Sei sulla pagina 1di 6

Journal of Materials Processing Technology 153154 (2004) 346351

Residual stress reduction in 7175-T73, 6061-T6 and 2017A-T4 aluminium alloys using quench factor analysis
G.P. Dolan, J.S. Robinson
Materials Science and Technology Department, University of Limerick, Limerick, Ireland

Abstract The quenching of aluminium alloys by immersion or spraying using cold water produces large thermal gradients in thick components such as forgings and plate. Whilst the rapid cooling ensures good mechanical properties the thermal gradients can be large enough to produce high levels of residual stress. Reducing the cooling rates during the quench can reduce the magnitude of the residual stresses, however it can also be detrimental to the mechanical properties, particularly for quench sensitive alloys. By generating time temperature property C-curves for each alloy and using quench factor analysis, it is possible to slowly cool the alloy from the solution heat treatment temperature to an intermediate temperature, above the critical temperature region of the C-curve, and then quench to room temperature. The result of this is a reduction in the residual stresses generated during the quench due to a reduced thermal gradient combined with a negligible effect on the mechanical properties of the alloys. The C-curves for the Vickers hardness have been generated for each alloy and temper and are presented. The residual stresses have been quantied for this alternative quenching route for each alloy and the Vickers hardness results are presented. 2004 Elsevier B.V. All rights reserved.
Keywords: Quench factors; Residual stress; Aluminium alloys

1. Introduction To achieve optimal mechanical properties many heattreatable aluminium alloys require a rapid quench from the solution heat treatment (SHT) temperature by either immersion in, or spraying using cold water. This rapid quench maintains a supersaturated solid solution, which is then allowed to decompose in a controlled manner through either natural or articial ageing. Whilst the retention of good mechanical properties is desirable, in thick sections the varying cooling rate can lead to large thermal gradients that not only cause property inhomogeneity in quench sensitive alloys [1], but can also lead to high levels of residual stress. These residual stresses can have a detrimental effect on the component leading to warping or cracking during the quench, dimensional instability during machining, reduced fatigue life and increased susceptibility to stress corrosion cracking. The thermal stress in a component is related to the alloys thermal expansion coefcient, elastic modulus and the temperature differential between the surface and the interior of the component. During the initial stages of the quench, the
Corresponding author. E-mail addresses: gerard.dolan@ul.ie (G.P. Dolan), jeremy.robinson@ul.ie (J.S. Robinson). 0924-0136/$ see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2004.04.065

surface is cooled more rapidly than the interior, which leads to a tensile stress at the surface and a compressive stress in the interior. This stress increases as the difference in the temperature increases between the surface and the interior. When the temperature difference reaches a maximum, the cooling rate in the interior becomes higher than at the surface and the stresses decrease, resulting in a tensile stress at the interior and a compressive stress at the surface [2]. Reducing the residual stresses can be achieved by reducing the temperature differential between the surface and the interior of a component. If the cooling rates during the quench can be reduced, then the high thermal conductivity of the alloy ensures that the temperature uniformity can be maintained between the interior and the surface. There are many factors relating to the quenchant that affect the cooling rate, such as the formation of a vapour blanket, the velocity, temperature, specic heat, heat of vaporisation, conductivity, density, viscosity and the wetting characteristics [3]. Alterations to the cooling rates during the quench can be achieved in many ways such as coating the component, adding polymers to the water or increasing the temperature of the water [4]. Care must be taken to ensure that the altered cooling rates do not adversely affect the mechanical properties of the alloy, particularly for quench sensitive alloys. An ideal quench will

G.P. Dolan, J.S. Robinson / Journal of Materials Processing Technology 153154 (2004) 346351

347

minimise residual stresses while maximising the mechanical properties of the alloy. To achieve this, an understanding of both residual stress formation and precipitation kinetics is required. At high temperatures, where residual stress formation begins, the cooling rates should be slow to minimise the thermal stresses. At intermediate temperatures, where the cooling rate has a signicant impact on the mechanical properties of the alloy, the cooling rate should be high to ensure a supersaturated solid solution is retained, and at lower temperatures the cooling rate can be reduced as it generally has little impact on the nal mechanical properties of the alloy [4]. When altering the cooling rates from that of a fast quench, it is important to be able to predict what the nal mechanical properties of the alloy will be. The best-known method was developed originally by Fink and Willey [5] and improved upon by Evancho and Staley [6]. Fink and Willey constructed time temperature property curves (TTP) C-curves, which are plots that show the times required to precipitate sufcient alloy content to change the strength by a certain amount. The characteristic shape of the C-curve can be explained as follows; at temperatures just below the solution heat treatment temperature it takes a considerable amount of time for precipitation to occur because the driving force for transformation is low due to small undercooling. With increased undercooling, the driving force for precipitation is high, and the time required for precipitation is low. This temperature region is commonly known as the critical temperature range as the time spent in this region will have a major impact on the mechanical properties of the alloy. At lower temperatures, the time required for precipitation increases again as the driving force for precipitation is high but the solid-state diffusion rates are low [7]. Initially Fink and Willey used average cooling rates through the critical temperature range to predict the mechanical properties and corrosion resistance of 7075-T6 [5]. This approach was less successful at predicting properties if the cooling rates varied during the quench. Evancho and Staley [6] managed to improve the accuracy of the predictions by using quench factor analysis as described below. The properties of heat-treatable aluminium alloys are dependant of the amount of precipitation that occurs during the quench. Precipitation kinetics for continuous cooling can be described by the equation = 1 exp(k1 )
tf t0

The TTP C-curve can be described by an equation of the form: C(T) = k1 k2 exp
2 k3 k4 k5 exp RT RT(k4 T)2

(3)

where C(T) is the critical time required to precipitate a constant amount of solute (s); k1 the constant which equals the natural logarithm of the fraction untransformed during quenching; k2 the constant related to the reciprocal of the number of nucleation sites (s); k3 the constant related to the energy required to form a nucleus (J mol1 ); k4 the constant related to the solvus temperature (K); k5 the constant related to the activation energy for diffusion (J mol1 ); R the gas constant (J mol1 K1 ); T the temperature (K). The quench factor is calculated by referring to a cooling curve and a TTP C-curve. An average temperature (T) between data points on the cooling curve is calculated and the C(T) value is then determined from the TTP C-curve. An incremental quench factor (q) is then calculated for this time step t, where t q= (4) C(T) This process is repeated through the critical temperature range and the incremental quench factors are progressively summed to yield the cumulative quench factor . The mechanical properties of the alloy can then be predicted from the quench factor using: y min max min = exp(k1 ) (5)

(1)

where is the fraction untransformed; k1 is a constant and = dt C(T) (2)

where t is the time (s); t0 the time at the start of the quench (s); tf the quench nish time (s) and C(T) is the critical time as a function of temperature; the loci of the critical times is the TTP C-curve; the quench factor.

where y is the predicted property; max the maximum achievable property after a fast quench; min the minimum achievable property. Low values of are indicative of fast quench rates with low amounts of precipitation occurring during cooling and good mechanical properties. High values of indicate slower quenches and lower mechanical properties. The k2 k5 constants are estimated using an iterative procedure. A large number of different quenches are conducted, and the cooling rates are recorded for each one. Quench factors are then calculated for each of the cooling curves and the property of interest is estimated using the calculated quench factors and Eq. (5). The sum of the squares of the differences between the estimated and the measured property is then determined. The values of the k2 k5 constants are then varied iteratively until the sum of the squares is minimised [8]. It should also be noted that there is a relationship between the quench factor and the size of the time intervals. It has been recommended [9] that the time step interval should be selected so that the average temperature drop should not be more than 25 C during the critical cooling range of the alloy to improve the accuracy of the predictions. C-curves are generated for specic alloys and tempers and cannot be interchanged as the ageing conditions

348

G.P. Dolan, J.S. Robinson / Journal of Materials Processing Technology 153154 (2004) 346351

can vary considerably from one temper to another and will affect the shape of the C-curve. Therefore it is necessary to generate C-curves for each alloy and temper, which can be a time consuming process. Another limitation is that the current classical quench factor analysis is that the accurate predictions are generally limited to above 8085% of the maximum attainable strength of the alloy [6]. Most commercial situations are mainly concerned only with a percentage loss in properties that are less than 15% so the current quench factor analysis model provides satisfactory predictions of properties. For most industrial processes it is possible to predict the nal mechanical properties of an alloy so long as the C-curve for the alloy and temper is available and the relevant cooling curves are known. However the main limitations to the more widespread use of the quench factor analysis has been the effort required in generating C-curves and the lack of C-curve data in the public domain. The aim of this paper is to generate time temperature property C-curves for the Vickers hardness for three aluminium alloys for specic tempers. With the critical temperature region identied for each alloy, it should be possible to reduce the residual stresses by slowly cooling from the SHT temperature to temperatures above the critical temperature region and then quenching into cold water. Since the time required for cooling would be less than that required for precipitation to occur, there should be minimal effects on the Vickers hardness of the alloys.

Fig. 1. Quenching to isothermal holding temperatures for determination of C-curves.

used, 7175-T73, 6061-T6 and 2017A-T4. The temperature during the quench and the isothermal hold was recorded using a 1.5 mm K-type thermocouple combined with an Eagle Electronics PC73C thermocouple PC card. A time interval of 0.2 s was selected for recording the cooling curve. The hardness was then measured using an Instron Wolpert Testor 930/250 (HV20) calibrated with a test block to the requirements of ASTM E92-92. To determine the effect of the alternative quenching technique on the Vickers hardness, specimens were solution heat treated at their respective SHT temperatures for 40 min and slowly cooled at a rate of 1 C/min to predetermined temperatures, quenched into water and then aged. 2.2. Residual stress reduction

2. Experimental The nominal chemical composition of the alloys used in this report can be seen in Table 1. 2.1. C-curve generation Small test pieces with geometry of 25 mm 25 mm 4 mm were used to determine the Vickers hardness C-curves. These specimens were solution heat treated in a carbolite air-recirculating furnace for 40 min. The solution heat treatment (SHT) temperatures employed for each of the alloys were: 4755 C for 7175, 5305 C for 6061 and 5105 C for 2017A. They were then rapidly quenched into a Durferrit STA 35/60 salt bath and held isothermally at temperatures ranging from 190 to 440 C for set periods of time, quenched into water at room temperature, Fig. 1, and then aged. The tempers were selected to reect those that are commonly
Table 1 Chemical composition corresponding to aluminium alloy specications (wt.%) Alloy 7175 6061 2017A Si 0.15 max 0.400.8 0.200.8 Fe 0.20 max 0.7 max 0.7 max Cu 1.22.0 0.150.40 3.54.5 Mn 0.10 max 0.15 max 0.401.0 Mg 2.12.9 0.81.2 0.400.8 Cr 0.180.28 0.040.35 0.10 max Zn 5.16.1 0.25 max 0.25 max Ti 0.10 max 0.15 max 0.10 max

The geometry used for the residual stress measurements was, 25 mm 25 mm 160 mm for both 7175 and 2017A. The 6061 was received as 20 mm thick plate, therefore the specimens used were 25 mm 20 mm 160 mm. The specimens were solution heat treated for 40 min at the respective temperatures detailed previously and allowed to slowly cool at a rate of 1 C/min to predetermined temperatures and then quenched vertically into room temperature water. To ensure repeatability no agitation was used during the quench. The residual stresses were measured directly after quenching without ageing since the ageing process can lead to a reduction in the residual stresses [10,11]. The residual stresses were measured using a Philips Xpert X-ray diffractometer. Scan parameters were controlled using Phillips PC-APDW (Version 4.0c) software. The 2 values were chosen to encompass the Cu K doublet for the {4 2 2} plane: 136 < 2 < 139.5 .

G.P. Dolan, J.S. Robinson / Journal of Materials Processing Technology 153154 (2004) 346351

349

Table 3 Standard deviation between predicted and measured Vickers hardness Alloy 6061-T6 Vickers hardness (HV20) 2.7 7175-T73 3.6 2017A-T4 2.3

Fig. 2. C-curves representing 99.5% of maximum attainable Vickers hardness (HV20).

3. Results and discussion 3.1. C-curve generation Using the iterative non-linear tting procedure as described previously, the k2 k5 constants for the Vickers hardness were determined. For the purpose of these experiments the max value was taken as the Vickers hardness achieved after a rapid quench into room temperature water from the SHT temperature and aged. The min value was taken as the Vickers hardness after furnace cooling from the SHT temperature to room temperature, the annealed hardness of the alloy. The k2 k5 constants for the Vickers hardness are presented in Table 2. For 2017A-T4, isothermal holding temperatures below 265 C were not used in the construction of the C-curves. This is due to some articial ageing that was occurring at these temperatures during the isothermal holds, which lead to an increase in the Vickers hardness of the alloy with increasing holding time. Since the purpose of the C-curves is to model the reduction in mechanical properties with increasing quench factor, it was not possible to use this data in the analysis. The Vickers hardness C-curves for the three alloys can be seen in Fig. 2. As expected for three very different alloy systems, there is considerable difference between the C-curves for each alloy. The 7175-T73 would be considered to be the most
Table 2 C-curve constants for Vickers hardness Alloy Vickers hardness (HV20) k2 (s) 7175-T73 6061-T6 2017A-T4 7.6E10 5.1E08 6.8E21 k3 (J mol1 ) 412 978 5794 k4 (K) 750 822 900 k5 (J mol1 ) 112200 94182 206784

quench sensitive of the three alloys, followed by 6061-T6 and 2017A-T4. The standard error between the measured and the predicted properties for the specimens used in the construction of the C-curves can be seen in Table 3. The standard error for the hardness measurements is very low and corresponds well to the variation of three HV20 associated with measuring hardness. C-curve constants for hardness and strength are commonly interchanged for quench factor analysis, however care should be taken when doing this as although there is generally a good relationship between hardness and yield strength, it is not always a linear relationship [12,13]. Once the C-curves have been generated the critical temperature range, where precipitation occurs most rapidly, is easily identiable. Fig. 2 shows how the critical temperature range can vary from one alloy to another. For times less than 10 s, the critical temperature range for 7175-T73 is approximately 410210 C, for 6061-T6 it is 440220 C and for 2017A-T4 it is 410250 C. Time spent in this region must be minimised during the quench so that high mechanical properties can be achieved. Temperatures above the critical temperature range can also be identied where precipitation rates are low. In Fig. 2 it can be seen that for 7175-T73 it should take up to 3 h before precipitation occurs at 425 C. For the same amount of time, the temperature is approximately 470 C for 6061-T6 and for 450 C for 2017A-T4. Therefore, it should be possible to cool the alloys from their solution heat treatment temperatures to these temperatures without any reduction in the Vickers hardness of the alloys. Fig. 3 shows the effect of slowly cooling from the SHT temperature using a cooling rate of 1 C/min, quenching into cold water and ageing, on the Vickers hardness of the alloys. For 7175-T73, it is possible to slowly cool the alloy from a SHT temperature of 475 C to a temperature of 425 C before the Vickers hardness begins to decrease. For 2017A-T4, it is possible to slowly cool from 510 C to a temperature of 460 C before a reduction in the Vickers hardness is observed. This is exactly what you would expect when comparing the results with the C-curves. Unfortunately, for 6061 it was not possible to reduce the temperature at which quenching occurs signicantly before a reduction in the Vickers hardness was observed, 505 C compared to a SHT temperature of 530 C. The reduction in the Vickers hardness is higher than that which would be expected from the C-curve. It should have been be possible to cool to a temperature of

350

G.P. Dolan, J.S. Robinson / Journal of Materials Processing Technology 153154 (2004) 346351

Fig. 3. Effect of quenching temperature after slow cooling on Vickers hardness.

Fig. 4. Effect on residual stress with quenching temperature after slow cooling.

closer to 470 C before a reduction in the Vickers hardness was observed. It may be possible that the range of temperatures used in the construction of the C-curves was not large enough for 6061-T6. The upper limit on the isothermal holding temperature was 440 C, as this temperature approaches the upper safety temperature for the operation of the salt bath, so it is possible that the C-curve for 6061 needs to be revisited to take into account higher isothermal holding temperatures. While this will not affect the predictions for relatively rapid quenches, predictions resulting from slow cooling in the upper temperature regions will be affected. Quench factor analysis conducted on specimens that underwent the slow cooling of 1 C/min, quenched and aged, and are presented in Table 4. There is generally very good agreement between the predicted and the measured hardness of the alloys. However the accuracy of the predictions decreases as the cooling times increase. This may be a limitation of the quench factor technique used in this report. Much work has been done to try and improve the quench factor model [14] so that properties can be more accurately predicted, from high cooling rates that develop high strength levels to very slow cooling rates that develop annealed strength levels. Future work in this area will utilise this newer model to further improve the ability to predict the properties of the alloys.

3.2. Residual stress reduction As previously discussed, residual stresses can be a major problem for quench sensitive alloys. There are many ways of reducing these stresses, but there is usually some compromise in the mechanical properties of the alloy. Since the shape of the cooling curve in the upper temperature region has a signicant effect on the residual stresses reducing the temperature at which quenching begins should lead to a reduction in the residual stresses. The experimental results of varying the temperature at which quenching begins can be seen in Fig. 4. These results were generated using the same slow cooling rate of 1 C/min from the SHT temperature followed by a rapid quench into cold water. The residual stresses can be reduced if the temperature at which quenching occurs is decreased. Having already identied how far the alloys can be cooled before the Vickers hardness begins to decrease, Fig. 3, residual stresses were measured when quenched from these temperatures. For 7175-T73, a reduction in the Vickers hardness is observed when the alloy is quenched from temperatures below 425 C. The residual stresses generated when quenched from this temperature are approximately 15% lower than when quenched from the SHT temperature of 475 C. For 2017A-T4, the reduction in the Vickers hardness is observed when it is quenched at temperatures below 460 C, the corresponding residual stresses

Table 4 Measured and predicted Vickers hardness for specimens cooled at 1 C/min, quenched and aged (QT: quench temperature) QT ( C) 7175-T73 Measured (HV20) 475 450 425 400 172.6 172 172 158 Predicted (HV20) 172 172 172 150 530 505 480 455 QT ( C) 6061-T6 Measured (HV20) 120 120 116 95 Predicted (HV20) 120 120 120 120 510 485 460 435 QT ( C) 2017A-T4 Measured (HV20) 128.3 128.6 128.3 127 Predicted (HV20) 128 128 128 116

G.P. Dolan, J.S. Robinson / Journal of Materials Processing Technology 153154 (2004) 346351

351

when quenched from this temperature are approximately 13% lower than at the SHT temperature of 510 C. The Vickers hardness of 6061-T6 decreases when quenched from temperatures below 505 C, higher than that expected from the shape of the C-curve. Since the aim of this report was to show the reduction in the residual stresses, with no loss in the Vickers hardness, the corresponding reduction in the residual stresses achieved, when quenched from a temperature of 505 C, was approximately 5%. The reduction in the residual stresses was less for 6061 than for the other two alloys due to the smaller temperature reduction that was possible. The residual stresses generated for 6061 are also signicantly lower than those for the other two alloys. This can be attributed to a slightly smaller test block being used for the measurements and also due to the as quenched strength of 6061 compared to 7175 and 2017A. In the as quenched condition the Vickers hardness of 7175 and 2017A are 90HV20 and 85HV20, respectively, whilst the Vickers hardness of 6061 is only 50HV20. Since the residual stresses are related to the yield strength of the material, the residual stresses generated in 6061 during the quench will be signicantly lower than 7175 and 2017A. For large components where cracking during the quench is a problem, lowering the temperature at which the quench occurs should reduce the frequency of cracking due to the lower stresses that are generated during the quench while maintaining the mechanical properties. Future work will examine other methods of reducing the residual stresses when combined with this alternative quenching technique. One method that will be examined is increasing the temperature of the water used for quenching. Residual stresses can also be reduced through thermal relaxation [10], so the effect of the articial ageing process on the residual stresses for these alloys will also be examined.

abled residual stress reductions of close to 15% for 7175, 13% for 2017A and 5% for 6061 before a reduction in the Vickers hardness was observed. Quench factor analysis appears to accurately predict the Vickers hardness for relatively fast cooling rates however the current quench factor model is unable to accurately predict the Vickers hardness for very slow and long cooling rates. Future work will utilise the newer quench factor model to improve the accuracy of the predictions.

References
[1] J.S. Robinson, R.L. Cudd, D.A. Tanner, G.P. Dolan, Quench sensitivity and tensile property inhomogeneity in 7010 forgings, J. Mater. Process. Technol. 119 (2001) 261267. [2] P. Archambault, J.C. Chevrier, G. Beck, J. Bouvaist, Optimum quenching conditions for aluminium alloy castings, in: Proceedings of the 16th International Conference on Heat Treating, 1976, pp. 105109. [3] C.E. Bates, Selecting quenchants to maximise tensile properties and minimise distortion in aluminium parts, J. Heat Treat. 5 (1987) 27 40. [4] J.A. Nicol, E.D. Seaton, G.W. Kuhlman, H. Yu, R. Pishko, Advanced Materials and Processes, 40S-40V, 4/96. [5] W.L. Fink, L.A. Willey, Quenching of 75S aluminium alloy, Trans. AIME 175 (1948) 414427. [6] J.W. Evancho, J.T. Staley, Kinetics of precipitation in aluminum alloys during continuous cooling, Metall. Trans. 5 (1974) 4347. [7] D.S. MacKenzie, Proceedings of the 16th ASM Heat Treating Society Conference and Exposition, March 1996, pp. 213219. [8] L.K. Ives, L.J. Swartzendruber, W.J. Boettinger, M. Rosen, S.D. Ridder, F.S. Biancaniello, R.C. Reno, D.B. Ballard, R. Mehrabian, Processing/microstructure/property relationships in 2024 aluminium alloy plates, US Department of Commerce, National Bureau of Standards Technical Report NBSIR 83-2669, 1983. [9] G.E. Totten, G.M. Webster, C.E. Bates, Proceedings of the First International Non-Ferrous Processing and Technology Conference, March 1997, pp. 305313. [10] B. Cina, T. Kaatz, I. Eldror, The effect of heating shot peened sheets and thin plates of aluminium alloys, J. Mater. Sci. 25 (1990) 4101 4105. [11] J.S. Robinson, D.A. Tanner, Residual stress development and relief in high strength aluminium alloys using standard and retrogression thermal treatments, Mater. Sci. Technol. 19 (4) (2003) 512518. [12] P.A. Rometsch, M.J. Starink, P.J. Gregson, Improvements in quench factor modelling, Mater. Sci. Eng. A 339 (12) (2003) 2. [13] P.A. Rometsch, G.B. Schaffer, An age hardening model for Al7SiMg casting alloys, Mater. Sci. Eng. A 325 (12) (2002) 28. [14] M. Tiryakioglu, J.T. Staley, The use of TTP curves and quench factor analysis for property prediction in aluminum alloys, in: Proceedings of the Materials Solutions Conference, 2001, pp. 615.

4. Conclusions The C-curves for Vickers hardness have been constructed for 7175-T73, 6061-T6 and 2017A-T4. Using the C-curves the critical temperature range of each alloy was identied. It was then possible to slowly cool the alloys from the SHT temperature to temperatures above the critical temperature region and quench into water with minimal loss in the hardness of the alloy. The slow cooling prior to quenching en-

Potrebbero piacerti anche