1、13FTM24 AGMA Technical Paper Innovative Induction Hardening Process with Preheating for Improved Fatigue Performance of Gear Component By Dr. Z. Li, Dante Software 2 13FTM24 Innovative Induction Hardening Process with Preheating for Improved Fatigue Performance of Gear Component Dr. Zhichao (Charlie
2、) Li, Dante Software The statements and opinions contained herein are those of the author and should not be construed as an official action or opinion of the American Gear Manufacturers Association. Abstract Contact fatigue and bending fatigue are two main failure modes of steel gears. Surface pitti
3、ng and spalling are two common contact fatigue failures, which are due to the alternating subsurface shear stresses from the contact load between two gear mates. When a gear is in service under cyclic load, concentrated bending stresses exist at the root fillet, which is the main driver of bending f
4、atigue failures. Heat treatment is required to increase the hardness and strength of gears to meet the required contact and bending fatigue performance. Induction hardening is becoming more popular due to its process consistency, reduced energy consumption, clean environment, and improved product qu
5、ality. It is well known that an induction hardening process of steel gears can generate compressive residual stresses in the hardened case. Compressive residual stresses in the hardened case of tooth flank benefit the contact fatigue performance, and residual compression in the root fillet benefits
6、the bending fatigue. Due to the complex gear geometry, the residual stress distribution in the hardened case is not uniform, and different induction hardening process can lead to different residual stress pattern and significant variation of fatigue performance. In this paper, an innovative approach
7、 is proposed to flexibly control the magnitude of residual stress in the regions of root fillet and tooth flank by using the concept of preheating prior to induction hardening. Using an external spur gear made of AISI 4340 as an example, this concept of innovative process is demonstrated with finite
8、 element modeling, using commercial software DANTE. Copyright 2013 American Gear Manufacturers Association 1001 N. Fairfax Street, Suite 500 Alexandria, Virginia 22314 September 2013 ISBN: 978-1-61481-081-0 3 13FTM24 Innovative Induction Hardening Process with Preheating for Improved Fatigue Perform
9、ance of Gear Component Dr. Zhichao (Charlie) Li, Dante Software Introduction Residual stresses embedded inside a hardened gear are critical to its fatigue performance. There are two main failure modes for gear components: contact fatigue and bending fatigue. In service, one pair of gears transfer to
10、rque load through the contact of two teeth. High shear stresses co-exist with high hydrostatic pressure under the contact surface. Depending on the load magnitude and the gear size, the depth of the highest shear stress point varies. To improve the contact fatigue life, the hardened case depth needs
11、 to be deeper than the highest shear stress point. Compressive residual stresses located inside the hardened case benefit the contact fatigue performance 1. The bending fatigue failures are commonly found at root fillet location where tooth flank and root meet. Under the contact load between two gea
12、rs, the root fillet experiences cyclic stresses, which is the driver of bending fatigue failure. Compressive residual stresses from heat treatment or other surface processing can significantly improve the bending fatigue performance 2. Induction hardening is more environmentally friendly than conven
13、tional furnace heating and liquid quenching. It also provides flexibility in control of case depth, residual stresses, and part distortion. Due to these advantages, the induction hardening process is becoming more popular to harden steel gears. During induction heating, the energy to heat the part i
14、s generated inside the part by eddy currents in response to the imposed alternating magnetic field. The energy distribution in the part is directly related to the distance between the inductor and the part, the frequency and power of the inductor. Lower frequency and lower power heat the part deeper
15、 over longer time period. Higher frequency and higher power heat a shallower layer over shorter time period. The temperature distribution in the part is a combined result of induction heating and thermal conduction. During induction hardening of steel components, both thermal gradient and phase tran
16、sformations simultaneously contribute to the evolution of internal stresses and shape change. Recent developments in heat treatment modeling technologies make it possible to understand the materials responses during heat treatment processes, such as how the internal stresses and distortion are gener
17、ated. DANTE is a commercial heat treatment software based on finite element method 3-5, which was designed to model the responses of steel parts during heat treatment processes. The materials responses include phase transformations, deformation, residual stresses, and hardness, etc. Typical heat tre
18、atment process steps include austenitization, carburization, quench hardening, and tempering. Phase transformation kinetics and mechanical properties are required for modeling the heat treatment processes 6,7. DANTE has a validated database for most common low and medium alloy carbon steel grades, w
19、hich have been used successfully in the past to model induction hardening processes 8,9. With the help of computer modeling, engineers with DANTE Software have discovered that residual compression at the root fillet of a gear can be enhanced by applying preheating prior to induction hardening proces
20、s. The preheating process can be implemented either by furnace or induction heating. In this paper, this innovative process is demonstrated by computer modeling, using an AISI 4340 spur gear example. Phase transformation kinetics Phase transformations are involved in most heat treatment processes of
21、 steel components. During heating, initial phases transform to austenite, and carbides dissolve while being held at the austenitization temperature. During cooling or quenching steps, austenite transforms to ferrite, pearlite, bainite, or martensite, depending on the cooling rate and hardenability o
22、f the steel grade. At different heat treatment stages or at different regions in a part, the material can be composed of different phases, and the volume fractions of individual phases are functions of chemical composition and thermal history. To model the heat treatment process of steel components,
23、 accurate descriptions of material properties and process information are required. The basic material property data includes phase transformation kinetics, thermal and mechanical properties of individual phases. Phase transformations during quenching are classified as diffusive and martensitic tran
24、sformations. The diffusive transformation is time and temperature driven, and the martensitic transformation is mainly 4 13FTM24 temperature driven. The two types of phase transformation models used in DANTE are described in equations (1) and (2). 111ddd dadTdt (1) 122mm mdamddT (2) where dis the vo
25、lume fractions of individual diffusive phase and martensite transformed from austenite; dis a transformation mobility and is a function of temperature; 1, 1(superscripts) are constants of diffusive transformation; a is the volume fraction of austenite; mis the volume fractions of individual diffusiv
26、e phase and martensite transformed from austenite; mis a transformation mobility and is a constant; 2, 2(superscripts) are constants of martensitic transformation; is a constant of martensitic transformation. For each individual phase formation, one set of transformation kinetics parameters is requi
27、red. Different experiments can be used to characterize phase transformations, such as dilatometry test, Jominy End-Quench test, and metallographic characterization, etc. Among these experiments, dilatometry test is preferred due to its accuracy, economic advantage, as well as more useful data obtain
28、ed 7. Figure 1a is a strain curve of martensitic phase transformation dilatometry test for AISI 4340. The X-axis is temperature in Celsius, and the Y-axis is strain from the combined effects of thermal shrinkage and phase transformations. During this specific cooling test, the cooling rate of the sa
29、mple is fast enough to avoid diffusive phase formations. During cooling, the dilatometry sample shrinks with the temperature dropping. When the sample reaches the martensitic transformation starting temperature (Ms), martensitic formation starts with volume expansion. The strain change during transf
30、ormation is a combination of thermal strain and phase transformation volume change. The data obtained from this specific dilatometry test include coefficient of thermal expansion (CTE) for austenite and martensite, martensitic transformation starting and finishing temperature (Ms, Mf), transformatio
31、n strain, and phase transformation kinetics (transformation rate) from austenite to martensite. These data are critical to the accuracy of modeling the internal stresses and deformation during quenching. Figure 1. a) Dilatometry strain curve during continuous cooling; b) bainitic TTT diagram of AISI
32、 4340 generated from DANTE database a) b) 5 13FTM24 Diffusive transformations can also be characterized by dilatometry tests. In general, a series of dilatometry tests with different cooling rates are required to fit a full set of model parameters for diffusive and martensitic phase transformations.
33、 Once phase transformation kinetics parameters are fit from dilatometry tests, TTT/CCT diagrams can be generated for users to review. TTT/CCT diagrams are not directly used by DANTE models of phase transformation kinetics, but they are often useful because they represent the hardenability directly.
34、Figure 1b is the bainitic isothermal transformation diagram (TTT) created from DANTE database. The incubation times for ferritic and pearlitic transformations are much longer than that of bainitic transformation for this steel, so they are not discussed in this paper. Descriptions of gear model and
35、heat treatment process A spur gear shown in Figure 2a is selected to study the effect of preheating temperature on residual stresses. The outer diameter of this gear is 164.0 mm, the inner diameter is 75.7 mm, and the thickness is 15 mm. This gear has a total of 28 teeth. With the assumption that al
36、l teeth behave the same during the whole heat treatment process, a single tooth model with cyclic boundary condition is used to represent the whole gear. The finite element meshing of the single tooth model is shown in Figure 2b, with 106850 nodes and 98784 linear hexagonal elements. Fine elements a
37、re used in the part surface to catch the thermal and stress gradients. Instead of modeling the electro-magnetic field, the power distribution generated by inductor is applied directly to drive the heat treatment model. The power distribution can either be predicted by electro-magnetic modeling softw
38、are or be estimated based on the relations of inductor power, frequency and part geometry, and both methods have been successfully used in the past 8, 9. In this study, the induction hardening process is simplified as two steps: 1) induction heating the gear teeth for 3.5 seconds, and 2) spray quenc
39、h the gear to room temperature using 6% polymer solution. There is no dwell between heating and spray quenching. The induction hardening process is also compared with traditional oil quench. Two oil quench processes are modeled: 1) furnace heating and oil quench of AISI 4340 gear, and 2) furnace hea
40、ting, carburization, and oil quench of AISI 4320 gear. The carburization temperature is 900C, hold in 0.8% carbon potential atmosphere for 6 hours. Six induction hardening processes are modeled. The first model has no preheating, and the other five models assume uniform preheating temperatures of 20
41、0C, 250C, 300C, 350C, and 400C, individually. With preheating, the time duration and frequency of induction heating are kept the same, but the power of the inductor is reduced to avoid overheating the teeth. The powers of inductor are 80%, 75%, 70%, 65% and 60% for preheating temperatures of 200C, 2
42、50C, 300C, 350C, and 400C, relative to the inductor power without preheating. The temperature distributions at the end of the 3.5 s heating are shown in Figure 3 for all the six scenarios. The lower bounds of the legends in Figure 3 vary with preheating temperatures, and the upper bounds are the hig
43、hest temperatures at the root. The temperature at root for the process without preheating is about 1100C, comparing to 1050C for all the cases with preheating. Figure 2. a) CAD model of the spur gear; b) single tooth FEA model a) b) 6 13FTM24 a) without preheating b) preheating 200C c) preheating 25
44、0C d) preheating 300C e) preheating 350C f) preheating 400C Figure 3. Contours of temperature distribution at the end of induction heating For all the six induction hardening scenarios, the depths of hardened layer at the root are kept closely to 1.5 mm. However, the obtained martensite distribution
45、s at the tooth tip have significant difference due to the preheating effect. With higher preheating temperature, more martensite is formed at the tooth tip. In general, partially hardened tooth tip is preferred to reduce the possibility of brittle crack at the tooth tip edges. See Figure 4. Results
46、and discussions In this study, a global Cartesian coordinate system is used to calculate the stresses for all the heat treatment models. The origin of the global Cartesian coordinate system is located on the axis of the gear. Once the models are completed, two local cylindrical coordinate systems ar
47、e created to post-process the tangential stresses at tooth flank and root fillet regions, as shown in Figure 5a. The tangential stresses at the root fillet are plotted using the first local cylindrical coordinate system, and the tangential stresses at the tooth flank are plotted using the second loc
48、al cylindrical coordinate system. Figure 5b shows two highlighted lines representing the root fillet (CD) and pitch (AB), individually. Using the local cylindrical coordinate systems, stresses away from the root fillet or tooth flank are not tangential to the surface any more. Tangential residual st
49、resses predicted at the root fillet are used to evaluate the bending fatigue performance of the gear. 7 13FTM24 a) without b) preheating c) preheating d) preheating e) preheating f) preheating preheating 200C 250C 300C 350C 400C Figure 4. Contours of martensite distribution from different induction hardening scenarios Figure 5. a) Definitions of two local cylindrical coordinate systems; b) pitch and root fillet lines selected for post-processing tangential residual stresses The tangentia
