C. J. Winter is a world-class machine product vendor that supplies thread forming equipment to machine industries. A 234SA is a rotary radial threading attachment that is designed to fit into one station of a Hydromat 12-station rotary transfer production machine. A 234 slide arm within a C. J. Winter model 234SA thread rolling attachment has failed in service. Through various engineering diagnostic methods, C. J. Winter has determined that the Von Mises stresses in the area of fracture exceeded the maximum allowable stresses in the slide arm. Our task is to analyze the current design, and propose a redesign that is guaranteed not to fail under normal operating conditions. The redesign will be used in the existing attachment and may be scaled up for larger attachments. In addition, we are required to develop a new test fixture, or modify the existing test fixture to allow for repeatable testing of production slide arms.
Note: There are
two die axle holes; the base consists of the block of material containing the
two threaded holes; the head is the major portion of the part containing
everything surrounding the pocket except for the sliding surfaces.
A single 234 slide arm failed in service at a C. J. Winter customer site. The 234 slide arm is a single part within a 234SA rotary radial threading attachment. The purpose of each of the two slide arms within the tool is to hold one roller thread die. Each die is aligned with the other within the 234SA by an alignment chain to prevent thread mismatch on the part. There is a thick outer ring surrounding the tool. The ring contains two internal ramps oriented axially, and each ramp coincides with one of the two slide arms contained within the assembly. The outer radius of the slide arms is designed to clear the inner diameter of the outer ring. The entire 234SA attachment spins continuously during threading. As the outer ring is advanced axially by an air cylinder, the ramps press on the roller and pin. This forces the two slide arm heads, and therefore the roller dies, together around the unthreaded part stock. The part stock is threaded as it is pinched between the two roller dies. Once the part is threaded, the air pressure is released, each slide arm returns to its starting position, and the threaded part stock moves on to the next station of the Hydromat machine. A video of the 234SA rotary tool in action can be seen at http://www.cjwinter.com/products/rotaryradial.htm .
In an attempt to prevent future failures, C. J. Winter immediately contracted slide arm destructive testing to IMR test labs. Two production arms were sacrificed for mechanical destructive testing along with another slide arm that was completely machined from a block of steel. IMR labs built a makeshift compressive testing fixture to fit an Instron frame with the intent of breaking the slide arms. The two cast arms failed, and the machined arm showed signs of imminent failure beyond the capabilities of the Instron frame. This preliminary data, the three sacrificed arms, as well as a new production arm were presented to us at the introductory meeting. A Finite Element Analysis report of the current design was also presented.
Our preliminary issues concerned material, process, and finishing. Once these decisions were made, a reproduction of the 3-D computer model was created using I-DEAS software at RIT. Finite Element Analysis (FEA) was performed on the original slide arm design. It was necessary to calibrate our FEA results to the C. J. Winter FEA results by applying identical boundary conditions and mesh sizing. Our team was not safe to proceed until our preliminary FEA results matched exactly. Modifications consisted of strategic addition of material to the high-stress areas revealed during FEA analysis. Once complete, we needed to prove that our design is superior to the C, J, Winter original design by performing a secondary FEA analysis on the new model using identical boundary conditions and mesh sizing.
Preliminary Redesign Issues
The current 234 Slide arm is made from A2 tool steel that is cast, machined, and then heat-treated. C. J. Winter found it difficult to machine the roller and sprocket pocket within the head of the slide arm. It is cost prohibitive to machine the part from a single block of stock. Previously, an EDM (Electrical Discharge Machining) was used to melt and plunge material from the pocket using a formed die. The EDM process electrically melts the metal, and a die displaces the molten metal from the roller pocket. The EDM operation was removed from the process since it was costly, and would leave surface stresses in the pocket that were detrimental to performance. The molds were redesigned to allow for casting of the entire piece. Finishing machining operations remain necessary.
At the introductory meeting, C. J. Winter suggested that other materials might be considered as alternatives to A2 tool steel. The casting process limits the number of alloys that can be used. AerMet 100 and AerMet 310 were two suggestions. The AerMet materials exhibit superior qualities such as strength and stiffness. However, they are impossible to use in this application due to cost restrictions. Both materials are titanium alloys, making the base metal extremely expensive. In addition, molten titanium is explosive if oxygen is present, and must be cast in an altered environment adding further cost to the operation.
Machining the sliding surfaces, and drilling and tapping the holes remains necessary as finishing operations. The slide surfaces are machined to an exacting 0.0001 tolerance. These dimensions can only be achieved with a grinding operation. There are six holes that must be honed to exacting tolerances with four separate drilling operations. Two of these holes are on the top surface and in the base of the pocket for the die axle, two holes on the outer radius hold the roller pin, and the remaining two holes are threaded within the head of the slide arm. One of these holes is a threaded blind hole used to house the taper-adjust screw, and the other threaded hole contains a setscrew. Both threaded holes lie in the highest stress areas in the slide arm. We employed design for manufacture and assembly (DFMA) concepts to eliminate one of the screws.
There are a number of outsourced parts contained within each slide arm. There is a roller and pin to transmit the pinch force from the outer ring, as well as the aforementioned die axle, threaded taper adjust screw, and the threaded setscrew. The only part known to fail during testing is the pin. After discussion with C. J. Winter, it became clear that these five parts were suitable for their respective applications.
Finite Element Analysis modeling in I-DEAS requires user specified boundary conditions. The model was constrained to sliding motion along the ground sliding surfaces, and forces applied to the roller pin, thread die axle, and taper adjust screw as specified below. At the suggestion of Dr. Kevin Kochersberger, a single point was selected on the minimally stressed slide surface and used as an anchor. This point was fixed in space before FEA analysis was performed to maintain the stability of our model. Boundary condition forces were placed on the model at the specified points, but there was no guarantee that they would be exactly equal and opposite. There is a small, high stress area on the sliding surface on the bottom of each model due to the anchor point.
In the current design, force transmission to the slide arm is via three separate parts. The pinch force of thread rolling is achieved when the outer ring of the assembly is advanced by air pressure. The roller pin transmits the pinch force to the slide arm. The capability of the air cylinder responsible for advancing the outer ring, and the six degree ramps dictate that the maximum pinch force will be 7,000 pounds. The roller is centrally located on the pin and the 7,000lb force is evenly distributed with 3,500lbs on either roller pin hole on the outer radius of the slide arm. The opposing force is taken up by a spring between the two slide arms until the thread dies contact the unthreaded stock. Once the unthreaded stock is contacted, the spring force becomes insignificant, and the pinch force is opposed by the threading operation and later by the incompressibility of the threaded stock.
The reaction forces are transmitted from the stock through the thread dies and into the thread die axle. On one end, the thread die is adjacent to the axle hole at the free end of the slide arm. Approximately 75%, or 5250lb, of the thread forming force is transmitted to the slide arm at this point. This percentage was determined by statically analyzing the thread die position on the die axle, and with a 7,000lb pinch force. Within the slide arm body, the axle rests on the taper adjust screw, which is contained within the threaded blind hole. The remaining 25%, or 1750lb, of the total force is transmitted from the die axle, to the taper adjust screw, and into the slide arm through the taper adjust screw.
Figure 2. Boundary conditions of original slide arm
Preliminary modeling showed minimal stresses in the sliding surface area. The most intense stresses occurred in the slide arm pocket walls. For this reason, the part was partitioned and meshed using different element sizes. Initially, a tetrahedron shaped mesh with sidewall nodes was used. The memory allocation limit of the RIT student lab computers was quickly determined. In order to use fine elements in the high stress areas of the slide arm, a pyramid-shaped element was chosen. A coarse mesh using 0.5inch elements was used in the low stress areas of the sliding surfaces increasing the available memory. A much finer mesh of 0.055inch elements was used to model the higher stressed end of the part. This mesh size and type allowed for a representative analysis without exceeding the memory allocation limits of the lab computers. The FEA analysis of the original slide arm was compared to the supplied FEA analysis printouts from C. J. Winter. Only after our model matched the C. J. Winter model, did we allow ourselves to proceed with a redesign.
Figure 3. Meshed original slide arm
Preliminary Finite Element Analysis
Finite Element Analysis was our roadmap to redesigning the slide arm. High stress areas of tension and compression were revealed and targeted as areas requiring additional material to increase the slide arm strength. Immediately the lack of radii and threaded holes were noted as stress concentration areas. The highest stresses occurred in compression surrounding the roller pin. The next highest stresses occurred in tension in the inner wall. Note that the original slide arm experiences 2.34x105psi maximum stress under the operating load of 7,000lbs.
Figure 4. Finite element analysis of original slide arm
The threaded blind hole was initially blamed for the part failure. Its location adjacent to the problematic inner wall area compounded by a stress concentration told us that high stresses are likely to occur. The sacrificed arms were visually inspected, and revealed river marks. These river marks are actual material grains. Fractured metals reveal a flowing grain pattern that tends to originate at the original point of fracture. The river marks visually led back to the threaded blind hole in both sacrificed cast slide arms. The re-creation of the analysis opened up possibilities for exploration of the model in much greater detail and we learned to perform a cross-sectional analysis in I-DEAS. The threads within the blind hole cannot be cut entirely to the closed end of the blind hole due to the nature of the tapping process, and this end of the hole is simply required for tap clearance. The highest stresses seen in our Finite Element Analysis (FEA) occur at the closed end. Since there are no threads at this end, we chose to model the part without threads. C. J. Winter engineering concurred that the small elements required to perform FEA analysis on threads used up an amount of memory within the computer that exceeds the capabilities of RIT lab computers as well as C. J. Winter computers. While a suitable computer could probably be made available to us, it was deemed unnecessary due to the conditions previously mentioned.
Figure 5. Finite element analysis of original slide arm thread plane
C. J. Winter FEA analysis of the original slide arm revealed the high stress areas of the part to us. Our model was tweaked to match, so there was nothing revealed by the outer view of the model. However, the re-creation of the analysis opened up possibilities for exploration of the model in much greater detail. I-DEAS software allows for cross sectional analysis across a specified plane. One cross-sectional plane was placed along the threaded blind hole, which lies directly below the highest stressed pocket sidewall. Previously, our team had focused on the threaded blind hole as the original fracture point. Cross-sectional analysis created the theory that the threaded blind hole simply offered a convenient fracture line after primary fracture had already occurred at the pocket sidewall. Further visual inspection of the sacrificed slide arms revealed that this theory was correct. The highest tensile stresses occurred at the inner sidewall. Compressive stresses were noted, but were initially ignored since they were not likely to cause failure.
The static factor of safety was a quick and easy way to verify our improvement in slide arm design. The factor of safety was calculated using the formula:
Static factor of safety = Yield Srength/Maximum Von Mises Stress
C. J. Winter original slide arm factor of safety = 1.8x105/2.11x105 = 0.85
This could be an indication of why the original slide arm failed.
Primary Redesign factor of safety at 7,000lb load = 1.8x105/1.16x105 = 1.55
Exceeds the static requirement.
Primary Redesign factor of safety at 7,000lb load = 1.8x105/1.72x105 = 1.05
Has a factor of safety greater than one at 1.5 times the actual load during service.
Beam theory was applied in the
pocket area to assist us while reducing the stresses within the arm during
thread forming. We assumed that the maximum stress occurs at the thinnest cross
section within the head of the slide arm. The cross section has an area of 1.06
in2. All equations and factors determined from tables contained in
“Mechanical Engineering Design,” Fifth Edition; Shigley
sMAX = 129 ksi (from FEA analysis)
sA = sMAX / 2 = 129 ksi / 2 = 64.5 ksi
sA = sM
SUT = 240 ksi
SE’ = 120 ksi
SE = SE’ KAKBKCKDKEKF = 68.43 ksi
KA = a*SUTb = 0.6318
Where a = 2.70, b = -0.265
KB = 0.95 (from tables)
KC = 0.95 (from tables)
KD = KE = KF = 1
Given: 1 = (sA/ sF) + (sM/ SUT)
SF = 88.2 ksi = sA NEW
a = (0.9*SUT)2/ SE = 681.8 ksi
b = -1/3*log(0.9*SUT/SE) = -0.166
Substituting sF for sA in the equation gives the estimated part life, N:
N = (sA/a)1/b = 224,068 cycles
This indicates that the design does not have infinite life. However, the redesign met the goal for the factor of safety for a static loading.
At this point it is important to mention that C. J. Winter mistakenly gave us drawings of their own redesign instead of the drawings of the original failed part. The drawings were used to create a third model file in I-DEAS. Comparison of the donated slide arm, and the image created in I-DEAS revealed that there was a discrepancy. C. J. Winter confirmed that there were two sets of drawings. We used the C. J. Winter redesign to our advantage in that we did not have to ask as many questions, such as those concerning allowable radii dimensions.
Most corners on the existing part are near sharp, and only have radii due to the limitations of the casting process. Near sharp corners cause stress concentrations, especially under near-failure conditions. There are two areas on the part that scream for the addition of a radius. The first opportunity occurs where the sliding surfaces meet the head of the part next to the die axle set screw. This area was given a radius of 0.08inch. The other obvious radius requirement occurs where the sprocket pocket meets the base of the slide arm head. The pocket in the slide arm serves two functions. First, at the upper end, there is clearance for a thread die. Since there are two thread dies used in the thread rolling process, and they must be aligned to prevent mismatch, a sprocket for the alignment chain is contained in the lower half of the pocket. Since this is the critical failure area, material must be added, and an adequate clearance must be provided for the alignment chain. This area was given a radius of 0.125inch. This radius allows for a significant decrease in stress concentrations, while maintaining an adequate clearance for the alignment chain.
The paragraphs contained in the “Preliminary Redesign Issues” section of this report were addressed. The casting process was retained due its ability to create the necessary part geometry. A2 tool steel is the material of choice for its ability to be cast, machined, and heat treated. None of the sourced parts were eliminated since they all perform well in their respective applications. After speaking with C. J. Winter engineering about the elimination of one of the two threaded elements, we agreed to retain both threaded holes due to separate functionalities. We were more comfortable with this after learning that the threaded blind hole was not the origin of fracture. Since the material, production, and finishing process cannot be altered, strategic material addition is the key to strengthening the slide arm.
The slide arm is a single part contained within a threading tool, so we needed to consider part-to-part clearances within the tool. The two slide arms within the tool slide together along the sliding surfaces. Luckily, the bases do not meet in the center, so there is a space left for the addition of material. The original design had a head depth of 0.9945 inches. The new design has a head depth of 1.1195inch providing an additional 0.1250inch of material that was previously wasted space inside the tool. Along this same surface, the original slide arm had an angle that originated between the roller axle set screw and the sliding surfaces and tapered to the outer edge of the head. Both threaded holes are located in the base, and a large amount of material addition would strengthen the slide arm. We decided to eliminate the angle, and add an additional 1/8inch. By doing this, we have also eliminated the need to maintain the angle tolerance, and the slide arm will be that much easier to manufacture.
Now we will move up to the high tensile stress area of the inner pocket sidewall. The intense stresses in this area are due to bending of the slide arm head. The inner sidewall is the original fracture location, and requires the addition of material. After consulting the glossy product literature about the location of parts within the threading tool, we noticed that the alignment chain tensioner has very little clearance with this side of the slide arm pocket. However, the other side of the slide arm has a huge clearance. We suggest that the part be mirrored to allow for additional material in the problem area without a clearance issue. One sidewall needs to remain thin due to the tight clearance of the chain tensioner, so we left the outer sidewall alone. The tensile stresses in the outer sidewall were much less than the tensile stresses in the inner sidewall, and they remained within the acceptable limits set by the factor of safety calculations. The problematic inner sidewall was extended 0.230inches, making the head even sturdier, and therefore removing stress from both sidewalls.
Bending requires that a complimentary compressive stress oppose the pocket sidewall tensile stresses. This compressive stress obviously occurs on the slide arm outer radius, but is concentrated at the roller pin cutout directly behind the inner pocket sidewall. While compressive stresses are not as likely to cause failure, they must be considered. Since the pin is a press fit, there is no need to have two sides open. Neither the roller pin nor the roller have failed in service, and should never require removal. We suggest pressing the pin from the outer side, and filling in the inner recess. This decreases the compressive stresses dramatically, but moves the stress concentration to the other recess that opposes the non-critically stressed pocket sidewall. We thought this was a fair trade off. The FEA results show that the maximum 2.34x105psi stress of the original C. J. Winter design was reduced to 1.28x105psi under the normal operating force of 7,000lbs. These stresses were compressive, and occurred at the recess for the roller pin on the outer radius.
The major difference in our proposed redesign is that the entire slide arm is mirror imaged. The sliding surfaces are opposite the original sliding surfaces. This is necessary for tensioner clearance, but should not affect the slide arm performance. The tensile stresses at the inner sidewall were shown to be 1.16x105psi. The corresponding factor of safety was calculated to be 1.55, which exceeds our static requirement.
Figure 6. Primary redesign with 7,000lbs.
Even considering the 1.5 factor of safety load requirement of 10,500lbs, the redesign experienced a maximum tensile stress of 1.72x105psi. This equates to a factor of safety of 1.05.
Figure 7. Primary redesign with 10,500lbs.
In the event that our primary redesign was rejected, we provided an alternative solution. Our secondary redesign included the original part configuration with the addition of material, reversing the chain tensioner. Once reversed, the tensioners will have a tight clearance with the non-critical side of the slide arm pocket, and leave plenty of room for the addition of material to the critical side of the slide arm. In this case, the part would not need to be mirror imaged. All else remains the same and the FEA results are similar to those of our primary redesign.
Figure 8. Secondary Redesign
Fall Quarter Conclusion
Two redesigns were presented to the CJ Winter engineering department for consideration. A superior design requiring extensive modification to existing components and a design with slightly lower capacity, but more suited to the application. The secondary redesign could be used in the 234SA rotary radial attachment with modification to the chain tensioner assembly. The primary design would require complete reorganization of the slide arm casting and machining process. The secondary redesign would require smaller changes to the slide arm production process. Both redesigns employ the concept of thickening the pocket sidewalls. This concept was not incorporated in CJ Winters proposed redesign.
The approved redesign contains qualities suggested by team 09F and was modified by CJ Winter to fit the existing 234SA rotary radial attachment with minimal modification of existing parts. The major additions can be seen in figure 9, below. Major additions include the closing of one side of the roller pin holes since a press fit is used and there is no forseen desire to remove the roller pin; the addition of material in the sprocket pocket; the addition of a radius near the sliding surfaces and the enlarging of existing radii in the sprocket pocket; the extension and widening of the outer wall material where the highest tensile stresses occur during loading.
Thickened sidewall Filled in hole Added radii Extended wall Added to base
Filled in hole
Added to base
Figure 9. Original and redesign; major design changes to slide arm geometry
Since the redesign was different then the original or either proposed redesign, a Finite Element Analysis needed to be performed on the approved redesign. Modification of the proposed redesign was fairly straightforward using the model files from the end of fall quarter. FEA was performed using the same parameters as all of the previous FEA models to preserve integrity.
Figure 9. FEA analysis of approved redesign
Fatigue Test Fixture
The aforementioned test fixture created by IMR labs was provided to the team. In an attempt to perform a fatigue test, an Instron frame was located in the SIMS building on the RIT campus. A new fixture was developed to fit the Instron test frame. The base of the fixture was milled from a block of solid steel with the same basic dimensions as the IMR frame. A tab was left on the bottom of the block to allow the lower mandibles of the Instron frame to grip the block. An anvil was machined with a tab for the upper mandible to grab, and a hole to allow the load to be transmitted to the roller pin within the test slide arm.
Figure 10: FEA analysis of the new fixture anvil and block.
FEA models of the fixture elements were created to determine if the fixture would maintain its integrity throughout the fatigue test. The bottom block was fairly simple to model since it was mostly blocks and circles. The toughest part of the model was the machine cutouts near the lower holding tab. The anvil was a little more difficult to model due to the angles required for fit. The holding tab was necessary as was clearance around the roller pin so as not to induce any non-vertical forces. All of the force on the anvil would be compressive from the upper jaw to the roller pin. The FEA analysis of the anvil indicates deformation occurs in the upper portion of the hole that accepts the pin. The deformation of the anvil may contribute to premature failure of a slide arm using the anvil made of A36 steel. Any variation in rigidity of the anvil may have produce non-vertical forces in the slide arm, and the additional non-vertical component of the 7,000 lb load may cause a premature failure. The FEA model of the anvil indicates a maximum Von Mises stress of 56 ksi, greater than steel’s yield strength of 30 ksi. However, the A2 tool steel used for the slide arm has a yield strength that is six times greater than A36 (or 181 ksi), and would withstand the stresses experienced during the fatigue testing without deforming. In the event that any future fatigue testing is performed, it is recommended that the anvil be made from A2 tool steel.
Figure 11: Old test fixture vs. New test fixture with Instron grip tabs.
Fatigue Testing Original Design
The subject of the fatigue test was an original-design slide arm. Due to scheduling conflicts of CJ Winter’s casting manufacturer, the prototype redesigned slide arm was still unavailable to us at the close of spring quarter. Testing an original-design slide arm served to test the fatigue test fixture and to confirm the conclusions drawn from analysis of the design.
The slide arm was placed into the new fatigue test fixture with a dummy thread roller die installed. The sliding surfaces were constrained to sliding in the vertical direction only. The roller die rested upon the bottom fixturee as a lower constraint. The anvil was installed by sliding the roller pin through the anvil and into the slide arm. The anvil was pinched by the upper mandible of the Instron frame. A compressive force varying between 1,200 lb and 7,000 lb was placed on the slide arm at a frequency of 10 Hz.
Figure 12: Slide arm loaded in test fixture, test fixture loaded in Instron frame.
Before announcing the results of the fatigue test, it should be noted that the tested slide arm is an original design, and not the redesign. In addition, there was a constant load provided from maximum possible force and backed off to below normal operating force. Also, since we did not have access to 1,000,000 unthreaded material blanks, or a Hydromat machine, it was physically impossible to fatigue test the slide arm under actual conditions. The rolling friction and any axial loads experienced by slide arms under actual use conditions were not considered during this fatigue test. Fatigue was modeled by simply loading and unloading the slide arm in a suitable position under maximum force conditions. This said, the team performed the test in what we considered to be very similar to actual conditions.
The result of the testing was roughly 2,100 cycles. The Instron data provided was very vague in that the number of data points was very low. In the expectation of infinite life, the data point collection was backed off to only 1 in every 80 cycles to conserve computer resources. The data can be seen below in figure 13.
Figure 13: Instron computer generated data
The redesign of C. J. Winter’s 234 slide arm proved to be a complex task. While modeling and analysis of the original design revealed the weakest areas of the slide arm, the complex configuration of the assembly prohibited a straight forward addition of material in the suspect locations. Mirroring the slide arm so that added material would not interfere with the chain tensioner, improved the part to satisfy the sponsor’s design criteria (a factor of safety = 1.5 for a static load application of 7,000 lb), but would require the complete reconstruction of a casting model and a re-oriented machining process for the slide arm. The team agreed that a less aggressive redesign be proposed concurrent to the primary redesign, as an option to the sponsor that was weaker than the primary redesign, but easier to incorporate into the current production line.
The team’s conception of a secondary redesign showed foresight towards the sponsor’s desired direction. When choosing between redesigns, strong consideration is given to a redesign that can improve part functionality while minimizing changes to assembly configuration and construction. Ultimately, C. J. Winter decided on a redesign that was less resistant to the stresses experienced in use, but would not require alteration of any other parts within the assembly, in lieu of the primary redesign that met the redesign goal.
The approved redesign is superior to the original design, as shown in FEA analysis. Application of a 7,000 lb static load to both designs, showed a 43% improvement in the factor of safety for the redesign. While not achieving the design goal, the sponsor appeared to be satisfied with the improvement.
An additional complication to the project resulted in the testing phase. C. J. Winter offered to have prototypes cast of the redesign for testing purposes. The lead time required for casting was expected to be 2 – 6 weeks, so the redesign was sent to the casting vendor before the start of the spring quarter. At the conclusion of the 7th week of the quarter, the sponsor and team concluded that the castings would be unavailable for testing, so alternative arrangements were discussed. Since the fatigue test fixture was designed and built, it was a logical progression to fatigue test an available slide arm with the fixture, so an original slide arm was tested.
The fatigue test performed on the original design demonstrated the use of the test fixture, while reaffirming the need to improve on the original design. C. J. Winter initially specified that the slide arm should be rated to withstand a 7,000 lb load. FEA analysis indicated that the static application of a 7,000 lb load provided a factor of safety below 1.0 (approximately 0.8), and the fatigue testing of a cyclic loading of 7,000 lb fractured the arm in under 2,200 cycles. Inspection of the fracture provided evidence of the source of fracture at the location of the maximum Von Mises stress (in tension) in the FEA model.