ࡱ > W { P Y bjbjWW }G = = ] < < < d P 4 L + L ` ` ` ` 2 ; O + + + + + + + $ , . + < [ a [ [ + ` ` [ . ` < ` + [ + J U! d) < + ` D N/5 \+ 8 ET409C Project Outline Randy Renwand 2-14-01 History: As the Process Engineer and Product Test Manager at a furniture manufacturer I am not at liberty to discuss every aspect of this issue however a general description is: We have been alerted to the intermittent breakage of torsion bars used to assist unit transformation. This report will include data and observations taken over the past several years leading up to some form of resolve by May 2001. I chose this topic because throughout my involvement with it, I have drawn upon most of the diverse material that has been presented in the Engineering Technology curriculum. I see this project as a unique opportunity to exercise and display skills from almost every major discipline of Engineering with particular emphasis on Manufacturing Engineering. Title: Torsional Fatigue in Alloy Steels Purpose: To increase understanding of torsional fatigue in order to improve the life span of the torsion spring. Objective: Obtain design specification for a lift assist device that will perform, without failure, through the warranty period of the product. Scope: This study will explore all feasible design and material possibilities restrained to the improvement of the lift assist function. Methodology: I. Research of General Metallurgy Overview of common Metallurgy concepts that apply to this case Characteristics of viable alloys Desirable characteristics II. Research of Metal Failure Modes and Causes A. Identify known failure modes Performance of relevant materials under the following conditions Tensile Strength Elongation Fatigue Oxidation Various surface qualities Analyze broken torsion bars from the field Determine cause of failure 1) Independent Lab Testing: Results primarily Hydrogen Embrittlement IV. Test current and proposed samples Axial Torque Machine Design Requirements Required Maximum torque with Safety Factor Adjustable speed Cycle stop on specimen failure 4) Capable of Cyclic Fatigue, Twist and hold and Ultimate yield B. Test current bar to establish baseline / control 1) Description of bar a) Extruded 8740, Heat treated 2) Data / Notes C. Test proposed bars 1) Various bar materials / surface qualities a) Chinese high elastic(through heat treat)8740, 8650(Hex Wrench Material) b) Extruded bars with improved, extruded surface finish Various corrosion prevention methods D. Investigate Torque-to-Yield performance 1) Data Observations exceeding of elastic limit V. Explore alternative designs A. Gas Spring Cyclic Machine Results VI. Conclusion Torsional Fatigue in Alloy Steels By Randy Renwand For Dr. Tom Robinson Engineering Technology 409C May 7th 2001 All components cease to function at some point in time. The two primary causes of this loss of function are wear and fracture. An engineers job is to design products that will perform over an acceptable life span for an acceptable cost. Everyone involved with a product has unique motives and expectations for that product. The customer desires low cost and long life span. The manufacturer has individual departments that accept compromise so that the company as a whole may prosper from the production of this product. The Accounting Department may prefer a higher cost and shorter life span. The Production Department wants a part that is easy to manufacture given that departments capabilities. The Engineering Department also has individual functions that have unique challenges. The Design Engineer is responsible for turning the concept of a product into a functional object. During the required research and development, prototypes are fabricated and tested until the design is proven. Next the Industrial Engineer is called to implement this new product into the production floor operations in the most efficient manner possible. Finally, the company is ready to sell the product for profit so that the process may begin again with the next concept. This study will provide a detailed look at the operations that take place in an Engineering Department tasked with solving a component failure problem. In this case, the design of the final part is constrained to retrofit into thousands of existing products, when service becomes necessary, therefore radical changes to the original structure are unlikely solutions. The part in question is a torsion spring, or bar, (See Figure 1) that is used to assist the conversion of a table into a mobile A frame. One end of the bar slides into a 1 long hex receiver that is fixed to the table top. Then a similar housing is installed on the other end of the bar and fixed to the table leg. When the table is in use (flat), the bar is under strain, the restoring force in the torsion bar(s) helps the user to lift the table into the mobile configuration. The specified life span of the bar is congruent with the five year warranty of the table. Figure 1 Torsion Bar Reference Drawings, Machined (Top) and Pre-Torqued (Bottom) Overview of Metallurgical Failure Analysis The solution process begins by developing a complete knowledge of why metal parts, such as these, fail. The torsion bar must meet two criteria throughout its time in service. The first, and foremost, is to resist fracture. The second is to possess an initial, acceptable spring constant value and retain it for as long as possible. The breakage problem is more important because a broken torsion bar is much more obvious than a worn out bar. Over time, the table may slowly become harder to lift, which is difficult to detect, but a broken bar becomes obvious right away. Properties that contribute to success for one of these criteria tend to inhibit performance for the other. A metals resistance to fracture is increased with an increase in a property termed ductility. Ductility is defined as the ability of the material to be deformed without fracture. Sufficient spring constant is achieved through the proper level of yield strength. Yield strength is a materials resistance to deform under load. Therefore, Ductility and Yield Strength oppose each other so a compromise must be found in order for the torsion bar to perform as required. When the benefits of both of these properties are considered together a level of toughness is achieved. A material has a high toughness value when both its strength and ductility are above average. Metallic grain size is an important factor that controls the mechanical performance of all metals. Smaller grains minimally increase fatigue life and can affect crack initiation and propagation independently, depending upon the material. This minimal increase seems to be the best solution because most other changes negatively affect fatigue life. In addition, if tensile strength changes with grain size, then fatigue properties change in the same direction Smaller grain = higher fatigue strength and lower creep strength (defined later). Heat Treatment is used to change grain size and other characteristics of a material. For fatigue tests in bending, research shows that the heat treatment given to a steel does have a slight effect on the Fatigue Ratio. The Fatigue Limit was from 5 9% greater when the steel had a tempered martensite structure (rapid quench from critical temperature then tempering) as compared with steels having pearlitic and ferritic structures (slow quench then temper). (Metal Fatigue in Engineering 1980) Fatigue Testing Fatigue performance is particularly difficult to evaluate. A precise association between theory and practice has generally eluded a solution for fatigue, like other properties can only be assessed by the practical test. (Designing Against Fatigue, 1962). Many products can, or must, be designed properly without testing them, especially if testing of these products is difficult or impossible. The designs for which any type of testing is optional, are usually comprised of materials and techniques for which there is a wealth of knowledge and experience regarding their performance. Some examples of these are: non-stressed parts, wooden parts and decorative parts. Actual testing of other man-made objects, such as buildings and bridges, is simply not possible. The torsion bars under analysis here are stressed parts, which do not have a long history of ideal service. To add to the complexity, fatigue strength values for materials are uncommon relative to the popular criteria such as tensile strength or modulus of elasticity. Since the torsion bar design is not satisfactorily proven and information relevant to its usage is not readily available, a testing program had to be created specifically for this case. But first, the available fatigue information can be studied to reveal any useful links to the issue at hand. A traditional material fatigue test apparatus is depicted in Figure 2. EMBED Paint.Picture Figure 2 Material Fatigue Test Apparatus Results from this type of test would allow a comparison of different materials or samples with different dimensions but the link between a solution to the torsion bar breakage problem and the data generated would be difficult to establish. Typical tensile fatigue failures are shown in Figure 3. EMBED Word.Picture.8 Figure 3 Tensile Fatigue Failure Types Stress values at one point on the sample, relative to time, from this test are depicted in Figure 4. Figure 4 Standard Fatigue Machine Stress Output Nevertheless, the fundamental relationships exhibited here are pertinent to the issue. The torsion bars referred to are the current production parts unless otherwise specified. The affect of alternating stress on a material with a given tensile strength is given by: (a / (t = (a / (t [1 ((m / (t)^c], where (a = Alternating Stress at (m, (t = Tensile Strength, (a = Alternating Stress at (m = 0, (m = Mean Stress and c = Constant for each material. This calculation is only accurate at the endurance limit, not sooner in the life span. With it, the Maximum Alternating Stress (torque) allowed for the torsion bar is 78.206 ft-lb. It was proven in torsion bar testing that the maximum torque initially developed was 90.5 ft-lb. So the capacity of the current design (material and/or dimensions) is exceeded by 15.7%. This excessive use is also evident from the reaction of a non-pretorqued bar to the pre-torqing operation*. These bars generally showed a 15( permanent set after one 90( twist, which is a 16.7% change. Pre-torquing overloads the material to induce favorable self stress, which increases the yield strength but has little affect on the total stress range. Another valuable figure is Maximum Torsional Stress (() which is equal to: (Ty) / j, where T = Torque, y = distance from neutral axis to where the stress value is desired and j = Polar Moment of Inertia. This works out to 534,376.931 psi, which is double the tensile strength of this material. This shows that the bar will not fracture when one large force is applied (See Figure 5), however it will yield then eventually break. This conclusion was also reached in Figure 5 Torque to Yield Sample After Test testing where a peak torque range from 170-250 ft-lb, depending upon the material, was recorded but the bar did not break until extreme deformation had taken place See Figure 6. Figure 6 Static Torque to Yield Example The term Fatigue Strength is given to the nominal alternating stress (load) that will just produce failure for a specified mean stress (load) and cycle count. The torsion bars under examination are subjected to a constant nominal alternating stress (100lb-ft) and mean stress (50lb-ft) so the cycle count becomes the variable. However, a speculated Fatigue Strength works out to 100lb-ft for 13,000 cycles. The Fatigue Ratio = Fatigue Strength / Ultimate Tensile Strength = .000377. This value is proportional to the cycle count because the Fatigue Strength is constant. Therefore, a nominal, acceptable cycle count, which is higher than the value of the current torsion bar must be determined. Cycle Ratio is the name given to the relationship between the number of cycles applied and the number expected to cause failure. Torsion Bar Design Criteria Cycles to Fracture As mentioned before, the life span to fracture of all torsion bars must exceed the five year warranty on the table. There are four standard Design criteria that an engineer may consider for a fatigue resistant part: Infinite Life Design Design stresses safely below the fatigue limit (example; engine valve springs). We conclude that an intrinsic fatigue strength limit exists in all materials below which fatigue can never occur (Achievement 1970). Plastic strain must occur in order for fatigue to begin (In Germanium, a material in which dislocations will not move, fatigue fracture is non-existent) Safe Life Design The component should last long enough relative to its time in service (example; transmission reverse gear). Fail Safe Design Cracks cannot propagate to cause failure and can be repaired when expected (example; aircraft engine parts). Damage Tolerant Tests whether cracks can or will grow large enough to cause harm before being detected. This needs a material with slow crack growth and high fracture toughness. Figure 7 represents a chart that was made possible through experimentation and can be used to estimate the necessary tensile strength of a material for use as a spring. Figure 7 SAE Fatigue Stress Factor Diagram for Helical Compression Springs Compression springs undergo twisting deformation similar to torsion springs. From the graph, the 8740 currently in use has a tensile strength of 265,000 psi, when this is multiplied by .6, a value of 159,000 psi, for fatigue, is reached. The difference between designing for fatigue resistance and designing for a few loadings are greater attention to details of shape and treatments, and the need to decide on a required lifetime of the part. (Designing Against Fatigue 1962). For these torsion bars, once the lifetime is established, the treatments (and material) can be established. For the shape would be the most difficult parameter to change due to retrofit limitations. The designer should prevent serious consequences of failure by making the part accessible for inspection and replacement, by Fail Safe design or by using large safety factors (Designing Against Fatigue 1962). Access to the torsion bars is by removal of two housing bolts, Fail Safe design is not used but large safety factors are (see below). In order to determine the maximum number of cycles that a bar could go through in five years, the following speculation took place. The table could possibly be cycled three times per day, 365 days per year for five years. This works out to 5,475 cycles, which represents an unlikely maximum with considerable safety factors (approximately 2.0) built in. However, an additional 525 cycles were added for an increase of 1.1 to the overall cycle count safety factor. A value of 6,000 cycles was proposed as the minimum, acceptable cycles to failure. Yield Strength and Hardness The yield strength value is not under direct evaluation because the current torsion bars give acceptable performance. Therefore, the improved bars should have a yield strength that is approximately equal to the current bars. If this value is considerably lower than the current bar then the manufacturer will not be able to install enough torsion bars in a table for the table to function the same. If this value is too high then one torsion bar may be too strong for the smaller tables to function properly. Test data shows, in Figure 8, that the hardness of the material is proportional to the yield strength (spring constant). Figure 8 Yield Strength Relative to Hardness Current bars are at Rockwell C 49-51 which equates to 230 lb-ft yield strength so proposed bars should not deviate from this torque value by more than 10%. This relationship is echoed when comparing the Fatigue Strength to the Hardness in Figure 9. Figure 9 Fatigue Strength Relative to Hardness Unfortunately, Fatigue Strength and Cycles to Fracture are not directly related. The final relationship involving initial material hardness is that with Tensile Strength, depicted in Figure 10. Figure 10 Ultimate Tensile Strength Relative to Hardness Tensile Strength (not Hardness) is linked to Cycles to Fracture, so an increase in Tensile values usually help with Fatigue Life. Figure 11 brings together Fatigue Strength, Ductility and Cyclic Strain Hardening* It shows that Strength and Ductility are directly related and inversely proportionate. The factor that controls the rate of crack propagation is the strain hardening behavior of the material at the crack tip. In the case of a torsion bar, Fatigue Strength is required over Ductility but, as mentioned before, a decrease in Ductility leads to an increase in the likelihood of fracture. Figure 11 Fatigue Strength Relative to Ductility Innovative Procedures and Equipment All of these differences between theory and practice make the need for a specialized approach obvious. For example, a simulation of the effects of the mobile table on a torsion bar would yield results that could be directly applied to corrective action. Proprietary testing apparatus was designed and constructed to fill this need. Figure 12 is presented here for comparison of the new machine to the standard fatigue test apparatus and Figure 13 is an actual photograph. EMBED Paint.Picture Figure 12 Concept of Cyclic Torsion Machine Figure 13 Cyclic Torsion Machine The Cyclic Torsion Machine (CTM) is capable of mimicking the effect of the mobile table on the torsion bar. Figure 14 allows a comparison of the stress exhibited on the CTM for comparison to the stress from a traditional Material Fatigue Test (Figure 3). Figure 14 CTM Stress Output CTM Design Criteria Twist the bar from 0( to 110( repeatedly Automatic stop upon bar breakage Machine should twist to the same 110( and 0( point regardless of bar influence Speed and Time of rotation should be controllable and variable Machine should be capable of holding the bar at any desired angle for any length of time Safe unattended operation Cost that justifies testing in house rather than contracting an independent laboratory Available Power 110, 240 & 480V Electrical power 100psi Pneumatic power Continuous supply of this power is not reliable so the machine should be capable of handling random power downs and ups. The 0( to 110( twist is 20( more than the bar goes through in actual use. This parameter is designed to accelerate testing and will provide a safety factor of 1.2 to the proposed solution. The machine is capable of cycling from any low point to any high point but the standard test goes from 0( to 110(. The middle and right green screens (in the picture) display and control low and high degree limits respectively. To expedite the testing program, the machine runs 24 hours a day, 7 days a week and even during working hours (24% of a week) it is not practical for an employee to monitor the condition of the torsion bar during testing. Therefore, the CTM is equipped with a limit switch mounted at the fixed (left) end of the bar. This fixed end will not allow the bar to rotate in torsion but it is hinged to allow the bar to fall slightly when it breaks, which trips the switch that shuts off motor power so the Cycle count at Fracture may be recorded. These torsion bars are capable of considerable amounts of resistance to deflection, which acts to overpower the theoretical, no-load movement of the CTM. To compensate for this, it is equipped with a limit switch for the low degree limit that resets the degree counters at the low point (0( standard) when that physical position is reached, to account for encoder step loss or gain. The high degree limit (See Figure 15) is controlled by the right green screen, which simply gives an intermittent output to a magnetic latching relay, when the preset value is reached. Figure 15 Torsion Bar held at 110( showing low degree switch (upper right) The latching relay holds one circuit closed until it is intermittently energized to open it (at which time it closes and holds the opposite circuit). Both low and high counters sense signals from an encoder mounted to the moving coupling. The encoder gives 360 signals per revolution = 1 degree per signal, which means that the value on the counter screen is the angle of the moving coupling. Forward and reverse speed, acceleration and torque are independently controllable via the motor controllers (clear panel below motor) and dials (below degree counters). Timer based relay circuitry ensures that only one motor controller has power at a time which affords the necessary control over time for the test. Essentially, the Latching Relay controls the direction that the motor turns (by reversing the motor leads) and the Timer controls when each of the two motor controllers is on. A three position switch at the bottom left controls the 15 ft-lb clutch brake, mounted to the rear of the motor (brown housing to the left of the black motor, upper right). The switch allows; 1) Normal cyclic operation 2) Angle hold 3) Manual angle set. Angle hold is accomplished by removing electrical power from the motor (which stops it) and the brake (which engages it). Manual angle set removes power from the motor but keeps the brake disengaged so that the coupling orientation is free to move for adjustment etc. All sensors remain active in each position. Unattended stop at fracture is achieved with the limit switch mentioned above but safe operation, attended or not is handled by the following. 1) A 2 amp circuit breaker located near the main on/off switch in the upper left corner 2) A wall thickness polycarbonate tube to shield torsion bar explosions (bottom) 3) A 1/8 thick ABS guard for the chain drive (black, right side). After determining the components that were required to meet the specifications, an estimated cost of $2000.00 was proposed. This is well within the means of the department for this purpose and is a considerable savings given the length and complexity of this project. 110V AC power is the most readily available electricity and once it was determined to be sufficient, it was chosen as the primary power. 240 and 480V AC power were not necessary and would have decreased electrical component choices. Pneumatic power was quickly found to provide insufficient torque unless excessive gear reduction was used which would put speed into the prohibitive range. The most common form of torque multiplication is gear reduction but if the increased torque is doing the same amount of work then it will need more time to do it, which means decreased speed. Random electrical power supply is safely managed by the brake (which engages by spring pressure when power is removed from its solenoid) and by the low degree limit switch that resets the sensors to 0( as soon as the moving coupling returns that position. With these measures in place, the CTM can stop and start safely and automatically in the event of intentional (Facilities Management) or unintentional power downs and ups. Alternating Current motors are more efficient and operate better at high speeds but they cannot match the torque available from Direct Current motors. A high speed, high gear reduction AC motor scenario was compared to a low speed, low gear reduction DC motor setup to find the best of the two. A 90V DC horsepower gearmotor was chosen with a 60:1 reduction box then a 9:1 chain drive pinion and sprocket was attached to the output shaft to yield 265 ft-lb of torque and 7 revolutions per minute (at continuous operation levels). It is bolted to a welded frame. The extra torque provided by the DC motor minimizes the need for numerous torque or speed multiplying mechanisms that could fail or wear. The maximum torque to yield measured from a torsion bar was just below 240 ft-lb so the machine is capable of yielding a bar with a safety factor of 1.1. Keep in mind that yielding a bar was not in the design requirements, nevertheless the CTM shows no difficulty in yielding numerous bars so far. Appendix D contains additional CTM details. All components were sourced locally and are available within 2 days. The Testing Effort Once the CTM was operational the business of torsion bar testing could begin. Appendix A contains the concentrated data output from over one year of various torsion bar analyses though this effort spans several years in all. A summary of Torsional Fatigue results, with emphasis on comparison to known phenomena, follows. Initial Surface Quality Any surface irregularity greater than the materials natural surface is controllable. The characteristics of the surface play a vital role in determining the fatigue strength, since failures usually start from the surface rather than from within (Metal Fatigue in Engineering 1980). Crack initiation occurs at regions where strain is concentrated. Early in the process, the concern of surface irregularities raised the most concern among the experts. Whether inclusions were caused by extruding the hex shape, heat treating to arrive at required properties, or other factors, they present stress risers that concentrate forces which aim to deform the bar more at the locations where they exist. Torsion bars are usually shot-peened and preset (Fatigue Fracture Mechanics 1997). Shot-peening reduces stress risers and induces favorable stress in the surface, preset also induces favorable stress. Bar 3 was given a stress riser by hand filing a (.020 deep groove (far deeper than any true inclusion) at 5 from the fixed end, it fractured at 12,458 cycles 4 away from the filed groove. Before it fractured, Bar 3 also under went a series of torque-and-hold cycles where it was twisted to 110( and held there for various time spans between 1 and 21 hours to investigate if creep fracture was responsible. Creep Fracture occurs when a crack propagates due to a sustained load that deforms the sample. The bar did not fracture due to creep so this cause was not the focus of further investigation at this point. The only other investigation regarding induced surface damage was with Bar 4 which had a notch milled in increments of .010. After each depth change the bar was cycled 50 times. This continued until a depth of .120 was reached at which point the bar broke during the first cycle. This test did not show anything other than the cross-sectional area at which the bar fractures and initial surface quality concerns were abandoned. Corrosion A sub-category of surface quality is the effects of corrosion on metal fatigue. This is a bigger enemy to the performance of the torsion bars because corrosion has the capability of growing inclusions and chemically affecting the metal where initial surface irregularities might be infinitely tolerable and do not have chemical affects. Repeated stressing accelerates the corrosive action and corrosion accelerates mechanical fatigue mechanisms. Bar 7 was placed in a moist, salt environment for 7 days then removed at which point it exhibited extreme corrosion. It was cycled until it failed at 18,260 cycles, the 84th percentile (See Figure 16). Figure 16 Bar 7 Extreme Corrosion Specifically corroded bars were not tested again although some future samples did show advanced corrosion for one reason or another. Figure 17 charts the affect of corrosion on general fatigue performance. Figure 17 Affect of Corrosion on Fatigue Properties Although testing did not show that corrosion was the primary factor in premature bar failure, the designer cannot allow corrosion of the surface to take place for obvious reasons. Suggestions for protection against corrosion are: A corrosion resistant material, such as stainless steel Surface treatments include shot peening, cold working and nitriding Anodic coatings work well, zinc or cadmium Cathodic coatings do not work, chromium, nickel, etc Electrolytic Plating causes surface cracks and possible Hydrogen Embrittlement* (See Figure 18) Figure 18 Hydrogen Embrittlement Example and Evidence on Torsion Bar Surface Coatings only protect if they are continuous (paint, grease, etc.) The concept of Fretting deserves mention here. Fretting Corrosion occurs when two surfaces interact (rub) so as to accelerate corrosion of one or both (See Figures 19 &20). The fretting corrosion value for steel and zinc is low at 1.25 where this value for two chromemoly parts is 4.14. EMBED Paint.Picture Figure 19 Figure 20 Fretting Corrosion Example and Evidence on Torsion Bar Fretting Fatigue takes place when a part subjected to cyclical loading suffers damage by reason of contact with a neighboring part. This acts as an initiating source of fatigue cracks and could be happening at the ends of the torsion bars where they meet the broached couplings. However, the fracture locations seem to be random and not enough of them occur at the ends to find this as the primary cause of premature failure. A synthetic, water repelling grease was chosen as the best corrosion prevention method for its lack of chemical reaction with the metallic surface. Testing of bars 9 through 18 established an average performance of the current production parts for statistical analysis (See Appendix B). Figure 21 displays the common fracture region of these. Further comparison can be drawn from Figure 3 C (to A) and E (to B). Figure 21 Typical torsional fatigue failures, spiral A, cone B Bars 19 through 28 were produced with experimental processes that proved to be unfeasible from a cost or lead time standpoint, though some results were favorable. The Chinese bars 29 through 38 finally showed a considerable improvement and were the best solution up until this point. Of particular interest is bar 38 which sustained torque to a 110( permanent set (the two ends were orientated 110 degrees apart at rest) then was cycled 7,512 times from 110( to 0(. Figure 22 displays fractures from two types of tests. Figure 22 Fracture from Fatigue Fracture from Yield Bars 43 through 57 represent a suggestion from Pacific Metallurgical Company to try AISI 8650 steel (See Figure 23). Various heat treatments of this material provide unsatisfactory results with an average of 6,495 cycles. Figure 23 Typical 8650 Fracture Alternate Design Exploration The torsion bar approach to mobile table conversion is not the only feasible method. A gas spring integration was briefly entertained thus the Cycle count to Yield had to be determined for that design as well. Figure 24 shows the machine used to perform the test. Figure 24 Gas Spring Cyclic Test Machine This machine simulates the angle change and compression that the gas spring would go through when installed on a table. The 6 diameter Pneumatic cylinder (on the left) is capable of over 2,500 lbs of force at 100 psi but the only spring tested put out a maximum of around 300 lbs. As the cylinder extends it pushes the linear bearing (bottom) which has one end of the gas spring attached to it, thus compressing the spring. The other end of the gas spring is connected to a fixed, 1,000 lb. Max load cell that monitored the performance of the spring throughout the test. A timer relay and counter controlled and monitored the action of the cylinder. A degradation of the springs performance began after a few hundred cycles and fluid leakage became apparent. By 15,000 cycles the load capacity of the gas spring had decreased by over 30% and it was obvious that it could not match the performance of even the current torsion bar design and components. The gas spring concept was abandoned for these reasons as well as retrofit complications and sales and marketing difficulties with an unusual design. Conclusion All but one proposed solution to the premature torsion bar failure performed worse than the current assembly. The Chinese equivalent 8740 material with heat treatment for increased elasticity is the superior choice in this application. The details of this component are as follows; Material 40 (Carbon) Cr Ni Mo A Steel Chinese equivalent 8740 Steel Chemical Composition Before Heat Treating C .37-.44, Si .2-.4, Mn .5-.8, Cr .6-.9, Ni 1.25-1.75, Mo .15-.25, Cu .3 Mechanical Properties After Heat Treating Yield 220,000 psi, Tensile 240,000 psi, Elongation 13%, Rockwell Hardness(C) 49-51 Heat Treatment Harden Part is to be held in the vertical position in a controlled air, deep well style oven at 830( C for 30 minutes then removed for oil quenching Annealing Hardened part will be loaded into a box type annealing oven at 300( C for 3 hours then removed for air quenching Straighten Annealed bar will be restraightened using a proprietary machine Tension Release Straightened bar is loaded into a baking oven for 5 hours at 200( C Rust Preventative Water resistant synthetic grease is applied to the bars AISI 8740 is a structural quality steel, that is often used in fatigue applications, one example of this is a piston rod bolt in an internal combustion engine that sees millions of extreme tensile cycles without failure. The heat treating process given to the new 8740 bars creates particularly small grain sizes and homogeneous (same material throughout) dispersion. The incredible similarity between the fractures (Figure 22) created by completely different stresses is proof that this material is unique. The shiny fracture surface, particularly on the fatigued bar is an indication that fatigue crack propagation was not initiated on the surface and did not cause the failure. This type of propagation occurs when grain boundaries are not capable of stopping it so that it causes total dislocation of the grains (notice the grainy appearance of figures 21 and 23). If a material fails in this fashion then less force is required to do so. The new 8740 however, exhibits pure shear failure without grain boundary propagation. Slip and Cleavage are two terms given to fracturing that shears each individual grain apart, leaving a mirror like finish. Clearly this is the optimum failure because it shows that the material is performing at the limit of its chemical composition. In less than ideal materials, it is much easier to displace one grain from another than it is to bisect individual grains. A component made of a single grain would be ideal for exhibiting the theoretical properties of a material but this is impractical from a manufacturing standpoint. Since grain boundaries due to processing must exist, they can be used to halt crack propagation and thats exactly what they are doing in the new 8740. Finally, a summarized comparison of the previous 8740 and the new 8740 is helpful in assessing the value of the outcome. The most important factor, fatigue life, of the new bars is up by 88% placing the minimum cycles to fail well beyond the five year warranty. Yield strength of the new bars is within 10% of the previous bar so no adjustment to table design is required. Rockwell hardness, and bar dimensions are the same. A corrosion prevention operation has been added to the process but this will be performed at the heat treating facility for a minimal up charge compared to performing it in house. And, the cost of the new bars is lower than domestically sourced goods. All of these advantages add up to a superior torsion bar, that will require a minimum of effort to phase into production. In fact, it will be transparent to the Mobile Table Assembly Cell and to the customer. Appendix A In House Test Dataoil 'coating' was removed from all bars before test unless noted.permanent set values increase with time held at 110, very high values could be due to machine over travel.bar 118,000 cycles, no breakage, permanent set at 50 degreesbar26,000 cycles, no breakage, permanent set at 50 degreesbar 3filed notch approx. .020" deep to simulate surface pitting/corrosion, 6,000 cycles no break, perm set CONTINUED BELOWbar 4ball milled 3/16"notch in .010" & 50 cycle increments until breakage, break @ 1 cycle @ .120" deep bar 5aunheat-treated bar, corrosion evident, broke @ 5,011, 10" from fixed end, one end frayed other pointed, shiny surface from scrubbing, black pits observed inside, cracks on middle of all flats, bowing is evident, fracture is magnetic from rubbingbar 5bunheat-treated bar, static torque to yield, 94 ft-lb max., no breakagebar 6afrom table w/ broken bars, 6000 cycles - no apparent damage, 110 hold 9:00am - 4:30pm return to 23, 5:00pm Friday - 9:00am Monday return 17, no apparent damagebar 6bfrom table w/ broken bars, 6000 cycles - no apparent damage, 110 hold 9:00am - 4:30pm return to 0, noapparent damagebar 7extreme corrosion, 6000 cycles - no apparent damage. CONTINUED BELOWbar 8White lacquer no primer, paint cracked/flaked approximately 1" from both ends, moving end worse, no paint damage at other locations, 270 twist - 1 cyclebar 9normal - pretorqued bar, 6000 cycles, no apparent damage, return to 14bar 1026,202 cycles, spiral fracture into 2 pieces, magnetic, at 2" from moving end, at 110, bar 1114410 cycles, longitudinal cracks observed with penetrant inspection, approx. 12" form moving end, bar removed and sent for further analysisbar 1211812 cycles, same mode as #10, at edge of moving broached piecebar 1313141 cycles, fracture is spiral at ends - planar for 1" in the middle, at approx. 6" from fixed endbar 1420819 cycles, same mode as #13, at edge of stationary broached piecebar 1518100 cycles, same mode as #10, at edge of fixed broached piecebar 1617219 cycles, same mode as #13, at 2" from fixed endbar 3continue to failure, 12,458 total, same mode as #10, at fixed broached piece, 4" from filed groovebar 7continue to failure, 18,260 total, same mode as #10, at fixed broach piece, at 81 degrees bar 176,205 cycles, same as #15bar 1811,248 cycles, same as #15bar 1911,295 cycles, 'new' vendor, same failure mode as #13, in middle of barbar 20static torque to yield, 'new' vendor, 187.4 lb-ftbar 21static torque to yield, 'new' vendor, 175.2 lb-ftbar 22static torque to yield, 'new' vendor, 184.4 lb-ftbar 2310,015 cycles, case hardenedbar 24static torque, case hardened, 192.8 lb ftbar 2529,215 cycles, '3rd drawback' to improve surface, initial 23,357 cycles, continue after a few months, more corrosion, additional 5,858 cycles, 1/2" brittle fracture in middlebar 26static torque, 3rd drawback, 216.8 lb ftbar 273,250 cycles, 8650 alloy, 30% of normal length (8.375"), 30% twist = 30 degrees, spiral break at moving endbar 288650 alloy, 29,000 cycles, no breakage, same test as #27, 30 degree permanent setbar 2931.603 cycles, 8740 Equivalent from China, uncoated, initial 25,000 cycles, continued after 1month to 6,603, 1/2" brittle/smooth fracture, 2" from moving endbar 30static torque, 8740 Equiv. uncoated, 199.6 lb-ftbar 3119,765 cycles, 8740 Equiv. Black Oxide, clean break w/ point at moving endbar 32continuous twist to break, 8740 Equiv. BO, clean, shiny break, flat and straight, 6" from moving endbar 33continuous twist to yield, 8740 EquivBObar 34cont. twist to break, 8740 EquivBO, 2,077 degrees, clean, at 8" from moving endbar 35cont. twist to break, 8740 EquivBO, at 6" from moving endbar 36cont, twist to break, 8740 Equivraw, 780 degrees, at 10" from fixed endbar 37cont. twist to break, 8740 Equivraw, 2,108 degrees, at 12" from fixedbar 387,512 cycles, 8740 Equiv raw, induce 110 degree permanent set then cycle from 110 to 0 degrees,1" failure at 8" from moving endbar 399,663 cycles, standard bar, Tectyl corrosion prevent coating, 1" brittle fracture, 10" from moving endbar 4015,418 cycles, standard Tectyl, 1" brittle at 8" from movingbar 418,421 cycles, standard Tectyl, extreme corrosion, 1/4" brittle fracture, 1" from moving end, same as #7bar 42stopped at 4.759 cycles, 8650 un-heat-treated, un-coated, noticeable twist/crack region at 10" from moving endbar 434,604 cycles, 8650 heat-treated 50 HRC, un- coated, pointed fracture at 1" from moving endbar 44static torque, 8650, 245 lb-ftbar 452,299 cycles, 8650, 1/2" spiral fracture at middlebar 467,887 cycles, 8650, 1" spiral at 10" from moving endbar 47before 13,271 (machine did not stop), 1/2" point at moving endbar 487,237 cycles, 8650, 1" spiral at 8" from moving endbar 4910,199 cycles, 8650, 1" spiral at 2" from moving endbar 507,285 cycles, 8650, 1" spiral at 8" from fixed endbar 517,337 cycles, 8650 annealed heat treat, 41 HRC, point-cone fracture, 10" from moving endbar 52static torque, 8650 annealed heat treat, 171.6 lb-ftbar 53static torque, 8650, 46 HRC, 201.6 lb-ft bar 5412,766 cycles, 8650 46 HRC, yielded long before test stoppedbar 556,850 cycles, 8650 46 HRC, 2" brittle fracture at 2" from fixed endbar 56before 10,385 cycles, 8650 46 HRC, 2" brittle fracture, failed region observed at 4" from moving at 4,567 cyclesbar 578,587 cycles, 8650 46 HRC, failed region at 10" form fixed end observed at 7,737Appendix B: Analysis of Bars that are comparable due to same testing scenarioIn order testedBar #Cycles to FailDescription312458.020" file groove5a5011unheat-treated6a8479from field table w/ broken bars 6b6831from field table w/ broken bars 718260Extreme Corrosion1026202Standard (8740) bar1114410Standard 1211812Standard 1313141Standard 1420819Standard 1518100Standard 1617219Standard 176205Standard 1811248Standard 1911295 'new' vendor 87402310015case hardened 874025292153rd draw back 874029316038740 Equivalent China, heattreat for max elasticity31197658740 Equivalent China, heattreat for max elasticity399663Standard w/ Tectyl corrosion inhibitor4015418Standard Tectyl418421Standard Tectyl, Extreme Corrosion4247598650 unheat-treated4346048650 heat-treated4522998650467887865048723786504910199865050728586505173378650 annealed HT5568508650 46HRCMean of All12389Median of All16951Mean of Std.13668Median of Std.15606.5Mean of China25684Median of China25684Mean of 86506495Median of 86506249Standard Deviation of All7298Ave Mode of All7000Standard Deviation of Std5761Ave Mode of Std.11000Standard Deviation of China8371Ave Mode of China25000Standard Deviation of 86502344Ave Mode of 86507000FrequencyIn order of cycle countCycle GroupFrequency of Standard BarFrequency of 865045 8229910043 8460420142 847593005a S501140217 S62055106b S683162155 8685070448 8723782050 8728591051 87337100146 8788711241 S84211216a S847913139 S9663141231001515149 81019916018 S11248171191129518212 S118121903 S1245820113 S1314121011 S1441022040 S1541823016 S1721924015 S181002507 S1826026131 C197652714 S208192810 S262022925292153029 C3160331 Appendix C Independent Test Laboratory Report SummariesLaboratory Cause of FailureNotesIndustrial Wire (bar vendor)surface pitting due to corrosioncurious about notch or corrodedbar testscompetitior=8640, ours=8740ranges of moly overlap, each could be the otherValley Metal Treating (heat treating facilityZezula & Associates (reportnot hydrogen embrittlement, Reports on filefurnished by Valley MT)excessive decarburizationDurkeefailure due to stress risers Reports on filecaused by surface corrosionpossible hydrogen embrittlementfrom hardening processPacific Metallurgy Companyhydrogen embrittlementSuggests 8650 (hex wrench)materialReports on fileIn house Penetrant Inspecton notes1. Static twisted bars from 2 years ago currently have no rust, very good surface and no cracks2. All cleaned 'dry' current bars have rust and surface pitting3. Most tested and field bars have numerous small lengthwise cracks Conclusion : a) fatigue causes cracks, not yieldingb) current bars have more corrosion and poorer surface quality4. Current bars have gouges for the length of most flats and scale build upThe two primary tests are : 1) Cyclic, 110 degrees, approx 20 sec cycle time2) Static torque to yieldA third is Rockwell HardnessIf bar is cycled then allowed to sit, corrosion can propegate during this time then acceleratedamage in later testing. Example: bars 6a, 6b, 7 etc. Appendix D: Torsion Bar test station BOM with troubleshooting and operation information 2-18-00Operation: Turn on, Set cycle count, Set low & high angle, Reset cycle counter, Set timer Some components were provided by the Tool Crib, all components not listed were fabricated Part NumberDescriptionQuantitySourceMechanical System in order of use6L9279 tooth ANSI. 50 sprocket1Grainger6L074Nickel roller chain10ft"1L21360 tooth ANSI 50 sprocket1"3A1513/4" split taper bushing1"3/4" modified shoulder bolt13/4" pillow block2Electrical System in order of useWire, white=neutral, black=line, yellow=signal, red +4 ftMcMasterCarr1A824On / Off Switch, lighted1Grainger5X852ESocket, Octal, for relay1"5X827ERelay, DPDT, 8pin, octal base 1"Micro switch1XC74Counter, predetermining1"6B0238A circuit breaker1"5X852ESocket, Octal, for relay1"1A366ERelay, time delay, repeat cycle1"4A161Socket, 11 pin, for latch2"3A965Relay, latching2"7397K21Momentary push button switch2McMasterCarr4A024Power relay, 125VDC, 110VAC coil1Grainger4A079Dust cover, power relay1"1XC74Counter, predetermining2"59515K51Encoder, 360 counts / rev., 24V1"4C5494" Axial fan1"2M510Motor controller2"6Z416Gearmotor 90V, 6A, 60rpm, 1/2hp 1"5U28910ft-lb brake, 110V solenoid1"4X851DPDT tab switch1"100 ft/lbs, 110 degrees for test200 ft/lbs, 200 degrees to fail6000 cycles in 24 hrs, 1 or 2 bars / week, 2 seconds tension - 10 seconds relaxPriceNotes$10$35$64$15110V operation$518 AWG$3$4Screw terminals$14Off at cycle limitoff if bar breaks$180Set to stop at pre-d. #$32in AC system$4Screw terminals$60 .1 - 10 seconds, on/off$16$64control from encoder$3to manually change DC polarity$26to pad high in rush$14$360low & high angle settings$255$20$128control from timer$489$237manually switched to hold at any angle$6for brake, run & angle setting$2,044Total Bibliography All Research and Development, including the design and construction of the CTM, for this project was performed by the author, unless otherwise noted. Materials Science for Engineers, Third Edition, James F Shackelford, Macmillan Publishing Company, New York, 1992 Metallurgical Failure Analysis, Charlie R. Brooks, Ashok Choudhury, McGraw Hill Inc., San Francisco, 1993 Machinerys Handbook, 24th Edition, Robert E. Green, Industrial Press, New York, 1992 Jorgensen Steel & Aluminum Stock List and Reference Book, Catalog, Earle M. Jorgensen Co. 1988 Designing Against Fatigue, R.B. Heywood, Chapman & Hall, London, 1962 Metal Fatigue in Engineering, H.O. Fuchs, R.I. Stephens, Wiley InterScience, USA 1980 Fatigue and Fracture Mechanics of High Risk Parts, Bahram Farahmand, Chapman and Hall, 1997 Achievement of High Fatigue Resistance in Metals and Alloys, ASTM Special Publication 467, Baltimore, 1970 Metal Fatigue: Theory and Design: Angel F Madayag, John Wiley & Sons, 1969 Solutions to Hydrogen Attack in Steels, P.F. Timmons, ASM International, 1997 Point Defects in Metals, A.C. Damask, Gordon & Breach, New York, 1963 The Physical Basis of Metal Fatigue, P.E. Forsyth, American Elsevier Publishing, New York, 1969 Fatigue Failure of Metals, S. Kocanda, Warsaw, Poland, 1978 * Pre-Torquing or Pre-Setting is an operation of twisting a new bar once before installation * Hardening as a result of Cold Working during twisting. CW greatly improves tensile strength with a minimal improvement to fatigue strength * As atomic hydrogen accumulates in steels via cathodic absorption the steel becomes embrittled, that is it shows a loss in ductility and toughness. EMBED MSGraph.Chart.8 \s EMBED AutoCAD.Drawing.15 EMBED AutoCAD.Drawing.15 EMBED MSGraph.Chart.8 \s EMBED MSGraph.Chart.8 \s EMBED MSGraph.Chart.8 \s EMBED MSGraph.Chart.8 \s EMBED MSGraph.Chart.8 \s EMBED MSGraph.Chart.8 \s EMBED MSGraph.Chart.8 \s EMBED Excel.Sheet.8 EMBED Excel.Sheet.8 EMBED Excel.Sheet.8 EMBED MSGraph.Chart.8 \s h v v x : G O \ q! w$ )' *' @' A' B' C' ( ( ( ( ( ( ( ( B) C) * * * * * * * * * * * * * * * * * * * * + + + + - - >- ?- Z- j 0J j s j UmH 5 jֶ Uj<~> CJ UVmH j` U j Uj>> CJ UVmH >* >*6B* jP U j UH* E % - . G 4 5 ^ _ - c d " a 0 &