HEAT TREAT SIMULATION USED TO IMPROVE GEAR PERFORMANCE

Transcription

1 HEAT TREAT SIMULATION USED TO IMPROVE GEAR PERFORMANCE Heat treat simulation using a finite element based tool offers the ability to compare competing heat treat process routes and material selection to predict dimensional changes, metallurgical phases and hardness, and residual stress state. B. Lynn Ferguson, FASM** Zhichao Li** Andrew M. Freborg** Deformation Control Technology Cleveland, Ohio Four main performance problems for gearing are fatigue due to cyclic loading, wear, excessive noise, and corrosion characteristics that are affected by heat treatment. Of these, the first three can benefit directly from selection and control of the heat treat process, and corrosion can be indirectly enhanced. Heat treatment process simulation represents a new, powerful tool for process design, eneabling improved fatigue life, wear resistance, noise reduction, and even in some cases, improved corrosion resistance in engineered steel parts. Heat treat simulation using a finite element based tool such as DANTE offers the capabilities of comparing competing heat treat process routes and material selection to predict dimensional changes, metallurgical phases and hardness, and residual stress state. With accurate simulations, information is gained concerning fatigue life, wear resistance, noise, and possible corrosion resistance improvements. The key to these performance enhancements is the accuracy of the simulations. DANTE Background A collaborative project involving government, industry, and academia led to the development of DANTE heat treat simulation software [1]. The program was specifically driven to address quality and test benefit concerns for heat treated automotive parts, particularly with respect to distortion. Figure 1 shows a diagram of the flow of information and the various modules involved in heat treat simulation calculations. Typical results predicted by DANTE are volume fraction of metallurgical phases, hardness, residual stress, and part dimensional change. DANTE consists of a thermomechanical model, phase transformation kinetic models for steel phase transformations during heating and cooling processes, a material database for steel grades, and a database of heat transfer coefficients for common heating and cooling processes. The models are implemented as subroutines that are accessed by the ABAQUS/STANDARD (a registered trademark of ABAQUS Inc., Paw- **Member of ASM International and member, ASM Heat Treating Society Fig. 1 DANTE chart showing the flow of data, module usage, and typical output. HEAT TREATING PROGRESS JULY

2 Fig. 2 A 40-tooth spur gear. Carbon, wt% Fig. 3 Predicted case profiles. Root center Root corner Tooth flank Depth from surface, mm Fig. 4 Predicted dimensional changes for baseline processed gear at the indicated process steps and temperatures. tucket, R.I.; finite element solver during execution. The material model implemented in DANTE is a variation of the Bammann-Chiesa-Johnson (BCJ) model, an IR-100 Award winner in 2000 [2]. The model is based on an internal state variable approach to mathematically describing the mechanical behavior of the individual metallurgical phases. The mechanical behavior of the steel as the phases evolve is then calculated from the behaviors of the individual phases. This model is well-suited for use in applications such as heat treatment where stress reversals due to thermal and/or transformation induced stress occur. DANTE has been used to simulate a variety of heat treat processes including gas and vacuum carburizing; immersion quenching in oil, salt, or polymer baths; spray quenching; intensive, or rapid, quenching (quenching in a liquid medium that is so highly agitated that no vapor phase is formed in the liquid and even nucleate boiling is eliminated or at least suppressed); and highpressure gas quenching. Improving Gear-Tooth Bending Fatigue Strength DANTE was used to compare two quench hardening processes to improve the tooth bending fatigue strength of carburized Pyrowear 53 (Carpenter Technology Corp, Reading, Pa.; steel gears used in helicopter transmissions [3]. The baseline process relied on conventional oil quenching of the carburized gear to develop hardness and residual surface compressive stress. The comparison process substituted intensive quenching using high velocity water in place of oil to substantially increase the magnitude of residual compression. Test pieces included 34 gears machined by Aero Gear (Windsor, Conn.; a 40- tooth spur gear is shown in Fig. 2. After carburization and a subcritical anneal, half of the gears were quenched hardened, deep-freeze treated, and double tempered by Aero Gear, and half of the gears were intensive quenched, deep-freeze treated, and double tempered by IQ Technologies Inc., (Akron, Ohio; using a patented process [4]. Aero Gear finish ground the gears removing in. (0.13 mm) from the baseline gear set and in. (0.18 mm) from the intensive quenched gears. Prior to manufacturing the gears, computer simulation was used to investigate the effects of heat treatment on the two process routings. DANTE models predicted that the intensive quenched gears would grow in. (0.05 mm) more than the baseline gears when starting from identical green dimensions. The models also predicted that the intensive quenched gears would have a higher magnitude of surface compression than the baseline gears. Furthermore, 30 HEAT TREATING PROGRESS JULY 2007

3 Fig. 5 Predicted dimensional changes for intensive quenched gear at the indicated process steps and temperatures. the simulations predicted that the difference in residual stress states was large enough that it would remain after finish grinding, so the deeper compression predicted for the intensive quenched gears would be present during single-tooth bending tests. These initial predictions were important because they provided confirmation and confidence that the study should proceed as planned. The aim case depth after carburization was in. (1.0 mm). Figure 3 shows predicted carbon profiles at the pitch diameter, root fillet, and center of root locations. The effect of geometry on carbon profile is evident; the root fillet location has a slightly shallower profile due to the smaller surface area-to-volume diffusion zone at this location. A comparison of Figs. 4 and 5 (in which one half of the gear has been removed to show the mid-tooth height, internal plane) shows predicted dimensional differences between the baseline and intensive quenching processes. Initial green dimensions were identical for the two processes, and the dimensional change is in reference to the starting green dimension. Intensive quenching results in a predicted final growth of in. (0.076 mm) per side, which was in. (0.043 mm) greater than the growth predicted for the baseline process. In fact, a in. (0.051 mm) thick layer of nickel was plated onto the bore of intensive quenched gears to accommodate the fatigue test support mandrel. Figures 6 and 7 show the predicted minimum principal stress contours for the two heat treatments at the indicated process steps. This stress is the highest compressive stress and serves as a good comparison of the two processes. At the surface of the gear root fillet and at mid-height, the final predicted minimum principal stress values are -845 MPa (-122 ksi) for the baseline process and -1,080 Fig. 6 Predicted minimum principal stress profile for the baseline gear at the indicated process steps and temperatures. Fig. 7 Predicted minimum principal stress profile for the intensive quenched gear at the indicated process steps and temperatures. HEAT TREATING PROGRESS JULY

4

5

6 Fig. 8 Typical fatigue crack for tooth bending. MPa (-157 ksi) for the intensive quench process. The stress at the center of the root is lower than at the root fillet, with values of -560 MPa (-81 ksi) for the intensive quenched gear and -410 MPa (-59 ksi) for the baseline gear. Residual stresses were calculated using x-ray diffraction data for the center of the root and are in agreement with these predictions [5]. The difference in residual compression at the root locations is due to the geometry of the gear root. Predictions based on these data show the potential benefit of the intensive quenching route (when properly applied) over the baseline heat treat process in terms of higher residual compressive stress. Simulation provides the reason for the differences in final dimensions and residual stress for the two quench methods and shows that these responses arise from the difference in microstructural evolution. Although both heat treating routes produce nearly identical hardness profiles and martensite content (>97%), the timing and location of martensite formation is markedly different. Martensite forms more quickly at the case-core interface and in the case itself during intensive quenching than it does for the baseline oil quench. As a result, thermal contraction of the core austenite induces higher compression in the case of the intensive quenched gear than for the baseline gear. The difference in residual compression remained after final heat treating steps as shown in Figs. 6 and 7. Reference 5 provides a more detailed explanation of these differences. Fig. 9 Maximum principal stress predicted for gear tooth bending at a load of 6,228 N; IQ = intensive quenched gear, OQ = baseline oil-quenched gear, and Stress Free = gear that had no residual stress. Maximum principle stress, MPa 2,000 1,600 1, IQ gear OQ gear Stress free IQ initial OQ initial Depth from surface, mm Fig. 10 Maximum principal stress profiles at the root fillet for single-tooth bending; initial profiles are also shown for the heat treated conditions. Gear Tooth Loading In the single-tooth bending test, a load is applied near the tooth tip, producing a maximum tensile stress near the root fillet. Simulation shows that as the tip load is increased, the location of the maximum stress shifts slowly outward around the root fillet and toward the tooth. Figure 8 shows a typical fatigue crack location and path for this gear tooth shape. The location of the crack initiation is near the top of the root fillet. Because tooth cracking is mainly a function of maximum tension and not maximum compression, it is of interest to compare the maximum and minimum principal stress profiles under no tooth loading and for the tooth loaded case. Figure 9 shows the predicted contours of maximum principal stress for a bending load of 6,228 N (1,400 lbf) for finite element stress models with the initial residual stress states for the baseline and intensive quenched conditions and for the case of no residual stress. Figure 10 shows a single profile line for these 34 HEAT TREATING PROGRESS JULY 2007

7 Fig. 11 Fixture used at Gear Research Institute for gear tooth bending loads two teeth at a time; no teeth needed to be removed for testing. cases, and the benefit of residual compression in terms of lowering the maximum principal stress is evident. Furthermore, even though the initial maximum principal stress profiles are similar for the baseline and intensive quenching conditions prior to loading, a difference develops during tooth loading, with the intensive quenched condition having a lower maximum principal stress value. Maximum bending stress, MPa 1,800 1,700 1,600 1,500 1,400 1,300 1,200 DCT: bending fatigue testing Load Intensive quench Baseline Near tip Runout 1,600 lb 1,500 1,400 1,300 1,200 1,100 1,000 1, , ,000 1,000,000 10,000,000 Cycles to failure Fig. 12 Single-tooth bending fatigue test results to date. The number of runouts (10 7 cycles) for each condition is shown at the right side. The ordinate includes both the maximum applied load and maximum root stress using an AGMA standard method to compute root stress. Gear Tooth Bending Fatigue Single tooth bend tests were conducted at Gear Research Institute ( In the fixture as shown in Fig. 11, the bending load was apllied on two teeth simultaneously. Tension-tension conditions were used, with a major-to-minor load ratio of 10. Runout was considered to be 10 7 cycles, and each runout means that both loaded teeth successfully endured this number of cycles. Figure 12 shows single-tooth bending-fatigue test data. The number of runout tests at each load is indicated at the right side of the figure for both baseline and intensive quenched conditions. Intensive quenching provides approximately 10-12% improvement in fatigue life; testing is still in progress. However, the benefit of higher surface compression for the intensive quenching condition is evident. The values on the S-N curve ordinate axis include both the actual applied maximum load and the corresponding maximum root stress calculated using a standard AGMA method. The values are similar to the finite element model for the case of no residual stress. The AGMA method does not include residual stress, and consequently, the maximum stress values are very high relative to the strength of the carburized case. The stresses reported in Figs. 9 and 10 are more realistic, and demonstrate the value of residual compression in reducing the actual stress realized at the root during bending. Summary Heat treating adds considerable value to the fatigue life of gears because it can be used to establish desirable residual surface compressive stress. A stress-free gear has much higher stress in the root fillet during tooth bending, and residual surface compression significantly lowers the maximum root stress. Although single-tooth bend testing has not been completed, a qualitative view of the fatigue data bears out the benefit of residual surface compression, with the intensive quenched gears having higher fatigue strength and higher surface compression. Statistical analysis will be performed after testing is completed. Heat treat simulation was used to help make decisions about the project directions and the probability of success. More specifically, simulation showed that a reasonable difference in stress state should be achieved in gears quenched in oil and by intensive quenching. Differences in gear dimensional growth were predicted, and these were in agreement with growth measured in the gear sets. Also, differences in the two quench methods were examined effectively with simulation in terms of microstructural evolution and residual stress distributions. Acknowledgement: DCT recognizes the technical and financial support of Clay Ames and Bert Smith of the U.S. Army AATD, Ft. Eustis, Va., under Contract W911 W6-05-C References 1. National Center for Manufacturing Sciences, Predictive Model and Methodology for Heat Treatment Distortion - Phase 1 Project Summary, Sept., Sandia National Laboratories, Microstructure-Property Model Software, Research & Development, Deformation Control Technology Inc., High Strength, Affordable Helicopter Gears, U.S. Army AATD Contract W911 W6-05-C-0017, U.S. Patent # 6,364,974, April B.L. Ferguson, A.M. Freborg and Z. Li, Residual Stress and Heat Treatment Process Design for Bending Fatigue Strength Improvement of Carburized Aerospace Gears, Quenching and Control of Distortion, Proc. of the 5th Intl. Conf. and European Conf. on Heat Treatment, Berlin, Germany, p , For more information: Lynn Ferguson is president, Deformation Control Technologies Inc., 7261 Engle Rd., Ste 105, Cleveland, OH 44130; tel: ; fax: ; lynn.ferguson@ deformationcontrol.com; Web site: www. deformationcontrol.com. HEAT TREATING PROGRESS JULY