Modeling the Heat Treatment Response of P/M Components STATEMENT OF WORK

Transcription

1 Modeling the Heat Treatment Response of P/M Components Research Team Makhlouf M. Makhlouf, Professor Richard D. Sisson, Jr., Professor Jiankun Yuan, Research Associate Virendra S. Warke, Ph.D. Student Focus Group Members David Au Quebec Metal Powders, Ltd. Ian Donaldson GKN Sinter Metals Worcester John Fulmer Nichols Portland Bill Jandeska General Motors Chaman Lall Metal Powder Products Co. Jean Lynn Daimler-Chrysler Corporation. Stephen Mashl (Chair) Bodycote IMT, Inc. Sim Narasimhan Hoeganaes Corporation Renato Panelli Mahle Metal Leve S.A. Rocco Petrilli Sinterstahl G.m.b.H. Sylvain St-Laurent Quebec Metal Powders, Ltd. S. Ryan Sun Borg Warner, Inc. STATEMENT OF WORK Introduction P/M components experience considerable changes during heat treatment that include changes in mechanical properties, in dimensions, in magnitude and sense of residual stresses, and in metallurgical phase composition. Since the quality assurances criteria that heat-treated P/M components must meet include prescribed minimum mechanical properties and compliance with dimensional tolerances, it is necessary for P/M producers to be able to accurately predict these changes in order to take appropriate measures to prevent their harmful effects and insure the production of good quality parts. Satisfactory response to heat treatment is often gauged by the ability of the component to be heat treated to a desired microstructure, hardness and strength level without undergoing cracking, distortion or excessive dimensional changes. In addition to the completely reversible changes that are caused by thermal expansion and contraction metallic components experience other permanent dimensional changes during heat treatment. These permanent changes can be classified into three groups based on their origin: 1

2 1. Dimensional changes with mechanical origins, these include dimensional changes caused by stresses developed by external forces, dimensional changes arising from thermally induced stresses, and dimensional changes caused by relaxation of residual stresses. 2. Dimensional changes with metallurgical origins, these include dimensional changes caused by recrystallization, solution and precipitation of alloying elements, and phase transformations. 3. Dimensional changes due to quenching, these are dimensional changes that occur during quenching or that result from stresses induced by quenching. Residual stresses often adversely affect the mechanical properties of P/M components. They are caused by differing rates of cooling during quenching and depend on the differential rate of cooling, section thickness, and material strength. Decreasing the severity of the quench results in a lower level of residual stresses but with a correspondingly reduced material strength of solution heat-treated materials. Residual stresses may also arise from phase transformations during heat treatment that result in differential volumetric changes in the material. Objective The main objective of this project is to develop and verify a computer simulation software and strategy that enables the prediction of the effects of heat treatment on powder metallurgy components. The simulation should accurately predict dimensional change and distortion, residual stresses, type and quantity of metallurgical phases in the microstructure, and hardness. Research Plan A commercially available software, Dante, which is a finite element (ABAQUS: Hibbitt, Karlsson & Sorensen Inc., RI, USA) based tool for analyzing metal heat treatment processes and marketed by DCT, Inc., will be used. This software can perform all the required simulations, but its materials properties database was not designed for P/M materials. Consequently, the earlier phases of the project will focus on assessing the capabilities of Dante and the possibility of adapting it to the specifics of powder metallurgy. Once, this is accomplished, later phases of the project will commence and will focus on using the modified software to predict the heat treatment response of powder metallurgy components. The predicted responses will be compared to experimentally measured responses and a modeling/prediction strategy will be formulated and recommendations will be made to the consortium members. In the early phases of the project, the heated parts will be a simple right cylinder and a thin rectangular sample with a central hole. The simple right cylinder sample will be used for characterizing heat transfer, hardness, and microstructure, while the rectangular sample 2

3 will be used to characterize dimensional distortions and residual stresses. At later phases of the project, production parts chosen by the focus group may be modeled. Methodology The project will be divided into four main tasks as follows: TASK 1: Assessment of Dante s ability to predict heat treatment response of wrought components. Subtask 1.1: Computer simulations Dante will be used to numerically simulate the heating and quenching processes on a simple right cylinder and a thin rectangular sample with a central hole. The simple right cylinder sample will be used in modeling heat transfer, hardness, and microstructure, while the rectangular sample will be used in modeling dimensional distortions and residual stresses. Both samples will be made from wrought 5160-carbon steel for which a comprehensive material properties data set is available. Input to Dante will include the 3D geometrical model and finite element grid, material properties, and heat treatment schedule. Output from Dante will include (1) Resultant volume fraction of metallurgical phases. (2) Hardness distribution after heat treatment. (3) Dimensional changes and distortion. (4) Magnitude and sense of residual stresses. Subtask 1.2: Experiments and measurements Dante s predictions will be verified by comparing them to measurements of corresponding parameters for specimens obtained using processing conditions similar to those used in the simulations. Measurement of dimensional changes and distortion A Starrett coordinate measuring machine will be used to measure the dimensional changes and distortion caused by the heat treatment process. The thin rectangular sample with a central hole is suited for this measurement, as it will experience excessive dimensional distortion particularly in the central hole. Sufficient measurements will be made in order to obtain accurate representation of the part before and after heat treatment. Coordinate measuring machines are a fast, accurate and more convenient alternative to conventional methods for measuring complex parts. Measurement of residual stresses The standard x-ray diffraction method for measuring residual stress in metallic components will be used. Line shifts due to a uniform strain in the component is measured and then the stress in the component is determined either by a calculation 3

4 involving the elastic constants of the material or by a calibration procedure involving measurement of the strain produced by known stresses. Again, the thin rectangular sample with a central hole is suited for this measurement, as it will exhibit excessive residual stresses. Measurement of volume fraction of metallurgical phases Standard metallographic sample preparation techniques will be used to prepare specimens from the heat-treated right cylinder samples. Samples will be prepared from three different cross-sections of the cylinder that are equally spaced along the length of the cylinder. Optical and scanning electron microscopy together with automated image analysis and energy disperssive x-rays will be used to characterize and quantify the various phases in the specimens. In addition, x-ray diffraction analysis will be used to measure and calculate the retained austenite whenever applicable. Measurement of hardness Standard Rockwell hardness and microhardness measurements will be performed on the heat-treated cylinders. Measurements will be performed across three different crosssections of the cylinder that are equally spaced along the length of the cylinder. TASK 2: Adapting Dante to modeling the heat treatment response of fully dense P/M component Once Dante s predictions are validated for wrought 5160 carbon steel components, the software will be used to model the heat treatment response of a 4600 series P/M alloy cylinder that has been pressed to its theoretical density. In this case, a critical input to the model is the properties of the material. The development of this data and its incorporation into the Software is the major focus of this Task. The WPI team will work with the software producers to accomplish this task. Subtask 2.1, 2.2, and 2.3 provide details of this task. Subtask 2.1: Input data generation Extensive material data is required as input to Dante s models, this data includes transformation kinetics data from dilatometry experiments, phase specific mechanical and physical properties, and heat transfer coefficients for various process steps as a function of temperature. The team at WPI will work closely with Oak Ridge National Laboratory to conduct dilatometry experiments. The mechanical, physical and process data may be generated at WPI facilities. Subtask 2.2: Computer simulations Dante will be used to numerically simulate the heating and quenching processes on a right circular cylinder made from 4600 series P/M alloy in the fully dense condition. Subtask 2.3: Experiments and measurements Dante s predictions will be verified by comparing them to measurements of corresponding parameters for specimens obtained using processing conditions similar to those used in the simulations. 4

5 Measurement of dimensional changes and distortion A Starrett coordinate measuring machine will be used to measure the dimensional changes and distortion caused by the heat treatment process. Sufficient measurements will be made in order to obtain accurate representation of the part before and after heat treatment. Measurement of residual stresses The standard x-ray diffraction method for measuring residual stress in metallic components will be used. Line shifts due to a uniform strain in the component is measured and then the stress in the component is determined either by a calculation involving the elastic constants of the material or by a calibration procedure involving measurement of the strain produced by known stresses. Measurement of volume fraction of metallurgical phases Standard metallographic sample preparation techniques will be used to prepare specimens from the heat-treated components. Samples will be prepared from three different crosssections of the cylinder that are equally spaced along the length of the cylinder. Optical and scanning electron microscopy together with automated image analysis and energy disperssive x-rays will be used to characterize and quantify the various phases in the specimens. In addition, x-ray diffraction analysis will be used to measure and calculate the retained austenite whenever applicable. Measurement of hardness Standard Rockwell hardness and micro hardness measurements will be performed on the heat-treated cylinders. Measurements will be performed across three different crosssections of the cylinder that are equally spaced along the length of the cylinder. TASK 3: Adapting Dante to modeling the heat treatment response on porous P/M components Subtask 3.1: Input data generation In order to accommodate the effect of density, similar data sets as described in Subtask 2.1 have to be generated for mechanical and physical properties, transformation kinetics data, and processing data as a function of density as well as temperature. This data will be used as input to the Dante model to accommodate density variability in P/M components. Subtask 3.2: Computer simulations Dante will be used to numerically simulate the heating and quenching processes on a right circular cylinder made from 4600 series P/M alloy in a porous condition. Subtask 3.3: Experiments and measurements Dante s predictions will be verified by comparing them to measurements of corresponding parameters for specimens obtained using processing conditions similar to those used in the simulations. 5

6 Measurement of dimensional changes and distortion A Starrett coordinate measuring machine will be used to measure the dimensional changes and distortion caused by the heat treatment process. Sufficient measurements will be made to obtain accurate representation of the part before and after heat treatment. Measurement of residual stresses The standard x-ray diffraction method for measuring residual stress in metallic components will be used. Line shifts due to a uniform strain in the component is measured and then the stress in the component is determined either by a calculation involving the elastic constants of the material or by a calibration procedure involving measurement of the strain produced by known stresses. Measurement of volume fraction of metallurgical phases Standard metallographic sample preparation techniques will be used to prepare specimens from the heat-treated components. Samples will be prepared from three different crosssections of the cylinder that are equally spaced along the length of the cylinder. Optical and scanning electron microscopy together with automated image analysis and energy disperssive x-rays will be used to characterize and quantify the various phases in the specimens. In addition, x-ray diffraction analysis will be used to measure and calculate the retained austenite whenever applicable. Measurement of hardness Standard Rockwell hardness and microhardness measurements will be performed on the heat-treated cylinders. Measurements will be performed across three different crosssections of the cylinder that are equally spaced along the length of the cylinder. TASK 4: Computer experimentation to characterize the effect of various processing parameters on the heat treatment response of P/M parts Subtask 4.1: Computer experiments Dante will be used to predict the effect of processing parameters on the heat treatment response, i.e., distortion, residual stress, hardness, and microstructure evolution, of industrial P/M part. Simple P/M part such as a gear made of 4600 alloy will be used for demonstration, and the effect of the following parameters will be characterized: Green density: (1) Low density, (2) full density, and (3) highly variable density across the length of the cylindrical component. Heating method: (1) Conventional heating, and (2) induction heating. Quenching method: One, which is normally adopted in the P/M industry. Subtask 4.2: Experiments and measurements to validate the computer experiments Software predictions obtained in subtask 4.1 will be verified for the average density and two different heating mechanisms by comparing them to measurements of corresponding parameters for specimens obtained using processing conditions similar to those used in the simulations. These measurements are outlined as follows. 6

7 Measurement of dimensional changes and distortion A Starrett coordinate measuring machine will be used to measure the dimensional changes and distortion caused by the heat treatment process. Sufficient measurements will be made to get accurate representation of the part before and after heat treatment. Measurement of residual stresses The standard x-ray diffraction method for measuring residual stress in metallic components will be used. Line shifts due to a uniform strain in the component is measured and then the stress in the component is determined either by a calculation involving the elastic constants of the material or by a calibration procedure involving measurement of the strain produced by known stresses. Measurement of volume fraction of metallurgical phases Standard metallographic sample preparation techniques will be used to prepare specimens from the heat-treated cylinders. Samples will be prepared from three different crosssections of the cylinder that are equally spaced along the length of the cylinder. Optical and scanning electron microscopy together with automated image analysis and energy disperssive x-rays will be used to characterize and quantify the various phases in the specimens. In addition, x-ray diffraction analysis will be used to measure and calculate the retained austenite whenever applicable. Measurement of hardness Standard Rockwell hardness and microhardness measurements will be performed on the heat-treated cylinders. Measurements will be performed across three different crosssections of the cylinder that are equally spaced along the length of the cylinder. Deliverables The main deliverable from the project will be a verified predictive strategy and software that enable a complete assessment of the response of powder metallurgy components to heat treatment. The characteristics predicted by the software will include part distortion, residual stresses, microstructure evolution, and hardness. An additional deliverable from the project will be documentation of the effect of processing conditions, including green density, heating method, and severity of quench on the response of 4600 series P/M alloy parts to heat treatment. 7

8 Schedule TASK 1: Assessment of Dante s ability to predict heat treatment response of wrought components. (7/1/2003 to 12/31/2003) Start End Subtask 1.1: Computer simulations 7/1/ /1/2003 Subtask 1.2: Experiments and measurements Measurement of dimensional changes and distortion Measurement of hardness Measurement of volume fraction of phases Measurement of residual stresses 7/1/2003 7/1/2003 7/1/2003 7/1/ /1/ /1/ /1/ /31/2003 TASK 2: Adapting Dante to modeling the heat treatment response of fully dense P/M component (1/1/2004 to 12/31/2004) Start End Subtask 2.1: Input Data generation 1/1/2004 5/30/2004 Subtask 2.2: Computer simulations 6/1/2004 9/1/2004 Subtask 2.3: Experiments and measurements Measurement of dimensional changes and distortion Measurement of hardness Measurement of volume fraction of phases Measurement of residual stresses 9/1/2004 9/1/2004 9/1/2004 9/1/ /1/ /1/ /1/ /31/2004 TASK 3: Adapting Dante to modeling the heat treatment response on porous P/M components (1/1/2005 to 12/31/2005) Start End Subtask 3.1: Input Data generation 1/1/2005 5/30/2005 Subtask 3.2: Computer simulations 6/1/2005 9/1/2005 Subtask 3.3: Experiments and measurements Measurement of dimensional changes and distortion Measurement of hardness Measurement of volume fraction of phases Measurement of residual stresses 9/1/2005 9/1/2005 9/1/2005 9/1/ /1/ /1/ /1/ /31/2005 8

9 TASK 4: Computer experimentation to characterize the effect of various processing parameters on the heat treatment response of P/M parts (1/1/2006 to 6/30/2006) Start End Subtask 4.1: Computer Experiments 1/1/2006 3/31/2006 Subtask 4.2: Experiments and measurements to validate computer experiments Measurement of dimensional changes and distortion Measurement of hardness Measurement of volume fraction of phases Measurement of residual stresses 4/1/2006 4/1/2006 4/1/2006 4/1/2006 6/30/2006 6/30/2006 6/30/2006 6/30/2006 STATUS Task 1 Summary of accomplishments Task1.1: o DANTE and ABAQUES software have been installed successfully on WPI computers. o o The learning curve on both programs has been started and we are now comfortable with the two programs. A DANTE/ABAQUS model has been developed for the thin rectangular part made from wrought 5160 steel plate. Task1.2: o Samples of hot rolled 5160 steel with the same dimensions as those used in the model have been prepared and have been heat treated and quenched in Houghton-G oil. Hardness and dimensional changes have been compared with the model s predictions. Task 1 Progress The DANTE/ABAQUS Model Overview DANTE software is basically a set of user-defined subroutines developed by Deformation Control Technology Inc., Cleveland, Ohio (DCT) that is linked into the ABAQUS standard solver. These subroutines contain a mechanics model, a phase transformation model, and an element diffusion model coupled with stress/displacement, and thermal and mass diffusion solvers. 9

10 The mechanics model is based on internal state variables and describes the mechanical behavior of each metallurgical phase over a wide range of temperatures, deformation levels, and deformation rates. This model includes the effect of phase transformations and transformation induced plasticity in the phases during heat-treating [1]. The phase transformation model is also based on an internal state variable framework where the volume fraction of phases is tracked with changing time and temperature. In this model, formation of ferrite, pearlite, and bainite is assumed to follow diffusive transformation kinetics. The martensitic transformation is assumed to be athermal, however the kinetics equations, are which written in rate form have an explicit dependence on cooling rate [2]. The material data used in the models is determined from various mechanical and thermal tests. Data for the mechanics model is obtained from temperature and rate-dependent tension and compression tests, while data for the phase transformation model is derived from heating and cooling dilatometry measurements. Other mechanical and thermal properties of the materials are extracted from the open literature and are implemented in the models as functions of temperature [3]. Solution Procedure and Boundary Conditions The general solution procedure for the DANTE/ABAQUS model is shown in Figure1 and consists of two models: (1) The thermal model, and the stress model. First, the thermal model is setup to solve a heat transfer problem for each step of the heat-treating process, such as furnace heating, immersion into the quench tank, and quenching. The output file generated by the thermal model contains mainly the thermal history of the part during the various process steps. The stress model accesses this files and calculates residual stresses, displacements, volume fraction of metallurgical phases, and hardness for the entire temperature history of the part. Boundary Conditions Thermal Model Boundary Conditions Stress Model Post Processing Figure 1 Solution procedure for the DANTE/ABAQUS model. 10

11 The part under consideration is rectangular in shape with a central hole. The purpose of the central hole is to exaggerate dimensional changes that may occur from thermal treatment of the part. Since the part will also be used in determining the heat transfer coefficient, a threaded hole is machined into the cross section of the shorter arm of the rectangular piece. The 3-D geometry appropriately meshed is shown in Figure 2. Figure 2 was created using the ABAQUS pre-processor. The meshed geometry contained 5118 hexahedral elements and 6685 nodes. In order to capture the steep temperature gradient caused by quenching, the mesh was generated such that the density of nodes is higher near the surface of the part than in its interior. The dimensions of the part are given in Table I. Figure 2 Geometry and mesh used in the model Table I Dimensions of the part. Dimension Magnitude Length mm Height mm Width mm Diameter of center hole mm 11

12 The Thermal Model: The following heat treatment process was modeled for wrought 5160 steel: (1) Furnace heating, followed by (2) Immersion into a quench tank, followed by (3) Quenching The initial conditions relevant to the thermal model are as follows: o o o The nodal carbon level: In our model, the carbon content of the steel is set constant at 0.59 wt. % C. The initial temperature of the part: In our model, the initial temperature of the part is set constant at 20 C. The heat treatment modes: The heat treatment modes are used to select the appropriate kinetics input from the different kinetics models in DANTE to match each of the process steps as they change from furnace heating to sample immersion in the quenching fluid to quenching. In our model, the heat treatment mode is initially set to invoke the heating kinetics, and then it is set to invoke the cooling kinetics. The boundary conditions relevant to the thermal model are as follows: o For the furnace-heating step: The heat transfer coefficient used in this step is obtained from one of the example problems in DANTE. The solution to this step proceeds until the temperature at all the nodes in the part reaches 850 C, which is the austenitizing temperature of 5160 steel. o o For the immersion step: The direction and velocity of immersion of the part into the quench tank is defined in the immersion step. This step is important in order to capture the temperature gradient along the immersion length of the part. In our model, the part is immersed along its length, with a velocity of 100 mm/sec, and the process time for this step is set to 0.9 seconds. For the quenching step: The heat transfer coefficient shown in Figure 3 was used. This data was obtained by the Center for Heat Treating Excellence at WPI for 4140 steel quenched in Houghton-G oil. When using this data to simulate the thermal behavior of wrought 5160 steel, it is assumed that the heat transfer coefficient is a function of temperature and surface finish only and is independent of material properties. It I assumed that the part is quenched in oil until the temperature at all the nodes drops to the temperature of the oil bath. 12

13 C) o Heat Transf Temperature ( o C) Figure 3 Heat transfer coefficient as a function of temperature for 4140 steel quenched in Houghton-G oil. The Stress Model: This model uses mainly the time-temperature history of the part generated by the thermal model in order to calculate nodal displacement, stress, volume fraction of metallurgical phases and hardness. The initial condition relevant to the stress model is as follows: o The magnitude of stress at all nodes is set to zero. The boundary condition relevant to the stress model is as follows: o Nodal constraints are required in order to prevent rigid body displacement and rotation. This requirement applies to all the process steps, and is defined only once in the model input file. In our model, 3 nodes around the center of the X-Y plane of the part (1-2 plane in Figure 2) are constrained in displacement and rotation. It is important to note that the stress model must be similar to the thermal model in its number of process steps and process time for each process step, as well as the number of elements and nodes in the part. 13

14 Results and Discussion Figure 4 shows the experimental device developed by CHTE for measuring the heat transfer coefficient of solids during quenching. The apparatus consists of a box furnace, a data acquisition system, a pneumatic controller, and an oil bath (not shown). The connecting rod attached to the pneumatic controller has a K-type thermocouple going through it. Once the sample is heated to the required temperature in the box furnace, the connecting rod is lowered to immerse the part in the quench tank located below the furnace table. Pneumatic on/off switch Pneumatic cylinder K-type thermocouple Computer with Data Acquisition Card Connecting rod Probe tip Furnace Oil beaker Thermocouple for Oil temp. Figure 4 Apparatus used for measuring the heat transfer coefficient during quenching. Samples, with the dimensions shown in Table I, are machined from a 5160 steel plate. A hole that is mm in diameter and mm in depth for inserting the thermocouple was drilled at the center of the X-Y plane of the samples (the 1-2 plane in Figure 2). The top 9 mm of the hole was bored and tapped with internal threads to allow screwing the sample to the connecting rod, which holds the part in an upright position during heating and quenching. Graphite powder was packed into the hole before the thermocouple is inserted in order to ensure intimate contact between the sample and the thermocouple. The sample is then heated to 850 C and time vs. temperature data is recorded while the sample is quenched. Figure 5 shows a comparison of the computed and measured cooling curves. In the early stages of quenching, the measured cooling rate is somewhat slower than the predicted cooling rate. There could be two possible reasons for this: (1) Contrary to our initial assumption that the heat transfer coefficient during quenching is a negligible function of 14

15 material properties, the use of the heat transfer coefficient for 4140 steel in simulating quenching of 5160 steel may introduce an error. (2) The hole machined at the center of the part entraps part of the vapor blanket that is formed by the hot oil and retains it covering the internal surface of the hole for a relatively long time thus effectively reducing the overall heat transfer coefficient of the sample. Figure-5 Comparison between measured and predicted cooling curves for wrought 5610 steel. Figure 6 shows a comparison between the measured (Fig. 6(a)) and computer-predicted (Fig. 6(b)) dimensions of the central hole in the part before and after quenching. The measurements were performed on a Starrett coordinate measurement machine at WPI. The machine can measure the coordinate with the least count of 10 microns. Since machining a hole in a flat part invariably involves some error, the radius of the the machined hole does not match perfectly the radius created in the model. Therefore, an accurate quantitative comparison between the measured and predicted part distortion is not possible. Moreover, a very small, unwanted, rigid body motion has occurred in the modeled part. This can be avoided in future models by constraining one more node in the stress model. Nevertheless, Figure 6 provides a useful qualitative assessment of the ability of the software to simulate dimensional distortion in wrought steel. 15

16 0.5 Before quenching After quenching Radius (mm) (a) Before quenching After quenching Radius (mm) (b) Figure 6 Coordinates of the circular hole before and after quenching. (a) Measured using CMM, and (b) predicted by the model. 16

17 Hardness (HRC) Location from one end (mm) Measured Predicted Figure 7 Comparison between measured and predicted hardness values for wrought steel. Figure 7 shows a comparison between the measured and computer-predicted hardness values across the sample s cross-section. There is very good agreement between the computer-predicted and measured hardness values except at the part s edge. This discrepancy at the part s edge is attributed to the loss of carbon at the surface due to decarburization. Decarburization is not accounted for in DANTE and is an issue that should be addressed. Figure 8 shows computer-predicted contour plots of volume fraction of martensite (Fig. 8(a)) and retained austenite (Fig. 8(b)). The average volume fractions of phases are % martensite and 0.100% austenite. For comparison purposes, x-ray diffraction measurements of retained austenite as well as computer assisted phase analysis of optical photomicrographs of the sample are currently underway. 17

18 (a) Figure 8 Contours of volume fraction of (a) martensite and (b) retained austenite. (b) 18

19 References 1. Bammann D. J., Chiesa M. L., Johnson G.C., Modeling Large Deformation and Failure in Manufacturing Process, proceeding of 19 th International congress of Theoretical and Applied mechanics,1996, pp Lusk M. T., Lee Y. K., A Global Material Model for Simulating the Transformation Kinetics of Low Alloy Steel, 7 th International Seminar on Heat Treatment and Surface Engineering of Light Alloys, 1999, pp Furguson B.L. et al., 3 rd International Conference on Quenching and Control of Distortion, 1999, pp