EVALUATION OF THE EFFECTIVENESS OF

API-MODIFIED GOODMAN DIAGRAM

IN SUCKER ROD FATIGUE ANALYSIS

by

M. P. PARAMESWARAN NAMPOOTHIRI, B.Tech.

A THESIS

IN

PETROLEUM ENGINEERING

Submitted to the Graduate Faculty

of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of

MASTER OF SCIENCE

IN

PETROLEUM ENGINEERING

Approved

Accepted

August, 2001
 ACKNOWLEDGEMENTS

I am grateful to the Department of Petroleum Engineering at Texas Tech

University for providing me with competitive departmental graduate scholarship and

giving me the opportunity to reach this stage in my education and development as an

aspiring engineer. I am also thankfiil to Amerada Hess Corporation and other members

of the Production and Lift Optimization Consortium for providing funding and assistance

to carry out this research work.

To my thesis committee chairman, Dr. James F. Lea, I would like to express my

gratitude for all your valuable suggestions and guidance. Your assistance on the

problems I faced to complete the thesis project was extremely helpful. Also, 1 am

thankful to you for providing me with a Research Assistantship all through my graduate

studies.

Most importantly, I am thankful to Dr. James Christian Cox for providing me

with all the help and guidance to conduct the experiments and write this thesis. It is only

with your timely advice and directions that I could finish this thesis project in a timely

and professional maimer. Also, I am thankful to Dr. Javad Hashemi for agreeing to be on

my thesis committee as well as providing me with valuable suggestions and patiently

answering all my questions.

I would also like to express my thanks to Dr. Paulus Adisoemarta for all his

assistance in conducting the experiments. His timely suggestions and guidance on the

experimental equipment can never been forgotten.

II
 I would also like to acknowledge Dr. Akanni Lawal's assistance in my efforts to

complete my duties as a graduate student. To Dr. Akanni Lawal, I am grateful to you for

assisting me as a graduate student advisor and providing me with all the guidance to

successfully reach this stage.

I am thankful to Joan Blackmon, Jeimifer Weitman, Joseph Mclnemey, Margaret

Ceja, and my friends at Texas Tech University for their cooperation.

I am so grateful to my wife, Devakikutty Leena, and son, M. P. Anand, for all the

continued support and encouragement and for strongly standing by me to achieve our

goals in life.

Finally, I am dedicating this thesis to my parents for all their encouragements and

blessings and for making me realize the value of education.

Ill
 TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii

ABSTRACT vi

LIST OF TABLES ix

LIST OF FIGURES x

NOMENCLATURE xi

CHAPTER

L INTRODUCTION 1

n. LITERATURE REVIEW 8

History of Fatigue Testing 8

History of Goodman Diagram 13

Evolution of API Modified Goodman Diagram 16

Hardy's Paper on Goodman Diagram 17

A New Look at Sucker-Rod Fatigue Life 19

Low-Cycle and High-Cycle Fatigue 20

Fatigue Loading 21

Application of API Modified Goodman Diagram

in Sucker-Rod Selection 23

Factors Affecting Fatigue Strength 28

Types of Sucker Rods 30

m. MTS FATIGUE TESTING MACHINE 31

IV
 rv. PROCEDURE 36

Overview 36

Specimen Preparation 39

Specimen Storage 41

Specimen Dimensions 41

Inspection Prior to Testing 41

Specimen Loading 41

Defining Test Parameters and Running the Test 43

Removal and Reloading of the Specimen 45

Test Control 46

V. RESULTS AND DISCUSSIONS 50

VL CONCLUSIONS 53

Vn. SUGGESTIONS AND RECOMMENDATIONS 54

BIBLIOGRAPHY 56
 ABSTRACT

In petroleum engineering applications an understanding of the fatigue behavior of

metals is of great importance primarily in the selection of sucker rods. Of all the

components of the oil-well pumping mechanism, sucker rods have the greatest effect on

the action and performance of the whole pumping system. The behavior of the sucker

rods in the transfer of forces and loads determines the action of the pump. Their inherent

characteristics affect the loads imposed on the surface equipment. For years, the

Goodman diagram is the basis of analyzing the endurance (fatigue) life of ferrous

material. Goodman published the diagram during the year 1908. An American

Petroleum Institute (API) Committee suggested several design factors for Goodman

Diagram in order to adapt it specifically to specifically sucker-rod pumping installations.

The diagram is known as the API Modified Goodman diagram. The committee had

suggested the following revisions to original Goodman Diagram. They are: (1) The apex

of the diagram should be set at the tensile strength of the material. (2) The >^-intercept

should have a factor of safety of two. (3) The safety factor on the material tensile strength

apex should be 1.75.

The result was the creation of API Modified Goodman Diagram. The design

formula for the allowable stress line is given as:

S,={TIA^MS^JSF, (1)

where

VI
 Sg = Allowable maximum stress, psi,

T= Minimum tensile strength, psi,

M= Slope of the allowable stress line = 0.5625,

S^ = Minimum stress, measured or calculated, psi,

SF = Design Safety Factor.

Since this diagram is used for the design of permissible stresses for sucker rod

installations, it is imperative that the design engineer understands the loads imposed on

the sucker rods; however, conventional polish rod dynamometer analysis does not

provide sufficient information regarding down-hole load conditions.

Since the rods available today is superior due to the improved manufacturing

processes and testing facilities, this thesis is an experimental study the effectiveness and

accuracy of the "API Modified Goodman Diagram." Tests conducted at 160% of the

stress range as specified in the API Modified Goodman Diagram on 5/8 inch. Grade 'D'

sucker rod specimens resulted in the following findings:

1. The Y-intercept of the diagram can be changed to r/2.5 as compared to T/4

given by the API Modified diagram.

2. The slope of the allowable stress line is found to be 0.3 as compared to the

0.5625 given by the Modified Goodman diagram.

In effect, the test results indicate that the present API Modified Goodman diagram is

conservative and the formula for the allowable stress line could be revised to

VII
 5, =(7/2.5+ M5^)SF (2)

where,

S„ = Allowable stress, psi,

T = Minimum Tensile Strength, psi,

M= Slope of the allowable stress line = 0.3,

-^min ^ Minimum Stress, psi,

5F= Safety Factor.

VIll
 LIST OF TABLES

Sample Test Results of Krupp's Axle Steel 14

Typical Sucker-Rod string Analysis 27

Experimental Test Data 37

IX
 LIST OF FIGURES

1. Original Goodman Diagram 3

2. Hardy's suggested modifications to Goodman Diagram 5

3. Modified Goodman Diagram suggested by Hardy 6

4. API Modified Goodman Diagram 7

5. Wohler's test Arrangement and S-N Diagram 9

6. Typical Rotating-Bending Test Arrangement 11

7. Typical Rotating-Bending Test Machine 12

8. Several Constant Amplitude stress-Time Pattems 23

9. Typical Dynamometer Card 26

10. MTS Fatigue Testing Machine 32

11. Closed Loop Control Diagram 34

12. Diagram Showing the Stress Range 38

13. Diagram Showing the Load Range 39

14. Figure Showing the Test Definition Windows 47

15. Figure Showing the Setting of Execution Parameters 48

16. Figure Showing the Execution Display 49

17. Suggested Revision to API Modified Goodman Diagram 52

 NOMENCLATURE

cr^^ - Maximum stress in the cycle, psi,

cr^„ - Minimum stress in the cycle, psi,

cr^ - Mean stress = —^^ ^^, psi.

cr„ - Ahemating stress amplitude = ^ ^^ ,psi,

Aa - Range of stress = a^^ - a^ ,psi,

T - Tensile strength ofthe material, psi.

R - Stress ratio = - ^ ^
C7
max

S^ - Allowable stress, psi,

SF - Safety Factor.

^niin " Minimum/static load applied to a specimen, lbs,

W^^ - Maximum load applied to a specimen, lbs,

^maxi " Maximum load applied at 160% stress range, lbs,

W^ - Equivalent static load, lbs,

S^^ - Maximum allowable stress, psi,

S^^ - Minimum stress, psi,

xi
 CHAPTER I

INTRODUCTION

Materials subjected to fluctuating or repeated loads tend to develop a

characteristic behavior, which differs fundamentally, in certain respects, from their

behavior under steady loads. This behavior is called fatigue and is distinguished by three

main features: loss of strength, loss of ductility, and an increased uncertainty in both

strength and service hfe. All material, even the most perfect crystals, is full of

imperfections of many kinds so that no material is perfectly homogeneous. Under repeat

loading, these imperfections in the material results in premature failure.

Metal fatigue can be defined as a general term used to describe the behavior of

materials under repeated cycles of stress and strain which cause a deterioration ofthe

material that results in a progressive fracture. Factures resulting from fatigue are among

the most difficult to foresee because conditions producing fatigue are frequently not

clearly recognizable. Fatigue occurs at stresses well within the ordinary elastic range as

measured in the static tension tests. Like other forms of brittle behavior, it is strongly

influenced by what may appear to be minor discontinuities in structure ofthe material.

Because ofthe difficulty of recognizing fatigue conditions, fatigue failures comprise a

large percentage ofthe failures encountered in engineering.

In petroleum engineering applications, an understanding ofthe fatigue behavior of

metals is of great importance primarily in the selection of sucker rods. Of all the

components ofthe oil-well pumping mechanism, sucker rods have the greatest effect on

the action and performance ofthe whole pumping system. The behavior ofthe sucker

1
rods in the transfer of forces and loads determines the action ofthe pump. Their inherent

characteristics affect the loads imposed on the surface equipment.

Permissible stresses, range of stresses, and speed of operation, which a given

sucker-rod string can handle without too frequent breakage, limits the pumping depth and

the economics ofthe installation. Because ofthe important role the rods play in the

system, the study ofthe fatigue strength and factors goveming the selection of sucker

rods is of great importance.

Most sucker rod failures are due to the result of overstressing. The mechanism of

sucker rod failure is that of fatigue and corrosion fatigue. Under conditions of repeated

loadings, a material will fail under a load smaller than that which will cause failure on

one or a few applications. The ability of a material to resist fatigue failure is measured by

the "endurance limit." Endurance limit is defined as the maximum stress, which may be

applied indefinitely, under conditions of repeated loadings, without causing the failure.

The endurance limit of material is determined on a testing machine by subjecting the test

specimen to reverse loadings from maximum tension to maximum compression. Several

specimens are tested to failure under gradually decreasing stresses until the stress is

found under which the specimen can withstand an indefinite number of cycles of

reversals. In the case of steel it has been found that a stress, which permits 10 million

reversals without failure, can be safely assumed as the endurance limit.

Fatigue and endurance limit, assuming noncorrosive conditions, are affected by

maximum stress, range of stresses, and number of reversals. When we consider sucker

rods, this statement can be interpreted as meaning that, for a given set of non-corrosive

conditions, the frequency of failures of rods depends on peak polished-rod load, range of
loadings, and speed of operation. The latter govems the number of changes of stresses

from minimum to maximum tension.

The Goodman diagram as shown in Figure 1 illustrates the principle that the

minimum stress must increase with increase in maximum stress in order to maintain the

allowable range of stresses required to offset fatigue failure.

Figure 1

Original Goodman Diagram. (Source: Goodman, J.: Mechanics Applied to

Engineering, Longmans Green and Co., New York, (1908) p. 539.)

The theory behind Goodman Diagram is discussed in detail in the coming chapters. The

diagram shows only the pattern ofthe allowable stresses, emphasizing the importance of

the range of stresses. It is of interest to note that in deep well pumping, with weight of
sucker rods representing a very large part ofthe polished-rod load, conditions are

favorable for maintaining a narrow range between the upstroke and dowm stroke loadings.

In the medium and shallow depth pumping of large volumes of fluid, particularly with

high speeds of operation, the down stroke load may be low compared to the upstroke load

and caution must be exercised in selecting the allowable maximum stress to maintain the

proper range of stress.

In order to take care of these types of conditions, an API committee had suggested

modifications to the Goodman Diagram. Hardy^ explains the factors which led to the

conclusions made by the API committee in modifying the Goodman Diagram as

illustrated in Figure 2. Figure 3 shows such a modified Goodman Diagram constructed

for hypothetical steel. Hardy's explanations are illustrated in detail in the coming

chapters. Because ofthe compressive stresses in rods, if they do occur, have highly

damaging effects; the line ofthe minimum stress is started at the zero stress line. The

corresponding point of maximum stress is found to be located at half the laboratory

determined air-fatigue limit ofthe steel. Hardy considered different safety factors for

constructing the diagram, as the modem day manufacturing standards and quality control

are superior to the standards followed during the Goodman time and to represent various

service conditions pertaining specifically to sucker-rods.

Sucker rod manufacturers had different opinion regarding the safety factors.

Some manufacturers used minimum yield strength for the apex whereas others chose
 ULTIMATE STRENGTH 100,000 PSI
1 1 1 -
^ U
1 —1—
tJ
_L . 1 ^ ^ ^ >

^
I 6 8 , 4 2 0 PSI
^ ~^
•y^~
'
50,0uu Hbl
— t — 1 — 1 — 1 ^ — ,
^ B ^P==T"
>M /
/
^ ^
1 ! ^ "1^=^^
^ 1^ TENSION , B^
^•-h- 33.00Q/PS'| / V- 1
1
1 I
" ^ ^
/ I1 —'—^ / \ j
1 : j '
1

\/ 1
1 ^ i-^ Tt" j —f—1

l-U
^—i— 1 1 /' ——
^ COMPRESSION

1
1 i
• y / I1
1
i t

— j — A 33,000 PSI 1 \/
f^^-n^
•' ; 1
1 1 ! i 1 > 1 1 1
A'
A B

Y Y
! I u I
1
(5 8 A5>n P<=: 1 —^
Ll n1 '^ 6 5 % ULIINIA I t STRENGTH-
> —•

^ 1
!
-t /
^^t^/ /
^7 3 /
^rfj

S MAX
D —7
V,'
^
D
\^ 1
1 fA
A
1
S MIN- M N STRE 58

A" 1
C D

Figure 2

Hardy's Suggested Modifications to the Goodman Diagram. (Source: Hardy, A.

A.: "Sucker Rod String Design and the Goodman Diagram," paper 64-Pet2,
presented at the 1964 Petroleum Mechanical Conference ofthe ASME, Los
Angeles, Sept. 20-23.)

too use the ultimate tensile strength. In order to have a consensus in the usage of

Goodman diagram, the American Petroleum Institute task force put forward several

suggestions and decided upon the Modified Goodman diagram given in 'API

recommended practice-11 BR (RP -1IBR).

This diagram is shown in Figure 4. This diagram is the engineering design basis for

sucker rod selection and is currently used by the oil industry for determining the

permissible loads for sucker rod used in beam pumping installations.

1
i
y ^ /
60
\ /
o
o
o

/
/
OL cir-
10
/
tn
UJ (y
(T
I- ^^ /
30
^> /
s /
D
ro _H if^
^
2 /
X /
<
z /
I
X
< /
2 10
V) /

/
/
10 20 30 40 50 60

S MIN- MINIMUM STRESS PS.I X 1,000

Stress range diagram AISI 1036 sieel (non-corrosive service)

Figure 3

Modified Goodman Diagram Suggested by Hardy. (Source: Hardy, A. A.

"Sucker Rod String Design and the Goodman Diagram," paper 64-Pet2,
presented at the 1964 Petroleum Mechanical Conference ofthe ASME,
Los Angeles, Sept. 20-23.)

The purpose of this research project is to evaluate the validity ofthe Modified

Goodman Diagram as per today's standards. The testing method used to arrive at the

Modified Goodman Diagram was not the stress controlled axial cyclic testing but with a
reverse side bending tests. In addition, we are depending on some unverified safety

factors for correcting for the test method used.

Because ofthe above reasons, it is thought that the testing ofthe rods in true

varying axial loading will resuh in much better evaluation ofthe fatigue life of sucker

rods as well as a possible revision to the Modified Goodman Diagram.

This thesis presents experimental results conducted on 5/8-inch Grade D sucker

rod specimens and interprets these results to demonstrate how these results affect the

Modified Goodman Diagram.

T MINIMUM TENSILE STRENGTH T

T/4 + MSMIN)SF 1
SA = {0.25T + 0.5625 SMWISF 2
AS^ = S^ -
where
SA = maximum allowable stress, psi
^SA = maximum allowable range of stress, psi
M = slope of SA curve = 0.5625
S/vf//v = minimum stress, psi (calculated or measured)
SF = service factor
7" - minimum tensile strength, psi

Figure 4

API Modified Goodman Diagram. (Source: Donnelly, R. W.: Beam

Pumping, Petroleum Extension Service, U. of Texas at Austin, (1986) p.
43.)
 CHAPTER n

LITERATURE REVIEW

History of Fatigue Testing

The first fatigue investigations seem to have been reported by German mining

engineers, who in 1829 performed some repeated loading tests on iron chain. Some of

the earliest fatigue failures in service occurred in the axles of stagecoaches. When

railway systems began to develop rapidly about the middle ofthe nineteenth century,

fatigue failures in railway axles became a widespread problem that began to draw the first

serious attention to cyclic loading effects. This was the first time that many similar parts

of machines had been subjected to millions of cycles at stress levels well below the yield

point, with documented service failures appearing with disturbing irregularity.

Wohler, a German railway engineer, set up and conducted the first systematic

fatigue investigation. He conducted tests on fiill-scale railway axles and also small-scale

bending, torsion, and axial cyclic loading tests for several different materials. Some of

Wohler's original data were reported as shown in Table 1. These data for Krupp axle

steel are plotted in Figure 5 on what has become known as the S-N Diagram. Fatigue

data today are often presented in much the same way.

Typical arrangement ofthe ahemating bending cantilever tester introduced by

Wohler is shown in Figure 5. This arrangement has only undergone minor changes since

it was introduced a century ago. The maximum fiber stress occurs at the point where the

8
cylindrical part of specimen B merges with the enlarged tapered end T. The bending

My
moment being, F^, the stress is computedfi-omequation a^ = —^ where

Zeitschr. f Bouwesen I860 Apporat zum Probiren der Widerstondsfohiqkeil

von Woqen-Achsen gegen xiederholte Biequngen

! a

Wohler's machine for fatigue testing of railway axles.

800 - ^

Unnotched
[Steel supplied in 1862)
600

400 Shorp Shoulder

(Steel supplied in 1653

200
10" to' 10*
Cycles to Foilure

Wohler's S-N cur\'es for Knipp axle steel.

Figure 5

Wohler's Test Arrangement and S-N Diagram. (Source: Handbook of Fatigue

Testing, S. R. Swanson (ed.), American Society of Testing and Materials,
Philadelphia, (1974) p. 1.)
M is the bending moment, y is the distancefi-omthe load, and / is the moment of inertia.

The force, F, is developed by adjustable loading springs, P. Today, dead weights are

used in modem machines in order to induce the lateral force. Noting that a^^ depends

on both the bending moment and the section modulus, it is possible, by using a suitably

tapered specimen, to locate cr^^ at some distancefi-omthe clamped end. The concem

with the stress concentration due to the fillet is thus eliminated.

It is to be noted that the original Goodman Diagram was prepared on the basis of

experiments conducted in the above maimer.

Another arrangement of rotating-bending machines ofthe constant bending

moment type is shown schematically in Figure 6. A general view ofthe machine and test

arrangement is shown in Figure 7. With this type of device, the region ofthe rotating

beam between the inboard bearings is subjected to a constant bending moment all along

the specimen length.

While under the influence of this constant moment, the specimen is caused to

rotate with the drive spindles about a longitudinal axis. Any point on the surface is

thereby subjected to a completely reversed stress-time partem.

The cantilever type of rotating-bending fatigue testing machine is also shown

schematically in Figure 6. It is very similar to the rotating-bending machine mentioned

above except that the bending moment varies along the beam, making the axial location

ofthe critical section important in applying the proper stress level. The stress-time

pattem is completely reversed unless some special device is added to produce an axial

stress component. The machine is basically ofthe constant load variety.

10
 All the above arrangements show that the data obtainedfi-omthese types of tests

are the basis ofthe original Goodman Diagram; however, this does not represent the real

values, which can be deduced by testing the samples in constant amplitude, stress

controlled cyclic fatigue testing. Axial load fatigue testing systems subject the

. L»«d O M ' " ^ '"^ $%«OOOrt btvr^

SuOPOn tM«/>n9 /
/
So#Cim«o

a ^ L r
t^A^
\J

/7777777 /77777777

i
mm _.
jg^i^^rr^-^ M

Figure 6

Typical Rotating-Bending Test Arrangements. (Source: Collins, J. A.:

Failure of Materials in Mechanical Design, T^ Edition, John Wiley &
Sons, New York, 1993) p. 189.)

11
Figure 7

Typical Rotating-Bending Test Machine. (Source: Wohler's Test

Arrangement and S-N Diagram. (Source: Handbook of Fatigue Testing, S. R.
Swanson (ed.), American Society of Testing and Materials, Philadelphia,
(1974) p. 66.)
12
test specimens to a stress (or strain) distribution that, for an tm-notched specimen, is

reasonably uniform throughout the cross section. The major advantage ofthe axial-load

system is that it produces a relatively uniform load distribution over a relatively large

volume of material in the specimen, thus eliminating the perturbing effects ofthe stress

(or strain) gradient present in the bending tests. Even though MTS type testing machines

were developed, there are no published data pertaining to the sucker-rod testing.

History of Goodman Diagram

The Goodman Diagram was developed by the interpretation of results of

experiments conducted by Wohler on the repetition of stresses on ferrous materials. In

these experiments, various materials were subjected to tensile, compressive and torsional

stresses, which were wholly or partially removed, and in some instances were even

reversed from tension to compression in the same bar. The intensity ofthe stresses were

at first large, almost approaching the static tensile strength ofthe material. Such stresses

caused fracture after a small number of repetitions, but as the intensity ofthe stresses was

reduced, the number of repetitions beforefi-acturerapidly increased. After a time a low

limit of stress was reached, it appeared that the bar was capable of withstanding an

infinite number of repetitions. This stress appears to depend upon the ultimate tensile

strength ofthe material and the magnitude of stressfluctuations,often termed as the

"range of stress," to which the material is subjected.

Goodman"* used these test results to develop a permissible stress diagram that

bears his name. The experimental results used to develop the Goodman Diagram are

shown in Table 1.

13
 Table 1

Experimental Data Set Used by Goodman. (Source: Original Goodman

Diagram. (Source: Goodman, J.: Mechanics Applied to Engineering,
Longmans Green and Co., New York, (1908) p. 538.)

MATERIAL.
A'ru/'/s AxU SUel.
Tensile strength, varying from 42 to 49 ton, per sq. inch.
1
Tensile stress applied Nominal bending stress
in tons per square Number of Number of
repetitions before in tons per square
inch inch repetitions before
from I to fracture. fracture.
from I to

0 38-20 ; i»,74i O 26-25 1,762,000

0 33-40 i 46,286 O 25-07 1.031,200
0 28-65 1 170,170 o 24-83 1,477.400
0 2614 i 123,770 o 23-87 5,234,200
0 23-87 473,766 o 40,600,000
23-87
0 22-92 1 13,600,000 (unbrolcen)
(unbroken)

Nonunal bending stress in ;

revolving axle Number of repetitions
from to before fracture.

55.'oo
•27,775
797,525
642,675
1,665,580
3,114,160
4-163,375
45,050,640

MATERIAL.
Krupp's spring Stftl.
Tensile strength, 57-5 tons per sq. inch.

Tensile stress applied Number of Tensile stress applied

repetitions before in tons p«r square repetitions before
in ch fracture. in ch
from fracture.
from to

4775 7-92 62,000 38-20 4-77 99,700

1^ 15-92 149,800 9-55 176,300
,. 23-S7 400,050 '433 619,600
M 27-83 376,700 »» j M 2,135,670
»' 31-52 19,673,000 1 i9'io 35,800.000
(unbroken) (unbroken)
4295 9-55 81,200 33-41 4-77 286,100
M 14-33 1,502,000 >, 9-55 701,800
»» IQ-IO 225,300 n-94 36,600,000
M 23-87 1,238,900 (unbroken)
•1 )» 300,900
•1 38-65 33,600,000
(unbroken)

Goodman had based his arguments on the "dynamic theory" which assumes that

the varying loads applied by Wohler were equivalent to suddenly applied loads, and

14
consequently a piece of material will not break under repeat loadings if the momentary

stress of sudden applications, does not exceed the static yield strength ofthe material.

Goodman attempted to show, through his diagram, the results of all the

experiments reduced to a common scale. Goodman's"^ explanation regarding his diagram

is:

In every case represented, the material stood over foiu- million repetitions before
fi*acture. The horizontal scale is immaterial; the vertical scale shows the ratio of
applied stress to the static breaking stress. The minimum stress on the material is
plotted on the line aob, and the corresponding maximum stress, which may be
repeated four million times, is shovm by the small circles above. If the dynamic
theory were perfectly true, all the points would lie on the line marked "maximum
stress"; for then the minimum stress (taken as being due to a dead load) plus twice
the range of stress {i.e. maximum stress - minimum stress) taken as being due to a
live load should together be equal to the statical breaking strength ofthe material,
(p. 540)

The results ofthe tests of revolving axles are shown in-group A; the dynamic theory

demands that they should be represented by a point situated 0.33fi-omthe zero stress

axis.

Likewise, when the stress variesfi-omzero to a maximum, the results are as

shown at B', by the dynamic theory, they should be represented by a point situated 0.5

fi-om the zero axis. For all other cases, the upper point should lie on the maximum stress

line.

So, Goodman concludes that when designing a member which will be subjected

to both a steady load (W^^) and a fluctuating load (W^^ - W^^), the equivalent static

load (^0) is given by

l^=f^„±2(^x-^.„) (3)

15
The plus sign is used when both the loads act together (when both are tension or both are

compression), and the minus sign when they act against one another.

Evolution of API Modified Goodman Diagram

Hein and Hermanson^ describe the evolution ofthe modified Goodman diagram

in their paper " A New Look at Sucker Rod Fatigue Life."

The Goodman Diagram shown in Figure 1 compares any conditions of ahemating

load or stress between the extremes of complete reversal to a static stress equal to the

tensile strength ofthe material. A curve generated usually shows the expected stress

range for a given life or cycle time ofthe material. This diagram shows that for every

minimum load, there is a maximum load to provide the required service life. The smaller

the load range, the higher the maximum allowable load.

The basic diagram was developed using short, polished metal specimens to reduce

the variables and achieve relatively consistent results. This was done in a short period of

time by running tests at very fast cycle rates of approximately 1750 rpm to develop an

endurance limit-fatigue life expectation of 10 million cycles. Fatigue diagrams normally

assume an infinite life if the threshold of 10 million cycles is achieved.

As the original tests were conducted with short, polished specimens, the

application ofthe Goodman Diagram in designing sucker rods caused factors of safety to

be considered. Sucker rod manufacturers had different opinion regarding the safety

factor to be applied. Some manufacturers used yield strength for the apex whereas some

have chosen to use the tensile strength. Various factors were used for both the tensile

strength and left hand portion ofthe diagram instead of using the K-intercept of one-half

16
ofthe tensile strength. Each manufacturer vigorously defended their companies

approach.

In order to have a consistency in the application ofthe Goodman diagram to

sucker rods, an American Petroleum Institute task force was formed to develop a

consensus to this issue. The committee met in the early 1960s at the Mayo hotel in Tulsa,

Oklahoma. The following were the conclusionsfi*omthe committee meeting.

1. The committee had agreed that the apex ofthe diagram should be the tensile

strength.

2. The y-intercept should have a factor of safety of two. This reduced the

intercept to the tensile strength divided by four.

3. The factor of safety on the tensile strength apex was recommended to be 1.75.

The result ofthe above was the API Modified Goodman Diagram, which is published in

the API recommended practice on care and handling of sucker rods.

Hardy's Paper on Goodman Diagram

The original Goodman diagram as shown in Figure 1 represents the combined

effect of steady and ahemating stress on the fatigue limit. In constructing the diagram it

is assumed that the fatigue limit for complete reversal is one-third ofthe ultimate static

tensile strength.

Hardy' in his paper "Sucker-Rod String Design and the Goodman Diagram"

presented the reasons for the modification ofthe Goodman Diagram by the API

committee for usage in the Sucker Rod selection. To illustrate the derivation of a usable

stress-range diagram for sucker rod applications. Hardy selected AISI C-1036 steel, a

17
material used by all rod manufacturers and designated as Grade C in the API standard

BRl 1. Summary of his explanations on the reasons for the modification ofthe diagram

are detailed below.

The manner in which the endurance limit varies with different ranges of stress

was first illustrated by the original Goodman Diagram. The steels of Goodman's day

were not forged to the standards as present day manufacturing processes. Depending on

the alloy and heat treatment, it is found that the endurance level is varyingfi-om48 to 56

percent ofthe tensile strength as against 33 percent ofthe tensile strength, which was

used for the original Goodman Diagram. Using fatigue endurance tests carried out in the

lab for normahzed AISI C-1036 steel with an average ultimate tensile strength of 100,000

psi, an endurance limit of 52,000 psi is found. With this information, a diagram is

prepared as shown in Figure 2. Here A'B' represents the laboratory determined

endurance limit of 52,000psi and AUand B'Urepresent the stress range limits. The

original Goodman Diagram is superimposed for comparison purpose.

As these diagrams are drawnfi-omthe results of laboratory tests on carefiilly

polished specimens under carefiilly controlled conditions, it is suggested that a suitable

safety factor shall be applied in deriving the modified diagram which will illustrate

usable working stress levels and ranges for designing sucker rod strings.

First of all, as sucker rods normally do not operate in compression, the portion of

the diagram lying to the left ofthe zero ordinate is eliminated as shown in Figure 2. This

leaves triangle OCU . Next, since operation at or near the ultimate strength is

impractical. Hardy suggests to reduce it by 35 percent, using 65 percent ofthe ultimate

strength as the apex ofthe triangle. This 65 percent was based on the years of experience
18
of Hardy and others with the knowledge that rod string would hang in a non corrosive

well at that stress with no cyclic load variations indefinitely. The following method was

used to find out the working y-intercept.

S l o p e ^ ' ^ = '""'Q°Q-"'Q"Q=0.31578
100,000 + 52,000

The y-intercept at C',0C = 52,000*1.31578 = 6S,420psi.

Creech^ states that the reverse bending values ofthe fatigue endurance limit as

determined by the rotating-beam method must be reduced by a factor of 0.72 for pull-pull

endurance limits.

As the surface ofthe rods is shot-blasted, Jones^ suggests that a stress

concentration factor of 1.25 to be used which ftirther reduces the above figure by a factor

of 0.80. Also, a safety factor of 1.5 is applied for operating unknowns, which ftirther

reduces the figure by 0.66. Applying these factors to the y-intercept at C , the Y-

intercept for the modified stress range diagram becomes,

y-intercept = 68,420 * 0.72 * 0.80 * 0.66 = 26,000psi (4)

These suggested modifications are shown in Figure 2 and Figure 3.

A New Look at Sucker Rod Fatigue Life

Hein and Hermanson^ have addressed the necessity of revising the API Modified

Goodman Diagram for Sucker Rod selections. As per them, the following factors

contribute to their argument.

19
 The current sucker rod bar stock is a hot rolled product whereas thirty years ago

when API has recommended the usage ofthe Modified Goodman Diagram, bars were

produced by major steel millsfi-omingots. Today, domestic manufacturers use mini-

mills with continuous casting and up-graded rolling processes. This produces a

smoother, more consistent rod product with less inclusions or discontinuities.

Additionally, improved workmanship requirements were developed by an API task force

group during 1983. This provides tighter requirements for straightness, smoother surface,

and limited surface discontinuities. Current day manufacturers also have better control

on the sucker rod chemistry, increased cleanliness ofthe steel, finer grain size and better

heat treatment requirements. They speculated that all these improvements should result

in the minimum cycles to failure increasefi-om10 million cycles to over 50 million

cycles.

Hem, and Hermanson also argued that, although the Modified Goodman

Diagrams is considered a usefial tool to determine the maximum allowable stress for a

given minimum stress, it is a very conservative theory ofthe effect of mean stress on the

fatigue life. They predict that the Gerber theory may be more reliable if all the applicable

strengths are accurately known.

Low Cycle and High Cycle Fatigue

Fatigue failure investigations over the years have led to the observation that the

fatigue process embraces two domains of cyclic stressing or straining that are

significantly different in character, and in each of which failure is probably produced by

different physical mechanisms. One domain of cyclic loading is that for which

20
significant plastic strain occurs during each cycle. This domain is associated with high

loads and short lives, or low number of cycles to produce fatigue failure, and is

commonly referred to as low-cycle fatigue.

The other domain of cyclic loading is that for which the strain cycles are largely

confined to the elastic range. This domain is associated with lower loads and long lives,

or high number of cycles to produce fatigue failure, and is commonly referred to as high-

cycle fatigue.

Low-cycle fatigue is typically associated with cycle livesfi-omone up to 10"* or

10 cycles, and high-cycle fatigue for lives greater than 10^ cycles. As the sucker rod

lives are expected to be above 10 million cycles, we are concentrating on the high-cycle

fatigue criteria.

Fatigue Loading

There is different fatigue response of a material to various loading that might

occur throughout the design life of a part under consideration. A designer is interested in

the effects of various loading spectra and associated stress spectra, which will in general

be a fiinction ofthe design configuration and the operational use ofthe part.

The simplest fatigue stress spectrum to which an element may be subjected is a

zero-mean sinusoidal stress-time pattem of constant amplitude andfixedfi-equency,

applied for a specified number of cycles. Such a stress-time pattem, often referred to as

completely reversed cyclic stress, is shovm in Figure 8(a).

A second type of stress-time pattem often encountered is the nonzero mean

spectrum shown in Figure 8(b). This pattem is very similar to the completely reversed

21
case except that the mean stress is tensile or compressive, in any event different fi-om

zero. The non-zero mean case may be thought of as a static stress equal in magnitude to

the mean cr^, with a superposed completely reversed cyclic stress of amplitude G^ .

Another case of nonzero mean stress is illustrated in Figure 8(c). In this special

case, the minimum stress ( T ^ is zero. The stress rangesfi-omzero up to the tensile

maximum and then back to zero. This type of stressing is often called released tension.

For released tension it may be noted that the mean stress is half the maximum stress, or

Sucker rods are subjected to a pull-pull stressfi-oma maximum in tension on the

upstroke to a minimum tension on the down stroke, and back again. This follows a

pattem as shown in Figure 8(c).

22
 ha
in

^ mm

(a)

•^-Time

(b)

t/j

Om.n = 0 - ^ Time

Figure 8

Several Constant Amplitude Stress-Time Pattems

(a) Completely Reversed (b) Nonzero Mean Stress (c) Released Tension
(Source: Collins, J. A.: Failure of Materials in Mechanical Design, 2"^* Edition,
John Wiley & Sons, New York (1993) p. 185.)

Application of API Modified Goodman Diagram in Sucker-Rod Selection

The most basic property of any sucker-rod string is its elastic behavior, which is

responsible for the complexity of its operation. The forces that excite the string at its two

ends produce elastic force waves that travel in the rod material at the speed ofthe sound.

These waves are of different magnitude and phase, and their interference and reflection
23
can greatly affect the actual forces that occur in any rod section. Due to the complexity

of describing these force waves, most rod string design procedures disregard the rod

loads arisingfi-omsuch effects.

The possible rod loads during a complete pimiping cycle at any depth in the string

can be classified into the following groups:

1. Weight of rods. This force is distributed along the string. At any section, it is

equal to the weight ofthe rods below the given section. It is positive for both

the upstroke and the down stroke.

2. Buoyant force. This force always opposes the rod weight and is equal to the

hydraulic lift caused by immersing the rods into the produced liquid. It is

customary to handle the sum ofthe rod weight and buoyant force by using

specified rod weights in fluid.

3. Fluid Load. This is concentrated force acting at the bottom ofthe string only

during upstroke, and it equals the force resultingfi-omthe net hydrostatic

pressiu"e ofthe fluid lifted, acting on the area ofthe pump plimger. It is always

positive.

4. Dynamic Loads. They are the results of changes in acceleration, during the

pumping cycle, ofthe moving masses (rods and fluid column). The

magnitude and direction of these forces is constantly changing during the

pump stroke, but generally dynamic loads result in a positive net load for the

upstroke and a negative load for the down stroke.

5. Friction Forces. These forces are of two kinds: fluid ftiction and mechanical

fiiction. Fluid is moving with the rods during upstroke and against the rods

24
 on the down stroke. Mechanical fiiction forces oppose the movement of rods;

they are positive during upstroke and negative during down stroke.

An examination of these forces during a complete pumping cycle shows that the

rod string is exposed to a cyclic loading. Although the upper rods are always in tension,

the tension level considerably increases during the upstroke due to the load ofthe fluid

lifted, the dynamic loads, and the fiiction forces. The down stroke load consists of only

the buoyant weight ofthe rods minus dynamic loads and fiiction forces. Thus the loading

ofthe sucker-rod string is pulsating tension, which must be accoimted for its mechanical

design. This is why the rod string has to be designed for fatigue endurances, as done in

most ofthe present-day procedures using the API Modified Goodman Diagram. Figure 9

and Table 2 presents an example of analysis of a sucker rod string.

API RP UL gives details about different steps of calculations to be performed in

analyzing a pumping system. Dynamometer cards as shown in Figure 9 is prepared based

on the APIRP IIL specification. Peak Polished Rod Load {PPRL) and Minimum

Polished Rod Load {MPRL) are calculated. PPRL represents the maximum load and

MPRL represents the minimum load. Analysis is made to see the suitability ofthe

specific rod as per these conditions using the "API Modified Goodman Diagram" as

shown in Table 2. Percentage Goodman range of stress is given by.

( S -S ^
% Goodman = max( actual) min * 100, (5)
C _ c-
V^ '^ imx,allowable ^ mm J

where

S^^,„o^>,,=(.T/4 + 0.5625S^)SF (6)

25
 PolishedRod

Effective Pump Load

Load (1000 Ibf) Perm. Load - In Balance

20r

—I
50 75 100
150
Position from Bottom of Strolte (in)

Figure 9

Typical Dynamometer Card.

26
Table 2

Typical Sucker-Rod String Analysis.

Counterbalance Weight 5000 Ibf

Move Counterbalance weight 0 in IN to balance torque
Average Pump Speed 6.00 SPM
Polished Rod Stroke 145.87 in
Maximum load (PPRL) 17639 Ibf
Minimum Load (MPRL) 7418 Ibf
Beam Loading 41.3% Rated
Required Motor Input Power 17.2 hp
Rotation CCW
Polished Rod Power 12.7 hp
Maximum Torque-upstroke 448.2 M in-lbf
Maximum Torque-downstroke 440.7 M in-lbf
Gearbox loading 70 % Rated
Required Motor Name Plate 21.0 hp

ROD STRING ANALYSIS

Rod Diameter Length Max. Stress Min.Stress % Goodman Range

(in) (ft) (psi) (psi) 1 0.95 0.9 0.9 0.8 0.6
160 170 180 190 200 240
1C 1 1357 22459 9445 71 77 83 92 102 180
2C 0.875 3000 21538 6848 76 81 87 95 104 164

Rod Node Depth(ft) Stroke Power (hp) Max. Load Min. Load Min.Eff.Load
Length (in) (Ibf) (Ibf) (Ibf)

Polished Rod 0 145.87 12.7 17639 7418 7508

Top- Rod 2 1357 12978 4118 4491
Pump 4357 133.98 10.3 4015 -1047 0

27
The Percent Goodman as calculated above will show whether rod size is suitable for a

given beam pump installation. When the value ofthe percent Goodman is below 100%,

it implies the rod is capable of withstanding the predicted operating stress ranges.

Factors Affecting Fatigue Strength

The general factors that affect the fatigue strength of materials are illustrated

below.

Load Spectrum

Fatigue strength of any given material is influenced by the load spectrum to which

the material is exposed. It is always not possible to conduct fatigue tests that will

represent the exact behavior ofthe machine part while in service. The idea of any test is

to have a close match ofthe load spectrum to which the material is exposed. Some parts

are miming in varying amplitude spectmm with both tension and compression and some

other parts will be exposed to only a tensile stress during the operation which is the case

for sucker-rods. The present test method is designed to have constant amplitude; stress

controlled loading whichftillyrepresent the behavior of sucker rods during the pumping

operation. This method is described in ASTM E 466-96.

Mean Stress

It is possible to have stress pattems, which have the same stress amplitude but

differ in mean stress. It is a proven fact that cycles with higher mean stress will produce

failure in the shortest time.

28
Effect of Rest Periods

It was believed that, for some materials, rest periods between groups of cycles

have no effect on the fatigue life; while in other cases, the rest periods considerably

improve the fatigue life. Sines states that for the same size of specimen with same test

conditions, the rest periods don't have any apparent effect on the fatigue life of materials.

Present tests are conducted with minimal rest periods imtil the 10 million cycles are

reached.

Effect of Frequency

A factor related to rest periods is the effect offi-equencyof cycling on the fatigue

life. An increase infi-equencywill produce two simultaneous changes: the speed of stress

application will increase and the time the specimen is allowed to dwell at low stresses

will decrease. A number of experiments have been conducted previously to determine

the effect of varying thefi-equencyat room temperature on the endurance limit of

different materials. The obtained data shown that there is no effect when thefi-equencyis

varied up to 5000 cpm. The present tests are conducted at afi-equencyof maximum 20

cycles/second, which is less than the specified 5000 cycles per minute.

Corrosive Environments

The simultaneous action of corrosion and fatigue is well known. When conditions

conducive to fatigue and to corrosion are superimposed, the resultant failure may occur

many times more quickly than the cumulative effects acting separately. Many materials

show an appreciable lowering ofthe fatigue strength, even in the normal atmosphere,

because of corrosion. Present tests are conducted in a non-corrosive atmosphere, as the

29
objective ofthe project is to validate the Goodman diagram, which was based upon tests

conducted in non-corrosive atmospheric conditions.

Temperature

There is significant effect of temperature on fatigue life of materials. It is a proven

factor that the fatigue life of a given steel material reduces with increase in temperature.

Surface Coatings

The application of any coating on a metal may affect the fatigue life. The effect is

rather complex that the coating application may make itself felt in any or all of a number

of different ways. Cleaning prior to coating may either smoothen or roughen the surface.

The application process may induce tensile or compressive stresses in the base metal and

in the coating. This is the reason why a safety factor is considered in defining the API

Modified Goodman Diagram.

Types of Sucker Rods

There are various grades of sucker-rods used in the beam pumping applications.

The major grades used are Grade D, Grade C, Grade K, etc. For this experiment, all

fatigue tests were performed on Grade D rods with a minimum tensile strength of

115,000 psi.

30
 CHAPTER m

MTS FATIGUE TESTING MACHINE

The machine being used to conduct the testing is MTS series 810.13 Material Test

System rated at 55 kips. The major components ofthe machine are:

a. Two column load firame

b. Actuator rated at 55 kip

c. Service Manifold

d. Hydraulic power supply rated 10 GPM at 3000 psi, for 460 volt, 3 phase

operation

e. Console including the following items-

1. Servo-controller

2. Valve driver

3. DC Transducer Conditioner,

4. AC Transducer Conditioner,

5. Feedback Selector,

6. Limit Detector,

7. Control Panel,

8. Digital fimction generator,

9. Coimter Panel.

The machine is a standard pre-engineered product designed to perform a wide

variety of material tests. The machine uses standard interchangeable assemblies for great

flexibility and reliability. The machine is capable to operate in force ranges from 1,000

31
pounds to 400,000 pounds with precise displacement and speeds of up to 30 inches per

second. The detail of entire machine assembly is shown in Figure 10.

Figure 10

MTS Fatigue Testing Machine.

The standard components ofthe MTS system are divided into three categories as

follows:

a. A Load Unit that allows application ofthe desired test loads to the specimen

and defines the test force limits. These include the load frame, load cell and

hydraulic actuator.

32
 b. A Performance Package that provides energy for the test in the form of

controlled pressurized oil flow. These include the hydraulic power supply,

hoses, servo-valve and service manifold.

c. A Control Package that provides precise command and dynamic programming

control and parameter limits for the tester, as well as readout and failsafe

devices.

The machine uses a closed-loop control as shown in Figure 11. The force, strain, or

displacement imposed on the test specimen is measured and continuously compared with

the desired command input. The difference between the measured value and the

command value is used to provide a continuous correction signal to the servo valve,

which causes this difference to be minimized. Transducer conditioners are used to excite

the transducer, and provide conditioning for the output signal from the transducer, which

may then be used, for control and readout. Force, strain or displacement may be used as

a control parameter during the test.

33
 Strain

^~^^~- Transducer
I —I signal
Readout |-* conditioners
instrumentation Load

Displacement
Loading
frame

Displacement
Specimen
strain

Load

i Mode selector

Digital
processor
interface
Analog
controller
^

Digital
• ^ Servo
valve
IZF Displacement
computer
Hydraulic
actuator

Figure 11

Closed-Loop Control Diagram. (Source: Collins, J. A.: Failure of

Materials in Mechanical Design, 2"^^ Edition, John Wiley & Sons, New
York (1993) p. 194.)

The system maintains the command value throughout the test by continuously

driving the servo valve to provide a precisely controlled test under static or dynamic

conditions. Closed-loop control using feedback selection, allows the test engineer to

select control variables for the test and to use a variety of command devices.

Servo-controlled hydraulic actuators apply a time-varying displacement to fatigue

specimen, but pressure sensors (or a load cell) can detect changes in the maximum or

minimum force and through feedback circuits, automatically adjust the controls so that

the forces are maintained at the desired level, regardless ofthe change in specimen

34
stiffiiess associated with crack growth. The whole system has to be calibrated as per the

instructions given in the machine operation manual. Thefrequencyof calibration is one

year.

35
 CHAPTER W

PROCEDURE

Overview

This procedure covers the present test method to analyze the fatigue life of grade

D sucker rods. As API Modified Goodman diagram is considered as the basis for

selection of sucker rods, the aim ofthe present test is to find out the range of stresses,

which can be applied to the grade D rods considering 10 million cycles. It is a well-

accepted norm that if the material does not fail until 10 million cycles, it can be

considered as having infinite life. The test is designed to apply 160% ofthe load range

specified in the Modified Goodman diagram and run samples for 10 million cycles.

Upon completion of 10 million cycles for the first sample, the maximum and minimum

load range will be changed for the subsequent samples as per the tabulation presented in

Table 3. Figure 12 and Figure 13 shows the present test ranges in comparison to the API

Modified Goodman Diagram.

36
Table 3

Experimental Test Data

Sucker Rod Grade

Tensile Strength (T) 115 ksi

Diameter of Rod 0.625 inches

Length of Specimen 12 inches

Cross sectional area 0.306795898 sjq. inches

of Sucker Rod
Sa = (7/4 + 0.5625 *S„JSF

Max.
Minimum Min. Stress IVIax. Allowable Allowable Max. Load
Load (Smin) Stress (Smax) Load (@160%)
(kip) (ksi) (ksi) (kip) (kip)

0 0.0000 28.7500 8.8204 14.1126

2 6.5190 32.4169 9.9454 14.7126
4 13.0380 36.0839 11.0704 15.3126
6 19.5570 39.7508 12.1954 15.9126
8 26.0760 43.4177 13.3204 16.5126
10 32.5950 47.0847 14.4454 17.1126
12 39.1140 50.7516 15.5704 17.7126
14 45.6329 54.4185 16.6954 18.3126
16 52.1519 58.0855 17.8204 18.9126
18 58.6709 61.7524 18.9454 19.5126
20 65.1899 65.4193 20.0704 20.1126
40 130.3798 102.0887 31.3204 26.1126

37
 Stress Range Diagram

70.0000
f *

60.0000

50.0000

m
=•
(/)
40.0000
^.^ M
V)
in

w 30.0000 if^

/ •
20.0000
M

10.0000

^ ^
0.0000
0.0000 10.0000 20.0000 30.0000 40.0000 50.0000 60.0000 70.00^
Stress (ksi)

- • — M a x . Stress (existing) — • — M i n . Stress Max. Stress (160%)

Figure 12

Diagram Showing the Stress Range.

38
 Load Range diagram for 5/8 inch , Grade "D", Sucker Rod

-• M in. Stress Line Original Modified Goodman Une •Present Test Une<160°/4

Figure 13

Diagram Showing the Load Range.

Specimen Preparation

Sucker rods upon receipt from the manufacturer is shop tested as per the

following minimum test requirements. The tests are carried out at "Permian Rod

Operations" as per the present industry norms as follows.

a. Inspection of the rods as per API recommended practice as detailed in API

RPllBR-"Recommende Practice for care and handling of Sucker Rods" and

in API spec. 1 lB-"Specification for Sucker Rods"

b. Shot blasting to clean down to bare metal prior to inspection.

39
 c. Straightening of rods-if required-using hydraulic cross-roll straightening

system

d. Fluorescent magnetic particle inspection utilizing an exclusive atomization

process in an isolation cabinet

e. Thorough visual examination

f Electromagnetic inspection utilizing eddy current and flux leakage detection

system

g. Verification of dimensional tolerance with NIST certified gauges

h. Corrosion inhibitor application prior to storage.

Upon completion ofthe initial Non-Destmctive Inspections, a Specimen Data Record to

be prepared as follows.

a. Details of the material as supplied: specification or designation, heat number,

chemical composition, surface condition and cleanliness.

b. Details ofthe material in the specimen - ultimate tensile strength, yield point,

tensile strength, percent elongation, percent reduction of area, hardness.

c. Detail of specimen design drawing showing the shape, size and dimensions of

specimen including details ofthe test section and gripping arrangements.

d. Details of specimen preparation.

e. Details of pre-test specimen surface preparation.

40
Specimen Storage

The specimens, after completion ofthe mitial testing and recordings, are stored in

a clean air-conditioned atmosphere as per the normal norms as specified in ASTM STP

566, "Handbook of Fatigue Testing."

Specimen Dimensions

The 5/8-inch sucker rods are cut using band saw machine to pieces having length

of 12 inches. These are used as the test specimen. The minimum grip length is 3 inches

at both ends.

Inspection Prior to Testing

Visual inspections with unaided eyes or low power magnification up to X20 are

conducted on all specimens. Obvious abnormalities, such as cracks, machining marks

gouges, imdercuts, and so forth, are not accepted. Specimens are cleaned prior to testing

with solvent(s) noninjurous and nondetermental to the mechanical properties ofthe

material in order to remove any surface oil films, fingerprints, and so forth. Dimensional

analysis and inspection is carried out in a manner that will not visibly mark, scratch,

gouge, score, or alter the surface ofthe specimen.

Specimen Loading

The following procedure is used to load the sample on the machine.

MTS Load Unit Control:

1. Set the following by using the "Fl" and "F2" controls. (For this, the

actuator positioning control has to be switched between on/off modes).

41
Controller Current Mode POD Mode

Axial Disp_SG Disp_POD

2. Switch on the hydraulic pump first by pressing the "Low" mode on the

Load Unit Control and subsequently switching over to "High" mode.

3. By keeping the "Actuator positioning control" to "ON" position, the

actuator position can be lifted up or brought down by turning the knob on

the control unit. (It is advisable to change the actuator position after every

test run).

4. After adjusting the actuator position, keep the Actuator Position Control to

the "OFF" mode during the entire test duration.

5. Unlock the "Load Frame" control. By opening the "UP" valve on the

machine imit, lift the load frame to a desired position for easy insertion of

the specimen.

6. Open the hydraulic grip control valve.

7. Mark the grip borderline on the specimen (In this case minimum 3 inches

from both ends ofthe specimen).

8. By keeping the marked "border line" in line with the grip, actuate the

"Lower Grip" control. The specimen will get gripped firmly. Keep the

lower grip control lever to the maximum position.

9. Bring down the load frame by opening the "Down" valve on the machine

unit imtil the "border line" marked on the specimen matches with the grip

line.
42
 10. Open the "Upper Grip Control" lever to the maximum position. This

makes the upper grip tight on the specimen.

11. Check the overall specimen loading.

12. Lock the load frame by the control on the machine.

13. Press "Fl" on the Load unit control. This will change the Current mode to

and POD mode to "Load_SG" and "Load_POD" modes. Press "F2" and

switch on the "Actuator Position Control", which will make the control

accept the change of load modes. Switch Off the "Actuator Position

Control" so that the current changes are set. Now the grip is in the "Load

Control" mode which means the grip will keep on applying a constant

force on the specimen. The load control imit will display as given

overleaf

Controller Current Mode POD Mode

Axial Load_SG Load_POD

These are the preliminary steps to be followed for the safe loading ofthe

specimen to the unit.

Defining the Test Parameters and Running the Test

The computer control allows different settings ofthe test parameters as required.

The following steps are to be followed for the test set up.

1. Open the MTS "Test Star" program

2. Select "MTS Fatigue Execution" mode.

43
3. Select "File, Select Procedure."

4. Select the "Sucker Rod" procedure as set in the system.

5. In the main menu, select "Select Batch" mode.

6. Select the batch name. (In this case Rod_0.625).

7. In the main menu, select "Select Specimen" mode.

8. Select the specimen identification (sample_01, sample_02 etc).

9. In the main menu, select "Parameters" which will open up the "Execution

Parameters" window.

10. Add values to "End level 1", "End level 2" and "Test Frequency" as per the

values defined in Table 3.

11. Open the "MTS Fatigue Test" menu.

12. Select the "Define specimen" mode.

13. A new window will open up as "MTS Cyclic fatigue Test: Batch Edit."

14. Add details like specimen information, specimen diameter. Elastic Modulus,

Yield Strength, Gage length and temperature.

15. In the menu, select "Define Test" mode.

16. In the new window, select execution parameters and enter the End level 1, End

level 2 and Test Frequency values.

17. Select "Test Termination Parameters" and define the required test parameters.

18. Select "Display Detector Actions." Add interlock criteria as required.

19. Now, select the "Execute" mode in the "MTS Cyclic Fatigue Test" menu.

20. The system is ready to start the test.

21. Start the test by actuating the "Run" button on the "Execution" menu.

44
 22. Watch for the smooth running ofthe machine. If any unnatural behavior like the

noise level from the servo valve etc is observed, stop the run and check for the

parameters. It may be required to start the test with a lowfrequencyand

subsequently raising the values. Properfrequencylevel can be achieved by

looking at the "Load vs. Time" display. The display can be selected from the

main menu. A smooth sine curve indicates that thefrequencyand loading is

proper.

23. When the machine starts running the test, change the "Display Detector Actions"

criteria from "disabled" to "indicate". This will indicate whether the present run

is above or below the defined parameters. If there are no indications, set the

parameters to "interlock" and keep the test continuing.

24. Have constant watch on the test run. The test conditions can be observed

through the web site http://129.118.21.55/.

25. Upon completion of 10 million cycles, stop the test run. Record the values and

add on to the "Test History".

Removal and Reloading ofthe Specimen

The following steps are to be followed for removal and reloading ofthe specimen.

1. Change the "Current Mode" and "POD-Mode" to "Disp_SG" and "Disp_POD" as

described in "Specimen Loading" above.

2. Release the hydraulic grip by controlling the levers. It is advisable to release the

"Lower Grip" first followed by the "Upper Grip".

3. Lift the load frame and remove the specimen.

45
 4. Tag the specimen showing the specimen identification, number of cycles, load

ranges applied and the frequency.

Follow the steps as detailed in "Specimen Loading" and "Define Test Parameters"

above for loading ofthe new specimens and setting the test criteria.

Test Control

There are various "Display" modes available for proper control and display ofthe

test. Samples ofthe displays like "Load vs. Time," "Max/Min Load Vs Cycles," etc., are

shown in Figures 14 through 16.

46
[Hi MTS Fatigue Execulion ZMxi
Actions Parameters Display Event Log Preferences Help
Select Procedure... 1 Cycle Fatigue
Select Batch..
ker_Rod vi
Select Specimen.
L0.B25
Exit
05

Event Log - Cycles-

PeakA'al reverse retention data to disk. Cycle = 107055GG
PeakA^al reverse retention data to disk. Cycle = 10705567
"3 Run 0

End of Test Status ~

Template: High Cycle Fatigue I
Stopped
Pfoceduie: Sucker Rod
Batch: Rod_0.G25~ Mode—
Specimen: Cut_05
User defined cycle storage data file not found. Disp_POD
Data will be stored using default cycles.

Home
^

Select Procedure Name

Select Batch Name

Figure 14

Figure Showing the Test Definition Windows.

47
Execution Parameters

Endlevel 1 18.31 I kip

Endlevel 2 14 I kip
Test Frequency 25 \TT:
I HZ

l^ Amplitude Control

Test Termination Parameters

Stable Cycle

Stable Cycle Method Reference Cycle 3

Reference Cycle 10000000

Stabilii
Z]

Termination Criteria

W Cycle Limited

Test Duration (Cycles) I 50000000

Control Amplitude Threshold 30" h zl

Failure Threshold
r 99 1^ zi
Stiffness Failure Threshold I 99" V'' d
r~ Hydraulic Shutdown at Termination

Figure 15

Figure Showing the Setting of Execution Parameters.

48
 [Hi MTS Fatigue Execution rg]x]
File Actions parameters Display Event Log Pteferences Help

Template: High Cycle Fatigue

Piocedue: Suckef_Rod Vi •
Batch; Rod_0.625
Specimen; Cut_04

Event Log Cycles—

PeakA'al data to disk. Cycle = 5000000
Timed data to disk. Cycle = 6000005
:3 9407525

PeakA'al data lo disk. Cycle = 6000000 Status—

Timed data to disk. Cycle = 7000037
HoU Running
PeakA/al data lo disk. Cycle = 7000000
Timed data to disk. Cycle = 8000021
Mode —
PeakA/al data to disk. Cycle = 8000000
Timed data to disk. Cycle = 9000005 Stop Load SG
PeakA/al data to disk. Cycle = 9000000

J Home

1 7t: Display Detector Actions |.D|x|

u6(6C(ors
C Control
(* Limits
Signal Upper Limit Lower Limit

Displacement | 1.60000 | Interlock dl -0.50000 1 Interlock d in

Load 1 20.00001 Interlock dl 8.00000 P|!p9§|| •d kip

" ^ •

Figure 16

Figure Showing the Execution Display.

49
 CHAPTER V

RESULTS AND DISCUSSIONS

Tests are conducted for all the load ranges as given in Table 3. Specimens were

changed upon completion of 10 miUion cycles to different load settings. All the

specimens successfully completed 10 million cycles without interruption and without

actuating the interlock. The tests proves that, the present day sucker rods are capable of

withstanding higher stress ranges compared to the ranges specified as per the API

Modified Goodman diagram.

Results ofthe tests are tabulated and a new "Modified Goodman" Diagram is

prepared as shown in Figure 17. By referring to the diagram it can be implied that the

present API Modified Goodman Diagram is conservative. The safety factor applied for

pull-pull test seems to be higher than what is required. This results in the re-definition of

entire API Modified Goodman Diagram as follows.

1. The Y-intercept shall be T/2.5 as against T/4 specified by the API

committee.

2. As a result, the slope ofthe allowable stress line changes to 0.3 fi-om

0.5625

3. The revised definition ofthe Modified Goodman Diagram becomes,

S„={T/2.5+ MS^JSF (7)

where

S^ = Allowable stress, psi,

50
 T = Minimimi Tensile Strength, psi,

M = Slope ofthe allowable stress line = 0.3,

iS^ = Minimum Stress, psi,

SF = Safety Factor (dimensionless).

The behavior ofthe specimens suggests that, the material will be able to withstand even

higher ranges of stresses. There were no visible damage to any ofthe specimens and the

interlock settings points to the fact that there were no specific deviations of loads during

the test process. So it has to be imphed that even the initiation ofthe crack growth had

not taken place during the tests. All the above confirms the conservativeness ofthe

present API Modified Goodman diagram.

51
 S, = (T 12.5 + MS „,„ )SF

0.8

0.6 T/1.75

T/2.5
0.4 1 K ^

a
in
0.2

<0
^ ^ ^ "
0 -I 1

0.2 0.4 0.6 0.8

Stress (psi)

Minimum Stress Allowable Stress

Figure 17

Suggested Revision to API Modified Goodman Diagram.

52
 CHAPTER VI

CONCLUSIONS

The API Modified Goodman diagram as used today for selection of sucker-rods is

conservative. Tests conducted on the grade 'D' sucker-rods prove that current day rods

are capable of withstanding higher stress ranges up to 160% ofthe stress range specified

by API committee. The proposed revision to the "API Modified Goodman diagram" is

presented which requires the re-definition ofthe allowable stress line as:

S, ={T/2.5+ MS^)SF (8)

where,

S^ = Allowable stress, psi,

T = Minimum Tensile Strength, psi,

M = Slope ofthe allowable stress line = 0.3

'^min ^ Minimum Stress, psi,

SF = Safety Factor (dimensionless).

The purpose of this thesis is to alarm the industry and the operators about the

conservative nature ofthe API Modified Goodman Diagram so that more economical

ways can be adopted while deciding about the selection of sucker-rod string.

53
 CHAPTER v n

SUGGESTIONS AND RECOMMENDATIONS

As the present tests are carried out at 160% ofthe stress range as specified by the

API Modified Goodman diagram, it is recommended that tests shall be conducted for

different ranges of stresses like 180%, 220%, etc., until an optimum stress range is

derived. This will require testing of different specimens at different percentages of loads

until the load range at which the fatigue occurs within 10 million cycles is determined.

Then the stress range has to be reduced stage by stage until the optimum stress range at

which the material will withstand 10 million cycles is determined.

Another area to be looked into is the validity ofthe theory that material should not

be exposed to the near ultimate strength region. The apex ofthe triangle (at tensile

strength) is considered as r/1.75. This value shall be changed to 771.2, r/1.3, etc., and

tests shall be conducted to see the ranges of stress, which can withstand 10 million cycles

without failure.

It is recommended that tests be conducted to find out the optimum number of

cycles a specific rod grade can withstand. It is possible that the specimens will withstand

even up to 50 million cycles without failure and this will enable us to redefine the entire

fatigue analysis concepts.

Since hardness is a direct indication of strength, it is advisable that hardness ofthe

specimens be noted prior to testing of future samples.

Another important suggestion is that even though the present tests are conducted to

redefine the API Modified Goodman diagram, future tests be carried out in relation to the

54
original Goodman Diagram. The slope ofthe Modified Goodman Diagram is not based

on any test results and so, tests conducted at stress ranges in relation to the maximum

allowable stress line ofthe original Goodman Diagram will result into a more reliable

modification ofthe Goodman Diagram.

It is also recommended that, in order to have repeatable test procedure, future

tests should be conducted on rods that have perpendicular shoulders.

For the future tests, which will be directed towards finding out the exact stress

range at which the specimens will fail before 10 million cycles, it is recommended that

higher diameter screw-in end connections be used on specimens as per ASTM standards

in order to avoid failure at the grips.

55
 BIBLIOGRAPHY

1. Hardy, A. A.: "Sucker Rod String Design and the Goodman Diagram," paper
64-Pet2, presented at the 1964 Petroleum Mechanical Conference ofthe
American Society of Mechanical Engineers, Los Angeles, Sept. 20-23.

2. API RP-1 IBR, Recommended Practice for Care and Handling of Sucker Rods,
American Petroleum histitute. Eighth Edition, Washington D. C. (1989).

4. Goodman, J.: Mechanics Applied to Engineering, Longmans Green and Co.,

New York (1908).

5. Creech, M. D.: Fatigue failures of Sucker Rods, Heldies Portable Pipe

Service, Oklahoma City, OK (1957) p. 7.

6. Jones, D. M.: "Considerations in Sucker Rod Joint Preload, "included in letter

to API committee Circa, spring 1959, USI Axelson, Division of U.S.
Industries, Inc.

7. Hein Jr., N.W. and Hermanson Jr., D. E.: "A New Look at Sucker Rod
Fatigue Life"; paper SPE 26558 presented at the 1993 Annual Technical
Conference and Exhibition, Houston, Oct. 3-6 (1993).

8. Cedric, R. W.: Engineering Material Science, Wadsworth Publishing

Company Inc., Belmont, CA (1961).

9. Frost, N. E.: Metal Fatigue, Clarendon Press, Oxford (1974).

10. Saba, J.: Modern Oil-Well Pumping, The Petroleum Publishing Co., Tulsa,
OK (1962).

11. Bid, N. M. A.: "Sucker Rods Failure Mechanisms, Method of Avoiding

Them," paper SPE 9639 presented at the Middle East Oil Technical
Conference, Manama, Bahrain, March 9-12 (1981).

12. Steward, W. B.: "Sucker Rod Failures," The Oil and Gas Journal, (April 9,
1973) Vol. 71, No. 15, p. 53.

13. Hardy, A. A.: "Why Sucker Rods Fail," The Oil and Gas Journal, (August 12,
1963) p. 109.

14. Paul, T. W.: "Improving Sucker Rod String Design," Petroleum Engineer,
(July 1973) p. 68.

56
15. Neely, A. B.: "Sucker Rod String Design," Petroleum Engineer, (March
1976) Vol. 48, No. 4, p.58.

16. Powers, M. L.: "Optimization of Sucker Rod Replacement," paper SPE 3470
presented at the 46th Armual Fall Meeting, New Orleans, Oct. 3-6 (1971).

17. Moore, K. H.: "Stop Sucker Rod Failures To Save Money," Petroleum
Engineer International, (July 1981) Vol. 53, No. 8, p. 27.

18. Collins, J. A.: Failure ofMaterials in Mechanical Design, 2"^ Edition, John
Wiley & Sons, New York (1993).

19. Polakowski, N. H. and Ripling, E. J.: Strength and Structure of Engineering

Materials, Prentice-Hall Inc., Englewood Cliffs, NJ (1966).

20. ASTM STP 566, Handbook of Fatigue Testing, by American Society for
Testing and Materials, Library of Congress Catalog Card Number: 74-83946
(1974).

21. Sines, G. and Waisman. J. L.: Metal Fatigue, McGraw-Hill, New York,
(1959).

22. Handbook of Fatigue Testing, S. R. Swanson (ed.), American Society of

Testing and Materials, Philadelphia, (1974).

23. Donnelly, R. W.: Beam Pumping, Petroleum Extension Service, U. of Texas

at Austin, Austin (1986).

57
 PERMISSION TO COPY

In presenting this thesis in partial fulfillment ofthe requu-ements for a master's

degree at Texas Tech University or Texas Tech University Health Sciences Center, I

agree that the Library and my major department shall make it freely available for

research purposes. Permission to copy this diesis for scholarly purposes may be

granted by the Dkector ofthe Library or my major professor. It is understood that

any copying or publication of this thesis for financial gam shall not be allowed

v^dthout my further written permission and that any user may be liable for copyright

infringement.

Agree (Pennission is granted.)

Disagree (Permission is not granted.)

Student Signature Date