USFDC Home  USF Electronic Theses and Dissertations   RSS 
Material Information
Subjects
Notes
Record Information

Full Text 
xml version 1.0 encoding UTF8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchemainstance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd leader nam 22 Ka 4500 controlfield tag 007 crbnuuuuuu 008 s2010 flu s 000 0 eng d datafield ind1 8 ind2 024 subfield code a E14SFE0003480 035 (OCoLC) 040 FHM c FHM 049 FHMM 090 XX9999 (Online) 1 100 Porteiro, John. 0 245 Spring design optimization with fatigue h [electronic resource] / by John Porteiro. 260 [Tampa, Fla] : b University of South Florida, 2010. 500 Title from PDF of title page. Document formatted into pages; contains X pages. 502 Thesis (M.S.M.E.)University of South Florida, 2010. 504 Includes bibliographical references. 516 Text (Electronic thesis) in PDF format. 538 Mode of access: World Wide Web. System requirements: World Wide Web browser and PDF reader. 3 520 ABSTRACT: The purpose of this work is to look into the fundamental issues regarding spring design and develop a new, easy to use software program that would allow for optimal, flexible spring designs. Most commercial programs that address this function are basic and do not allow the designer much control over the variables hindering design. This is so because most programs start from the premise that the spring is a general purpose part of the system or that other design parameters can be altered to accommodate the chosen spring. In cases where this is not so, such as in hydraulic cartridge valves, where the geometric constraints are severe, spring design may be a cumbersome process. This is particularly true when fatigue life is taken into account. The solution chosen here is to tailor the software application to these particular design constraints, incorporating some ideas about spring optimization. In addition to this, a concerted effort was made to make the subject more accessible to the engineers using the program by automating the more technical aspects of the process allowing the designers to make intelligent decisions based on how the variables would affect design. To this end currently existing software was evaluated to determine where it was lacking and a new program was written and painstakingly tested. Finally, it was used to correct flaws identified in existing springs. 590 Advisor: Rajiv Dubey, Ph.D. 653 Helical springs Design Stress Constraints Cartridge valve 690 Dissertations, Academic z USF x Mechanical Engineering Masters. 773 t USF Electronic Theses and Dissertations. 4 856 u http://digital.lib.usf.edu/?e14.3480 PAGE 1 1 Spring Design Optimization With Fatigue by John L. Porteiro A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Mechanical Engineering Department of Mechanical Engineering College o f Engineering University of South Florida Co Major Professor: Steven T. Weber Ph.D. Co Major Professor: Rajiv Dubey Ph.D. Alex A. Volinsky, Ph.D. Date of Approval: April 1, 2010 Keywords: helical springs, design, stress, constraints, cartridge valve Copyright 2010 John L. Porteiro PAGE 2 2 D EDICATION I w ould like to dedicate this paper to my father and mother, who are a constant source of inspiration and encouragement, without whom this document would probably not come to be. Their love and patience has nurtured and supported me through my many years, an d I hope that I have done the same for them. I also acknowledge the various other teachers that I have had and who dedicated time to my growth. PAGE 3 3 ACKNO W LEDGEMENTS I would like to acknowledge all those that encouraged me to write this thesis. I would a lso like to thank Dr. Steve Web er and Dr. Rajiv Dubey for their patience and knowledge in guiding me on this project. I am also grateful to Dr. Alex A. Volinsky for his helpful comments and careful review of the manuscript Along the way I also had the help of the engineering department of Sun Hydraulics, who were all very helpful and kind, providing valuable data and resources. The Spring Manufactures Institute, for providing research materials and basic information that forms the basis for much o f the paper PAGE 4 i i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES iv ABSTRACT v CHAPTER 1. INTRODUCTION 1 1.1 What is a S pring ? 1 1.1.1 The History of Springs 2 1.1.2 Current Spring Developments 3 1.2 Literature R eview 4 1.3 Problem Statement 8 CHAPTER 2. SPRING THEORY 9 2.1 Basic Models: Dimensioning 9 2.2 Fatigue Design 14 2.3 Spring Design Due to Existing Constraints 22 CHAPTER 3. OPTIMUM SPRING DESIGN WITH FATIGUE 2 6 3.1 Proposed Model 2 6 3.1.1 Spring Variables and Constraints 2 7 3.1.2 TK Solver 3 0 3.1 3 Numerical Procedure 3 1 PAGE 5 ii ii 3.2 Numerical Results 3 5 3.3 Comparison with Test Results 4 3 CHAPTER 4. CONCLUSIONS AND RECOMMENDATIONS 47 4.1. Conclusions 4 7 4.2. Recommendations for Future Work 4 8 REFERENCES 50 BIBLIOGRAPHY 53 APPENDICES 5 4 Appendix A. Program C ode 55 A.1. Main Program 5 5 A.2. Solver Subroutine 60 Appendix B Governing E quations 6 3 Appendix C Test D ata 6 4 C.1. Calibration Data 6 4 C.1.1. 100 lb S ensor 6 4 C.1.2. 1000 lb S ensor 6 5 C.2. Sample S pring Plots 66 C.2.1. Spring 2 25 6 6 C.2.2. Spring 2 38 6 7 C.3. Sample Excel Data Sheet 6 8 PAGE 6 iii iii LIST OF TABLES Table 1. Spring 238 3 6 Table 2 Spring 225 3 8 Table 3 Spring 176 40 Table 4 Spring 239 4 2 PAGE 7 iv iv LIST OF FIGURES Figure 1 Compression Spring s 2 Figure 2 Leaf Spring 3 Figure 3 T orsion S pring 4 F igure 4 Wahl Stress Correction Factor. 6 Figure 5 Manual Spring C oiler 7 Figure 6 Simple Spring Model 9 Figure 7 Schematic of Shear Stress Distribution in a Wire Cross Section 1 1 with Wahl Correction Factor Figure 8. Fatigue Failure of a S pring 1 4 Fig ure 9 Comparison of Goodman Gerber, Soderberg and Morrow Lines 1 8 Figure 10 Goodman Diagram 18 Figure 11 Modified Goodman Diagram 1 9 Figure 1 2 Residual Shear Stress Distribution After Pres etting 2 1 Figure 1 3 Loaded Shear Stress Distribution After Presetting 2 1 Figure 1 4 Screen Shot of Software P rogram 31 Figure 15. Program Flow Diagram 33 Figure 1 6 Spring 238 Redesigns 44 Figure 1 7 Spring 255 Redesign 45 Figure 1 8 Spring 239 Redesign 4 6 PAGE 8 v v Optimization of Spring Design with Fatigue John L. Porteiro ABSTRACT The purpose of this work is to look into the fundamental issues regarding spring design and develop a new easy to use software program that would allow for optimal, flexible spring designs. Most commer cial programs that address this func tion are basic and do not allow the designer much control over t he variables hindering design. This is so because most programs start from the premise that the spring is a general purpo se part of the system or that other design parameters can be altered to accommodate the chosen spring. In cases wh ere this is not so, such as in hydraulic cartridge valve s, where the geometric constraints are severe, spring design may be a cumbersome process This is particularly true when fatigue life is taken into account. The solution chose n here is to tailor the software application to the s e particular design constraints incorporating some id eas about spring optimization. In addition to this a concerted effort was made to make the subject more accessible to the engineer s using the program by automating the more technical aspects of the process allowing the designers to make intelligent decisions based on how the variables would affect design. To this end currently existing software was evaluated to determine where it wa s lacking and a new program was written and painstakingly tested Finally, it was used to correct flaws identified in existing springs. PAGE 9 1 1 CHAPTER 1 INTRODUCTION 1.1 What is a Sp ring? What is a spring? According to A.M. Whal ( 1991) : may be defined as an elastic body whose primary function is to deflect or distort under load (or to absorb energy) and which recovers its original shape when released after being H e goes on to classify the main functions of springs as one of fo ur things: to absorb shock, to apply force, to support a structure, or to provide load control. This broad definition includes things that people do not normally think of as springs. Under this definition aircraft wings, the chassis of a car and even the shoes we wear will be cons idered springs. These items all distort under load and return to their original shape once the load is released. A shoe will absorb the impact of the foot fall and the bending of the arch of the foot and ret urn to its normal stat e when the s e inputs are r emoved. Aircraft wings must take the loading and unloading of the wings on takeoff and landing in addition to any turbulence that the plane may encounter. Obviously such springs have wildly different properties and functions, an d so can not be all analyzed with the same techniques so, thus, the purposes of this paper, we will consider only helical compression springs (Figure 1) These springs are by far the most common type and are useful in the operation of many devices due to s everal desirable PAGE 10 2 2 properties, such as a near linear rate (particularly after the first 20% of deflection), the many materials that can be used to make them, and the ease of manufacture. For this reason helical compression springs have been in use for some time. Figure 1 Compression Springs. Note that the end condition on these springs is that of closed and ground. You can see that the spring needs about a quarter turn to separate from the bottom coil. 1.1.1 The History of Spr in gs The history of springs is very long. No one can be sure when they were first created since they were probably part of some of the most basic tools. Relatively sophisticated devices, such as eyebrow tweezers have been dated as early as the Bronze Age. Ct esibius of Alexandria developed a process to manufacture springier bronze in the third century B.C. E xample of military applications would be the bow and arrow as well as the more powerful catapult More modern s prings made of metal s came about with the in ventions of locks and clocks around the 15 th and 16 th centuries and were also used later in the suspensions for carriages (Figure 2). In the eighteenth century, the Industrial Revolution brought about methods, materials and techniques for the mass producti on of springs PAGE 11 3 3 (Figure 3) Springs today are made out of various materials and have different ways of achieving the same goals of storing and absorbing energy. Some late advances in spring technology include the development of springs made of plastic, Bell eville springs, machined springs, and new nanotechnology developments that have resulted in springs that have the size of a molecule The properties of a spring can be altered to suit the needs of a particular situation. A number of variables can be used t o affect the properties of a spring. Among them we can include the way the ends of the spring are produced, the material the spring is made of, the type of surface treatment, and the design stress levels. All these variables can be used to alter the perfo rmance of the spring and a good designer must balance the variables to get the desired results. Figure 2 Leaf Spring (courtesy of Tubal Cain) 1.1.2 Current Spring Developments Spring research carried out today involves the d evelopment of new materials in an attempt to create stronger, smaller springs. These efforts include the improvement of the basic materials currently used as well as the use of entirely new ones. Some advances in the technology of spring materials have re s ulted in the Stelmor process (controlling cooling of spring steel wires at the factory), the development o f micro alloys (like SiCr) PAGE 12 4 4 and in better quality controls procedures that allow a much better surface finish of the spring wires. An example of a new material is the use of carbon fiber springs in the suspension of ra ce cars. As stated above, the ad vent of nano science has resulted in the creation of springs the size of a molecule. Taking advantage of the different material properties and type s of s pring these new developments have different ways of achieving the same goal of storing and absorbing energy ; however they are still springs that obey the same laws as the more traditional springs. Figure 3 Torsion Spring (Court esy of Tubal Cain) 1.2 Literature R eview R obert Hooke in his as extension so goes force. This simple statement, today known for spring design. Two year s later Lagrange (1770) analyzed the proportion of loads to deflection in springs. Practical applications of PAGE 13 5 5 springs such as their advantages in improving the riding quality of carriages were discussed by Gilbert (1825) while the laws of isochronism of watches and chronometers was studied by Frodsham (1847). This last work also describes the design and materials used in spring watches. St. Venant (1847) studied the problem of torsion in rectangular prisms In the following 50 year s there was a tremendous growth in number of publications in spring design ranging from topics such as railway springs (Adams, 1850 ; Anonymous, 1858; Rey, 1876; Nadal 1896 ; McCarty, 1898 ), watch springs (Young, 1852; Phillips, 1861; Caspari, 1875) Bellev ille Springs (Morandiere, 1866; Resal, 1888) spring motors (Doubler, 1876; Leveaux, 1876) and conical springs (Resal, 1892) among others. A description of Belleville springs and their advantages was given by Morandiere (1866) and Resal ( 1888) while the ef fects of steady and sudden loads on helical springs were investigated by Rankine (1866) The deflection and carrying capacity of spiral springs was studied by Begtrup (1892). French (1902) published formulas for the calculation of the torsional elasticity and safe stress levels for different ratios of wire to coil diameters. A. Roever (1813) proposed a pioneering theory for the description of the maximum bending and shear stresses on spring coils. He was the first to realize that the standard formulas for d eflection and force exerted by springs were inaccurate because the stress distribution in the spring, due to pure torsion, did not have a linear relationship indicates that additional stresses appear due to the curvature, coil pitch angle and wire cross section, and a stress correction factor must be applied. For springs with coil to wire diameter ratios between 3 and 10, not using a correction factor underestimates maxi mum stresses between 14 and 30%. PAGE 14 6 6 Arthur Wahl (1929) proposed a stress correction factor to take into account those additional stresses using torsion theory and carried out tests that were in good agreement with the values theoretically predicted. Finnieco new correction factor that included the displacem ent of the actual center of rotation of the fibers from the geometrica l center of the cross section, carrying out tests to confirm his predictions. Goehner (1932) analyzed the maximum stresses in springs of circular and square cross section. The resulting formula was extremely complex and was simplified and found to be in excellent agreement. Wahl (1939, 1949, 1963) analyzed the use of the curvature correction factor in spr ings with large deflections, with free and fixed ends, under fatigue loading and endurance. An excellent analysis of when and when not to use the stress correction factor was presented by Carlson (1985). Figure 4 Wahl Stress Correc tion Factor (Courtesy of Tubal Cain) PAGE 15 7 7 Figure 4 shows the correction factor plotted out against the ratio of the diameter of the coil to the ratio of the wire diameter ( ). This was the key innovation, in t hat the stresses are affected by the amount of curvature in the wire used to form the helix. The graph indicates that the stresses increase as C becomes smaller. This was found to be due to effects of curvature on the inner surface. The inner surface is sh orter when compared to the outer surface and this means that it undergoes more deformation, increasing the stresses. The spring industry has been a mature industry for some time now, still there is cons tant progress in all areas such as the studies in lateral deflection of free end springs (Wolansky, 2005), the innovative application software for spring analysis (Zubek, 2010), the development and processing of new materials and the advent of newer faster more efficient automatic coiling machines that re placed the old arbor t hat was used for manual coiling (Figure 5) Figure 5 Manual Spring Coiler (Courtesy of Tubal Cain). PAGE 16 8 8 1.3 Problem Statement Any company that uses springs as part of an assembly must fit a spring into a specific space. In the initial design process this space can be altered to allow for th requirements. However this can be very expensive later, when the product i s in production. Often a product based on the original design is desired, but t may require a sp ring with different properties. An engineer is then tasked with re designing a spring tha t must fit into the same space. Usually he will not have a particularly deep knowledge of springs, and so this can make the re design more difficult tha n it should be. In the past the process consisted of calculating multiple designs and look ing at charts to estimate their performance. This is a l ong and time consuming process. Further more if the engineer is not experience d it is possib le to pursue a design that is feasible for the given space. This problem is particularly difficult when lon g life is needed. For long fatigue life no suitable substitute exists for the chart method of design. To make such processes easier an engineer needs some way to compare potential designs quickly with minimal knowledge of the underlying math and theory. To date there are programs that will calculate the properties of a spring given specific data. Some even include rud imentary fatigue life analysis. However there has been no effort to create a program that will help design a spring that would be a suita ble substitute to an existing design, taking into account the fatigue properties. Such a program would have to look at several springs and use a set of criteria to suggest whether to accept or reject the spring. PAGE 17 9 9 CHAPTER 2 SPRING THEORY 2.1 Basic Mode ls: Dimensioning Basic spring theory state s that a spring can be approximated b y a (circular) bar in torsion. The derivation of this model can be seen in most text books o n mechanics of solids and the twist in the wire is roughly equivalent to the deflec tion in the spring. Figure 6 Simple Spring Model The load from adjacent coils applied at A and B exerts a twisting force at C (with permission of Tubal Cain) For an open coiled helical spring subject to an axial load W, the external work can be written as: PAGE 18 10 10 [1] where is the spring deflection, T and M are the torsion and bending moments and and are the wire rotation and bending angles respectively. Substituting the values of T and M as a function of the load and the constant angle the coils make with the planes perpendicular to the axis we obtain: [2] or: [3] where we have also substituted and by their values as functions of the bending and torsional moments as well as the elastic and torsional modulus, (E, G) and the moments of inertia. We can write T and M as functions of : [4] Here, l is the wire length For n turns of radius R its value is 2 Rn/cos Substituting : [5] As a function of the coil and wire diameters, D and d the expression is: [6] When the angle is small cos 1 and sin 0, leading to the standard form of the equation for deflection in a spring as seen here. [7] PAGE 19 11 11 This model applies to small angles of pitch wh ich are common in most springs. It does not account for the bending of the material, as can be seen from the derivation where bending is the second term In most practic al applications bending can usually be ignored. The correction factor developed by Wahl compensates for the added stresses due to the curvature absent in the torsion bar model of the spring and fo r the effects of direct shear. Figure 7 shows the stress dis tribution with the Wahl correction factor. There is also a correction for deflection but it is seldom used due to the small effect that it has on the deflection (for most springs around 3%). These are the most important attributes of springs, other attribu tes like the diameter of the spring, the spring wire, and the number of coils (or turns) in a spring, will also be discussed. Figure 7 Schematic of the Shear Stress Distribution in a Wire Cross Section with Wahl C orrection Fac tor The diameter of a spring and the w ire diameter are related in that t hey affect the same properties. Increasing the diame ter of the coil by a factor of two will half the spring rate, (the ratio of load to deformation). D oubling the w ire diameter will result in an eight fold PAGE 20 12 12 increase in spring rate. One of the consequences of this is that a spring can not be scaled down and maintain the same properties. Consequently the wire diameter and the diameter of the coils have been found as useful in describing the springs properties when combined as C (the ratio of coil to wire diameter). This ratio is used in several equations including with a low C (around 4) are very st iff and suffer from greater effects due to curvature, while higher values (20 or so) tend to buckle under load. Increasing t he number of coils also alte rs the spring rate. It does so because each individual coil d eflects a certain amount under load (stres s) independent ly of the other coils. This is bec ause the load is distributed through out the spring wire evenly. Thus the more coils present the more th e spring deflects For most materials it is desired to maintain a number of coils that keeps the pitch fa irly low. This increases stability and more uniformly distributes a load. This can be seen in the equation for eccentricity of loading: [8] [9] Where: The more coils in the spring th e closer to unity the load will be. The distribution of load also affects the fatigue life, so the number of coils is more important than it is generally assumed. PAGE 21 13 13 spr ing. This refers to the way the spring makes contact with the surfaces around it. There are four basic conditions: open, closed open and ground, and closed and grou nd. The open condition occurs when the wire is cut to the desired length of spring and no c hange in pitch or special machining is made in order to make the spring more stable. This condition would be used if there is a special fitting designed to hold the spring in place. Open and ground is similar to the op en condition with the except ion that t he end is ground flat so that when placed on a surface the wire e nd surface will be parallel to it. A closed end is such that the last coil is decrease d in pitch so that the two last coils are touching. This allows for a more stable spring that will stand on its own with out any special fitting. The closed and ground ends are similar t o the closed, but in this case, it is ground flat, which allows the spring to be much more stable tha n the other types and to stand up level with the ground. The closed end ty pes are desirable because of the stability against buckling that they provide, but in order to do so extra material is used that will not deflect. In addition the closed end conditions introduce a non linear spring rate at the beginning of deflection sinc e t he coils close to the end s do not have full pit c h to deflect and will touch the end coil sooner, becoming inactive. In F igure 1 all springs except for the one on the far right have about 2.5 coils inactive due to the end coil condition (1. 25 per end). PAGE 22 14 14 Figure 8 Fatigue Failure of a Spring 2.2 Fatigue D esign The fatigue life is an important part of any design and springs are no different. Springs are often in service in cri tical applications where fa il ure is unacceptable Sinc e the most po pular material for springs is steel the appropriate fatigue cal culations use metals as the material Metal fatigue is the failure of a component as a result of cyclic stress below the yield stress The failure occurs in three phases: crack in itiation, crack pro pagation, and catastrophic failure. The duration of each of these three phases depends on many factors including fundamental raw material characteristics, magnitude and orientation of applied s tresses, processing history, ambient conditi ons and excitation frequency Fatigue failures often result from applied stress levels significantly below those necessary to cause static failure (elastic regime) PAGE 23 15 15 On a microscopic scale failure occurs along slip planes in the crystall ine structure of t he materials. Most metals with a body centered cubic crystal structure have a characteristic response to cyclic stresses. These materials have a threshold stress limit below which fatigue cracks will not initiate. This threshold stress value is often refer red to as the endurance limit. In steels, the life associated with this behavior is generally accepted to be 2x10 6 cycles. In other words, if a given stress state does not induce a fatigue failure within the first 2x10 6 cycles, future failure of the compon ent is considered unlikely. For spring applications, a more realistic threshold life value would be 2x10 7 cycles. Metals with a face center ed cubic crystal structure (e.g. aluminum, austenitic stainless steels, copper, etc.) do not typically have an endur ance limit. For these materials, fatigue life continues to increase as stress levels decrease; however, a threshold limit is not typically reached below which infinite life can be expected. In steels the cause of this endurance limit is generally attribute d to the presence of interstitial elements such as carbon or nitrogen that pin dislocations preventing the slip mechanism that leads to micro cracks (Bannantine, 1990). Endurance limits are affected by the conditions and will disappear if the metal element is subjected to periodic overloads, corrosive environments, or elevated temperatures. T he charts used to predict the fatigue life of a metal component are based on empirical data from a large numbers of tests and provide us with the most reliable method for fatigue life prediction. The stress li fe or S N (Stress Number of cycles) method is one of the preferred methods and it has been in use for over 100 years The S N approach is valid in applications where the stress stays within the elastic region, a nd life PAGE 24 16 16 expectancy is long. As a result this method will not apply to static loading conditions or situations where the life expectancy is less then 1000 cycles as this tends to involve the plastic region. This is due to the fact that the stress life appr oach ignores the true stress strain behavior treating a ll strains as elastic. Fortunately the S N curves for different st eel alloys are similar and they allow the use of a single curve, modifying the results to account the difference in elastic modulus. To estimate the S N curve for steel the following power relation was found (Bannantine, 1990). [10] where S is the cyclical stress and N the number of cycles. If S e and S u are the endurance and ultimate stresses and S 1000 the stress after 1000 cycles: [11] [12] If one make s the assumption that and the preceding equations reduce to: [13] This assumption only works for steels as for other materials and are not as clearly defined. Because it is nearly impossible to predict when an individual component will fail one must rely on a statistical analysis and safety facto rs to acc ount for extreme cases. Th e data for the analysis has to be obtained from many tedious and expensive tes ts. Man y engine ers have looked for an empirical relationship between the applied stres s and the PAGE 25 17 17 life of the spring and o ver the years many such relation s have been proposed and found to be useful: Gerber (Germany 1874): [14] Goodman (England 1899): [15] Soderberg (USA 1930): [16] Morrow (USA 1960): [17] Where: is the stress amplitude. [18] is the stress mean. [19] All of these relations connect the endurance limit on the alternating stress (stress amplitude ) axis with the yield streng th ( ), ultimate strength ( ), or true fracture stress ( ), on the mean stress axis ( ) Using the alternating and mean stresses is one of the more convenient ways to re present fatigue loading conditions. T he Soderberg relation is considered to o conservative and is seldom used. The actual test data falls between the Goodman and Gerber lines, with Gerber being closer to the results and at times overestimating, and Goodman being more conservative. For most situations the mean stress is small compared to the alternating stress and t here is little difference in the theories. When the ratio of maximum to minimum stress is close t o one the theories are PAGE 26 18 18 distinct; but this is a r egion for which little data is available Figure 9 shows these various plots in a chart. Figure 9 Comparison of Goodman, Gerber, Sod erberg and Morrow Lines The cyclical s tresses are usually used to create a Goodman diagram. In th is diagram the alternating stresses are on the y axis and the mean stresses a re on the x axis. The value for the endurance limit is then placed on the alternating stress axis and the ultimate tensile strength on the mean stress axis. These are then connect ed with the Goodman line (infinite life). A line drawn from the origin with the slope of the alternating stress to the endurance limit is the load line. To find the life of the spring one finds the point located buy the mean stress and the alternating stre ss (this should be on the load line). Figure 10 shows a basic Goodman diagram Figure 10 Goodman Diagram PAGE 27 19 19 Another popular way of predicting life is the modified Goodman di agram. It is different in that it is usually normalized so that it cover s a range of similar materials. As stated in the Spring D esign M anual of the American Society of Mechanical Engineers (Warrendale, 1996 p.90) to construct a modified Goodman diagram the initial and maximum stresses are normalized by dividin g them by the ultimate tensile strength of the material ( ) This allow s the diagram to be used for various similar materials with different tensile strengths. Starting at the initial load point a vertical l ine can be drawn. Along this line are the various loading ratios that the spring can achieve The data from S N curves is usually incorporated as life lines at the top of the graph showing expected life. This allows for the inclusion of som e common surface treatments such as shot peening, because shot peening does not reduce stresses, like presetting, and so will not alter the stress values. The modified Goodman diagram presented in Figure 11 includes an S N curve showing the effects of shot peening as well as an S N curve without shot peening. Figure 11 Modified Goodman diagram PAGE 28 20 20 The manner in which a spring is stressed also affects its fatigue life. Springs that are cycled over a narrow stress range (10,000 p si) generally have long li fe, and if the range is small enough (5,000 p si) the stress correction factor can be ignored (Carlson, 1969). As noted above the lo ad is seldom distributed evenly and it has been found that springs that do not have whole numbers of active coils will ha ve longer fatigue lives as these distribute the loading stresses better (Carlson, 1969). From this it is easy to see that lateral loading of a spring (as in a vibration damper) will also reduce its fatigue life. The models for predicting spring failure ca n not be general ized to all springs because of the many factors that affect it therefore specific conditions must be used to get an accurate prediction. Unfortunately st ress in a sp ring is not uniform. T he stress is greater on the inner surface and the coil surface itself is vul nerable to imperfections in the material that serve as stress concentration point s. These inevitably lead to cracks that cause catastrophic failure in a sprin g, reduci ng its effective fatigue life. For this reason fatigue failures always start at the inner surface and proceed outward (Carlson, 1969). To prevent cracks from forming and spreading spring makers have found several solutions: protective coatings, s urface treatments after the spring was coiled, and treatments to induce residual stresses in the spring that would be beneficial. An example of surface treatment is case hardening, which makes the surface harder to scratch, and more resistant to cracks. This is done by heating the material and allowing t he grain structure on the surface to reorganize, changing its mechanical properties. Shot peening is an example of inducing residual stresses control. PAGE 29 21 21 Shot peening is a process where a small round sphere or other object is propelled with enough velocity to create local yielding on the surface (a small dent). This local yielding causes compressive stresses to be built up on the surface around the impact area. Once this stress is there it will automatica lly act to close or prevent a small crack from spreading on the surface. It has been shown that any process that creates compressive stresses on the surface of the spring will increases the life of the spring. Surface coating can be as simple as painting the spring protecting it from the environment and potential crack producing abra sions or corrosive chemicals or it can be as complex as electroplating. Figure 12 Residual Shear Stress Distribution After Pres etting Figure 13 Loaded Shear Stress Distribution After Presetting PAGE 30 22 22 These various processes are all useful in increasing the life of springs, and a re usually used in combination. Shot peening and heat treatments are particularly useful, where plat ing is more difficult to achieve because the bond between the two metals must be vary strong to prevent separation during torsion of the material. A typical spring might go through the following processes: after coiling an d annealing (to remove stresses f rom the coiling process), a dip in anti corrosive liquid, a pre set to induce beneficial residual stresses, and shot peening. Figure 12 shows the residual stresses that are induced by pre setting. Figure 13 shows the new stress distribution when the spring is loaded. All these work together to preserve and enhance the fatigue life of the spring. When properly done this can increase the stress ranges up to thirty percent or increase the life up to ten fold (Carlson, 1969). 2.3 Spring Design Due to Existing C onstraints The springs that are considered here are for pre existing shapes that can not or would be prohibitively expensive to be altered. As such there are limits to what can be done to the spring design. The object then is to find a way to extract th e needed force out of the spring while maintaining the same form factor. O ne helpful approach is minimize the use of resources to allow for the best utilization of the available space. In an article in Springs Henry Sweiskowski (1995) explains that there are three basic way s to minimize the amount of material. The first uses the initial load as the variable for optim ization. He states that retainer springs (l ock washers) are in this group. The second optimizes the final load T he author did not s pecify wh ich type of spring this group was best suited for but static ally loaded springs seems to be a good fit. The third method is the PAGE 31 23 23 optimization of the energy sto red for a given working stroke. H e commented that springs used for stopping or acceleratin g a mass belong in this group. This case is the one that interests u s the most but all are useful. The core of this work involves taking the derivative of the appropriate variables (wire diameter, coil diameter and spring index) in the spring equations to minimize spring volume Of the three cases, we will derive the third case as it is useful for our purposes We start with five equations: We can now combine these equations and solve for the volume: Taking the deriva tive of the volume with respect to the spring index and setting it to zero will yield the minimum volume : PAGE 32 24 24 [27] Solving for C: Finally if [29] is substituted for into [26] we obtain Sw eiskoski also notes that when Eq. 29 is used to choose the wire diameter the minimum value for volume is independent from the coil diameter and stroke. I t is also worth noting that Eq.3 0 can be solved for the energy stored (E) showing the direct relation between the torsional modulus and the energy capacity of a spring, while solving for S will shows that the stress is proportional to the square r oot of the volume of material. This may be part of the reason for the push for better materials in springs rat her tha n new theories and design techniques. There has always been a drive in for smaller springs that are able to store more energy, as can be seen from one of the first uses of springs in the pocket watch. Research in this area continues to this day, al though most of it deals with new alloys or different processes after winding to allow the spring to withstand gre ater stresses, or last longer. Despite this long history it is hard to find information about maximizing the amo unt of energy in a given vo lum e or for a given deflection and the majority of programs that exist only calculate the spring rate and deflection based on par ameters submitted by the user. PAGE 33 25 25 The energy is stored in the spring essentially as stress T he desired fatigue life is often the lim iting factor when designing for optimum efficie ncy. This means that a wire thicker th a n the optimum value is often needed in order to keep stress es within acceptable levels. Central to the ideas of spring design just examined is the contention t hat there is more tha n one configuration that will achieve the desired results, with one the most desirable solutions being that the amount of material used to be a minimum To find the optimum spring design one could reason that the design that stores the maximum e nergy U for a given space a nd material would be the best. But as we see when deriving the equation for stored energy the energy stored is essentially the stress on the spring. This means the design is also constrained by the m aximum shear stress f s tha t the material can support : Th e amount of energy stored by the spring can be increased by adding material increasing d D and n or increasing its maximum shear stress by changing materials or vario us treatments after coiling. New alloys provide more strength and thus more energy storage but at a greater cost and often with other limitations as well. Extra processes aimed at increasing the life or strength of a material also add cost and complexity to a sprin g. For this reason it is difficult to know which design will be c heaper as the design may minimize the amount of material and then a host of processes could be used to improve its basic c a pa bilities at added cost, a different material and different processes could be used t o also produce an optimal design. With this in mind the engineer must contribute to the final decisions, but the process can be made easier by speeding up the spring design. PAGE 34 26 26 CHAPTER 3 OPTIMUM SPRING DESIGN WITH FATIGUE 3.1 Proposed Model Current co mmercially available spring design programs allow for the determination of spring parameters typically as a function of free length, coil diameter, spring rate and spring material. Other parameters such as number of coils, stress factors, wire diameters et c. are chosen by the program and not available to the designer as a design variable. In a good number of cases this is undesirable because this makes it difficult to find acceptable designs when volumetric constraints are present such as in hydraulic cartr idge valve spring designs. In dev eloping the model our main objective is t o addr ess this problem by optimizing the design of a spring using a greater number of chosen spring parameters in a way that requires a minimum of knowledge and experience from the user The calculation procedure that the program uses automates the process of trial and error t o selec t a spring that will satisfy the design requirements This saves time and allows the designer to eliminate many designs that would not work and reduces design time The program incorporates a subroutine designed to expedite this process. The subroutine is not meant to produce a final spring design as it is not possible to evaluate whether the spring will satisfy all the criteria tha t the designer has in m ind, instead it produce s a design that PAGE 35 27 27 satisfies all major design criteria and constraints and is a robust starting point to a final custom izing process. In contrast, the models that are currently available in most programs will calculate a spring once the designer has specif ied enough criteria This is despite the f act that some of the criteria like the number of coils, the diameter of wire, the diameter of the spring coil, and the free length, are almost never critical and can be varied in almost all desi gns. This process allows the engineer greater flexibility in designing a spring and allows greater customization t han currently available 3.1.1 Spring Variables and Constraints Most spring design programs assign equal importance to all variables Becaus e of that, variables that are important to most designers such as spring rate and free length are given the same weight as variables that are not as important to the design such as the wire diameter and the number of coils which are seldom important facto rs in designs. Another parameter of secondary importance is the coil diameter which can usually be varied somewhat with out significantly affecting spring performance so that it can be increased or decreas ed to suit a particular design. By acknowledging th e s e facts, the designer will find it much easier to create a design that will work Most engineers do not have experience that would allow them to realize this. This becomes critically important when the fatigue life is included in the design. This difficu lty is due to the relationship between physical constraints that affect the design and the ability of the designer to correctly anticipate operating load conditions. A spring can be designed such that given the same critical properties: spring rate, free l engt h and approximate coil diameter, it will have different fatigue life because of the variation of other parameters This fatigue life varies PAGE 36 28 28 directly with the amount of material in the spring in general, more material means longer life As noted previou sly, the most comm on constraints to designs are f ree length, spring rate, and coil diameter. Free length is important because most spring designs try to avoid having any type of gap between the spring end and the load because a gap would cause an impact l oading condition. Thi s would reduce fatigue life and require a greater safety margin. In addition such impacts usually affect the part in which the spring is installed causing unforeseen stre sses and vibrations. Increasing the length to eliminate a gap wi thout changing the number of coils will also change the spring rate The number of turns is altered to allow the spring to reach the new length and to maintain the same rate Because all the coils deflect th e same amount for a given load if more coils are added the result is a larger def l ection S ince the spring is longer it needs more deflection to keep the same deflection to length (rate) ratio If the number of coils is kept constant (so the pitch is altered to reach the new length) the rate will rema in the same. Altering the number of coils is not usually done in commercially available programs because the general spring theory applies only to small variations in pitch and it becomes possible to violate this condition when the spring pitch becomes la rge. When the pitch is large the individual coils can deflect a greater distance before the spring goes soli d (the condition when the spring is fully deflected and there is space between coils) If so the material can be subject to plastic deformation d u e to excessive stresses This will cause a permanent set, lo wering the original free length and greatly reducing fatigue life. Spring rate is one of the most obvious design parameters A s it is usually the most important property of a spring it is almost always specified by the designer. A spring with a spring rate that is too we ak for the given load will compress until solid. When this PAGE 37 29 29 point is reached the spring is simply deforming the material and not storing energy nor can it absorb i mpact If the spr ing rate is too high the spring will not deflect it will effectively behave like a solid bar and will be of little use. Spring r ate is affected by wire and coil diameter s and to a lesser extent by the number of coils (or turns) S o altering these allows t he engineer to change other spring properties while maintaining the sa me rate. In addition to this spring rates are affected by the tolerances of the design and the quality of the material used. Because of this springs usually vary by about 3% from the calculated values. The number of coils, al though comple tely irrelevant to the designer most of the time, is n ot a user set parameter in most commercial programs This variable is usually dictated by the mechani cs of the spring. As noted previously it aff ects the spring rate but it is not used to alter the design. The risks of reducing the number of coils are : overstressing ( possibly taking a permane nt set ) and go ing solid It can however be used in applications where there is a large available deflecti on before the spring becomes solid Also there are secondary effects that would push a designer to look for designs with a specific number of coils (related to load distribution) T hese effects are generally small and can be ignored in most designs. Coil diameter is more useful as a variable tha n the number of coils, but it is not one that can be greatly altered. Changing the coil diameter will affect the stresses and the spring rate but less so tha n changing the wire diameter. It also has the advantage o f a wider range of available values (unlike wire diameter) and it wil l not a ffect the available deflection making it less complicated to use as a variable. The coil diameter is limited by the fact that springs usually have to fit in a cavity or over some o ther object (sometimes PAGE 38 30 30 another spring). Designs must give the spring a clearance between it and its housing and any object that it may fit over. This somewhat lity to alter the coil diameter an d its usefulness as a variable. The va riable that is most useful in terms of design is the wire diameter It has a strong impact on the attributes of a spring Despite the limited number of wire diameters available, t his is somethi ng that seldom hinders a design Alter ing this value has to be done carefully. Increasing the wire diameter will result in slightly lager corrected stresses and less deflection. F rom a practical point of view the wire diameters that can be used are limited to the standard siz es available from the distributo rs C hangi ng the diameter can sometimes alter the strength of the material as is the case with music wire. This means that it is often necessary to make adjustments to the coil diameter to achieve the rate that is desired. 3.1.2 TK Solver The author used TK solve r to develop and run the spring design program. TK Solver is a powerful modeling software program that uses a rule based language and a robust engine for solving systems of equations either algebraically or by iteration It can skip equations for which it needs information, and continue on to others. It then returns to the bypassed equation s and tries to solve them now with the new information. This is done vi a multiple passe the main sheet containing al l the eq uations that the program will use. This sheet can include subr outines and procedures, but these are not treated in the same multiple pass system; rather they are solved line by line. The sub routine codes are complete in that loops and function calls PAGE 39 31 31 and er ror messages within the program can be used, much like in other programming languages. A screen shot of the program is shown in Figure 14 Figure 14 Screen S hot of Software Program 3.1. 3 Numerical Procedure The numeri cal proce dure is one of iterative redesigns, t hat is the design parameters obtained in one iteration can be reapplied as starting parameters of the new spring design. Once this is done and the new criteria selected the program then goes about finding a solution t hat meets the criteria se t forth. Some of the criteria are internal and kept with in the program. The s e criteria are related to stress levels and how close to solid the spring will be once loaded to the determined height. The program requires four variabl es to generate a sprin g design. A n additional five four for fatigue and one for the buckling condition, are required to determine the full fatigue lif e The program can solve for any variables but it will generate an error if the PAGE 40 32 32 coil diameter is not one of the parameters given. This is due to the rules in the program When the solver subroutine is used the wire diameter no longer needs to be spe cified The two settings of the solve r subroutine as well as its internal procedures are of interest so the y will be examined in detail here. On the first setting the solver tries to find a spring that is at lest 30% below the stress needed to make the spring yield. It also looks to use 85% of the available deflection. These values were chosen to avoid the prob lems of non linearity that occur from the spring going solid. The encyclopedia on spring design by the Spring hen it is necessary to specify a rate it should be specified between two test heights which lie within 15 to 85% o f the full These values allow the s pring a good margin of safety of operation. The program also works to make the spring stay below 85 % of the maximum fatigue stress. The process that the solver uses is a nested loop. The solver starts by using the given outer diameter s ubtracting a tenth of an inch f r o m it, and choosing the smallest avail able wire size. It then checks the spring to see if it satisfies th e criteria that the designer has in mind. If the criteria are not met, it increases the coil diameter by a thousandth of an inch. By increasing the coil diameter, the spring rate and stress will be lowered. This means the higher stress solutions will be eliminated first. When the coil diameter has been increased by two tenths of an inch the inner loop will exit and the wire diameter will be increased by one size. Then the inner loop starts again. The inner loop is where the checks for fatigue life and wo rking load are done. If a solution is found both loops are ended. From this point th e designer can turn the solver off and make small adjustments to the design that better suit s his purpose. The initial design is usually not usable for practical reasons. Because spring manufacturers will provide coil sizes in PAGE 41 33 33 quarter of a turn increments only, a case when the program specifies a spring with a number of total coils that does not conform to this fact is not acceptable Because of this, it is important to b e able to fine tune the design. The basic variables used by the program, initial and f inal load heights, spring rates, free length, wire diameter and outer coil diameter (or inner) are interchangeable When using the solver subroutine the wire diameter cannot be specified and an initial guess of the coil diameter must be supplied Figure 15. Program Flow Diagram PAGE 42 34 34 Following the flow diagram shown in Figure 15 w e will go through an example assuming that the free length, initial and final working loads and initial load height are specified. The buckling condition will be that of one end piv oted a nd one constrained, which is a common constraint for springs. After the program initializes the variables, it enters the first loop. This loop is set to cycle through every wire diameter available. It does this by taking the number of wire sizes in t he int ernal list and using this as an index or counter for the number of cycles. The loop counter is used to select the wire size from the list It then calculates the maximum amount that the diameter can be varied. It does this by using the hole and shaft parameters, subtracting the shaft diameter from the hole. This value is then divided by the variation increment (set to 0.001) and this is the number of cycles for the second loop Before entering the second loop the initial diameter of the spring is set This is done in the first loop since the diameter must be reset each time the second loop is completed. It is set to the shaft diameter plus twice t he wire diameter just selected. Once this is done the second loop is entered. In the second loop the sp ring is actually calculated. First the initial and final working heights are subtracted to obtain the change in height. T hen the initial and final loads are subtracted and divided by the change in height to calculate the spring rate. Once the rate is obtai ned the number of active coils is calculated using the shear modulus (G), the wire diameter the rate and coil diameter. T he solid height is then calculated using the number of turns and wire diameter. Then the solid height is checked to see if a solutio n is possible with the current wire diameter. After this various factors are calculated including basic working stresses at initial, final and solid loads. The n the program calculates the Whal correction factor, the spring index, an d the ultimate tensile strength There is a check PAGE 43 35 35 performed to see if a spring can be set (compressed to solid to induce beneficial residual stresses). Then the yield stress is calculated based on whether the spring is set or unset. The program has two different setting, each o ne of them using three cr iteria to finalize the design. For the reasons stated previously, t he first setting looks for a spring that uses at most 85 % of th e available deflection, is 15% above solid height and uses 85% of the yield strength. The second sett ing uses at most 90% of the available deflection, is at least 10% above solid height and uses 90% of the yield strength. If the spring can meet the three chosen criteria then the program proceeds to calculate the fatigue properties of the spring. The fatig ue section calculates a large number of parameters related to the fatigue life of the coil. Among them are the mean stress, alte rnating stresses, fatigue life and effects of shot peening. Then a final check is made and if the spring passes this check it se ts the exit variable to complete and the loops are both exited. The program then returns the new wire diameter and the di ameter of the coils 3.2 Numerical Results T o test and verify the program and its capabilities three Sun Hydraulics springs Spring s 238, 225 and 239 were used as a test for the program Spring 238 was one of the better documented springs having two tested revisions. The second revision was necessary because one of the first samples tested failed (yielded) under load. T he original des ign of Spring 238 was flawed in that the spring rate was too high, with a spring rate around 970 lbs/in, and it was intended to fit over another spring, 239, which had a diameter of 0.515 inches, not leaving any clearance. Studying th e basic information a b out it provided in T able 1 it is obvious that the spring had no fatigue life problems since without presetting or shot peening it still had a n infini te safety factor of PAGE 44 36 36 1.13, and was still 15% above the yield stresses at the fin al loading point This indi cates that shot peeing is probably unnecessary and was added more as an additional safety margin. In the first redesign ( see Table 1 ) the wire diameter was reduced to 0.177 inches Table 1: Spring 238 Original First Redesign Second Red esign Outer Diameter [in.] 0.889 0.875 0.900 Inner Diameter [in.] 0.515 0.521 0.526 Wire Diameter [in.] 0.187 0.177 0.187 Total Number of coils [1] 8 7 7.75 Number of active coils [1] 5.5 4.5 5.25 Rate [lb/in] 932 930 924 Free length [in.] 1.815 1.80 4 1.835 Solid height [in.] 1.496 1.239 1.449 End condition [1] Closed & ground Closed & ground Closed & ground Initial Load [lb.] 118 118 118 Final working height [in.] 1.688 1.677 1.707 Initial stress [lb/in 2 ] 46,327 53,300 46,800 Final load [lb.] 234.00 234.00 234.00 Final stress [lb/in 2 ] 1.563 1.551 1.581 Final working height [in.] 92,060 106,300 92,800 Percent below yield stress [%] 50.17 3.45 49.49 Estimated life [cycles] cycles] Infinite <100K Infinite Infinite safety factor [1] 1.85 0.92 1.96 the in order to allow for a larger inner spring clearance The diameter of the coil was then constricted to 0.875 inches to maintain the spring rate. The free length and the inner diameter were shortened to 1.804 and increased to 0.521 inches, respe ctively, to allow the spring to achieve the load targets and to fit over the inner spring. The number of coils was also reduce d to 7 from the original 8 ( increasing the spring rate at the same time ). As can be seen these measures increased the stres ses pl aced on the spring. T hree samples of the first redesign of Spring 238 T he high stresses caused one of the springs to fail during PAGE 45 37 37 testing and al though the other two springs performed well in terms of spring rate, a detailed exa mination of the test data led to the conclusion that this design was not acceptable because it was almost yielding at the final load The initial re design of S pring 238 was made with out the aid of the fatigue calculations. After the failure it became apparent that this was not pract ical and, as a result, fatigue calculations we re added to a later version of the program. Running the program again for this spring with the added fatigue calculations yielded a fatigue life value that indicated that the spring wa s suitable for static load ing only The spring life could be improved by setting which would boost the life to over 100 thousand cycles, but this would require a special setting apparatus since compressing it to solid (or shut), t he usual way of setting springs, would cause it to def orm or break. The yield height was just un der a tenth of an inch lower tha n the final load (with a corresponding load increase of about 90 lbs). Shot peening wou l d also increase the fatigue life and with the special set would give the spring a calculate d infi nite safety factor of 1.48. In addition, in this re design the yield point was extremely close to the final working height so it was decided to try a second redesign For the next redesign the wire diameter returned to its original size (0.187 inche s) and the outer diameter of the spring was increased to 0.900 inches. The number of active coils was also slightly reduced by a quarter turn. This allowe d the spring rate to remain as needed while reducing the stress levels. The free length was actually i ncreased i n this design, as there were no clearance issues with the adjoining part. Because of this the spring was given a tolerance that tended to reduce the free length PAGE 46 38 38 The problems with this spring that were related to the spri ng rate were corrected and the final spring had a slightly lower spring rate of 924 lbs/in while maintaining a calculated infinite safety factor above 1.5. The initial and final working heights are a little higher then the original design but that was fou nd to be acceptable Table 2: Spring 225 Original Redesign Outer Diameter [in.] 0.656 0.636 Inner Diameter [in.] 0.392 0.386 Wire Diameter [in.] 0.132 0.125 Total Number of coils [1] 7.5 7.5 Number of active coils [1] 5.75 5. 00 Rate [lb/in] 527.5 526 Free length [in.] 1.250 1.250 Solid height [in.] 0.990 0.937 End condition [1] Closed and ground Closed and ground Initial Load [lb.] 70 70 Final working height [in.] 1.117 1.117 Initial stress [lb/in 2 ] 45,700 52,300 Fina l load [lb.] 111 111 Final stress [lb/in 2 ] 72,500 83,000 Final working height [in.] 1.039 1.039 Percent below yield stress [%] 52.9 46.57 Estimated life [cycles] cycles] Infinite Infinite Infinite safety factor [1] 2.52 1.52 The springs that follow were redesigned once and tested They exhibited problems related to spring rate or the number of active coils. Th e next spring to be redesigned was the 225 spring shown in Table 2 This spring also had a rate problem Measurements showed a rate of 610 lb s/in while the design called for a rate of 5 27 lbs/in. The design PAGE 47 39 39 had a calculated infinite safety factor of 2.52 when set and shot peened, so fatigue life wa s not an issue. One of the probl ems that were identified was that the number of inactive coils (or turns) was inaccurate at 1.75 inactive coils. The closed and ground end condition means that at least one coil is inactive at both ends. In addition because of the helix angle, the spring does not separate from the ground coil until about a quarter turn. This adds up to about 2.5 inactive coils. The fewer coils d eflecting caused the spring rate to be higher tha n desired. Another problem was that the spring was very close to solid when the final load was applied, leaving 0.05 inches of deflection. This f orced the spring maker to be very careful with the tolerance on the wire diameter as typical manufacturing tolerances of the wire could cause trouble. The redesign decreased the coil diameter and reduced the wire diameter, while correcting the number of a ctive coils to 5. The decreased wire diameter allowed the spring to deflect more, easing the fears of the spring going solid because of an out of tolerance wire. Also, the calculated load height is exactly the same as in the original design. The largest change is the fact that this spring has a lower fatigue life safety factor, and this is in part because the spring is not shot peened in the design. However it is set and wit h out this the safety fa ctor drops to 1.45. For the case in which the spring is s hot peened and pre set the factor of safety jumps to 2.52 This did not seem necessary and because of the added cost to the spring manufacture it was removed. PAGE 48 40 40 Spring 176, shown in Table 3, the next to be studied had similar problems to Spring 225 in th at it had an impractical value for the number of inactive coils. This also caused the spring to have a higher tha n calculated spring rate value (284 lbs/in ). Table 3: Spring 176 Original Redesign Outer Diameter [in.] 0.462 0.470 Inner Diamete r [in.] 0.272 0.272 Wire Diameter [in.] 0.095 0.093 Total Number of coils [1] 9.5 10.25 Number of active coils [1] 8.75 7.75 Rate [lb/in] 271 264 Free length [in.] 1.313 1.313 Solid height [in.] 0.903 0.953 End condition [1] Closed and ground Closed and ground Initial Load [lb.] NA NA Final working height [in.] NA NA Initial stress [lb/in 2 ] NA NA Final load [lb.] 67.5 67.5 Final stress [lb/in 2 ] 104,000 90,500 Final working height [in.] 1.062 1.058 Percent below yield stress [%] 14.1 45.52 Es timated life [cycles] cycles] <100 K 100+ K Infinite safety factor [1] 0.63 .86 This spring has only three quarters of an inactive coil turn and this is not possible for the end condition given since for a closed spring one need s at last one full turn to form the spring base at both top and bottom. In addition, another quarter turn at each end is needed for the spring to reach its full pitch. Because of this the spring rate is too high Alt hough the infinite life factor was not listed as one of the pro blems the spring as designed was calculated as below infinite life. These calculations were done without shot peening or pre sett ing since that is what was specified for the spring. One of the reasons PAGE 49 41 41 the safety factor is so low is that the spring is using almost all of the available deflection causin g a very severe loading condition. The redesign corrected the coil problem and some other modifications were made namely adding shot peening and setting to improve fatigue life. The improvement in fatigue lif e was still not enough to give it infinite life as per specifications the spring rate was also lower. The coil diameter was slightly increased and the wire diameter decreased to 0.93. This had to be done as an increase in coil diameter would make the sprin g go solid. The spring had to gain an inactive coil and lose an active one. This makes the total active coils 7.75 and th e total number of coils 10.25. T his resulted in a solid height of 0.953 inches, still a reasonable distance from solid but not as much as the original. The infinite l ife factor of 0.86 is better tha n the original design, but due to constraints the wire diameter can not be increased. Increasing the wire diameter would improve the fatigue life of t he spring The next spring we will look a t is S pring 239. It also had problems with the number of inactive coils though not as much as S pring 176. This means that its rate was also off. Another concern was that the final working height was close to making the spring go solid. Of interest is that this spring was used inside the previously examined spring 238. Looking at the solid height and the final load height, it is apparent that only 0.03 inches separates them. The fat igue life on this spring is high enough so that even with out shot peening th e spring will have infini te life safety factor of 1.33, al t hough pre setting of the spring must be done since the spring is so close to solid at its final working height. The redesign decreased the number of inactive coils and again reduced the wire diame ter. The spring rate is slightly lower, an d the working heights are closer but a little PAGE 50 42 42 higher then the original. This is due to an increased free length. The details of the spring are shown in table 4 below. T he wire diameter was decreased to 0.095 and the coil diameter reduced to 0.498 inches. The free length was increased to allow the rate to be closer to the original. This spring was also shot peened and set, since with out shot peening the infinite life safety factor drops to 1.11. The fact that this spring is not quite as durable should come as no surprise since all of the modifications increased the stress levels. Table 4 : Spring 239 Original Redesign Outer Diameter [in.] 0.510 0.498 Inner Diameter [in.] 0.315 0.308 Wire Diam eter [in.] 0.100 0.095 Total Number of coils [1] 9.5 8.5 Number of active coils [1] 11.5 11.0 Rate [lb/in] 212 210 Free length [in.] 1.487 1.517 Solid height [in.] 1.150 1.045 End condition [1] Closed and ground Closed and ground Initial Load [lb.] 36.8 36.8 Final working height [in.] 43,100 49,200 Initial stress [lb/in 2 ] 1.313 1.342 Final load [lb.] 63.3 63.3 Final stress [lb/in 2 ] 74,100 84,700 Final working height [in.] 1.188 1.216 Percent below yield stress [%] 53.96 47.85 Estimated life [ cycles] cycles] Infinite Infinite Infinite safety factor [1] 2.16 1.89 PAGE 51 43 43 3.3 Comparison with Test Results The test results were compiled using a compression tester that produc ed a graph of the load versus the deflection curve The tester had two load c ells with 100 lb and 1000 lb rating s The springs were compressed at a rate of half an inch per minute. To compensate for the flex of the machine a calibration curve was created for both load cells and it is available in the appendix. The raw data was th en tu rned into a graph. This is accomplished by selecting points on the load deflection curve and placing them into E xcel. The points are then used to calculate rates by taking the first and second pairs of data and subtracting them. This gives the change in load and the change in distance. Then the change in load is divided by the change in deflection yielding the rate. This process is repeated with second and third points and so on until the number of desired points is accumulated. This raw data is furthe r altered by using the calibration curve to compensate for the deflection of the tester. The graph shows the spring rate as a function of the applied load. One thing that will be obvious from looking at the graphs is that springs are not linear in gener al, but they are very close to being so The spring rate changes the most at the initial deflection as some coils close to the ends go solid, and when the spring is near solid height as all the coils start to touch and go solid. The calculations then ref lect an average value, not any specific value. The graph below (Figure 1 6 two data sets, one with a spring that did not yield during testing and the other that did. The spring that yielded is easily picked out as the one that shows a dramatic drop in the spring rate when it reaches the end of the test. The point at which it failed was the final load PAGE 52 44 44 height specified in the spring design. The rate for that spring drops about 50 l bs/in when the other two sample s increase by that amount. As expected the two first revision springs have very similar loading curves, and an average spring rate of 891 and 895. This rate is a little low when compared to the 930 lbs/in the design calls out The second load curve is not as linear but it has an average spring rate of 921 lbs/in, closer to the 927 lbs/in calculated in the design. Figure 16. Spring 238 Redesigns S are fairly consistent. First r edesign was a weaker tha n expected, but it was the most linear of the groups that was tested. This makes it unfortunate that it would fail as it did. The second redesign is quite erratic in that the data points that were calculated seem to have fluctuating rates PAGE 53 45 45 S pring 22 5 had a predicted spring rate of 527 lbs/in close to the 526 lbs/in that was predicted. In addition the spring seams to have a good working region that has a linear spring rate as can be seen in the graph (Figure 1 7 ) The initial jump in rate is not Figure 17 Spring 225 Redesign. uncommon as the coils near the ends will very quickly touch the end coils, causing the spring rate to jump up as fewer coils deflect. Though there is a dip at the end this was not due to yielding. It is remarkable that the spring rate is so close to the predicted value, and some of the other springs that were tested did not have rates this close but they were in the range of 550 lbs/in which is stil l a fairly accurate prediction. Spri ng 239 was designed to fit inside spring 238. It was redesigned to achieve a spring rate of 210 lb/in. The test results are shown in Figure 18. PAGE 54 46 46 Figure 18 Spring 239 Redesign The test data shows that the spring rate is very lin ear with an average spring rate close to th e design value with a narrow range, less than 5% within the testing range It was therefore co nsidered a successful redesign. PAGE 55 47 47 CHAPTER 4 CONCLUSIONS AND RECOMMENDATIONS 4.1. Conclusions The program was succ essful in modifying existing spring designs producing springs that, when tested, were found to have spring rates equal, with in tolerances to those calculated by the program. Multiple springs were tested and so the program can be considered reliable Fatig ue life calculations w ere incorporated and proved successful No spring that wa s redesigned by this program had problems in terms of fatigue life when tested The addition of automation makes the process of spring design easier on the designer In additi on, the internal criteria for selecting springs such as checking for how much of the deflection is in the linear region of the spring, how close the spring is to yielding, and what the fatigue life of the spring is, prevent bad designs from being suggested. This reduces the level of expertise required to run the program, making it more accessible. This is important as most engineers do not have much experience designing springs, yet may be required to do so to incorporate a spring into a design. The freedom to alter any parameter of the spring allows the designer to create designs that other commercial ly available software will not. The automation makes it easier for the des igner to use the program, but it does so without removing the customizabi lity of PAGE 56 48 48 the spring de sign, a llowing the designer t o make adjustments as needed This for example means that the number of coils in a calculated design can be altered to allow the designer to provide a specification which a manufacturer can reliably produce The process of spring optimization depends ultimately on what the designer whishes to optimize. Most designers prefer designs that minimize use of resources and thus the expense of producing the design. This is a complex issue and the flexibility of th e program allows the designer chose how to go about it. The program provide s a solution that minimizes the use of materials by using the smallest wire and coil diameter for a given fatigue life, it then allows the user to optimize the spring to suit his ot her needs. 4.2. Recommendations for Further Work The program would benefit greatly from a graphical user interface. This would allow the user to see more clearly the relevant da ta. Currently this information is only available in a worksheet This detrac ts from the ease of use, and makes data analysis cumbersome Another modification to the program would be the ability to add loading characteristics in order to achieve more accurate fatigue calculations. This would allow for other uses of the spring besi des simple compression, such as use in vibrations mounts, or springs designed to absorb shocks, since these do not have a load that is evenly applied to the spring surface, which increases fatigue. This addition would also aid in the process of spring opti mization since the designer would have a better idea of the true values of the stresses. The current program will display only one result from all the possible results. In addition this result is the first res ult that the program finds. T he program could be PAGE 57 49 49 modified to provide several alternative possible designs and select a representative sample of the available designs T his would provide even greater flexibility and save even more time and effort for the spring designer. Finally it should be achiev able to solve the problem of obtaining a solution with a number of coils which is practical for the spring designer to produce. To accomplish this the number of coils suggested should have only whole, half, or quarter coils. This is a tolerance that is acc eptable for nearly all the spring manufactures and the automatic coilin g machines in use today Though the study of springs and their properties is a well established area, there is no real criterion for the evaluati on of a design. In general, optimizati on has not beet used by the spring makers to choose among possible designs as little information on what constitutes an optimal design exists and no formal way of deciding which of two designs is a better use of resources. The design and fabrication of sp rings is often driven by constraints, but this should not mean that there is no formal way to establish a more efficient design. With the wide use of springs and mass production techniques this should be a valuable procedure to have. I t would be worth the effort to study the relationship between the volu me of a material and spring performance. If a corre lation is found it could allow the designer to know the minimum material required to achieve the design requirements. PAGE 58 50 50 REFERENCES Adams, W. A., 1850, Instn. Mech. Engrs. Proc ., Jan., p 19 31 Instn. Mech. 8 Proc ., p 160 165 Bannantine, Julie A, 1990, Fundamentals of Metal Fatigue Analysis ,. p.2 and 5 Prent ice Hall Englewood Cliffs N. J. Am. Mach ., v 15, n 33, p 2 3 Z. VDI Bd 77, 198. Carlson Fatigue Testing of Springs Springs May, p. 18 19 Carlson Springs October, p. 47 53 Academie des Sciences Comptes Rendus V 81, p 1122 1123 Doubler, J. W. H., Sci. Am. Supp., v 2, p 791 Mech. World v. 121, n. 3149, 3150, 3151, 3152, 3153, May 23, 1947, p. 4937, May 30, p. 523 7 June 6, p. 558 63, June 13, p. 585 7, June 20, p. 610, 612 Instn. Civil Engrs Proc., p. 224 254 French, Am. Soc. Mech. Engrs Trans., v 23, p 298 312 PAGE 59 51 51 Journal of Science v. 18 p. 95 98 Goener, O., 1932, Die Berechnung zylindrischer Schraubenfedern, Z. VDI Bd76, 269 Honegger E., 1930, Zur Berechnung von Schraubenfedern mit Kreisquerschnittt P. Appl. Mech ., Vol. 12, 99. Brown B overi Rev ., vol. 18, no. 35 Mar. 1931, p 120 125 Hooke, R., 1678, De Potentia Restitutiva London Berlin, XXV Berlin. Leveaux, E. H., 1876, Sci. Am. Sup p., v. 2, p 741 743 il Springs for Freight Cars Railroad Gaz ., v 30, p119 120 Societe des Ingenieurs Civils de France p 629 642 Nadal, M. J., 1896, Theory of the Stab Annales des Mines ser 9, v 9, p 415 467 Annales des Mines: Memoires v 20, ser 5, p 1 107 Rankine, W. J. M., 1866, Engineer v 222, p 3 74 Academie des Sciences Comptes Rendus v 107, p 713 718 Academie des Sciences Compte s Rendus v. 114, p 147 152 Rey, L., 1876, Societe des Ingenieurs Civils de France p 845 855 Z. VDI Bd57, 1906. Sweisk Springs September 1995, p 39 59 PAGE 60 52 52 Transactions ASME v 51, also Am. So c. Mech. Engrs., Advance Paper no. 39 for mtg. Dec. 3 7, 1928 Am. Soc. Mech. Engrs. Trans. (J. Applied Mechanics) v. 6, n. 1 and 4, p. A 25 30 Wahl, A. M ., 1944, Mechanical Springs 1st edition McGraw Hill, New York. Wahl, A. M., 1963, Mechanical Springs 2nd edition McGraw Hill, New York. 33 34. ry of the Main Springs of a Watch Showing how to Select and Jl. Franklin Inst ., v 54, p 344 347 56 PAGE 61 53 53 BIBLIOGRAPHY Cain, Tubal, 1988, Spring Design and Manufacture, Argus Books, Swanley, Kent, England. Spring Manufacturers Institute, 2002, Handbook of Spring Design, Oak Brook, Il. Society of Automotive Engineers, 1996, Spring Design Manual Second Edition, Warrendale, Pa. Wahl, A. M., 1991, Mechanical Springs Spring Manufacturers Institute, Oak Brook Il. PAGE 62 54 54 APPENDICES PAGE 63 55 55 Appendix A Program Code A. 1 Main Program Rule DATE=DATE() TIME=TIME() ;Setting section set_check = (Ss*Kw Sys('unset)*Sut)/(Sys('uns et)*Sut)*100 If and (set_check >= 10, set_check<= 30) then OK = 'yes else OK = 'no If OK = 'yes then f = 'try_setting FinalLoadChk=(KwSf Sys)/Sys*100 ;Error section if and(solved(),FinalLoadChk > 10) then call errmsg("Final working Load is close t o causing the spring to yeild, consider a new design.") IF AND(solved(),solid/fwh=>1,SOLVED()) THEN CALL ERRMSG("SPRING WILL GO SOLID") IF d=>od/2 THEN CALL ERRMSG("WIRE DIA TOO LARGE, OR SPRING DIA TOO SMALL") if and(BuckleFlag=1,B=1) then call ErrMsg( "Spring may not be stable with given end condition Try to use a 1/4 or 3/4 turn for the ends.") IF and (solved(),set_check > 35,setting ='set,FinalLoadFlag=1) THEN CALL ERRMSG("Setting spring causes too much stress. Change setting to 'unset'.") ; Peening section Sew = TEL(peen) If (peen = 'unpeened) then Sfw = Sfw_unpeen(cycles)*Sut If (peen = 'peened) then Sfw = Sfw_peen(cycles)*Sut ;Spring design equantions PAGE 64 56 56 Appendix A.1. (Continued) od=id+2*d D=od d ; MEAN DIA c=od/d 1 ; SPRING INDEX delta_h=iwh fwh ; WORKING STROKE rate=(Lf Li)/(delta_h) ; RATE BASED ON LOAD AND DEFLECTION free=iwh+Li/rate ; FREE LENGTH if EndCon =1 then solid=Nt*d else solid = (Nt+1)*d ; modified to include end conditions Ls=rate*(free solid) ; SOLID LOAD ; note this equation uses a DEFELCTION so end condidtions are irrelevant L3 = rate*(free h3) ;Load at specified height rate=G*d^4/(8*D^3*Na) Nt=Na+Ni L100 = (free h100) *rate Fn=14000*d/(Na*D^2) ; Approx natural frequency (Lee spring catalog) p=(free 2*d) /Na ;calculates pitch Note: Makes no alowece for end conditions ODF=(D^2+(p^2 d^2)/(pi())^2)^0.5+d ; Solid hight outer diameter ;Stress equations Kw=(4*c 1)/(4*c 4)+.615/c ; WAHL STRESS CONCENTRATION FACTOR Ks=1+.5/c Si=8*Li*D/(PI()*d^3) ; STRESS A T INITIAL ;this equation was using a stress concentration factor for set springs. Why Sf=8*Lf*D/(PI()*d^3) ; STRESS AT FINAL ;this equation was using a stress concentration factor for set springs. Why? Ss=8*Ls*D/(PI()*d^3) ; STRESS AT SOLID ;this equatio n was using a stress concentration factor for set springs. Why KwSi = CorrectedStress(Kw,Ks,Si,setting) KwSf= CorrectedStress(Kw,Ks,Sf,setting) KwSolid=CorrectedStress(Kw,Ks,Ss,setting) S100=8*L100*D*Kf/(PI()*d^3) ; 100,000 psi PAGE 65 57 57 Appendix A.1. (Continu ed) ;Buckling equations Buck_x = free/D Buck_y = (free fwh)/free call Buckle(;buckdef,LB) ; buckling BK1 = 1/(2*(1 G/BMod)) BK2 = 2*pi()^2*(1 G/BMod)/(1+2*G/BMod) ;Fatigue calculations if Material =1 then Sut = 184649*d^( .1625) else Sut=239 000 Kf=Kw Sus = 0.67 Sut Sa=8*Fa*D*Kw/(PI()*d^3) ; alternating stress Fa = (Lf Li)/2 Fm = (Li + Lf)/2 Sm=8*Fm*D*Ks/(PI()*d^3) ; mean stress Ses = (0.707*Sew*Sus)/(Sus 0.707*Sew) Sfs = (0.707*Sfw*Sus)/(Sus 0.707*Sfw) Nfs_finite_check = Sfs* (Sus Kw* Si)/(Sfs*(Sm Kw*Si) + Sus*Sa) Nfs_infinite_check = Ses*(Sus Kw*Si)/(Ses*(Sm Kw*Si) + Sus*Sa) Sys = Sys(setting)*Sut Ns_shut = Sys/(Ks*Ss) Ns = Sys/(Ks*Sf) Sms = Sus*(Sfs^2 Sfs*Sa + Sus*Sm)/(Sfs^2 + Sus^2) Sas = Sfs/Sus*Sms + Sfs Z S = sqrt((Sm Sms)^2 + (Sa Sas)^2) OZ = sqrt(Sa^2 + (Sm Kw*Si)^2) Nf2 = (OZ + ZS)/OZ Sms_i = Sus*(Ses^2 Ses*Sa + Sus*Sm)/(Ses^2 + Sus^2) Sas_i = Ses/Sus*Sms_i + Ses ZS_i = sqrt((Sm Sms_i)^2 + (Sa Sas_i)^2) OZ_i = sqrt(Sa^2 + (Sm Kw*Si)^ 2) Nf2_i = (OZ_i + ZS_i)/OZ_i Tms = (Sys + Sm Sa)/2 Tas = (Sys + Sa Sm)/2 PAGE 66 58 58 Appendix A.1. (Continued) ZSy = sqrt((Sm Tms)^2 + (Sa Tas)^2) Nfy = (OZ + ZSy)/OZ If Ns < 1 Then Nfs_finite = f Else Nfs_finite = MIN(Nf2,Nfy,Nfs_finite_check) If Ns < 1 Then Nfs_infinite = f Else Nfs_infinite = MIN(Nf2_i,Nfy,Nfs_infinite_check) ;These equations were added into the program form other sources if solved() then call GetStandard(d,Material;Dws,Dwl,WireAvailability) ;ASD Lw = Lw() ; Length of Wire req uired to form the Spring ;ASD Ap = ATAN( 0.5 p / ( D PI() / 2 ) ) 180 / PI() ;Pitch Angle ASD Def=free solid ;Available Deflection Na = G*d^4/(8*rate*D^3) ;Number of Active coils ASD FWP=Lf/((Pi()*DIA1^2*.25) (Pi()*DIA2^2*.25)) ;imported from spr ing2 calculates pressure IWP=Li/((Pi()*DIA1^2*.25) (Pi()*DIA2^2*.25)) ; imported from spring2 ;JP's equns LinearRatio = (free fwh)/(free solid) ; raito for linearity if Material = 1 then G = given('G,G,11500000) else G = given('G,G,11000000) ;import ed from spring2 gets a modulus based on wire dia, or the modulus of 17 7 stainless steel if Material = 1 then BMod=given('BMod,BMod, 3E7) else BMod=given('BMod,BMod,2950000) ; gives default values for music wire or stainless steel cycles = given('cycles, cycles, 10E6 ) ; gives default values for infinite life test. C all ChrisGoodman(Ses,Sew,Sus,Sfs,Sfw,Si,Sm,Sa,Sys) C all BlankASDPlot() ; essentially from ASD rule sheet Just in a function now DIA2=given('DIA2,DIA2,0) if EndCon =1 then Ni=given('Ni,Ni,2 .5) else Ni=given('Ni,Ni,3) ;default number of inactive coils is 2.5 CG and 3C Material=given('Material,Material,1) S100=given('S100,S100,100000) EndCon=given('EndCon,EndCon,1) if solved() then Li=given('Li,Li,0.00000000000001) PAGE 67 59 59 Appendix A.1. (Contin ued) YeildHT= (Sys('unset)*Sut*PI()*d^3)/(Kw*8*D*rate)+free SetHT= (Sys('unset)*Sut*PI()*d^3)/(Kw*8*D*rate)*1.3+free BC=given('BC,BC,4) if given('d) then d=d else call Guess_Wire_dia(Solver;DiaWireTest,d,ODTest) ;guesses the wire dia if buckdef=("No Bu ckle") then Buck=Def else Buck =buckdef ; this set of equations are ment to reduce chance if and((Buck)/(Def) if Mod(Nt*10,5)=0 then B=1 ; effective this is so it is canceled by me but can be activeated if it is found to be accurate. Solver=given('Solve r,Solver,2) ; Default for the solver. Special_Set=given('Special_Set,Special_Set,"no") if Special_Set="yes" then FinalLoadFlag=0 else FinalLoadFlag=1 ;Spring2 Plots if or(Li=0,Material=2) then call blank('life_inf) if or(Li=0,Material=2) then cal l blank('life_100k) if or(Li=0,Material=2) then call blank('Ap_factor) if and (Lf > Li,Material=1 ) then call InfinateLifeII(Lf,Li,KwSf,d;life) UDP=L3/((Pi()*DIA1^2*.25) (Pi()*DIA2^2*.25)) ;imported from spring2 calculates pressure EndCond=given('EndCo nd,EndCond,1) ;Default is Ground ends ;ASD plots if solved() then call FillLists(;ZZZ) if solved() then call Fill1(Li,iwh,P1t) if solved() then call Fill2(Lf,fwh,P2t) P1t = Toload(Li) ; load tolerances P2t = Toload(Lf) ; load tolerances PAGE 68 60 60 Appen dix A.1. (Continued) Lft = Tolfree(rate) KE = CE*N^Y GE = 386 if (setting='set) then call CYCLES(;KE,CE,Y) else N = "N/A" if solved() then call Goodman() if solved() then call holeshaft(ShaftDia,id,od,HoleDia,ODF) ;Spring2 plots A. 2 Solver Subrout ine Statement ExitVar=0 LoopFlag=0 if Material=1 then Wires=145 else Wires=114 for x=1 to Wires if Pre=0 then ExitVar = 1 If Pre=0 then exit if given('d) then exit if x=Wires then LoopFlag=LoopFlag+1 If and(x=Wires,LoopFlag=1) then x=1 if x =Wires then d=0 if x=Wires then call errmsg("The guess function can not find a suitable wire diameter to make a spring with the parameters given. You may want to change the parameters or try a larger wire diamter then .283 in.") If Material=1 then d='New _Spring_Guess_list[x] if Material=2 then d= 'WDS2[x] if Pre= 1 then VarDia=.00 else VarDia=.0 if Pre=1 then DiaCount=1 else DiaCount=1 ODI=.001 ODTest=od VarDia D=ODTest d for y=1 to DiaCount Del_h=iwh fwh rate=(Lf Li)/Del_h PAGE 69 61 61 Appendix A.2. (Conti nued) Na=(G*d^4)/(rate*8*D^3) Nt=Na+Ni solid=Nt*d free=iwh+Li/rate LinRatio=(free fwh)/(free solid) Ls=rate*(free solid) Ss=Ls*8*D/(Pi()*d^3) Sf=Lf*8*D/(Pi()*d^3) C=ODTest/d 1 if and (C<=1,LoopFlag=0) then C=1.01 if C=1 then C=1.01 Ks=1+.5/C Kw =(4*C 1)/(4*C 4)+.615/C; Sut=Ultimate_Stress(d,Material) if LoopFlag=0 then Sys=Sys('unset)*Sut else Sys=Sut*Sys('set) if LoopFlag=0 then Stress=(Kw*Sf Sys)/Sys else Stress=(Ks*Sf Sys)/Sys S et_check = (Ss*Kw Sys('unset)*Sut)/(Sys('unset)*Sut)*100 IF and(set_check > 35,set_check if Pre=2 then R=.9 else R=.85 if Pre=2 then W=1.1 else W=1.15 if Pre=2 then S= .1 else S= .3 if and(LinRatioW,Stress If Pre =1 then goto Skip ;this block will look at fatigue charicteristics Si=8*Li*D/(PI()*d^3) ; STRES S AT INITIAL ;now uses the Whal concentration factor Fa = (Lf Li)/2 Fm = (Li + Lf)/2 Sus = 0.67 Sut if peen='unpeened then Sfw=Sfw_unpeen(cycles)*Sut if peen= 'peened then Sfw =Sfw_peened(cycles)*Sut Ns = Sys/(Ks*Sf) ; static safety Sa=8*Fa*D*Kw /(PI()*d^3) ; alternating stress Sm=8*Fm*D*Ks/(PI()*d^3) ; mean stress Sfs = (0.707*Sfw*Sus)/(Sus 0.707*Sfw) PAGE 70 62 62 Appendix A.2. (Continued) Nfs_finite_check = Sfs*(Sus Kw*Si)/(Sfs*(Sm Kw*Si) + Sus*Sa) Sms = Sus*(Sfs^2 Sfs*Sa + Sus*Sm)/(Sfs^2 + Sus^ 2) Sas = Sfs/Sus*Sms + Sfs OZ = sqrt(Sa^2 + (Sm Kw*Si)^2) ZS = sqrt((Sm Sms)^2 + (Sa Sas)^2) Nf2 = (OZ + ZS)/OZ Tms = (Sys + Sm Sa)/2 Tas = (Sys + Sa Sm)/2 ZSy = sqrt((Sm Tms)^2 + (Sa Tas)^2) Nfy = (OZ + ZSy)/OZ If Ns < 1 Then Nfs_f inite = f Else Nfs_finite = MIN(Nf2,Nfy,Nfs_finite_check) If Pre=1 then Finite_life= 1.2 else Finite_life=1.1 if Nfs_finite>Finite_life then ExitVar = 1 else ExitVar = 0 DiaWireTest:= d if ExitVar=1 then exit ODTest=ODTest+ODI Skip: Next y if Ex itVar=1 then exit Next x PAGE 71 63 63 Appendix B Governing Equations Load Stress Wahl correction factor Deflection Sp ring Rate Spring Index PAGE 72 64 64 Appendix C Test Data C. 1 Calibration Data C.1.1 100 lb S ensor PAGE 73 65 65 Appendix C.1.2 1000 lb Sensor PAGE 74 66 66 Appendix C. 2 Sample Spring Plots C.2.1 Spring 225 PAGE 75 67 67 Appendix C.2.2 Spring 238 Appendix C. 3 Sample Excel Data Sheet PAGE 76 68 68 