussion t h a t the collapse failures a r e in much
given careful consideration. As he pointed out, the closer agreement with the maximum strain-energy
pipe used in the pullout test investigations was probably equation than with the Hencky-von-Mises equation.
of average physical properties. How much reduction in The remarks offered by Professor Schwalbe will be
joint strength might be expected if the physical prop- helpful when the performance specifications a r e being
erties had been close to the minimum specified values? prepared.
Unfortunately, this question cannot be answered defi- The summary of replies to the questionnaire on field
nitely from the few test results available. Inspection practice in setting casing-prepared by Mr. Miller,
of the results given in the paper shows, however, t h a t
chairman of the drilling-practice committee-undoubt-
joint strength is not directly proportional to the physi-
edly will be of great interest to and receive the close
cal properties of the steel. Many examples a r e found in
consideration of operating engineers. Several different
which the reverse is true. Undoubtedly, this is the
practices a r e in use a s regards the release of tension,
resultant of many other factors which enter into joint
strength. The authors believe t h a t tests should be and from this summary each operator will be able to
made on pipe selected a t random, with the proviso t h a t fix upon a practice which best suits his individual
enough tests be made to cover all conditions f o r finished requirements.
casing delivered to t h e purchaser. They a r e not in IJrilling itnd I'roduction P m c t i c e , 483 (1939).