AM

E 2

0 03

1 2th

Prediction of mechanical properties and microstructure distribution of quenched and tempered steel shaft
B. Smoljan Faculty of Engineering, Chair of Material Science and Engineering, Vukovarska 58, 51000 Rijeka, Croatia An investigation of modeling of the quenching of steel workpiece of complex form has done. An algorithm of computer simulation of transient temperature fields is based on finite volume method. Heat transfer coefficient and heat conductivity coefficient values involved in mathematical model have been obtained by the inversion method, i.e., by the calibration. An algorithm for prediction of specimen hardness is based on Jominy test results. Hardness in specimen points was calculated by the conversion of calculated characteristic cooling time for phase transformation t8/5 to hardness. Mechanical properties are predicted based on calculated hardness. The inversion method of phase portion estimation based on calculated hardness in the quenched steel has been established. The designed method has been used in computer simulation of phase portion and mechanical properties in quenched specimens of steel 41 Cr 4 (DIN). 1. INTRODUCTION Steel quenching could be defined as "cooling of steel workpieces at a rate faster than still air" [1]. The cooling rate has to be so fast that austenite mainly will be transformed within martensite and bainite ranges. Simulation of steel quenching is a complex problem, dealing with estimation of mechanical properties and microstructure, and dealing with evaluation of residual stresses and distortions after the quenching. Research of numerical simulation of hardening degree, i.e. hardness and microstructure distribution in quenched steel specimen is one of with high priority research in simulation of phenomena of steel quenching [2]. Simulation of phase transformation is in the root of the simulation steel quenching. A model of quenching would not be considered representative of the actual process if it does not incorporate the effects of phase transformations. Phase transformation modeling is one of the main challenges in modeling of heat treatment [1]. Simulation of anyone process can be made successfully only if all mechanisms of process are well known and if the appropriate mathematical methods are used. Unfortunately the mechanism of phase transformations is not fully understood and interactive influence of different elements, austenitizing temperature, etc. usually are not taken in account. The errors

lii

B. Smoljan

in phase transformation calculation could be extremely great if a model is based only on grain size of prior austenite, austenitizing temperature and elemental composition of steel. Phase transformation kinetic depends also on the degree of solution of the carbides and it cannot be accurately predicted only from elemental composition. The grain size at the austenitizing temperature must be known in calculation of phase transformation kinetic. Moreover, in practice numerous phase transformation calculations are based just on statistic correlation between chemical composition and final microstructure as result of quenching. The investigations of steel quenching suggests that choosing a proper representative of the cooling phenomenon, which is relevant for structural transformation is one of the most important factor for a good simulation of hardening [3] [4]. 2. NUMERICAL SIMULATION OF SPECIMEN COOLING Transient temperature field in an isotropic rigid body can be defined by 2-D final volume formulation. The control volume for a 2-D situation is shown in Figure 1.

Figure 1. Control volume for 2-D situation Discretization equations by finite volume method formulation, i.e., algebraic equation system is equal [5][6]:
T ij (
1 2 m=1

b(i,i+n)j+

bi(j,j+n)+bij )=

2 m=1

1 ( b(i,i+n)jT(i,i+n)j + bi(j,j+n)T 1 j+n) ) + bij T0 i(j, ij

m=1

i = 1,2...imax; j = 1,2... j max n = 3 - 2m

(1)

Prediction of mechanical properties and microstructure

liii

In equation (1) bij = ( ij cij Vij)/ t; b(i,i+n)j = W(i,i+n)j-1 and bi(j,j+n) = Wi(j,j+n)-1. Variable W(i,i+n)j is the thermal resistance between ij and i+n,j volume and variable Wi(j,j+n) is the thermal resistance between ij and i,j+n volume (n=1):

W (i,i+n)j =

1 F (i,i+n)j 1 F i(j, j+n)

l (i,i+n)j

l (i,i+n)j

ij
l i(j, j+n)

l (i+n,i)j

i+n, j
l i(j+n, j)

(2)

W i(j, j+n) =

ij
+

j+n, j

(3)

Thermal resistances for boundary volume are: W (i,i+n)j = 1

F (i,i+n)j
1

ij
l i(j, j+n)

1 Ts (i,i+n)j cos 1 Ts i(j, j +n) cos

(4)
r (i,i+n)j

W i(j, j+n) =

F i(j, j+n)

ij

(5)
z i(j, j +n)

were is heat conduction coefficient Wm-1K-1, is density in kgm-3, and c is specific heat is heat transfer coefficient in Wm-2K-1, Ts is coefficient of heat capacity in Jkg-1 K-1 and transfer at the boundary temperature which is equal to Ts. Solution of temperature field change is determined by solution of discretization system for any time step t. Time of cooling from Ta to some temperature in particular grid point is determined by the summation of steps tm.

t ij =

M m=1

tm

(6)

In this way the cooling curve, i.e., time - temperature couples (ti,j,Ti,j) in each one gridpoint of the specimen can be predicted.

3. MATERIAL DATA INVOLVED IN MATHEMATICAL MODEL OF TRANSIENT TEMPERATURE FIELD

In the transient temperature simulation field is determined by specific heat capacity (c), density ( ), heat conduction coefficient ( ) and heat transfer coefficient have to be known. Heat transfer data can be experimental evaluated or they can be predicted by inverse method, i.e. by calibrating [7][8]. Surface heat transfer coefficient evaluation is the most difficult of the evaluation physical properties used in the mathematical modeling of steel quenching. First, magnitude of heat transfer coefficient remarkable depends on surface temperature and variations of the conditions in the quenching bath and the state of a specimen surface. Specimen shape and size also affect heat transfer coefficient. In quenching, it is a rule that different locations in a specimen have different values of heat transfer coefficient.

liv

B. Smoljan

Variable for steel is equal ~7800 kgm-3 and accepted values of variable c are shown in Table 1. Table 1. Accepted specific heat capacity of steel 0 100 250 370 500 700 760 1000 Temperature, / C

Transformation

30.6 22.5 19.8 18.0 16.5 16.5 17.1 18.0 Austenite to martensite Specific heat capacity/ Austenite to (bainite 30.6 26.1 21.6 18.0 16.5 16.5 17.1 18.0 -1 -1 Jkg K +martensite), or bainite 30.6 28.8 25.2 21.6 16.7 16.5 17.1 18.0 Austenite to (pearlite + bainite), or pearlite

Using the Crafts -Lamont diagrams optimization procedure for calibration of heat conduction coefficient ( ) and heat transfer coefficient () was done [8]. By varying both, values of heat transfer coefficient and heat conduction coefficient the calibration was provided. The time t8/5 distribution in cylindrical quenched steel specimen was predicted by the computer simulation. Time t8/5 vas calculated for large spectra of a specimen bar diameter (D=2R) at position of cylindrical specimens equal to r/R=0, r/R=0.5 and r/R=0.9. Corresponding distance from water - cooled end of Jominy specimen to time of cooling t8/5 was predicted by the diagram shown in figure 2. The optimal values of both heat transfer and heat conduction coefficients have been estimated by the comparison of estimated distance from water - cooled end of Jominy specimen predicted by computer simulation and by Crafts-Lamont diagrams. Values of and were accepted when relative differences between distance from water-quenched end estimated by computer simulation and by Crafts -Lamont diagrams were negligible. Accepted values of are presented in Table 2. Table 2. Calibrated values of heat conductivity coefficient of carbon and low alloyed steel Temperature, 0 100 250 370 500 700 760 1000 Transformation /C 30.6 22.5 19.8 18.0 16.5 16.5 17.1 18.0 Heat conduction coefficient, / Wm-1K-1 30.6 26.1 21.6 18.0 16.5 16.5 17.1 18.0 30.6 28.8 25.2 21.6 16.7 16.5 17.1 18.0 Austenite to martensite Austenite to (bainite +martensite), or bainite Austenite to (pearlite + bainite), or pearlite

Calibrated values of heat transfer coefficient ( ) of water with severity of quenching, i.e., Grossmans to H-value equal 1.4 vs. surface temperature is shown in Table 3.

Prediction of mechanical properties and microstructure

lv

Table 3. Calibrated values of heat transfer coefficient vs. surface temperature for quenching medium with severity of quenching H=0.45 Temperature 0 360 440 580 900 /C Heat transfer coefficient / Wm-1K-1 610 800 2830 790 610

4. PREDICTION OF HARDNESS AND MICROSTRUCTURE DISTRIBUTION PHASE PORTION

One of the most important factors for efficient simulation of hardening is the good selection of representative of the cooling phenomena that is relevant for phase transformation [8]. Characteristic cooling time, relevant for phase transformation in most structural steels is the time of cooling from 800 to 500 C (time t8/5) [4]. Everyone location of Jominy-specimen and simulated specimen has one characteristic time t8/5 [4]. The diagram of distance from the quenched end of Jomini-specimen (Jominy-distance) vs. cooling time t8/5 is shown in Figure 2. The hardness at grid-points can be estimated by the conversion of cooling time t8/5 results to hardness by using both, the relation between cooling time and distance from the quenched end of Jominy specimen and the Jominy hardenability curve.

Figure 2. Distance from the quenched end of Jominy-specimen vs. cooling time from 800 to 500 C Mechanical properties of quenched steel, and also quenched and tempered steel, directly depend on degree of quenched steel hardening [10]. Relation between hardness HV and ultimate tensile stress Rm is equal:

lvi
Rm= 3.3 HV, Nmm-2

B. Smoljan (7)

Yield strength Rp0.2, specific elongation A5, reduction of area Z, toughness Au could be estimated from the ultimate tensile stress Rm or hardness HV [6]: Re=Rp0,2=(0.8+0.1)Rm+170C-200, Nmm-2 Z=96-(0.062-0.029C)Rm, A5=46-(0.04-0-012C)Rm, Au =460-(0.59-0.29C)Rm, % % J (8) (9) (10) (11)

were C is a ratio between the actual hardness and martensite hardness in HRC. Structure composition of steel cooling depends of actual steel hardness. It can be written that the steel hardness generally is equal: HV=((% ferrite)HV(F)+(% pearlite)HV(P)+(% bainite)HV(B)+( % martensite)HV(M))/100 (12) Amount of phases portion is equal unity: ((% ferrite + % pearlite) + % bainite + % martensite) / 100 = 1 (13)

By the equations (12) and (13) is not difficult to predict phase fractions if the hardness (HV) of cooling microstructure is known and hardness of microstructure constituents separately is known. Results of austenite decomposition are depending on the chemical composition of steel, severity of cooling, austenitizing temperature and steel history. The austenite decomposition results can be estimated based on time, relevant for structure transformation. The characteristic cooling time, relevant for structure transformation for most structural steels is the time t8/5 (Figure 2) [4]. If other heat treatment parameters are constant, the austenite decomposition results in some location of a cooled specimen will depend only of the time t8/5. It could be written for Jominy test that phase hardness depends of chemical composition (CC) and cooling rate parameter (CRP) that corresponds to actual distance (d) of Jominy specimen quenched end. It was adopted that CRP = log t 8d/ 5 .
M HVdM = f M (CC , CRP ) = HVmax K M log B HVdB = f B (CC , CRP ) = HVmax K B log

t8M5d / t8M5 max / t8B/ 5d ;

(14) (15) (16)

t8B/ 5 max

P HVdP + F = f P + F (CC , CRP ) = HVN + F + K P + F log

t8P/+5FN ; t8P/+5Fd

Prediction of mechanical properties and microstructure

lvii

where N is normalizing, Bmax is lower bainite. Characteristic value of HV, K and t8/5 in equations (14), (15) and (16) has to be evaluated based on chemical composition for investigated steel combined by Jominy test results. Hardness of quenched structures with characteristic percentage of martensite can be predicted by using the diagram of hardness at different percentages of martensite vs. carbon content after Hodge and Orehoski [10] and Jominy curve, but the influence of chemical composition of steel has to be taken in account. The regression relations between the cooling time from 800 to 500C for cooling structures of 100%, 50%, 10% and 0% pearlite are established.

log

100%(P t8 / 5 x%(P t8 / 5

+ F) + F)

= f (CC , T a , t a ) ;

(17)

where (x) is pearlite contents in microstructure, Ta is austenitizing temperature in K; ta is austenitizing time in h; Characteristic Jominy distances for characteristic time t8/5 is estimated using the relation between cooling time and distance from the quenched end of Jominy specimen shown in Figure 2.

5. APPLICATION
The presented mathematical model of steel workpiece quenching has been applied in computer simulation of the mechanical properties and microstructure of a quenched shaft. The shaft is shown in Figure 3.

Figure 3. Shaft

lviii

B. Smoljan

The shaft was made of steel 41 Cr 4 (DIN). The Jominy-test data of investigated steels are done in Table 4. Table 4. Jominy-test data Distance, mm Hardness HRC 1.5 3 5 8 11 15 20 25 30 35 40 80

56

55

55

53

52

45

38

35

32

31

30

22

Heat treatment for quenching was heating to 850 C for 30 min and water quenching with slow agitation. Shaft was quenched in agitated oil with the severity of quenching, i.e., Grossmanns Hvalue equal to 0.45. The calibrated values of heat transfer coefficient vs. surface temperature of water with severity of quenching H=0.45 are shown in Table 3. Temperature of tempering was equal to 600 oC. The phase portion distribution in shaft of steel 41 Cr 4 (DIN) are estimated by computer simulation. Elemental composition of investigated steel was 0.41 % C, 0.28% Si, 0.68% Mn, 1.06 %Cr. Distribution of properties and microstructure fields of the quenched and tempered shaft is presented in Figure 4.

E B

Figure 4. Distribution of properties of quenched shaft of steel 41Cr4 In Table 5 and Table 6 the characteristic values of mechanical properties and microstructures portion are shown.

Prediction of mechanical properties and microstructure

lix

Table 5. Predicted properties of the quenched and tempered shaft of steel 41Cr 4 Position Hardness HRC As-quenched A B C D E F G 24 28 34 38 44 48 54 Quenched and tempered 17 20 23 24 26.5 28 32 Rm Nmm 683 737 797 818 861 887 997
-2

Rp0,2 Nmm-2 418 476 540 562 610 639 752

A5 % 21 19 18 17 16 15 12

Z % 59 57 56 55 54 53 50

Au J 117 101 84 78 69 65 37

When the hardness in quenched specimen points is known and when the phase distribution vs. Jominy distances is known it is not difficult to predict the phase distribution in quenched steel specimen (Table 6). Table 6 Simulated structure composition of quenched shaft Position As-quenched Portion of as-quenched microstructure hardness HRC A B C D E F G 24 28 34 38 44 48 54 100 % pearlite 4 % martensite+ 26 % bainite+ 70 % pearlite 42 % martensite+ 56 % bainite+ 2 % pearlite 49% martensite + 51% bainite 71 % martensite + 29 % bainite 98 % martensite + 2 % bainite 100 % martensite

6. CONCLUSIONS
A mathematical model of steel quenching has been developed to predict the distribution of mechanical properties in a specimen with complex geometry. The model is based on the finite volume method. The numerical simulation of quenching is consisted of numerical simulation of temperature transient field of cooling process and of numerical simulation of hardening. The characteristic cooling time, relevant for structure transformation for most structural steels is the time of cooling from 800 to 500 C (time t8/5). The austenite decomposition

lx

B. Smoljan

results were estimated based on time t8/5, relevant for structure transformation. The time t8/5 steel specimen was estimated by using the calibrated values of heat transfer coefficient. Hardness in specimen points was estimated on the basis of the time t8/5, i.e., by the conversion of mentioned specific time to hardness results. Mechanical properties are calculated based on hardness results. The inversion method of computer simulation of austenite decomposition and evaluation of quenched phase portion was established. For the calculation of microstructure composition, the hardness in specimen points, Jominy test results and chemical composition of steel must be known. The established mathematical model was applied in computer simulation of microstructure transformation in steel shaft. It can be concluded, that by proposed method mechanical properties and microstructure composition in quenched steel specimen can by successfully calculated.

REFERENCES
1. Theory and Technology of Quenching, eds B. Lii i , H. Tensi and W. Luty, SpringerVerlag, 1992. 2. E. Just, Vergten - Werkstoffbeenflussung durch Hrten und Anlassen, VDI - Bericht, No.256, pp.124-140, 1976. 3. B. Li i et al, Workshop Designed System for Quenching Intensity Evaluation and Calculation of Heat Transfer Data, Proc. of the 1st Inter. Conf. on Quenching & Control of Distortion, Chicago, Vol. 1, pp. 17-26, 1992. 4. A. Rose et al, Atlas zur Wrmebehandlung der Sthle I, Verlag Stahleisen, Dsseldorf, 1958. 5. S. Patankar, Numerical Heat Transfer and Fluid Flow, McGrawHill Book Company, New York, 1980. 6. B. Smoljan, Numerical Simulation of Steel Quenching, Journal of Materials Engineering and Performance, Vol. 11(1), (2002), pp. 75-80. 7. B. Li i et al, Workshop Designed System for Quenching Intensity Evaluation and Calculation of Heat Transfer Data, Proc. of the 1st Inter. Conf. on Quenching & Control of Distortion, Chicago, Vol. 1, pp. 17-26,1992. 8. B. Smoljan, The Calibration of the Mathematical Model of Steel Quenching, 5th World Seminar on Heat Treatment and Surface Engineering, Isfahan, Ed. M. Salehi, ISSST and IFHT, Vol. 1, pp. 709-715, 1995. 9. B. Li i , G. Totten, Controllable Delayed Quenching, Proc. of the Inter. Heat Treating Conf., Schaumburg, pp. 253-262, 1994. 10. S. Hoyt, Metal Data, Columbus, Ohio, 1952.