Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Chapter 5 Heat Conduction and Heat Transfer 5.1 Overview Cooling in the heat treatment is an important process for obtaining the required microstructure in steels. Different quenchants and its agitation methods for the process induce significant differences in temperature distributions in parts. Cooling rates, which differs according to locations, affects the phase transformation behavior and then contribute significantly to the final phases in steels. In addition, stress and strains in the parts, which are induced mainly by phase transformation and thermal expansions during cooling, finally produce distortions and residual stresses after the cycle. Heat conduction phenomena in parts have been expressed by an equation for achieving the heat treatment simulation. The equation considers characteristics of thermal properties in materials and heat transfer at surfaces, and also the effect of heat generation due to phase transformations, which depend on temperature. In this chapter, numerical models and characteristics for the heat transfer phenomena are described. Furthermore, the heat recovery and inverse hardening phenomena in steel cylinder are mentioned, which are produced by contributions of heat generations due to phase transformations.
5.2 Heat conduction equation in solid 5.2.1 Relation between heat flux and temperature The heat transfer phenomena can be descried analogically by the model for the diffusion phenomena, which is mentioned in the previous chapter. Historically, it is known Fick referred the Fourier’s law of heat conduction (Fourier, 1822; Jakob, 1949 & 1957, McAdams, 1954; Carslaw and Jaeger, 1959; Gröber, Erk and Griugull, 1961; Katto, 1964; Shoji, 1995; Lienhard and Lienhard, 2011) when formulating the diffusion phenomena (Koiwa, 1998).
qx
q x + dx
y
A
x z dx
Fig. 5.2.1 Heat conduction in infinitesimal volume of solid. As in the case of the diffusion phenomena, we consider an infinitesimal rectangle volume as shown in Fig. 5.2.1 in a solid. Here, the heat flux qx along the x axis direction at the point A, which is identified by a coordinate value x, in the rectangular volume, is assumed to be proportional to the gradient of the temperature T at the point. Then the Fourier’s first law is derived as follows: 1
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
qx = − k
∂T ∂x
(5.2.1)
where, k, T, and qx are heat conductivity, temperature and heat flux, respectively. Since the heat flux qx is the amount of heat passing through per unit area and time, and then its unit can be expressed as J/(m2s). When the units of T and ∂T/∂x are specified as °C and °C/m, respectively, the thermal conductivity k is measured in W/(m °C) based on Eq. (5.2.1). On the other hand, the temperature change at any point in a solid can be represented by the following partial differential equations.
ρC
∂T ∂ ∂T = k ∂t ∂x ∂x
(5.2.2)
The above relation is called as the Fourier's second law, and ρ and C are density, kg/m3, and specific heat, kJ/kg/°C, respectively, which can be derived based on the heat balance in the rectangle shown in Fig. 5.2.1. The thermal diffusivity, a=k/(ρC), m2/s, can be found in Eq. (5.2.2), which corresponds to the diffusivity in the diffusion phenomena in Chapter 4. During heat treatment processes, heat generations occur in steel parts by plastic deformation works and phase transformations, and also the Joule heating due to electric currents. Now, considering the above heat generations and heat flows along the y and z axis to Eq. (5.2.2), a general heat conduction equation is derived as follows:
ρC
∂T ∂ ∂T = k ∂t ∂x ∂x
∂ ∂T + k ∂y ∂y
∂ ∂T + k ∂z ∂z
W J L +Q +Q +Q
(5.2.3)
where Q W , Q J and Q L are the heat generation rates due to plastic deformation works, Joule heating and phase transformations, respectively. To solve the heat conduction equation for practical problems is needed to specify the initial and boundary conditions described in the next section. In addition, analyses of the phenomena in complex shapes are performed based on the finite element method described in Chapter 9. A significant fraction of the plastic strain energy caused by the work is converted to heat (Elam, 1935). For simplicity, assuming this conversion is completely done, the heat generation rate Q W due to the plastic deformation works is descried as follows (Mendelson, 1968):
Q W = σ ε P
(5.2.4)
where σ and ε P are the effective stress and the effective plastic strain rate, respectively. The unit of plastic deformation work is W/m3, when the units of stress and strain rate are Pa (= N/m2) and 1/s, respectively. The Joule heating is generated, for examples, by eddy currents during induction heating processes. Its rate Q J is derived by specifying the current density i and the electrical resistivity ρ as follows (Kinbara, 1972): 2
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Q J = ρ i 2
(5.2.5)
where the unit of the Joule heating is W/m3, when using units of the current density and the electrical resistivity are A/m2 and Ωm, respectively. The theory of high frequency induction heating is described in Chapter 7. In addition, the heat generation rate Q L due to phase transformations is determined as follows (Agarwal and Brimacombe, 1981):
Q L = ∑ LIJ ξIJ
(5.2.6)
where LIJ is the total amount of heat generation due to the phase transformation from phase I to J. In addition, ξIJ is the change rate of phase volume fraction due to the phase transformation from phase I to J, which is called as the transformation rate. The unit of the heat generation is W/m3 when using the units of total heat generation due to phase transformations and transformation rate are J/m3 and 1/s, respectively. If the unit of heat generation due to phase transformations is measured in the heat per unit mass, for example, J/kg, LIJ in Eq. (5.2.6) is obtained by converting of such units by multiplying the density of material to the above amount. 5.2.2 Initial and boundary conditions For solving heat conduction phenomena in practical parts, it is necessary to specify initial and boundary conditions for the heat conduction equation. As for a typical initial condition, a uniform temperature is specified throughout a solid at the starting of a cycle of heat treatment. As for boundary conditions, when temperatures on a surface are known as Ts, it is possible to directly specify the temperature as follows:
(T ) x=0 = TS
(5.2.7)
In the above equation, one-dimensional problem along only the x axis is used as an example for convenience, and x=0 corresponds to the surface. This is the same to different boundary conditions below. On the other hand, heat flux qx=0 at the surface may be specified directly as boundary conditions as follows:
∂T −k = qx = 0 ∂x x =0
(5.2.8)
However, a model of heat transfer boundary can often represent more realistic conditions, which is shown as follows:
qx =0 = α (Te − Ts )
(5.2.9)
where α is the heat transfer coefficient, while Te and Ts are the temperatures in an environment and at the solid surface. Unit of the heat transfer coefficient is W/(m2 °C), when heat flux qx=0 is J/(m2 s) and temperature T 3
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
is °C. In the heat treatment simulation, the heat transfer coefficient for cooling into quenchants depends commonly on surface temperatures of heat treated parts.
5.3 Thermal properties of steels The heat conduction analysis operated in the heat treatment simulation needs data of specific heat, enthalpy of transformations, thermal conductivity and density of steels. These thermal properties have been measured by an inherent method of each individual scientist through the ages; especially a considerable data has been accumulated for steels. Here, specific heat, thermal conductivity and enthalpy of transformation in pure iron, carbon steels and alloy steels are introduced from literatures, and then their adequacy are investigated for applying them to the current heat treatment simulation. The density is discussed in the next chapter. 5.3.1 Specific heat First of all, it should be confirmed what the specific heat is. If heat ∆Q is needed when changing the temperature ∆T in an object with mass M, and the specific heat C is expressed as follows:
C=
1 ∆Q M ∆T
(5.3.1)
For the unit of the specific heat, kJ/(kg °C) is often used. On the other hand, the specific heat in a mole of substance is measured in the unit of J/(mol °C). When a specific heat is derived based on Eq. (5.3.1), the temperature of a solid should be uniform. It is difficult to achieve exactly under the above condition in measurements; therefore, the obtained data would contain errors. In addition, the specific heat of solid is measured usually in constant pressure. A theoretical estimation of the specific heat of metal has been attempted to divide it to vibrational, electronic and magnetic effects (Nishizawa, 1973). The above effects on specific heat can correlate with crystal lattice vibrations, electronic excitation of free electron, and atomic spin equilibrium in ferromagnetic alloy. Even though the theoretical study is effective to understand this kind of phenomenon, specific heat data for individual substances are eventually confirmed by measurements. For the specific heat of pure iron, measured data in the 1930s have been assessed by Austin (1932). Subsequently, Griffiths-Awbery (1940) and Pallister (1949) reported specific heat data measured using the Sykes (1936) method and the electric current heating method, respectively. In more recent years, Guillermet and Gustafson (1985) obtained specific heat curves by the thermodynamic assessment based on past measured data. Figure 5.3.1 was made by the author to compare the above four kinds of specific heat data in pure iron. As shown in Fig.5.3.1, a peak appears in each specific heat curve in the range of from 600 to 800 °C, which is induced by magnetic transformations. As for the peak values, the measurement by Awbery and Griffiths agrees well with Guillermet and Gustafson’s. However, these values are quite smaller than the measured data by Pallister and Austin. In the lower range of temperature than the peaks, there are small 4
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
discrepancies in between the curves derived by the individual researchers. On the other hand, there are larger differences between the values in the austenite region.
1.4
1.4 Guillermet(1985) Austin(1932) Awbery(1940) Pallister(1949)
1.2 1.1 1 0.9 0.8 0.7 0.6 0.5
1.2 1.1 1 0.9 0.8 0.7 0.6 0.5
0.4
0.4
0
200
400
600
800
1000
1200
1400
400
500
600
700
800
900
Temperature, °C
Temperature, °C
Fig. 5.3.1 Specific heat of pure irons.
Fig. 5.3.2 Specific heat of eutectoid steels. 1
1 Pure Fe 0.06%C 0.23%C 0.4%C 0.8%C 1.22%C
0.9 0.8 0.7
Specific heat, kJ/kg/°C
Specific heat, kJ/kg/°C
Kramer; pearlite Kramer; austenite Hagel; pearlite Hagel; austenite Awbery Pallister
1.3
Specific heat, kJ/kg/°C
Specific heat, kJ/kg/°C
1.3
0.6 0.5 0.4
Pure Fe 0.06%C 0.23%C 0.4%C 0.8%C 1.22%C Pure Fe by Guillermet
0.9 0.8 0.7 0.6 0.5 0.4
0
100
200
300
400
500
Temperature, °C
600
700
800
900
1000
1100
1200
Temperature, °C
(a) Lower than 700°C. (b) Higher than 800°C. Fig. 5.3.3 Specific heat of carbon steels. 0.9
1.5%Mn 3.5%Ni 3.5%Ni, 1%Cr 1%Cr 2%Si, 1%Mn Pure Fe
0.8 0.7
Specific heat, kJ/kg/°C
Specific heat, kJ/kg/°C
0.9
0.6 0.5 0.4
1.5%Mn 3.5%Ni 3.5%Ni, 1%Cr 1%Cr 2%Si, 1%Mn Pure Fe by Guillermet
0.8 0.7 0.6 0.5 0.4
0
100
200
300
400
500
600
Temperature, °C
800
900
1000
1100
1200
Temperature, °C
(a) Lower than 600°C. (b) Higher than 800°C. Fig. 5.3.4 Specific heat of low alloy steels. For comparing the specific heat of different eutectoid steels, the author plotted the data by Awbery and Snow (1939), Hagel et al. (1956) and Kramer et al. (1958) as shown in Fig. 5.3.2. In the temperature range 5
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
of pearlite, there are discrepancies in between data measured by Hagel et al. and Kramer et al. using the Smith (1940) method, and by Awbery and Snow using the Sykes (1936) method. Especially, in the range from 600 to 700 °C, data by Awbery and Snow shows a lower value than others. Meanwhile, in the austenite region, a discrepancy in between data by Pallister and Kramer et al. is small. However, there is a discrepancy in between data by Kramer et al. and Hagel et al., both which used the same Smith method. Hagel et al. and Kramer et al. measured only for the eutectoid steels, while Awbery and Snow (1939) and Pallister (1946) broadened the range to the data for carbon steels and other alloy steels. The author depicted specific heat curves by selecting the carbon steel data up to 700 °C from the data by Awbery and Snow as shown in Fig. 5.3.3 (a). It is difficult to find some regularity of the carbon concentration dependency in the specific heat from this figure. The data for pure iron in this figure was obtained by Griffiths and Awbery (1940). Fig. 5.3.3 (b) was created for comparing the specific heat of different carbon steels in the austenite region. The specific heat curves included here were measured by Pallister(1946), except the data of pure iron by Gustafson and Guillermet (1985). As already mentioned, the specific heat curves of pure iron by Guillermet and Gustafson differ from Pallister’s. However, the specific heat curves of carbon steels becomes almost horizontal and their discrepancies are within the range of 0.05 kJ/(kg °C). On the other hand, the specific heat for various low-alloy steel up to 600 °C, which was measured by Awbery and Challoner (1946), were compared by the author as shown in Fig 5.3.4 (a). Carbon concentration of the steels in the figure is in the range of about 0.2 to 0.5% C. The Difference between the specific heat curves for the steels is 0.05 kJ/ (kg °C), therefore, carbon may not so affect to the specific heat when Si, Mn, Cr and Ni are within the range shown in the figure. Curves of the low-alloy steels are all located above the pure iron, which was drawn for comparison. In addition, the specific heat curves of alloy steels in the austenite region show substantially a horizontal distribution as depicted in Fig. 5.3.4 (b), which are within the range of about 0.005 kJ/(kg °C). To investigate the effect of alloying elements on the specific heat, Pallister (1946) plotted data of the specific heat in a variety of carbon steels and alloy steels at 1250 °C, specifying the number of atoms contained in 100 g of steel as the horizontal axis. As a result, he found a tendency that specific heat was raised with increasing number of atoms. For example, specific heat is higher in the steel which includes many lighter elements such as carbon; reversely steels including a few heavy elements such as tungsten make it lower. However, this trend appeared in the austenite region. The specific heat data measured by Awbery and Snow (1939), Awbery and Challoner (1946) and Pallister (1946) were published as the data book from British Iron and Steel Research Association (BISRA) (1953). Now, specific heat of steels with given chemical compositions can be predicted by thermodynamic software (Saunders and Miodownik, 1998). In addition, the prediction by such software needs the database of the free energy for each element and the interaction parameters between elements. Specific heat data of pure iron by Guillermet and Gustafson has been adopted by such a database (Dinsdale, 1991). In addition, Miettinen (1997) reported on the prediction method of specific heat based on the somewhat simplified thermodynamic approach. 6
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
5.3.2 Enthalpy of phase transformation Phase transformations induce heat absorption or release in materials, which is called as the latent heat due to phase transformation. The latent heat produces a discontinuity in an enthalpy temperature curve at the phase transformation point. Then, the latent heat has been referred to as the enthalpy of phase transformation in recent years (Nishizawa, 2008), also this text follows this terminology. In addition, the expression, the heat generated by the phase transformation, is used at some locations in the text. Gentle heating or cooling of specimens across the A3 point in pure irons or the A1 point in eutectoid steels produce a stationary point in temperature changes by the heat generated by phase transformations. On the other hand, there is no stationary point even in eutectoid steel in practical cooling processes, such as quenching, and then its pearlite transformation occurs over a temperature range of the process. In such a case, the phase transformation generates heat continuously in the temperature range. Enthalpy of transformation at the A3 point in pure iron has been measured, and their old results prior to the 1930s were introduced by Austin (1932) in his review. The value of about 16 kJ/kg for the enthalpy of transformation is observed in his report at that time, which is close to the value of 18.1 kJ/kg obtained by Gustafson and Guillermet (1985) in recent years, using assessments based on the previous experimental data. In contrast, Tajima (1998) reported the values of 16 and 19 kJ/kg, during for the heating and cooling processes, respectively, using the differential scanning calorimetry (DSC) system. As for the enthalpy of transformation for diffusion phase transformations in carbon steels, the author knows only the measurements by Snow and Awbery (1939), and Tajima (1998) for several carbon steels, and Hagel et al. (1956) and Kramer et al. (1958) for eutectoid steels. For comparing these enthalpy data, Fig. 5.3.5(a) was drawn by the author. Tajima used the DSC for his measurements. Meanwhile, Snow and Awbery (1939) recorded a total change in the amount of heat per unit weight for specimens, and then the difference of the total amount between before and after phase transformation was plotted as the enthalpy
90 80 70 60 50 40 30 20 10 0
Enthalpy change kJ/kg
Enthalpy change kJ/kg
change in Fig. 5.3.5(a).
Tajima Awbery Hagel Kramer 0
0.2
0.4
0.6
0.8
1
90 80 70 60 50 40
Tajima Krielaart
30 20 10 0 0
Carbon content, %C
0.2
0.4
0.6
0.8
1
Carbon content, %
(a) Phase transformation.
(b) Phase and magnetic transformations.
Fig. 5.3.5 Enthalpy changes due to phase and magnetic transformations. The enthalpy of transformation in eutectoid steel measured by Hagel et al. (1956) and Kramer et al. 7
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
(1958) were plotted in Fig. 5.3.5(a), which show a somewhat larger value than Awbery and Snow’s. In addition, these values measured by Hagel et al. and Kramer at al. are 82.1 and 77.0 kJ/kg, respectively. The reason why this difference in the measurements arises may be that Hagel et al. and Kramer et al. used the Smith method, which may be more accurate than the method used by Awbery and Snow. On the other hand, the data measured by Tajima or Awbery et al. showed almost a linear relation between the enthalpy of transformation and carbon concentration. Krielaart et al. (1996) also measured the enthalpy of transformation to the carbon steels, 0.17, 0.36, 0.57 and 0.8 %C, using DSC. However, this data contains both the effects of the phase transformation and the magnetic transformation. Since Tajima (1998) also reported the same enthalpy data, the author plotted them for comparison as shown in Fig. 5.3.5(b). This figure shows that the measurements by Tajima and Krielaart agree well in the vicinity of the eutectoid point, however their differences increase in the range of lower carbon concentration. As for the martensitic transformation, reports on the enthalpy of transformation are limited. However, the driving force of this phase transformation, as already mentioned in Chapter 3, had been estimated by the thermodynamic-based approach. For example, using such a method, Cohen and Kaufman (1956) predicted the enthalpy of the martensitic phase transformation in Fe-Ni alloys. A good agreement between these estimations and experimental results was confirmed by Normann and Scheil (1959). In addition, the enthalpy of transformation in martensitic transformation appears as a change in the cooling curve, which was used to measure the Ms as already mentioned in Chapter 3. For considering effects of alloy components to the enthalpy of transformations, its experimental data has not been reported so much. Here, the author shows only the measured values of the enthalpy, 95.9, 75.4 and 74.1 kJ/kg for the eutectoid steel of 1.91 %Co, 1.85 %Mn and 0.51 %Mo, respectively, by Hagel (1956). Since the measured value for eutectoid steel containing only carbon is 82.1 kJ/kg, ratios of the above eutectoid steels are 1.2, 0.92 and 0.90, respectively. Not only specific heat but also enthalpy of transformation of steels with given chemical composition now can be predicted by thermodynamic software (Saunders and Miodownik, 1998). On the other hand, Miettinen (1997) reported on the prediction of the enthalpy of phase transition using his simplified thermodynamic approach. 5.3.3 Thermal conductivity Powell (1939) assessed data on the thermal and electrical conductivities in steels, which was experimented from the 1900s to the 1930s. Based on his results, Powell and Hickman (1939, 1946a) measured the electrical resistivity in 22 different grades of carbon steels and alloy steels, in the range from 0 to 1300 °C. At that time, in the range below 300 °C, they measured not only the electrical resistivity but also thermal conductivity (Powell and Hickman, 1939). In addition, measurements of the thermal conductivity of the 13 %Mn steel (Powell, 1946a) and the 0.8 %C carbon steel (Powell and Hickman, 1946b) were reported in the range from 0 to 850 or 1000 °C, respectively. For obtaining the thermal conductivity, cylindrical and hollow cylindrical probes were used for measuring steady state temperature difference and heat flow along 8
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
the longitudinal or radial direction, respectively. The above measurements were applied to the heat conduction analysis, and then the data of the heat conductivity were obtained. Meanwhile, the electrical resistivity was calculated using measured longitudinal voltage drop in the cylinder induced an electric current.
40
0.06%C 0.08%C 0.23%C 0.4%C 0.8%C 1.22%C
60 50
Thermal conductivity W/m/C
Thermal conductivity, W/m/C
70
40 30
0.06%C & 0.08%C 0.23%C 0.4%C 0.8%C 1.22%C
30
20
20 0
100
200
300
400
500
600
700
700
800
Temperature, C
900
1000
1100
1200
Temperature, C
(a) α phase. (b) γ phase. Fig. 5.3.6 Thermal conductivity of carbon steels. 40
1.5%Mn 3.5%Ni 3.5%Ni, 1%Cr 1%Cr 2%Si, 1%Mn
60 50
Thermal conductivity W/m/C
Thermal conductivity W/m/C
70
40 30 20
30
20 0
100
200
300
400
500
600
700
700
Temperature, C
800
900
1000
1100
1200
Temperature, C
(a) α phase. (b) γ phase. Fig. 5.3.7 Thermal conductivity of low alloy steels. As mentioned above, electrical resistivity was measured for all the 22 steels in a wide range of temperature; on the other hand, thermal conductivity was obtained for almost steels only under the conditions below 300 °C. Thus, Powell (1946b) estimated thermal conductivity, not measured in the above test conditions, based on the obtained electrical resistivity. Basically the values in the higher temperature region were extrapolated based on data below 300 °C. In that case, the correlation between thermal conductivity and electrical resistivity in the range that both data were measured has been used for reference. The above mentioned thermal conductivity and electrical resistivity data by Powell et al. were published later by BISRA (1953) as the data book. Graphs shown in Figs. 5.3.6 and 5.3.7 were depicted by the author based on the values of thermal conductivity shown in the BISRA’s data book. Where, thermal conductivity of α and γ phases in carbon steels are shown in Figs. 5.3.6 (a) and (b),respectively, and the same data in low-alloy steels are in Figs. 5.3.7 (a) and (b), respectively. Comparisons between Figs. 5.3.6 9
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
(a) and 5.3.7 (a) show that alloy components decrease thermal conductivity. However, Powell (1946b) could not perform a quantitative assessment of the effect. Electrical resistivity data contained in the BISRA’s data book are discussed in Section 7.3. A measurement of thermal conductivity in steels after the works by Powell et al., which was noticed by the author, is included in a work by Kobayashi et al. (1987). Where, Fe-C alloys with different carbon concentrations were used to measure the thermal diffusivity by a transient heating technique using a square wave pulse. Alloys to measure were produced by dissolving electrolytic iron and carbon in a vacuum induction melting furnace, and then eight stages of carbon concentrations were adjusted in the range of 0 to 1.4 %C approximately. The measured thermal diffusivity was expressed as an empirical formula which may estimate a value in from room temperature to 400 °C, and in carbon concentrations from 0 to 1.4 %C. Influences of the microstructure were investigated by comparisons between measured values in the case of spherical or plate-like cementites, however significant differences were not been observed. Kobayashi et al. obtained thermal conductivity based on measurements of specific heat and thermal diffusivity. The results showed a trend similar to that of BISRA in Fig. 5.3.6, and are somewhat larger in perspective. The indicated value of the density used in obtaining the thermal conductivity from thermal diffusivity is likely to be measured at room temperature. In addition, electrical resistivity measured by Kobayashi et al. is introduced in Section 7.3. In late years, Miettinen (1997) derived an empirical formula on thermal conductivity, considering the dependence of chemical composition, including C, Si, Mn, Cr, Mo and Ni, based on past experimental data. In addition, this experimental expression is intended to obtain the thermal conductivity at 25, 200 and 400 °C. Furthermore, dependence of temperature as well as alloy components was represented by a polynomial in austenite region.
5.4 Heat transfer during heat treating Generally, waters, oils, polymer solutions and gases have been used for quenchants (Totten et al., 1993). Studies on the heat transfer in mainly liquid quenchants were reviews in this text. It is known that a cooling process in the liquid quenchants is usually divided into three stages, vapor film, boiling and convection. In addition, quenchants are often used with an agitation which affects their heat transfer in cooling. Phenomena of heat transfer during boiling of liquids have been studied for the heat exchange in equipments such as boilers, heat exchangers and unclear reactors, and then many significant results reported as a part of textbook (Jakob, 1949 & 1957, McAdams, 1954; Gröber, Erk and Griugull, 1961; Katto, 1964; Shoji, 1995) or for example a specific publication (Japan Society of Mechanical Engineers, 1965). In experiments on boiling heat transfer phenomena, a probe has usually a heating source in its inside, which controls temperature. On the other hand, a probe for the phenomena during heat treatment does not control their temperature in any way. It may be difficult that complex behaviors in quenchants during cooling in the quenching, such as vapor film generations and boiling, are simulated directly under modern technologies. Therefore, in the current heat treatment simulation, the model of heat transfer boundary described as Eq. (5.2.9) has been applied to 10
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
surfaces of heat treated parts. The following sections outline results of studies on cooling during heat treatment. History of researches on measurements of the cooling and cooling rate curves, and the H value, the severity of quench, derived from an analytical study of heat transfer are described. Meanwhile, the author outlines a platinum wire method used for the study of heat transfer through the ages, and its expansion to measure in the vapor film stage. In addition, the methods of lumped heat capacity, inverse analysis, CFD (Computational Fluid Dynamics), and so on, for predicting heat transfer coefficients are summarized. Finally, an outlook on deriving heat transfer coefficients for heat treatment simulation is discussed. 5.4.1 Cooling curves and cooling rate curves Studies on cooling phenomena during quenching were started by measuring temperature changes in a probe with a simple geometry, which is immersed into a variety of quenchants. A chart recorded a temperature change is called as the cooling curve. It is difficult to trace back to a beginning of the research on this curve. Here, from studies by Benedicks(1908), recognized as a pioneer in this field by many researchers in their literatures, to recent researches are introduced in chronological order. Furthermore, also the cooling rate curves are discussed, which were obtained from the cooling curves in early researches in this field. (1) From 1900 to 1929 Benedicks (1908) obtained cooling curves at the center of carbon steel cylinders with different carbon concentrations during water quenching. The thermocouple was made from 0.25 mm diameter platinum and platinum-rhodium wires. In the experimental apparatus, the probes were heated in a furnace and were motioned circularly in still water as shown in Fig. 5.4.1. 33 different tests were carried out under combined conditions, those are carbon concentrations of carbon steel: 0.21-1 .99% C, probe length: 15-50 mm, mostly 50 mm, probe diameter: 4.0-12 mm, mostly 6.5 mm, quenching temperature: 695-1000 °C, mostly about 850 °C, and water temperature: 4.5-16 °C, mostly 13-15 °C. His report shows not only some measured cooling curves, but also cooling times for all the conditions, which were obtained from differences between times at 700 and 100 °C during cooling. He pointed out that the cooling time is proportional to the ratio of mass and surface area of the probe. After about 10 years, Lynch and Pilling (1920) measured cooling curves quenching from 830 °C at the center of 5 %Si-Ni cylindrical probe, 6.4 mm in diameter and 50 mm in length, to investigate the properties of quenchants, such as water, salt water, soap water and oils. The thermocouple was made from 0.2 mm diameter platinum and platinum-rhodium wires. The probe performed a similar circular motion to in the apparatus by Benedicks (1908) in still water. On the other hand, the effect of temperature on the cooling properties was examined by quenching into quenchant controlled at several temperature levels. A cooling curve obtained from water quenching was drawn as shown in Fig. 5.4.2, which classifies three stages in the cooling process, as (A) vapor film, (B) boiling and (C) convection. Also cooling rate curves, showing a relation between cooling rates, °C/s, and temperatures in the probe, were produced in order to quantify the 11
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
cooling characteristics. In addition, the cooling rate was called as quenching power by Lynch and Pilling.
Furnace
Specimen
900
Tank
Temperature, °C
800 700 600 500 400 300 200 A
100
B
C
0 0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15
Time, s
Fig. 5.4.1 Benedicks’s apparatus.
Fig. 5.4.2 Cooling curve measured in water at 58 °C from 830 °C.
Scott (1924) measured cooling curves at the center of the Fe-32 %Ni alloy cylindrical probe, 1 in (25.4 mm) in diameter, during their quenching into water, several concentrations of glycerol solutions, oil-water emulsions and oils. Two quenching temperatures, 100 and 800 °C, were tried for these tests. He showed the effect of different kinds of quenchants and different concentrations of glycerol water on the cooling properties based on the shapes of obtained cooling curves. In his subsequent report (Scott, 1934a), cooling curves were measured during water quenching from 100 or 750 °C at the center and the position which is 0.84 in (21.3 mm) from the center, in the Fe-32 %Ni alloys cylindrical probe, 2 in (50.8 mm) in diameter. Then, he obtained theoretical results of heat transfer phenomena in terms of these measurement conditions, which were compared with experimental results. (2) From 1930 to 1939 Cooling curves were reported by French (1930a, 1930b), which were measured in steel cylinder, sphere and plate probes quenched into a variety of quenchants. This result is a compilation of studies over the past 6 years in the U. S. Bureau of Standards. In cylindrical and spherical probes, their diameters were varied step-by-step in the range 1/2-11 in (12.7-280 mm). Cooling curves were measured by the thermocouple, which was made from platinum and platinum-rhodium wires, on the surface or at the center of the probes. Probes were made of not only carbon steels but also Cr steels. The quenchants were selected from waters, sodium hydroxides, brines, oils, air, etc., and effects of the agitation on their properties were examined. The agitation was performed by rotating the cylindrical cooling tank at a constant speed, and its degree was adjusted by setting the distance between the probe and the center of cooling tank. On the other hand, French quenched probes, which were a different dimension of cylinders, spheres and plates, from 875 °C into many kinds of quenchants. The cooling rate V at 720 ° C were derived by cooling curves at the center of the probe, and then empirical formula: 12
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
VD n = c
(5.4.1)
was obtained commonly for various probe shapes. Where, D is diameter of cylinders or spheres, or thickness of plates, and also n and c are experimental coefficients. It was pointed out that n depends on the type of quenchants, not on the shape of the probes, while c relates to both types of quenchants and the shapes of probes. French examined effects of surface roughness and oxidation, gases contained in quenchants, agitation, cooling spray, etc. On the other hand, Obinata (1930) measured cooling curves at the center of a α brass cylindrical probe, 10 mm in diameter and 30 mm in length. The thermocouple was made from 0.5 mm diameter platinum and platinum-rhodium wires. Probes were quenched from 752 °C into water (at 0 to 100 °C), oils (Shirashime and transformer oils), liquid air and toluol. Not only cooling curves but also cooling rate curves were created according to the work by Lynch and Pilling (1920). Using stepped cylindrical probe, 45 or 75 mm in diameter, 150 mm in length each, made of austenitic stainless steel (14.8 Ni-7.8 Cr), Obata (1931) measured cooling curves at the center of the 75 mm diameter cylinder during quenching into rape seed oils. The thermocouple was made from platinum and platinum-rhodium wires. The oils were classified as old, new and their mixtures. Their properties, specific gravity, free organic acidity, flash point, viscosity, etc. were measured. Oil temperatures were controlled at 25 °C and the other values, which were set at 10 °C intervals in the range from 40 to 100 °C. Comparing the time required for cooling from 800 to 200 °C, it was reported that the fastest was new oil, and mixed and old oils were second and third, respectively. Also, it was pointed out that the cooling time was essentially increased with rising oil temperature in any oils. . Using Cr-Ni steel ball probes, 4 mm in diameter, Wever (1932) was performed quench experiments for different cooling conditions (Houdremont, 1956). Experimental results were plotted as relations between cooling rates and temperatures at several stages during cooling. Sato (1933) quenched probes coated on the surface from 800 °C for examining the effect of facing based on obtained cooling curves. A probe was cylinder, 6 mm in diameter and 70 mm in length, made of Fe-20 %Cr-20 %Ni-alloy, its surface coated with soup which was made as a mixture of clay, graphite powder, abrasive grain, baked borax and water. Waters with different temperatures, glycerols and several oils were used for quenchants. Except for some oil, it made clear that the cooling time of quenching was reduced by the presence of the coating. Examining movies on the cooling phenomena, it was confirmed that the absence of the coating produced a vapor film covering the probe just after cooling, while the existence made generating active fine steam foams on the probe surface from the beginning of cooling. In addition, the Sato’s apparatus measured temperature changes by converting it from the thermal contraction of the probe. The similar approach was succeeded by the apparatus by Ishihara and Ichihara (1942) for measuring length changes during quenching. Further studies (Narazaki et al., 1988; Inoue and Uehara, 1995) on the effect of the coating are seen in recent years. Using the apparatus devised by Sato (1933), except for 5 mm in probe’s diameter, Hara (1935) measured cooling curves during quenching from 830 °C into waters, rape seed oils, soybean oils, new and old fish oils, mixture of old fish oils and vegetable oils. All quenchants were investigated at temperatures 13
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
set 20 °C intervals in the range from 20 to 100 °C, and then their cooling curves and cooling rate curves were created. Moreover, behaviors of not only still but also agitated cooling were reported, however an agitation method was not described. Comparisons between characteristics of the quenchants based on average cooling velocities from 800 to 400 °C showed that the values in the agitation is larger than still. In addition, relations between temperature of quenchants and viscosity, cooling rate and viscosity, viscosity and cooling rate when adding water to the quenchants, quenching temperature and cooling rate, and also oxidation of quenchants and viscosity were examined. Lange and Speith (1935) quenched copper spheres, 10-20 mm in diameter, and photographed using the schlieren method for examining phenomena occurred in quenchants. Their interests were such as cooling by turbulent flow in the quenchants, a role of the turbulent or laminar boundary layer, heat dissipation through a closed vapor blanket and heat exchange on collapse of the vapor blanket. Based on the pictures to characterize these phenomena, their considerations were described. For the heat dissipation through the vapor blanket, the existence of a laminar flow of the quenchant around the vapor blanket was pointed out. In addition, a supporting device was provided at the bottom of the probe to observe clearly various phenomena in quenchants. Meanwhile, silver sphere probe were quenched waters, brines, lithium chloride solutions, Pektinit (mainly pectin ingredient) solutions, rape seed oils, sodium palmitate solutions and so on, and then measured cooling curves at the center were reported. The diameter of the silver sphere used in this study was 20 mm, which differs from 7 mm in the probe by Engel (1931). The reason for this change in diameter was instability in smaller probes at the immersion stage. The thermocouple was made from platinum and platinum-rhodium wires. Confirming further the characteristics of the Pektinit solution practically, carbon steel cylinders, 24 mm in diameter and 80mm in length, were quenched in the solutions of different concentrations, and measured surface hardness data were compared. Russell (1939) adopted a silver sphere to investigate cooling characteristics of quenching oils, which is similar to the probe used by Lange and Speith (1935). Cooling curves were measured using a thermocouple at the center of the sphere, 1 in (25.4 mm) in diameter, during quenching from about 850 °C. The thermocouple was made from platinum and platinum-rhodium wires, 0.006 in (0.15 mm) in diameter. Probes were made of not only silver, but also austenitic steel (20% Ni-25% Cr), with the same dimensions for comparisons. Eight kinds of quenching oils were tested, which had different cooling properties. Property values of the oil, i.e., saponification value, acid value, iodine value, flash point, density, viscosity, volatility, specific heat, thermal conductivity and so on, were reported by Jones (1939) separately. Measured cooling curves showed differences between materials of the probe. In the case of silver, a bending point of the cooling curves were appeared clearly by Russell when moving from vapor blanket to boiling stages, which was called as the characteristic point. However the austenitic steel did not show the point explicitly. The temperature at the characteristic point was called as the characteristic temperature in subsequent studies. Russell examined further cooling curves obtained at the different thermocouple positions, i.e. on the bottom or the side surfaces in the silver sphere probe. It was found that the cooling rate was larger on the bottom in the early stage of cooling, while the time to reach to the characteristic point was shorter at the side. 14
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Using a cylindrical steel probe, Stanfield (1939) examined cooling characteristics of some quenching oils, which were the same as used in the studies by Russell (1939), and water and air. Probes were made of four kinds of steels, i.e., low carbon steel, 0.87 %C steel, 3 %Ni-Cr steel and austenitic Cr-Ni steel. Cylindrical probes, 3 in (76 mm) in diameter and 6 in (152 mm) in length, installed thermocouples at two places, 0.3 in (7.6 mm) from the center and the surface. The thermocouple was made from platinum and platinum-rhodium wires, 0.35 mm in diameter. Quenching temperatures included often in the range from 820 to 830 °C were specified, and then their cooling curves were plotted using data obtained at two thermocouples in the probe. Curves showed differences between oils and also with or without surface scales, however could not be used to specify the characteristic points. (3) From 1940 to 1949 Using the 20 mm diameter silver spheres devised by Lange and Speith (1935), Rose (1940) measured temperature changes by a thermocouple at the center of the probe. The thermocouple was changed from a 0.3 mm platinum/platinum-rhodium to 0.5mm iron/constantine to prevent property changes after repeated heating. Spheres were supported by a thin-walled neck, 30 mm in length, with a small mass. After heating the silver sphere at 800 °C, was immersed in quenchants to a depth of about half of the supporting neck, and then uniformly moved at about 25 cm/s. All experimental results were reported in the form of the cooling rate curve rather than the cooling curve. In addition, as described later, heat transfer coefficients were calculated by the lumped heat capacity method. Reported cooling rate curves were obtained from, for example, airs (still and compressed), waters (temperature dependency), sodium hydroxide solutions, calcium hydroxide solutions, mineral oils (temperature and property dependencies), rapeseed oils, fish oils, mixed oils emulsions (oil volume and temperature dependencies), pectin solutions and water glass. Using a chromel - alumel sphere probe, Tawara (1941) measured cooling curves for waters, brines, soap waters, and 21 kinds of animal and vegetable oils. Chromel and alumel hemispheres, 4 mm in diameter, were welded for making this probe as shown in Fig. 5.4.3, which also acts as a thermocouple. The 0.2 mm chromel and alumel wires were welded to the corresponded hemispheres, which were used as not only the thermocouple lines but also support lines. After heating to 850 °C, the probe was immersed and rotated into quenchants using a mechanism to give it a circular motion about 7 cm/s. For calibrating the probes, Shirashimeyu, the refined raped seed oil, was selected as the standard oil. Almost quenchants were set at the specific temperatures, i.e., 20, 40, 60, 80 and 100 °C for cooling tests. The experimental results were reported as the characteristic temperatures, the cooling time required between 700 and 300 °C and the cooling rate curves, however the cooling curves were for limited quenchants. Cooling rate curves revealed a difference of cooling characteristics between animal, vegetable and mineral oils. Furthermore, the reason why vapor films are more stable in mineral oils was discussed based on specific heat of vapor, volatile fraction of oils and so on. In addition, Tawara examined about quenchant properties which are affected by water contaminations in animal-vegetable oils, quenching temperatures of refined rape seed oils, oxidation of mineral and refined rape seed oils and mixing of different types of oils.
15
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Furnace Specimen
Tank
Alumel
Chromel
Alumel wire (0.2 mm, Dia)
Chromel wire (0.2 mm, Dia) Welding
Pump
Welding
Fig. 5.4.3 Chromel - alumel 4mm dia. sphere probe.
Fig. 5.4.4 Apparatus by Schallbroch et al.
The test equipment reported by Rose (1940) was improved by Schallbroch et al. (1942) based on an advice from the Kaiser Wilhelm Institute. Fig. 5.4.4 show the schematic diagram of their apparatus, which was drawn by the author based on their original. Probe was the 20 mm diameter silver sphere installed a thermocouple at the center. After heating to 800 °C in an electric furnace, it was immersed in quenchants until a depth which was not mentioned. Quenchants were circulated normally in a tank by a gear pump at a flow rate which was not specified. As for using salt bath, a different kind of agitation method, which was not informed in detail, was used. They reported cooling curves for 4 types of mineral oils, a spindle oil and rapeseed oil at 80 °C. For a type of mineral oil and the rapeseed oil, effects of temperature on cooling characteristics were clarified based on the difference in their cooling rate curves. For tap waters, cooling and cooling rate curves were reported in several temperature levels in the range from 20 °C to the boiling temperature, which showed obviously the effect of temperature on cooling characteristics. In addition, the influence of a presence of water and stirring in the salt bath, temperature in calcium chloride solutions on the cooling curves were reported. Using cylindrical silver probe, Jones and Pumphrey (1947) obtained cooling curves of waters and oils. The probe was 3/4 in (19 mm) in diameter and 3 in (76 mm) in length, and had a conical tip of 90 degrees. A silver - platinum thermocouple was installed at the center of the probe. A cooling tank was installed a jet orifice to cause around 1 ft/s (30 cm/s) flow on the surface of the probe using a circulating pump. Cooling curves were obtained from five oils and waters at 20 and 80 °C in agitation, and also for water at 80 °C without agitation. On the other hand, quenching tests were added using austenitic stainless steel probes to compare between cooling rate curves obtained from silver and steel probes. Rose’s Experimental technique (Rose, 1940) using the silver sphere was applied by Peter (1949) to extensive conditions in a wide range of quenchants (Houdremont, 1956). For example, as for waters, cooling rate curves were measured for distilled waters, tap waters in Clausthal and Dusseldorf, well waters, 16
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
different distilled waters including nitrogen, air, oxygen and carbon dioxide, at the different temperature levels incremented with 20 °C in a range of from 20 °C to boiling temperature. In addition, cooling properties were examined for eight different quenching oils, various solutions, metal bathes, salt bathes and mercury. Meanwhile, Peter (1950) reported separately about effects of surface characteristics on cooling properties. In this case, using mild steel spheres, 19 and 40 mm in diameter, for example, it was revealed how cooling rate curves were influenced due to surface oxidations and salt-coated layers. (4) From 1950 to 1959 Using a silver cylinder, Tagaya and Tamura (1951a) measured cooling curves to examine quenchant properties. The cylinder, 10 mm in diameter and 30 mm in length, was installed thermocouples, 0.5 mm chromel-alumel wires, at the center, and a point on the side, which locates up 5 mm in height from the bottom of the probe. The thermocouples were replaced with the silver-chromel because of anticorrosion, especially on the surface (Tagaya and Tamura, 1952c). They remarked that the selection of silver was determined based on the literatures by Lange and Speith (1935), Rose (1940), Peter (1949) and Russell (1939). Accuracy and repeatability in measurements were cited as an advantage of using silver, while its higher thermal conductivity than steels was identified as a drawback. They described that the results obtained from the silver probe could be applied to the problems of quenching steels, under some considerations. In addition, it was pointed out that the Sato’s cooling apparatus (Sato, 1933) had a problem of accuracy. 800
Temperature, °C
Surface
I
700
Center
II
600 500 400
III
300 200
VI
100 0 0
0.5
1
1.5
2
Time, s
Fig. 5.4.5 Cooling curves measured at surface and center of cylindrical probe in distilled water at 20 °C from 800 °C. Tagaya and Tamura (1951b) revealed other stage at the very beginning of the cooling than the stages classified by Lynch and Pilling (1920) in the cooling curve of water using the silver probe. This was classified as stage I as shown in Fig. 5.4.5, which was followed by vapor film (II), boiling (III), and convection (IV) stages. Stage I was described as the process which was for reaching water around the probe to the boiling point, and can be identified only in cooling curves measured on the surface of the probe. Meanwhile, the cooling curves on the surface were reported as test results, since the curves at the center had time delays. In addition, it was described that the stage I was not significant enough to affect the 17
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
cooling power (Tagaya and Tamura, 1951c). Cooling curves measured on the bottom surface of silver balls, as already mentioned by Russell (1939), showed the portion of quick cooling corresponding to the stage I. For each stage of the cooling, Tagaya and Tamura (1951c) discussed how cooling properties of quenchants contribute their properties. For example, it was described that the existence of vapor film stage and its durations relate to molecular structures of quenchants, in particular their shape and polarity, vapor pressure and vaporization heat. On the other hand, they clarified that the temperature at the end of vapor film stage, the characteristic temperature in the distilled waters at 20 °C did not depend on their quenching temperature from the experimental results. As for LiCl, NaCl and KCl solutions, it was found that the characteristic temperature was raised by increasing their concentration, which was explained by the polarity of each solution. In addition, their cooling curves of distilled water at 20 °C showed that temperature decreased in the stage I became larger with increasing the quenching temperature. Tagaya and Tamura measured cooling curves for a variety of quenchants, i.e., water and liquids composed mainly water (Tagaya and Tamura, 1952a), animal and vegetable oils (Tagaya and Tamura, 1952b), concentrated salt solutions (Tagaya and Tamura, 1952c) and mineral oils (Tagaya and Tamura, 1953) in a wide range of their conditions. In addition, they summarized tabular forms in contrast to an average cooling rate in temperature ranges for pearlite (700-500 °C) and martensite (300-200 °C). Using cylindrical silver probes, 4 different diameters, 10, 15, 20 and 25 mm, Tagaya and Tamura (1956a) examined effects of their diameter on cooling properties. In this experiment, total 22 different cooling conditions were applied, and then as a result, it was revealed that characteristic temperature and the beginning temperature at the convection stage in cooling curves did not depend on their diameters. Therefore, they created the mother cooling curve by correcting the time axis unit systematically, which could apply to cooling curves from cylinders with any diameters. (5) From 1960 to 1979 Tagaya and Tamura (1962) created mother cooling curves using on cooling curves obtained from various diameters SUJ2 and SK6 steel probes. In addition, Tokihiro and Tamura (1974) assessed more generally mothers cooling curves from obtained curves in water , animal oils, vegetable oils and mineral oils using spheres, cylinders and square prisms made of silver, SUS27 or SK6 steels. They considered delays in cooling curves due to heat generation from phase transformations as a parallel movement of the curve. Meanwhile, Yamazaki and Okamoto (1967) obtained cooling curves from the 0.46 %C carbon steel wire, 2.2 mm in diameter and 70 mm in length, due to a variety of quenchant jets after heating by an electric current to 850°C. Quenchants were waters, oils, polymer solutions and so on, and their temperatures and flow rates were set to several conditions. The cooling curves were measured by a thermocouple, the 0.25 mm chromel - alumel wires, was installed on the cylinder surface. Mitsuzuka and Fukuda (1974) found an unstable region in the middle of vapor film and boiling stages, during quenching the silver cylindrical probe based on Tagaya and Tamura’s research into water at 60 °C. In the region, the vapor film broke locally and recovered quickly. In their consecutive study, Mitsuzuka and Fukuda (1977) examined characteristics of cooling when a low-carbon steel plate (28 × 220 × 220 mm), 18
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
which was immersed in still water of 27 and 65 °C. For two immersing conditions, vertical and horizontal orientations, of the plates, cooling curve measurements and phenomenon filming were made. As a result, it was clear that significant differences occur in symmetry of the cooling and a collapse of the vapor film in terms of both conditions. (6) From 1980 to 1999 Narazaki et al. (1988) obtained cooling curves at the center of a silver cylindrical probe coated with clay or vitreous substances, 10 mm in diameter and 30 mm in length, during quenching in water at various temperatures. The cooling curves show clearly effects of coating thickness and water temperature. In their subsequent studies, Narazaki et al. (1989) measured cooling curves at the center of the silver probe, which was immersed in waters at various temperatures. Probes were spheres, 10 mm in diameter, and also cylinders, 10 mm in diameter and 30 mm in long, which were provided a roundness at the corner of the end face of the cylinder, their radius Rc = 0, 1, 3 and 5 mm. The roundness, Rc = 5 mm, corresponds to a probe provided a hemisphere at the end. In addition, the probe was supported by the silver tube, 3 mm in diameter. First, for defining the immersion depth as the distance from the liquid surface to the top of the probe, these effects on cooling curves were confirmed. Using cylinder of Rc = 5 mm for test at 30 °C water temperature, generation time of the characteristic point was reduced from 17.5 to 5s by increasing the immersion depth from 5 to 25 mm. This origin was considered that the supporter induced a collapse of vapor film earlier than at the probe by increasing the immersion depth based on an observation of boiling. Finally, the experiments by Narazaki et al. showed, in the case of setting the immersion depth equal to and less than about 15 mm, the collapse of the vapor film from the support was not generated earlier than at the probe, regardless of its shape and temperature. The collapse of vapor film leading from a support of a probe was reported also by Nishio and Uemura (1986), which was found during cooling a platinum sphere probe in distilled water. The size of the sphere was 10 mm in diameter, and the outer diameter of the platinum tube to support it was 2 mm. On the other hand, Beck and Moreaux (1992) described that cooling curves measured from a silver cylinder, 16 mm in diameter and 48 mm in length, provided with a hemispherical end during quenching in 40 °C of still water from 850 °C, did not show good reproducibility. Narazaki et al. (1989) confirmed that when setting the immersion depth to 10 mm, no vapor film collapses earler at the support, as mentioned at the above. In this condition, cooling curves at the center of the silver cylinder, corner radius Rc=0 or 5 mm, were obtained by immersing it in water at temperature levels incremented with 10 °C in the range from 20 to 90 °C, and 95 °C. As a result, it was revealed that the case, Rc=0, no rounded corners, increases the characteristics temperature with decreasing water temperature. For example, waters at temperatures 50 and 80 °C showed characteristic temperatures of about 600 and 350 °C, respectively. On the other hand, in the probe with a hemisphere, the characteristic temperature was around 200 °C in waters at any temperatures of the probe. In addition, it was described that cooling curves obtained using a silver sphere showed almost the same tendency of the cylinder probe with a hemisphere end. The above trend in the temperature characteristics was appeared in experiments by Uemura and Nishio 19
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
(1986), which used a platinum sphere for distilled water quenching. However, characteristic temperatures in their experiments were affected somewhat from water temperatures by comparing the Narazaki et al.’s results. Narazaki (1995) reported on cooling curves obtained during immersing the silver probe, which was already mentioned (Narazaki et al., 1989), to polymer solutions. The results showed that as for still waters, cooling curves with different radiuses of rounded corners depicted large differences. That is, higher characteristic temperature was shown in the curves from the probes with the lower roundness. Shapes of cooling curves obtained under agitated cases, 0.2 or 1.0 m/s in velocity, were similar each other even for any roundness radius, by suppressing the vapor film stage. While silver spheres, 10, 16 and 20 mm in diameter, were immersed in still polymer solutions, the results showed that characteristic temperatures were increased using smaller probes. Jeschar et al. (1992) obtained cooling curves on the surface of the 30 mm diameter Ni sphere, during quenching in still waters at temperature levels incremented with 20 °C in the range of 20 to 100 °C. Rising water temperatures, the characteristic temperature fell, however cooling rates were not changed in the vapor film stage. Meanwhile, cooling curves were measured using the 40mm diameter sphere during quenching in 40 °C water agitated by flow, 0, 0.3, 0.5 or 0.7 m/s in rate. In this case, the characteristic temperature increases with higher flow rate, while cooling rate did not change in the vapor film phase. In addition, as for the 100 °C water, it was shown that flow rate did not contribute significantly on the characteristic temperature. Showing graphically a relation between surface heat flux and temperature in the Ni sphere, influences of the different probe diameters between 20, 30 or 40 mm were depicted in the case of water temperature 20 °C and zero flow rate. On the other hand, it was revealed in the graphs that there were flow rate dependences in the case of 30mm diameter and 20 °C water, and also temperature dependences in the case of 30 mm diameter and 0 flow rate. As a result, it was clarified that the characteristic temperature became higher, in cases of the smaller diameter of the sphere, larger flow rate and lower temperature. When quenching Cr - Ni steel cylindrical probe, 15 mm in diameter and 45 mm in length, into water, Tensi (1992a) pointed out the characteristic temperature tended to decrease with larger rounded corners. (7) Standardization of cooling curve analysis To clarify quench cooling characteristics of various quenchants, many researchers measured cooling and cooling rate curves and analyzed their shapes, as already mentioned. The shapes of these curves depend on generally a shape and a material of each probe, an immersion method, and so on, even when using the same quenchant. Therefore, for comparing these curves, it is desirable to measure in the same condition at all times. Moreover, these measuring devices are needed good reproducibility, and considered about its safety and economy. As mentioned earlier, cooling and cooling rate curves for various types of quenchant have been measured using some specific experimental methods. Some of them have been defined as national or international standards for cooling curve analysis as described below (Totten et al., 1997). In Japan, the measurement method used in the studies by Tagaya and Tamura, as mentioned already, 20
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
was established in 1965 as the Japanese standard, JIS K2242. In this system, cooling curves are measured on the surface of the silver cylinder, 10 mm in diameter and 30 mm in length. On the other hand, an apparatus using silver cylinder, 16 mm in diameter and 48 mm in long, was standardized as AFNOR NFT 60178 in France. This size of silver cylindrical probe was reported in the study by Moreaux and Beck (1992) as previously mentioned. Meanwhile, in China, a method using silver cylinder of the same dimensions as JIS K2242, 10 mm in diameter and 30 mm in length, was established as ZBE 45003-88, which measures cooling curves at the center, not on the surface of the probe. In recent years, in Japan, an apparatus to measure cooling curves at the center of the silver cylinder, which has the same dimensions as the JIS K2242, was standardized for water-solution quenchants. The specifications of this apparatus were employed to establish ASTM D7646 in 2010, which is to measure cooling curves of polymer solutions for quenching aluminum alloys. Rather than silver, Inconel 600 was used as a material for a cylindrical probe, 12.5 mm in diameter and 60 mm in length, for measuring the cooling curves at its center by a chromel - alumel thermocouple. The system using the Inconel probe was standardized as ISO 9950: 1995 (Tensi, 1995b; Totten et al., 1997). These standards defined material and shape of the probe, installation method of a thermocouple, and maintenance method for probes. In addition, a quenchant was also standardized to use for verifying the accuracy and repeatability of the tests. On the other hand, heat transfer coefficients were obtained by the lumped heat capacity method in the case of the silver probe. This procedure has not been included in the standard. In addition, the silver probes were Ni plated by Beck and Moreaux (1992) or iron plated by Ichitani (2004), which have not been described as specific provisions in the standards. Since the specifications above are for quenchants of the stationary state, it is not applicable for an agitation state. 5.4.2 Severity of quench H Cooling phenomena in cylinders during quenching were described analytically as a heat conduction problem in solid by considering a heat transfer on surfaces by Scott (1924). In his study, for solving the heat conduction equation of the infinitely long cylinder, the substitution was performed for a non-dimensional analysis:
h=
α
(5.4.2)
k
where h was called as quenching constant. On the other hand, α and k is the heat transfer coefficient on the surface and the thermal conductivity of the material for the cylinder, respectively. These properties were known as amounts depended on temperature, however it was difficult to consider the dependences in a theoretical calculation at that time. The unit of h is 1/m, when using units of k and α are W/(m2°C) and W/(m°C), respectively. A value of 1/2 h was defined by Grossmann (1940) as the severity of quench, H. In addition, he used 1/in as the unit of H. To represent cooling characteristics of the entire quenching process as a value of severity of quench was questioned by several studies as described below. Scott (1924, 1934a) calculated theoretically heat conduction problems in cooling experiments. For 21
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
example, he showed that the obtained cooling curves at the center and at the location 0.84in (21.3 mm) from the center in the 2in (50.8 mm) diameter Fe - 32% Ni alloy cylinder by setting thermal properties and h the e data appropriately, agreed well with experiments (Scott, 1934a). In addition, these experiments were performed under the condition that a cylinder at 100 or 750 °C was immersed into water of 0 °C. At that time, cooling rates at the specified temperature, obtained at the center of the cylinder, were compared with his theoretical calculations. Moreover, not only waters but also oils and sodium hydroxide solutions, even for conditions with or without agitation, were considered for his calculations, and the results were compared with the experiments. Scott (1934b) recognized that heat transfer coefficient α depended to temperature was better to quantitatively assess the three stages of cooling processes, vapor film, boiling and convective, which was clarified by Lynch and Pilling (1920), rather than an amount like h. He defined the problem whether only h can describe the cooling phenomena, although it was difficult to consider the temperature dependency of heat transfer coefficients in the theoretical calculation. He obtained heat transfer coefficients of waters, oils, glycerin solutions, airs, using cooling rate measured at the center of the steel cylinder and results from a graphical calculating method of the thermal conductivity. In particular, as for air cooling, he showed the heat transfer coefficients at multiple levels of surface temperatures in the range of 75 to 780 °C. Grossmann (1940) specified the values of severity of quench, H, 1/in, for various cooling conditions, when building upon his graphical calculation method for the hardenability of steel cylinders. For example, those values for quenchants were 0.02: stationary air, 0.3: still oil, 1.0: still water, still brines: 2.2 and so on. They had been considered as references for characteristics values of corresponding cooling conditions. Assuming the thermal conductivity of steel as 25 W/(m°C), values of H, 0.02, 0.3, 1.0 and 2.2 1/in corresponds to heat transfer coefficients, 39, 590, 2000 and 4300 W/(m2 °C), respectively. . Janulionis and Carney (1951) found that H was affected by the temperature of quenchants and the diameter of probe, based on their experiments, quenching a 9.76 %Ni-16.76 %Cr stainless steel cylinder into waters or oils. In particular, they noted that H increased in the boiling stages. In their studies for estimating H, first, a cooling time to reach from a quenching temperature to a specific temperature was read from cooling curves. Then H was determined from the cooling time using their balance sheet pre-determined theoretically. A similar trend for H was confirmed in cooling experiments, using silver cylinder by Pumphrey and Jones (1947) and eight steel cylinders by Carney (1954).. Tagaya and Tamura found a temperature dependency of H, which was obtained from the measurements during quenching steel probes (Tagaya and Tamura, 1956b) and silver probes (Tagaya and Tamura, 1956c) into waters, oils and brines. Meanwhile, they obtained corresponded heat transfer coefficients to H values of steel and silver probes, and then indicated that these were associated in the form of a polynomial (Tagaya and Tamura, 1956d). However, their heat transfer coefficients did not depend on the temperature. 5.4.3 Measurement of heat transfer coefficient by platinum wire methods: boiling curves As already mentioned, Benedicks (1908) obtained cooling curves from his cylindrical probe. In the same paper, he reported his experimental results on heat transfer characteristics between a current heated 22
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
platinum wire, 50 mm in long and 0.21 mm in diameter, and a fluid flow, both of which were in a glass tube, 70 mm in length and 3.5 mm in diameter. Quenchants, such as distilled waters, rapeseed oils, alcohols, ethers, benzenes and airs were used for cooling the platinum wire by flowing through the pipe. During the experiment, temperatures of platinum wire and quenchants, and heating amounts of wire were measured. A calorific value of the wire was determined from measured currents and voltages, under the assumption that temperature distributes uniformly in the cross section of platinum wire, because of its extremely small diameter. Maximum temperature of the platinum wire was about 150 °C for measurements of waters, from 20 to 57 °C. This test showed that differences in heat transfer characteristics occur between before and after the boiling point. Experiments to determine the heat transfer characteristics of quenchants using the platinum wire were also performed by Davis (1924). The platinum wires, 60.8 mm in length and 0.102, 0.152 or 0.204 mm in diameter, was cooled by circling it in a quenchant which was contained in a circumferential channel. Using distilled water, paraffin oil and three transformer oils, their relative speed levels were set by using a 10 cm/s increment in the range from 10 to 70 cm/s. The experimental results were organized by an empirical formula, which related heat consumptions in the platinum wire per unit of time and unit length to the thermal conductivity, kinematic viscosity, specific heat and relative velocity of the quenchants, and further the diameter of the platinum wire. Using the experimental results with platinum wire by Benedicks (1908) and Davis (1924), Scott (1934b) obtained heat transfer coefficients between the platinum wire and quenchants as shown in Table 5.4.1. He compared his obtained values with the case of a vertical steel plate shown in the book by McAdams (McAdams, 1933 first edition of 1954). On the other hand, he showed that experiments by Benedicks (1908) and Davis (1924) were clearly responding to the classified boiling and convection stages by Lynch and Pilling (1920), based on the comparison of temperatures at the wire and boiling point of quenchants. Therefore, at this point, the platinum wire method was not applicable to the vapor film stage. Around that time, Nukiyama (1934) investigated heat transfer characteristics on the surface of electric current heated platinum wire, 0.14 mm in diameter, in water at 100 °C. Heat flux at the surface was increased rapidly by rising the temperature, and then a transition phenomenon to the state at a temperature of 900 °C, shown as the red arrow, was appeared when heating it to the temperature of about 115 °C, as shown in Fig. 5.4.6. In addition, this figure was redrawn by the author after converting units of the heat flux in the original figure. It was revealed that maximum heat flux was produced at the point before this transition, while the results observed that the platinum wire was covered with a vapor film by separating it from water at the state after the transition. In addition, this kind of relation between the heat flux and the temperature is called as the boiling curve. Incidentally, the author created the diagram for the temperature dependence of heat transfer coefficients by converting from the boiling curve as shown in Fig. 5.4.7 for reference.
23
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Table 5.4.1 Heat transfer coeff., α, W/(m2K) of different quenchants using the results by platinum wire. Quenchant Name Liquid temp °C
Boiling point °C
Above boiling point Pt wire α temp °C
Below boiling point Pt wire α temp °C 50 80 50 50 50 50 50 50-300 50-300
Reference
0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21
Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908)
0.21 0.21 0.21 0.21 0.21 0.21
Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908) Bendicks (1908)
water water Methyl alc. Ethyl alc. Benzen Toluene Rape oil Air stream Air still
20 57 16 15 17 16 17
water Methyl alc. Ethyl alc. Toluene Benzen Ether
20 16 16 16 17 15
Water Paraffin oil Oil #1 Oil #2 Oil #3
11 11 15 14 16
46 61 55 64 56
14100 3150 3590 3530 3440
Aniline Toluene Glycerin Carbon tetrachloride Olive oil
20 20 20 20
40 40 40 40
1390 1320 1160 1110
0.15 0.15 0.15 0.15
Davis(1924) Davis(1924) Davis(1924) Davis(1924)
20
40
850
0.15
Davis(1924)
100 66 78 110 80 36
110-130 70-85 80-95 120-140 85-105 40-55
8000 10000 2800 2500 1800 1600 1400 420 150
Pt wire dia., mm
75000 59000 50000 42000 33000 21000
1600000
Davis(1924) Davis(1924) Davis(1924) Davis(1924) Davis(1924)
Heat trans. Coeff., W/(m °C)
100000
Heat flux, W/m
2
2
1400000 1200000 1000000 800000 600000 400000 200000 0
10000
1000
100
0
200
400
600
800
1000
1200
1400
0
Temperature at wire, °C
Fig. 5.4.6 Boiling curves by heated platinum wire.
200
400
600
800
1000
1200
1400
Temperature at wire, °C
Fig. 5.4.7 Heat trans. coeff. on heated platinum wire.
Then, Nukiyama raised temperature of platinum to about 1000 °C after the transition, and then confirmed that the heat flux increased during the process. After that, by gradually decreasing temperature, the heat flux fell to a smaller value than the maximum point, and then rapidly returned to the original state by a transition. Nukiyama considered the cooling transition corresponded to the minimum point of heat 24
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
transfer. In addition, he pointed out an analogy between a behavior of platinum wire during a temperature drop and a cooling phenomenon during quenching. In addition, the study by Nukiyama was introduced as a pioneering work in the field of boiling heat transfer over the textbooks on heat transfer (Jakob, 1949 & 1957; McAdams, 1954; Gröber, Erk and Griugull, 1961; Katto, 1964; Lienhard and Lienhard, 2011). Studies on the boiling phenomena using the platinum wire can be characterized by examining heat transfer properties of quenchants by a heated probe. After that, by obtaining pool boiling curves of quenchants with not only a wire but also other shapes of the probe, which were made of platinum as well as other metals, researches in this field had been developed rapidly (Jakob, 1949 & 1957; McAdams, 1954; Gröber, Erk and Griugull, 1961; Katto, 1964; Lienhard and Lienhard, 2011). In more recent years, Thompson and Bergles (1970) conducted a study to confirm the analogy between pool boiling and quenching, which was suggested by Nukiyama. They provided a copper pipes for a heated probe of pool boiling, and specified its dimensions to 0.5 in (12.7 mm) in diameter and 0.02 in (0.5 mm) in thickness for Freon-113 and distilled water, and the same outer diameter and 0.425 in (10.8 mm) in inner diameter for liquid nitrogen. On the other hand, the quenching test using a copper cylindrical probe, 0.5 in (12.7 mm) in diameter and 3 in (76.2 mm) in length, or a toroidal probe, 0.5 in (12.7 mm) in diameter and 5.5 in (139.7 mm) in toroidal diameter. Heat flux - temperature curves of Freon-113, distilled water and liquid nitrogen, which were obtained from both of steady pool boiling and quenching tests, were compared, and then it was suggested that the results have a practical implication although an agreement was not enough.
Heat flux, W/m
2
sub 30 °C
sub 20 °C
100000 sub 0 °C
sub 10 °C
Quenching Pool boiling
100000 sub 40 °C
2
sub 40 °C
Heat trans. coeff., W/(m °C)
1000000
10000
10000 sub 30 °C
1000
sub 20 °C
sub 0 °C sub 10 °C
Quenching Pool boiling
100
1
10
100
1000
0
Surface temp.-liquid temp, °C
Fig. 5.4.8 Boiling curves of ethyl alcohol.
50
100
150
200
250
Surface temp. °C
Fig. 5.4.9 Heat trans. coeff. curves of ethyl alcohol.
Research on the above analogy was performed by Tachibana and Enya (1972) by a different kind of laboratory equipments using Freon-113 and ethyl alcohol. Steady-state pool boiling curves were obtained during heating quenchants through a 0.1 mm thick copper sheet. On the other hand, cooling curves were measured using a copper disc probe, 50 mm in diameter and 2 mm in thickness, during cooling on a single side, which provided a thermocouple on the opposite side of the disc. In addition, copper or mild steel cylindrical probes, 8 mm in diameter and 70 mm in length, were used for comparison. By comparing between pool boiling and quenching curves of Freon-113 and ethyl alcohol, they concluded that the steady pool boiling data is used reasonably as a first approximation of the heat transfer characteristics during 25
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
quenching. The author created Fig. 5.4.8 by converting the units of the boiling curve of ethyl alcohol in their original figure. In addition, since the report noted the saturation temperature of ethyl alcohol is 78.4 °C, the boiling curves could convert to the heat transfer coefficient - temperature curves as shown in Fig 5.4.9. 5.4.4 Prediction of HTC based on lumped heat capacity method A method to obtain surface heat transfer coefficients (HTC) from cooling curves measured in a silver sphere probe was developed by Rose (1940). In this technique, temperature within the probe can be assumed to be uniform at any point during cooling process because of the high thermal conductivity of silver. Then the relative amount of heat contained in the probe, QV, is as follows:
QV = C p ρVT
(5.4.3)
where T is the uniform temperature of the probe, Cp and ρ are specific heat and density of silver, and V is the volume of a sphere. On the other hand, since the heat flux on a probe surface, qS, is obtained from the boundary condition in equation (5.2.9), which includes the heat transfer coefficient α, the total heat flowed out from the surface of a probe per unit time, Q S , is:
Q S = Sα (Te − T )
(5.4.4)
where Te is the temperature of a quenchant, and S is the total surface area of a probe. Since Q S obtained from Eq. (5.4.4) is equal to the time change of the heat in a probe, QV, in Eq. (5.4.3), the following relation is derived.
V
d ( ρ C pT ) dt
=Sh(Te − T )
(5.4.5)
By ignoring the temperature dependence of the density and specific heat, the heat transfer coefficient, α, is obtained as:
α=
dT ρ C pV dt S (Te − T )
(5.4.6)
from the Eq. (5.4.5). The above approach to determine the heat transfer coefficient is generally called as the lumped heat capacity method (Totten et al., 1993). As already mentioned, Rose (1940) measured cooling curves at the center of the 20 mm diameter silver sphere immersed in quenchants with a flow rate of about 25cm/s or sometimes 10 cm/s. Then, he applied the lumped heat capacity method to the cooling curves for obtaining heat transfer coefficients, and then made a table to compare with those by other researchers. The author reconstituted a part of the data in the Rose’s table as shown in Table 5.4.2. Here, heat transfer coefficients representing the vapor film, boiling and convection stages were shown after converting the unit from kcal/(m2h°C) to W/(m2K). Meanwhile, Rose described, the data in the original table by Engel (1931) was obtained from the sphere probe, its 26
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
diameter were 4 or 7 mm, corresponded to the materials, Cr-Ni steel or silver, respectively. Table 5.4.2 Heat transfer coeff., α, W/(m2K) of different quenchants using lumped heat capacity method. Quenchant Name Temp °C
Cooling Rate cm/s
water water water water water water
20–25 20–25 18 15.5 20 8
moderate moderate moderate moderate 25 10
rape oil rape oil rape oil oil oil oil heavy oil light oil heavy oil heavy oil light air air air air air air
20 20 20 20–25 20–25 20 20 20 20 20 20–25 20–25 20 20 20 20
moderate moderate 25 moderate moderate moderate moderate moderate 25 25 non non non non stream stream
Vapor blanket Temp α °C
Nucleate boiling Temp α °C
720 720 700 700
6300 3490 3110 1270
550 200 500 300
10100 25000 12600 6050
720 720 700
2110 1980 1690
550 550 500
2790 3490 3660
720 720 720 700 700
843 500 686 1420 779
600 500 500 550 450
3190 2970 3490 3020 3260
Convection Temp α °C
Probe
Reference
38–35 39
3500 6050
90
2440
32%Ni 1%C Cr–Ni Ag Ag Ag
Scott (1934a) Scott (1934a) Engel (1931) Engel (1931) Rose (1940) Rose (1940)
300 200 200 178–57 58 300–200 200 200 300–200 300–200 780–75 62 720–200 700–200 720–200 700–200
768 512 488 451–394 512 670–372 744 599 314–244 477–244 108–19 19 74–17 49–33 307–149 465–256
Cr–Ni Ag Ag 32%Ni 1%C Cr–Ni Cr–Ni Ag Ag Ag 32%Ni 1%C Ag Ag Cr–Ni Ag
Engel (1931) Engel (1931) Rose (1940) Scott (1934a) Scott (1934a) Engel (1931) Engel (1931) Engel (1931) Rose (1940) Rose (1940) Scott (1934b) Scott (1934b) Engel (1931) Rose (1940) Engel (1931) Rose (1940)
Krainer and Swoboda (1944) obtained temperature curves of the heat transfer coefficient for five different kinds of quenching oils and the two different temperature waters based on the Rose’s study. After converting the unit of heat transfer coefficients in the original figure in their paper to the SI system, Fig.5.4.10 was made by the author. As a result, a temperature dependence of heat transfer coefficients was shown more clearly than the form depicted in Table 5.4.2.
Oil A Oil D Water 60°C
2
Heat transfer coeff, W/(m °C)
100000
Oil B Oil E
Oil C Water 20°C
10000
1000
100 0
200
400
Temperature, °C
600
800
Fig. 5.4.10 Temperature curves of heat transfer coefficients by Krainer and Swoboda. 27
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Jones and Pumphrey (1947), as already mentioned, obtained cooling curves of waters and oils using a silver cylinder, 3/4 in (19 mm) in diameter and 3 in (76 mm) in length. Then by applying the lumped heat capacity method to the data, they got and showed temperature curves of the quenching constant h, not the heat transfer coefficient, for one type of oil and a water about 80 °C. A method which was similar to the lumped heat capacity method was applied by Shyne and Sinnott (1952) to a quenching phenomenon in KNO3–NaNO2 salt cooling, in spite of a significant temperature gradient in a probe. A cylindrical probe made of the 430 stainless steel, 1 in (25.4 mm) in diameter and 6 in (152 mm) in length, was heated to 845 °C, and then immersed in the salt that is set to temperature levels between 195 and 385 °C. Cooling curves are measured at the center and on the surface of the probe, and their average temperature was considered as the probe temperature for applying the lumped heat capacity method. A heat lost from the probe was determined by multiplying a mass and a specific heat of the probe to the temperature difference between initial and current probe temperatures. This time changes of the heat correspond to the left side of Eq. (5.4.5). A heat transfer coefficient, α, was calculated by Eq. (5.4.6) in which measured surface temperature of the probe was specified as the temperature, T, at the right side of the equation. For obtaining heat transfer coefficients of spray and air blast cooling, the lumped heat capacity method was applied by Shimada et al. (1976) to cooling curves measured from a silver disc probe. Dimensions of the silver discs were 2 mm in thickness and 20, 30 or 40 mm in diameter, which were provided a thermocouple at its center. Heat transfer coefficients of the air blast cooling did not show almost a temperature dependence, and had a value of about 230 W/(m2K). In contrast, the dependence appeared in the case of spray cooling, as similar to normal water quenching, and the heat transfer coefficients tended to increase after temperature decreasing less than 300 °C. Values of heat transfer coefficients were outlined, as 350 W/(m2K) in the high temperature range, and 3500 W/(m2K) in the vicinity of the peak position of 150 °C.
Heat transfer coeff., W/(m K)
1000000
Temperature, C
30C, 10% PAG
600
230C, molten salt 120C, JIS2-1 oil
400 30C, water 80C, JIS1-2 oil
200 30C, 10% brine
0
30C water 80C, JIS1-2 oil 230C, molten salt
2
800
2
4
6
8
10
100000
10000
1000
100 0
Time, s
(a) Cooling curves by silver probe.
30C,10%PAG 120C, JIS2-1 oil 30C, 10％ brine
200
400
600
Surface temperature, °C
800
(b) Heat transfer coeff. by lumped heat capacity method.
Fig. 5.4.11 Heat transfer coefficients by lumped heat capacity method using cooling curves by silver probe.
28
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Tajima et al. (1988) measured cooling curves of water, at various degrees of subcooling, difference between saturation temperature and the corresponded temperature, using a gold cylindrical probe, 4mm in diameter and 40mm in length, provided a chromel - alumel thermocouple in the center. Heat flux temperature curves were obtained by the lumped heat capacity method, which were compared with previous experimental results. Narazaki et al. (1997, 2002) measured cooling curves of various quenchants using a silver cylindrical probe as shown in Fig. 5.4.11 (a), and calculated temperature curves of heat transfer coefficients by applying the lumped heat capacity method as shown in Fig. 5.4.11 (b). Their lumped heat capacity method considered temperature dependences of thermal properties of silver. By applying the lumped heat capacity method to cooling curves measured from a cylindrical steel probe, Ma (2002) and, Sisson and Maniruzzaman (2004) obtained temperature curves of heat transfer coefficients. A probe, 9.525 mm in diameter and 38.1 mm in length, was provided a thermocouple in its center. Tow kinds of steels, the 4140 steel or the 304 stainless steel, were used for the probes, which could be replaced each other. That is, the probe was joined by a screw to a steel coupling, the same diameter of the probe and 63.5 mm in length. Furthermore, the coupling was joined to a threaded stainless steel rod, the same diameter and 355.6 mm in length. Probes were quenched into seven kinds of mineral oils after heating in air or argon. In Fig. 5.4.12, the temperature curves of heat transfer coefficients obtained for the T-7A oil were redrawn by the author. The different trend was shown in the curve obtained from the 4140 steel probe heated in air, because of oxidation.
2
Heat trans. coeff., W/(m °C)
2500
4140-air 4140-argon 304-air 304-argon
2000 1500 1000 500 0 0
200
400
600
800
1000
Tempearture, °C
Fig. 5.4.12 Heat transfer coefficients by steel and stainless steel probes. 5.4.5 Prediction of HTC based on analytical or numerical solutions A way to obtain directly heat transfer coefficients (HTC) using a cylindrical carbon steel probe was devised by Yoshida (1950). He provided thermocouples at the center and near the surface of the cylinder, 20 mm in diameter and 120 mm in length, which measured cooling curves during quenching from 720 °C to tap water at 15 °C as shown in Fig. 5.4.13(a). Then, heat transfer coefficients were derived by comparing the cooling curves obtained as experimental results and also by solving a heat conduction equation in the infinitely long 29
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
cylinder, as shown in Fig. 5.4.13(b).
600
2
Surface Center
700
Temperature,
Heat Trans. Coeff., W/(m °C)
800
500 400 300 200 100 0 0
5
10
15
0
20
200
400
600
800
Surface Temperature, °C
Time, s
(a) Cooling curves in steel cylinder.
100000 90000 80000 70000 60000 50000 40000 30000 20000 10000 0
(b) Heat transfer coeff. based on analytical solutions.
Fig. 5.4.13 Heat transfer characteristicss based on analytical solutions. On the other hand, Nakagawa and Yoshida (1950) derived empirical formulas of heat transfer coefficients corresponding to initial still fluid cooling, vapor film, boiling and convection stages, by considering heat transfer phenomena based on both theoretical and experimental approaches. In addition, Nakagawa and Yoshida (1952a) measured cooling curves and derived heat transfer coefficients using the cylindrical steel probe provided two thermocouples, as already mentioned, for not only waters but also glycerol solutions, various oils, oleic acid solutions and brines. Moreover, they derived empirical formulas for vapor film and boiling stages in quenchants, from initial cooling to the state reaching the maximum value of the heat transfer coefficient. At the same time, Nakagawa and Yoshida (1952b) also examined dependences of quench temperatures, surface roughness, coating and diameters of the probe on heat transfer properties. Meanwhile, they studied cases of sphere and cube shaped probes as well (Nakagawa and Yoshida, 1952c). Based on the approach by Nakagawa and Yoshida, Yamaguchi et al. (1969), first, measured cooling curves using thermocouples provided at the center and 1 mm in depth from the surface of carbon steel and stainless steel cylinders, 5 levels in the range of 25 to 75 mm in diameter. In addition, for the 75 mm diameter probe, thermocouples were added at a point between the center and the surface. For cooling by water, air and oil, temperatures at the center were solved analytically using an infinite long cylinder under thermal properties at 450 °C, and compared them with measurements for seeking cooling heat transfer coefficients by trial and error. A graph of heat transfer coefficients and temperature differences between in a quenchant and at center of steel cylinder showed that the results were independent of diameter cylinders. In addition, the temperature curves of heat transfer coefficients were approximated using lines corresponding to each cooling stage, which was proposed by Enya (1966). These lines were drawn based on a double logarithmic plot of heat transfer coefficients and the temperature difference between at the surface and a saturation temperature. Using cylindrical probe, 30 mm in diameter and 90 mm in length, made of austenitic stainless steel, X 8 Cr Ni Mo 18 10, Tensi and Steffen (1985) estimated heat transfer coefficients of waters, aqueous polymer 30
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
solutions, etc. Using cooling curves measured by thermocouples at near the surface, the center, and also the midpoint of them in the probe, radial temperature distributions were estimated momentarily. Furthermore, changes of surface temperatures were assumed by extrapolating the temperature distributions. Since temperature distributions were obtained in the probe, amounts of heat in the probe were calculated by integration over the total volume, not Eq. (5.4.3) which is used for the lumped heat capacity method. Using the surface temperature obtained already, the heat transfer coefficients were calculated from Eq. (5.4.6). Heat transfer coefficient curves of water, PAG polymer solution and sodium oleate solution were extracted by the author from their original paper, and then were redrawn as Fig.5.4.14. The above method can be positioned as the more precise approach than one by Shyne and Sinnott (1952).
3500
Water PAG-type polymer quenchant Sodium oleate solution
2
10000
Heat trans. coeff., W/(m °C)
2
Heat trans. coeff., W/(m °C)
12000
8000 6000 4000 2000 0
3000 2500 2000 1500 1000 500 0
0
200
400
600
800
1000
0
Surface tempearture, °C
200
400
600
800
1000
Surface tempearture, °C
Fig. 5.4.14 Heat trans. coeff. by Tensi and Steffen
Fig. 5.4.15 Heat trans. coeff. by Liscic.
Liscic (1992) devised a method to obtain a surface heat flux based on a difference of the cooling curves measured from thermocouples at the surface and 1.5 mm in depth from the surface in an austenitic stainless steel (AISI 304) cylindrical probe, 50 mm in diameter and 200 mm in length. This probe has a structure that can be installed internal thermocouples within a tolerance of less than ± 0.025 mm. Liscic and Filetin (2011) applied the above apparatus to a low viscous accelerated quenching oil of 50 °C with medium agitation, and obtained a temperature curve of heat transfer coefficients as shown in Fig. 5.4.15. Segerberg and Bodin (1992) obtained heat transfer coefficients at different surface positions of quenched cylinders and rings into still oils at 70 °C as shown in Fig. 5.4.16 and Fig. 5.4.17, respectively. Cooling curves of cylinders, 30 mm in diameter and 90 mm in length, were measured at near the upper (point 5), lower (point 6) and longitudinal surfaces (point 1-4). A ring, 110 mm in outer diameter, 70 mm in inner diameter and 30 mm in height, was quenched in a horizontal position, and cooling curves were measured at near the top, bottom and side surfaces of the probes. Another ring, 110 mm in outer diameter, 35 mm in inner diameter and 10 mm in height, was quenched in horizontal and vertical positions. The obtained heat transfer coefficient curves of these rings described that the positioning contributes to the characteristics of heat transfer coefficients, as shown in Fig. 5.4.18.
31
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
6000
5000
Point 2 Point 4 Point 6
Point 1
4000 3000 2000 1000
Point 2
5000
2
Point 1 Point 3 Point 5
Heat trans. coeff., W/(m °C)
2
Heat trans. coeff., W/(m °C)
6000
Point 3
4000
Point 4
3000 2000 1000
0
0
0
200
400
600
800
1000
0
200
Surface tempearture, °C
800
1000
Fig. 5.4.17 Heat trans. coeff. of a ring.
6000
6000
Point 1 Point 2 Point 3
2
5000
Heat trans. coeff., W/(m °C)
2
600
Surface tempearture, °C
Fig. 5.4.16 Heat trans. coeff. of a cylinder.
Heat trans. coeff., W/(m °C)
400
4000 3000 2000 1000 0
Point 1
5000
Point 2 Point 3
4000 3000 2000 1000 0
0
200
400
600
800
1000
0
200
Surface tempearture, °C
400
600
800
1000
Surface tempearture, °C
(a) Horizontal position.
(b) Vertical position.
Fig. 5.4.18 Heat transfer coefficient of ring. 6000
1200
Temperature, °C
1000
Heat trans. coeff., W/(m °C)
D=130 x L=390
2
D=30 x L=90 D=12.5 x L=60
800 600 400 200
D=130 x L=390
5000
D=30 x L=90 D=12.5 x L=60
4000 3000 2000 1000 0
0 0
10
20
30
40
50
0
60
200
400
600
Time, s
Surface tempearture, °C
(a) Cooling curves.
(b) Heat trans. coeff.
800
1000
Fig. 5.4.19 Heat transfer coefficient of cyrinders. In addition, Segerberg and Bodin quenched Inconel 600 cylinders with different diameters into the same quenching oil as the above, and obtained cooling curves with different shapes apparently, except only the 12.5 mm diameter cylinder was measured at its center, as shown in Fig. 5.4.19 (a). Then, they concluded that since the curves of heat transfer coefficient estimated based on the results agreed well each other as shown in Fig 5.4.19 (b), the value from the smallest diameter probe can be applied to larger 32
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
diameter cases. It was also described that a heat transfer coefficient by a silver probe was consistent with the results of the Inconel probe. In addition, the heat transfer coefficients in their paper were estimated based on cooling curves and numerical calculations of heat conduction problems. Using quenching test results from a hollow cylinder, 60 mm in outer diameter, 30 mm in inner diameter and 60 mm in height, made of Cr-Ni steel, X 6 CrNiTi18 19, Tensi and Lainer (1996) determined heat transfer coefficients on the inner and outer surfaces. In this study, based on the calculation of one-dimensional heat conduction and using cooling curves obtained at three points of probe, heat transfer coefficients were estimated by trial and error. In addition, the quenchant was oil at 60 °C, which was tested under still and 2 kinds of agitation conditions. Laumen et al. (1997) measured cooling curves at the center and a point near the surface, which is the position of 0.9R, R: radius, of an austenitic stainless steel cylinder, 28 mm in diameter and 112 mm in length, in order to determine heat transfer characteristics of a gas cooling. An actual temperature of the gas was measured at the gas entry and exit in a furnace. The radial temperature distribution of the probe was obtained by the finite deference method under a condition to match a calculated value at 0.9R with the measurement. Using the heat flux calculated by a similar manner to Liscic (1992) from the measured temperature at 0.9R and estimated surface temperature, and then heat transfer coefficients were obtained by applying the averaged of the measured gas temperatures additionally to Eq. (5.4.4). The heat transfer coefficient of the 10bar H2–N2 gas mixture, 0-100 % H2, were obtained as values in the range of from 400 to 600 W/(m2K). 5.4.6 Prediction of HTC based on inverse analysis As already noted, cooling curves obtained from low mass probes made of high thermal conductivity metals such as silver, gold and platinum can be used to predict heat transfer coefficients (HTC) simply by using the lumped heat capacity method. On the other hand, for conditions not applicable to this method, the inverse heat transfer analysis technique (Beck et al., 1985; Beck and Osman, 1992) has been developed. This can predict heat transfer coefficients for temperature dependent phenomena of the heat conduction. (1) One dimensional problem Wallis (1989) obtained temperature curves of heat transfer coefficients on upper and lower surfaces of the Alloy 718 disc during quenching it horizontally, using the CONTA program for one-dimensional inverse analysis, as shown in Fig. 5.4.20 (a). Units of temperature and heat transfer coefficient in this figure were converted from the original figure by the author. This inverse calculation used cooling curves measured at several thermocouples, the number of which was not stated in the report, located in the central part and along the thickness direction of the disc. Meanwhile, temperature curves of the heat transfer coefficients were obtained during quenching the disc vertically, and then the results are shown in Fig. 5.4.20 (b) for comparison with the case of quenching in the horizontal position.
33
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
3000
Heat trans. coeff., W/(m °C)
Top surface
Vertical Horizontal (top)
2
Bottom surface
2
Heat trans. coeff., W/(m °C)
3000
2000
1000
2000
1000
0
0 0
200
400
600
800
1000
1200
0
1400
Surface tempearture, °C
200
400
600
800
1000
1200
1400
Surface tempearture, °C
(a) Top and bottom surfaces in horizontal position.
(b) Vertical and horizontal positions.
Fig. 5.4.20 Predicted heat transfer coefficient of disc quenched into oil. Hernandez-Morales et al. (1992) predicted heat transfer coefficients by applying the CONTA program for one-dimensional inverse analysis, using cooling curves obtained from steel disc probes. Discs, 200 mm in diameter and 20 mm in thickness, were made of stainless or mild steels. Thermocouples were provided at only one point located 1.4 mm in depth from the surface on a single side of the disc. Using water, brine, oil and air as quenchants, discs were quenched vertically. Temperature curves of heat transfer coefficients were reported for various conditions, some of which were compared with the results in literatures. Felde et al. (2005) applied a one-dimensional inverse problem analysis to cooling curves obtained at the center of the Inconell 600 cylindrical probe, 12.5 mm in diameter, for predicting temperature curves of heat transfer coefficients. On the other hand, Felde et al. (2009) used the same method for cooling curves obtained from the JIS K 2242 silver probe. Obtained temperature curves of heat transfer coefficients were confirmed to be almost identical to those calculated from the lumped heat capacity method. (2) Two dimensional problem Trujillo and Wallis (1989) investigated heat transfer characteristics during oil quenching an Alloy 718 disc using the INTEMP program for two-dimensional axisymmetric inverse problem analysis. Cooling curves was measured at several points within the disc, 10.5 in (266.7 mm) in diameter and 2.75 in (69.9 mm) in thickness, during quenching it horizontally from 2150 °F (1177 ° C). Then applying the inverse analysis to this result, heat flux changes were obtained in 7 ranges on surfaces in an axisymmetric cross section. Heat transfer phenomena in Ni-based super alloy rings quenched in forced air and oil were examined by Ramakrishnan (1992) using the QUENCH2D program for two-dimensional axisymmetric inverse problems. Dimensions of the ring were 355.6 mm in outer diameter, 76.2 mm in inner diameter and 69.9 mm in height. Probes were heated to 1149 °C, and then were quenched horizontally into oils in the range of 21 to 24 °C temperature with agitation. In the case of air cooling, the heating was similar to the oil; on the other hand, cooling was done by high-speed airflow on the top and bottom surfaces of the ring simultaneously. Temperature curves of heat fluxes on ring surfaces for oil quenching and air cooling were obtained as shown in Fig. 5.4.21(a) and Fig. 5.4.22(a), respectively, by applying QUENCH2D to cooling curves 34
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
measured at 13 thermocouples installed inside the ring. The author converted these to heat transfer coefficients as shown in Fig. 5.4.21(b) and Fig. 5.4.22(b), respectively. In addition, these curves are corresponding to the four sides in the cross-section of the ring, upper, lower, inner and outer surfaces.
5000000
I. D. & O. D. surfaces Top surface Bottom surface
7000
2
Heat trans. coeff., W/(m °C)
2
4000000
Heat flux, W/m
8000
I. D. & O. D. surfaces Top surface Bottom surface
3000000 2000000 1000000 0
6000 5000 4000 3000 2000 1000 0
0
200
400
600
800
1000
1200
0
200
Surface tempearture, °C
400
600
800
1000
1200
Surface tempearture, °C
(a) Heat flux.
(b) Heat transfer coefficient.
Fig. 5.4.21 Predicted heat transfer characteristics of ring quenched into oil. 600
600000
Heat flux, W/m
2
2
500000
Heat trans. coeff., W/(m °C)
I. D. surface Top & Bottom surfaces O. D. surface
400000 300000 200000 100000
I. D. & O. D. surface Top surface Bottom surface
500 400 300 200 100 0
0 0
200
400
600
800
1000
0
1200
200
400
600
800
1000
1200
Surface tempearture, °C
Surface tempearture, °C
(a) Heat flux.
(b) Heat transfer coefficient.
Fig. 5.4.22 Predicted heat transfer characteristics of ring cooled by air. Sugianto et al. (2009) predicted heat transfer coefficients in different surface regions of a cylindrical stainless steel probe quenched into water using inverse analysis programs for axisymmetric two-dimensional problems. 6 thermocouples were provided at the center, at the upper and lower surface points, and at 3 points on the side in the cylinder, 20 mm in diameter and 60 mm in length. Three conditions of water quenching were still, and the 0.3 and 0.7 m/s agitation. Inverse calculations were attempted by two methods, one uses heat transfer coefficients of silver probe as initial values and another is software included in a commercial heat treating simulation code. Results from the still quench condition, which was reported overall, described that the predictions by the method using initial values of the silver probe agreed better with the experiments.
35
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
5.4.7 Prediction of HTC based on CFD The fluid flow simulation, which is also called as computational fluid dynamics (CFD), has recently become popular in a variety of fields. Although this is to be applied for quenchant flows, in general, its major role is for the salt or gas quenching. (1) Salt bath quenching The fluid flow analysis was applied to the salt bath quenching by Shick et al. (1996). In this study, first, the physical properties of the salt, such as viscosity, density, specific heat and thermal conductivity were reviewed. Cooling curves at the center of 1/4, 1/2, 1, 2 and 3 in diameter cylindrical probes made of stainless steel predicted by a flow analysis were compared with those described in past literatures. The 1/4 in diameter probes was quenched from 871 °C into the 258 °C salt bath. Meanwhile, the 1/2 in diameter and from 1 to 3 in diameter probes were quenched from 850 °C into 220 °C and from 900 °C into 200 °C, respectively. The predicted and experimental results agreed well. On the other hand, the CFD commercial code (STAR-CD) was applied to the fluid flow analysis of a steel ring during salt bath quenching. The ring, 56.7 mm in outer diameter, 35.3 mm in inner diameter and 21.0 mm in height, the upper surface of which is connected to the small ring, the same inner diameter, 43.3 mm in outside diameter and 3mm in height. The results of the fluid flow analysis estimated temperature curves of heat transfer coefficients on the four sides of the ring in the cross section as shown in Fig. 5.4.23.
2
Heat trans. coeff., W/(m °C)
7000
I. D. surface Top surface O. D. surface I. D. surface
6000 5000 4000 3000 2000 1000 0 0
200
400
600
800
1000
Surface tempearture, °C
Fig. 5.4.23 Predicted heat transfer coefficients of ring cooled in a salt bath. (2) Gas quenching Studies on a gas quenching by using the CFD software, ANSYS CFX, was reported by Macchion (2005). First, he summarized typical features and geometries of furnaces for the gas quenching, flow characteristics of pure and mixed gases, and turbulence models. As a matter of basic validation for CFD, wind tunnel experiments on cylinders (Wiberg, 2004), 150 mm in diameter and 300 mm in length, was applied. It was identified that surface heat transfer characteristics can be predicted well, when cooling the cylinder by a flow in the axial direction. Next, axial direction flows around cylinders, 28 mm in diameter and 110 mm in length, were predicted, and surface temperature distributions at a cylinder were shown. Furthermore, it was 36
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
made clear that different parts of a quenching rack, for example, a rod or a plate, affected flow conditions. Subsequently, impacts of different heat transfer from gas composition were examined by CFD. In addition, the CFD and measured results of flow rates inside the furnace were compared when it was actually empty. Applying CFD to flows around the cylinders inserted into a furnace, the results were illustrated as distributions of the heat transfer coefficients obtained in a typical cylinder. How differences of furnace and duct shapes affect the flow was investigated by CFD. Like Macchion’s studies, Fritsching and Schmidt (2007) performed a comprehensive work on flow phenomena in the gas quenching, and proposed that phenomena could be modeled in a phased manner as macro, meso and micro scales. A macro scale model corresponds to one focused on the upstream flow of a furnace. A meso scale model deals with a flow state in a furnace with several components. On the other hand, a micro scale model is used to study flow phenomena around individual parts. Examples in comparison between simulation and experiment were shown in the each scale. In the case of micro-scale, a comparison of the heat transfer characteristics of part surfaces was reported. For simulating a flow around a nickel cylinder in an experimental furnace for gas quenching, Douce et al. (2005) applied FLUENT, as the CFD software. The cylinder, 15 mm in diameter and 80 mm in length, was placed vertically, and measured a phenomenon around it due to a fluid flow from the top by the PIV (Particle Image Velocimetry). A simulation by FLUENT was performed for this flow condition, which was compared with an experimental velocity distribution. In addition, heat transfer coefficients were estimated based on the flow information, which were used for a distortion simulation. A nitrogen flow due to gas jet nozzles from four directions, for cooling an austenitic stainless steel cylinder, was analyzed by Frerichs (2004) using FLUENT. The velocity of the gas at the nozzle exit was from 100 to 255 m/s. As a result, heat transfer coefficient distributions were obtained on the surfaces of the cylinder. Heat transfer analysis performed based on these data showed that simulated cooling curves agreed well with experiments at the center and at points on the surface of the cylinder. On the other hand, Stratton (2007) applied FLUENT to nitrogen gas flows, induced from a multiple jet nozzle, 100 m/s in velocity, for cooling a steel disc. Heat transfer coefficients estimated based on the results were applied to heat conduction analysis of the disc, and a simulated cooling rate at a point was compared with an experimental result. (3) Liquid quenching CFD has been attempted to apply to flow simulations induced even by liquid quenching. In these cases, boiling phenomena of liquid quenchants are basically not considered. Ohki (2001) applied CFD to cooling by oil jets for inner and outer diameter surfaces of a ring, and then estimated heat transfer coefficients by referencing flow velocities around them. Predicted cooling curves at several points of the ring cross section based on the obtained heat transfer coefficients were compared with the experiments. Kumar et al. (2007) applied FLUENT to a flow around pinion gears on a rack during oil quenching. In addition, a shape before gear cutting was used for the modeling of gear. 37
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
To optimize a flow in a quenching oil path of a press quenching apparatus for a ring gear, Schüttenberg (2007) used CFD. Dimensions of the ring gear were 220 mm in outer diameter, 132 mm in inner diameter and 32 mm in height. Heat transfer coefficients on surfaces of the gear were estimated from the results by CFD. There are few theoretical studies to obtain heat transfer coefficients in a two-phase flow, gas and liquid, during quenching. The research by Yamagami (2003) found by the author simulated a cooling phenomenon by a water jet on a high temperatures stainless plate using CFD included a function of fluid phase change. Also surface heat transfer coefficients were estimated by using an inverse analysis based on temperatures in the steel plate. 5.4.8 Outlook on deriving heat transfer coefficients for heat treatment simulation Ideally, if a coupling simulation is applied to both phenomena in a heat treated object and a flow around it, it is not necessary to specify heat transfer coefficients for its model. However, at present, these cases to which such a simulation can be applied are very limited. Heat transfer analysis in the heat treatment simulation needs to set heat transfer coefficients on surfaces of parts. As already mentioned, some probes to obtain cooling curves and cooling rate curves were developed to understand the cooling power of quenchants, and then national standards and even international standards were established for the system (Totten, Tensi and Liscic, 1997). The specific cooling curves obtained from the individual probes can be used to make heat transfer coefficients by a lumped heat capacity method or an inverse analysis method. These coefficients may be useful for simulating practical problems as the first approximation. However, since these coefficients were obtained in terms of individual conditions, a large discrepancy with a reality may be made. For example, it is known the characteristic temperature of the probes depends on the type of metal. That is, Narazaki et al. (2002) noted that stainless steel probe showed the highest characteristics temperatures in oil quenching, and Ni and silver were the second and third, respectively, based on shapes of the temperature curves of the heat transfer coefficients. Beck and Moreaux (1992) also reported that the material dependence of the characteristic temperature could also be seen in the relationship between heat flux and surface temperature in 100 °C water quenching of Ni, Al and silver probes. In practice, it is believed that a good approach for estimating heat transfer coefficients close to reality is need by correcting the first approximation data obtained from the probes are by some approaches. The following gives an outlook how the data are collected and corrected. (1) Collection of first approximation data Cooling curves obtained from the standard measuring devices for the cooling power have provided heat transfer coefficients by a lumped heat capacity method or an inverse calculation, which were reported in literatures. These data are the first approximations, however may be effective to be used as is for practical purpose. Therefore, their database is significant by collecting from literature and new experimant. 38
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
Meanwhile, Peter (1949, 1950) reported a number of cooling rate curves, which can be converted into temperature curves of heat transfer coefficients using a lumped heat capacity method. It is considered that the same kinds of data as Peter’s are present. These data may be useful for making a database of temperature curves of heat transfer coefficients with the original data. On the other hand, heat transfer coefficients during cooling or in steady state, which were reported as empirical formulas, should be organized after clarifying the scope of these applications. (2) Correction methods on first approximation data In order to correct the approximate heat transfer coefficients more realistically, it is necessary to clarify the characteristics of the heat transfer at different locations on surfaces of simple shape components, such as cylinders, discs and rings in typical quenching conditions. This attempt was performed by Bodin and Segerberg (1992) as already mentioned. Also Trujillo and Wallis (1989), Ramakrishnan (1992) and Sugianto et al. (2009) worked on the same theme using two-dimensional inverse analysis software for some limited cases. These kinds of studies should be performed more precisely and systematically. Heat transfer characteristics at different locations on part surfaces, especially in case of the liquid quenchants, may be confirmed based on a collapse of vapor film, which corresponds to the end of the vapor film stage. This collapse in a simple shape probe such as sphere and cylinder was revealed by visualization experiments. For example, Tensi (1992a) showed it in the cases of quenching Cr-Ni steel cylinders, 25 mm in diameter and 100 mm in length, and triangular prism, 15×15×45 mm, from 860 °C into 35 °C distilled water. In the case of a triangular prism, boundary curves between vapor film and wetted parts were clearly photographed. Above systematic studies of positions dependences on the heat transfer should accompany visualization experiments on the vapor film collapse. In addition, there are questions how heat transfer characteristics are affected by part positions on a rack and also quenchant agitations. We will find the correction methods based on studies using experimental and simulation works steadily.
5.5 Temperature recovery and inverse hardening phenomena Descriptions on the heat conduction in a solid and the surface heat transfer have been terminated generally before this section, while the phase transformation has been already expressed in Chapter 3. Therefore, simulation examples on a phenomenon due to remarkably both heat conduction and phase transformation can be introduced here. That is a temperature recovery in a cooling curve during phase transformation by a heat generation, which may induce an inverse hardening phenomenon additionally. 5.5.1 Temperature recovery phenomenon When quenching a cylindrical steel specimen slowly, it has long been known that a temperature recovery called as recalescence phenomenon may occurs (Bain and Paxton, 1966). Takeo et al. (1974) quenched 0.84 %C-0.68% Mn steel cylinders with 11 different diameters, the range from 3.5 to 25 mm, into boiling water, and then measured cooling curves at its center. As a result, temperature recovery phenomena were 39
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
found in cooling curves from cylinders of more than 4.5 mm in diameter. In addition, this experiment was conducted in relation to a study on a patenting process of steel wires. As for the boiling water quenching, heat transfer coefficients were evaluated by Iwata et al. (1980) based on both the experimental and theoretical procedures. Experiments by Takeo et al. for the 5.5, 8.5, 13 and 25 mm diameter steel cylinders were simulated by Agarwal et al. (1981), and the temperature recovery generation was confirmed under the same experimental conditions. Simulated results of an air-cooled steel cylinder, 12 mm in diameter, showed a temperature recovery as well as an associated significant delay in the progress of the diffusion transformation at the center than near the surface. These experiments by Takeo et al. were also simulated by Arimoto et al. (1998), and it was confirmed simulated and experimental cooling curves agreed as shown in Fig. 5.5.1. In this simulation, a TTT diagram was used after time-shifting the data for 1080 steel by US Steel (1951) as shown in Fig 3.3.3 in Section 3.3.2, and data of the enthalpy of phase transformation was referred as 75.8 kJ/kg, which was derived from measurements by Kramer (1958). In addition, Fig. 5.5.1 depicted a simulated cooling curve using 83 kJ/kg as the enthalpy temporarily for reference.
Temperature, C
900
Measured 8.5mm 13mm 25mm Calculated 75.8kJ/kg 83kJ/kg
800 700 600 500 0
20
40 60 Time, s
80
100
Fig. 5.5.1 Temperature recovery phenomenon at center of 12 mm dia. eutectoid steel cylinder during 100°C water quenching. Otsuka et al. (1987) cooled a 1% Ni eutectoid steel cylinder, 3 mm in diameter and 10 mm in length, under about 10–6 Torr in a temperature – dilatation measuring apparatus, Formastor-F, after holding it for 15 min at 1000 °C. An obtained cooling curve showed a temperature recovery, which was predicted by considering a generated heat due to a phase transformation by the heat conduction analysis. 5.5.2 Inverse hardening phenomena After oil quenching of bearing races and rollers, an inverse hardening phenomenon had been found in the parts in the past. Tamura and Shimizu (1975) discussed that this phenomenon was produced by residual vapor films at some locations on surfaces during cooling. Then, they simulated experimentally this cooling delay due to the vapor films as air-cooling, and then defined a test using cylindrical specimens of simple 40
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
geometry for reproducing the phenomenon. In their experiment, SUJ2 steel cylinders with 6 different diameters, 10, 15, 20, 25, 30 and 36 mm, were held for 20 min at 840 °C, and then finally quenched into 5% brine after air cooling for various periods. A phenomenon in the 20 mm diameter cylinder, which was found in one of the tests by Shimizu and Tamura, was simulated by the authors (Arimoto et al., 2000). The simulation used the TTT diagrams shown in Fig 3.2.2 in section 3.2.5 for steel bearings, which were specified for the steels, 52100 (US Steel, 1951), En 31 (BISRA, 1949) and 100 Cr 6 (Max Planck Institute, 1954). Changes of temperature and volume fraction of pearlite at the center and on the surface of the specimen were obtained as shown in 5.5.2. The cooling curve at the center showed a similar temperature recovery phenomenon as shown in Fig. 5.5.1, while there was a delay of phase transformation progress at the center than the surface. These curves identified the different trends derived from each the TTT diagram. By neglecting the effect of the heat generation by phase transformation of steel 52100, the temperature recovery disappeared as shown in the figure. In this case, progress curves of phase transformations have the same shape at the center and the surface, that is, they are merely shifted as an interval of about 5 s. In this simulation, the enthalpy of phase transformation was referred as 75.8 kJ/kg from the measurements by Kramer et al. (1958). 700 0.5 0.9
680
0.9
0.9 0.5
Volume fraction
Temperature, C
800 750 700 650 600 550 0
S S C
Temperature
C
0.6
S C
C
S 52100 En 31 100 Cr 6 52100(without latent heat)
20
0.8
C
40 60 Time, s
0.4 C: Center S: Surface
80
0.2
0 100
Volume fraction transformed
1
Fig. 5.5.2 Predicted temperature and pearlite volume fraction in 20 mm dia. bearing steel cylinders.
Temperature, C
0.9
850
0.5
0.9
0.1
0.1
0.1
0.5
0.5 0.5
ξ=0.1
640
620
0.1
0.1
0.9
0.5
0.1
660
Lines of (dξ/dt)ξ=0.5 52100 En 31 100 Cr 6
0.9 Locus of dξ/dt 52100 En 31 (with latent heat) Center Center Surface Surface 100 Cr 6 (without latent Center heat) Surface Center Surface
600 0.01
0.02
0.04 dξ/dt, 1/s
0.1
Fig. 5.5.3 Loci of pearlite volume fraction rate at center and surface in 20 mm dia. bearing steel cylinders.
Based on the simulated results from the steel cylinder using the different TTT diagrams, Fig. 5.5.3 described loci of points which correspond to temperatures and transformation rates at the center and on the surface. In the figure, curves of the relationship between temperature and transformation rate at the 50 percent transformation, shown by dotted lines, which are hereinafter simply referred to as the transformation rate curve, are drawn for the 3 kinds of the TTT diagrams. On the other hand, loci of temperature and transformation rate points are created by connecting points from 10 to 90% of volume fraction with curve segments. Looking at the locus of the steel 52100, it is identified that their shapes at the 41
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
center and on the surface are quite different. The locus on the surface locates right than at the center, which means the transformation rate on the surface is greater than the center; therefore, the progress of phase transformation on the surface is faster. Incidentally, ignoring the effect of the heat generation due to phase transformation in the 52100 steel, the loci at the center and on the surface are almost same, therefore, they are overlaid as shown in the figure. The point on the locus at 50 % transformation contacts inevitably the transformation rate curve of the 50 % transformation. This means that constraints are imposed on existence ranges of the locus. For example, a comparison between En 31 and 52100 steels shows that transformation rate curve of the 50% transformation in En 31 is located in the lower left region than that of steel 52100. Thus, the locus of En 31 is also located in the lower left region inevitably than the 52100 steel. In addition, locating in the left region decreases overall an activity level of phase transformation, and locating in the lower region means phase transformations occur in lower temperature. Here, simulated results of the 52100 and 100 Cr 6 steels are compared. The transformation rate curve of 100 Cr 6 steel is located in left region than the curve of 52100 steel except the range higher than 680 °C. In addition, the curvature of the curve is smaller than the 52100 steel. Loci shapes at the center and on the surface in the 100 Cr 6 steel are similar, which may be contributed by the small curvature of the transformation rate curve. Meanwhile, since the En 31 and 52100 steels have larger curvatures in the transformation rate curve, tends are shown that loci at the center remains extremely in an area of smaller transformation rate than the surface. This is considered that a progress of transformation at the center decrease than the surface in the En 31 and 52100 steels. 70 50s
40s
Temperature, C
800
80s
Volume fraction
100s
750 700
90s 70s
0.8 0.6
80s Temperature
650
Center
600
Surface
550 0
1
90s
20
0.4 70s 60s 70s 80s 90s 100s 60s
40
60 80 Time, s
100
0.2
Pearlite volume fraction
850
Hardness, HRC
60 50
50s
60s
50s
60s 60s
65s
70s
70s
70s
80s 80s
80s 70s
40
90s
80s
90s
90s 100s
30
75s 100s
100s 80s with l. h. without l. h.
En 31 100 Cr 6
20 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
Depth, mm (i) Measured
0
120
Depth, mm Depth, mm (ii) Simulated (iii) Simulated 52100 En31&100Cr6
Fig. 5.5.4 Predicted temperature and pearlite volume Fig. 5.5.5 Experimental and simulated fraction in 20 mm dia. bearing steel cylinders hardness distribution in 20 mm dia. during air and brine cooling. bearing steel cylinders. In the experiments for the inverse hardening by Shimizu and Tamura, specimens were quenched into 42
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto
5% brine after air cooling for 60, 70, 80, 90 or 100 s. Fig. 5.5.4 shows simulated results by the authors, which include temperature and pearlite volume fraction changes on the surface and at the center during the two-stages cooling. In this simulation, the TTT diagram for the 52100 steel was used. As shown clearly in the figure, temperature in the specimens decrease rapidly during the 5% brine quenching and then the pearlite transformation on the surface is stopped almost instantly. On the other hand, pearlite transformation in the core is stopped after some continuity. In Fig. 5.5.5, simulated hardness distributions in the 20 mm diameter cylinder are compared with the experimental results by Shimizu and Tamura. Shapes of the simulated hardness distributions show the inverse phenomena, which are similar to the experiments. In addition, predicted values of hardness were derived by the weighted mean method using the harnesses and volume fractions of each martensite and pearlite. Experimental hardness distributions show the inverse phenomena in the cases of the air-cooling for from 50 to 100 s, while simulation predicted the phenomena in the cases of air cooling for from 60 to 100 s. In addition, predicted hardness distributions have different shapes, which occurred due to the differences of the TTT diagram for it. Incidentally, in the simulated results without heat generations due to phase transformations using the TTT diagram of steel 52100 show the inverse hardening in cases of air-cooling for from 60 to 75 s, which is obviously different from the experiments. As mentioned above, the simulated results did not agree fully with the experiments. The reason is that the TTT diagrams for the simulation did not correspond directly to the SUJ2 steel which was used for the specimens. Since it is clear that a distinct difference in the simulated results was induced from the different TTT diagrams, the simulation should use the TTT diagram of the SUJ2 steel used for the specimens. Other cause of the discrepancy may be produced from the data of the phase transformation enthalpy, since the data from the eutectic steel was used instead of the SUJ2 steel. In addition, experimental and simulated results for the cooling processes with temperature recovery phenomena could not be directly compared, since Shimizu and Tamura did not report their cooling curves. Then, the authors (Arimoto et al., 2004a) reproduced experimentally the inverse hardening phenomena using the 20 mm diameter cylindrical specimens, and confirmed to appear a similar temperature recovery in a cooling curve as shown in Fig. 5.5.2. This experiment was simulated based on the above mentioned approach, and then a tendency of simulated temperature changes during the temperature recovery agreed well with the experiments. However, they did not agree precisely, since the TTT diagram was not produced from the same grades of the actual specimen. Similarly, the exact agreement for the hardness distributions was not obtained.
Reference Agarwal, P. K. and Brimacombe, J. K., 1981, “Mathematical Model of Heat Flow and Austenite– Pearlite Transformation in Eutectoid Carbon Steel Rods for Wire,” Metall. Trans., Vol. 12B, pp. 121–133. Arimoto, K., Huang, D., Lambert, D. and Wu, W. T., 2000, “Computer Prediction and Evaluation of Inverse Quench–Hardening of Steel, 20th Heat Treating Conference, ASM International, pp. 737–746. Arimoto, K., Jin, C., Tamura, S., Funatani, K. and Tajima, M., 2004a, “Verification of Inverse Quench– Hardening
43
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto Phenomena in Bearing Steel Specimens by Experiment and Computer Simulation,” Transactions of Materials and Heat Treatment, Vol. 25, No. 5, pp. 465–470, (14th Congress of International Federation for Heat Treatment and Surface Engineering, October 26–28, 2004, Shanghai, China) Arimoto, K., Lambert, D., Li, G., Arvind, A. and Wu, W. T., 1998, “Development of Heat Treatment Simulation System Deform–HT,” Proc. 18th Heat Treating Conference, ASM International, pp. 639–644. Austin, J. B., 1932, “Heat Capacity of Iron – A Review,” Industrial and Engineering Chemistry, Vol. 24, pp. 1225–1235. Awbery, J. H. and Challoner, A. R., 1946, “The Physical Properties of A Series of Steels II – Specific Heat up to about 900C,” Alloy Steel Research Committee Paper No. 23/1946, JISI, Vol. 154, pp. 84–90. Awbery, J. H. and Griffiths, E., 1940, “The Thermal Capacity of Pure Iron,” Proc. R. Soc., Vol. 174, pp. 1–15. Awbery, J. H. and Snow, A., 1939, “The Physical Properties of A Series of Steels I – Total Heat at Various Temperatures up to 950C,” Spec. Report No.24, “Second Report of the Alloy Steels Research Committee,” Iron and Steel Inst., London, pp. 216–236. Bain, E. C. and Paxton, H. W., 1966, “Alloying Elements in Steel,” Second Edition, Third Printing, ASM, p251. Beck, J. V. and Osman, A. M., 1992, “Analysis of quenching and Heat Treating Processes Using Inverse Heat Conduction Method,” Proc. First Int. Conf. Quenching and Distortion Control, Chicago, pp. 147–153. Beck, J. V., Blackwell, B. and St. Clair, Jr., C. R., 1985, “Inverse Heat Conduction: Ill–Posed Problems,” Wiley–Interscience, New York. Benedicks, C., 1908, “Experimental Researches on the Cooling Power of Liquids, on Quenching Velocities, and on the Constituents Troostite and Austenite,” J. Iron & Steel Inst., Vol. 77, pp. 153–257. Bergles, A. E. and Thompson, W. G., 1970, “The Relationship of Quench Data to Steady–State Pool Boiling Data,” Int. J. heat Mass Transfer., Vol. 13, pp. 55–68. BISRA: British Iron and Steel Research Association, 1949, “Atlas of Isothermal Transformation Diagrams of B.S. En Steels,” Special Report No. 40, The Iron and Steel Institute, London. BISRA: The British Iron and Steel Research Association, 1953, “Physical Constants of Some Commercial Steels at Elevated Temperatures,” Butterworths Scientific Publications. Carney, D. J. and Janulionis, A. D., 1951, “An Examination of the Quenching Constant, H,” Trans. ASM, Vol. 43, pp. 480–496. Carney, D. J., 1954, “Another Look at Quenchants, Cooling Rates and Hardenability,” Trans. ASM, Vol. 46, pp. 882–927. Carslaw H. S. and Jaeger J. C., 1959, “Conduction of Heat in Solids,” 2nd Edition, Clarendon Press Oxford. Davis, A. H., 1924, “Convective Cooling of Wires in Streams of Viscous Liquids,” Philosophical Magazine, Vol. 47, pp. 1057–1092. Dinsdale, A. T., 1991, “SGTE Data for Pure Elements,” CALPHAD, Vol. 15, pp. 317–425. Douce, J. F., Bellot, J. P., Denis, S., Lamesle, P., Gouhinec, F., Chaffotte F., and Pellegrino, G., 2005, “Coupled Numerical Model for Gas Quenching and Experimental Validation,” 1st Int. Conf. on Distortion Engineering, Bremen, Germany, pp. 437–444. Elam, C. F., 1935, “Distortion of Metal Crystals,” Oxford University Press, London. Engel, N., 1931, “Untersuchungen über die Stahlhärtung,” Ingeniorvidenskabelige Skrifter. A. no. 31. (has not been gotten by the author) Enya, S., 1966, “Cooling Curves in Quenching: Unsteady Heat Conduction Problem with Boilng Heat Transfer as a Boundary Condition,” Seisan Kenkyu, Vol. 18, pp. 198-200. (in Japanese) Felde, I., Réti, T., Sarmiento and G., Smoljan, B., 2005, “A Novel Evaluation Method of Quenchants by Using Computer Simulation,” Proceedings of the 1st International conference on heat treatment and surface engineering of tools and dies, Pula, Croatia. Felde, I., Réti, T., Sarmiento, G., Smoljan, B. and A. Deus, 2009, “Analysis of a numerical method developed for estimation of the heat transfer coefficient obtained during quenching,” Proceedings of the 17th IFHTSE Congress (Journal of The Japan Society for Heat Treatment Vol.49 Special Issue). Fourier, J. B. J., 1822, “Theorie Analytique de la Chaleur,”. French, H. J., 1930a, “The Quenching of Steels,” ASST. (The same contents were in: French, H. J., 1930b, “A Study of the Quenching of Steels I & II,” Trans. ASST, May, pp. 646–727, June, pp. 798–888.) Frerichs, F., Lübben, Th., Fritsching, U., Lohner, H., Rocha, A., Löwisch, G., Hoffmann, F. and Mayr, P., 2004, “Simulation of gas quenching,” J. Phys. IV France, Vol. 120, pp. 727–735. Gröber, H., Erk, S. and Griugull, U., 1961, “Die Grundgesetze der Wärmeübertragung,” 3rd edition, Springer-Verlag. Grossmann, M. A., 1940, “Principles of Heat Treatment,” revised, ASM, p88. Guillermet, A. F. and Gustafson, P., 1985, “An assessment of the thermodynamic properties and the (p, T) phase diagram of iron,” High Temperature–High Pressures, Vol. 16, pp. 591–610. Hagel, W. C., Pound, G. M., Mehl, R. F., 1956, “Calorimetric Study of Austenite: Pearlite Transformation,” Acta Meta., Vol. 4, pp. 37–46.
44
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto Hara, O., 1935, “ON the Quenching Medium,” Tetsu- to- Hagane, Vol. 21, pp. 802–811. (in Japanese) Hernandez–Morales, B., Brimacombe, J. K., Hawbolt, E. B., and Gupta, S. M., 1992, “Determination of Quench Heat–Transfer Coefficients Using Inverse Techniques,” Proc. First Int. Conf. Quenching and Distortion Control, Chicago, pp. 155–164. Houdremont, E., 1956, “Handbuch der Sonderstahlkunde,” Springer–Verlag. Ichitani, K., 2004, “Improvement of Heat Transfer Accuracy in Heat Treatment Simulation,” Industrial heating, Vol. 41, pp. 29–35, -09 Ishihara, Y. and Ichihara, M., 1942, “Study on Temperature and Dilatation Curves with the Braun Tube,” J. Japan Inst. Metals, Vol.6, pp. 511–516. (in Japanese) Iwata, H., Satomi, Y., Ogita, H., Nakata, H. and Mizuhara, M., 1980, “Heat Transfer Coefficient in Patenting Process with Hot Water,” Tetsu – to – Hagane, Vol. 66, pp. 1327–1332. (in Japanese) Jakob, M., 1949, “Heat transfer”, Vol. 1, J. Wiley, New York. Jakob, M., 1957, “Heat transfer”, Vol. 2, J. Wiley, New York. Japan Society of Mechanical Engineers: Division Committee of Heat Transfer and Thermodynamics, Subcommittee on Boiling Heat Transfer Research, 1965, “Boiling Heat Transfer,” Japan Soc. Mech. Eng. (in Japanese) Jeschar, R., Specht, E. and Kohler, C., 1992, “Heat Transfer During Cooling of Heated Metallic Objects with Evaporating Liquids,” in “Theory and Technology of Quenching,” Springer–Verlag, pp. 73–92. Jones, J. A. and Stevenson, W. W., with Appendices by Heselwood, W. C. and Griffiths, E., 1939, “The Properties of Quenching Oils,” Spec. Report No.24, “Second Report of the Alloy Steels Research Committee,” Iron and Steel Inst., London, pp. 269–282. Jones, F. W. and Pumphrey, W. I., 1947, “Some Experiments on Quenching Media,” JISI, Vol. 156, pp. 37–54. Katto, Y., 1964, “Introduction to Heat Transfer,” Yokendo. Kaufman, L. and Cohen, M., 1956, “The Martensitic Transformation in the Iron–Nickel System,” Trans. AIME, Vol. 206, pp. 1393–1401. Kinbara, T., 1972, “Electricity and Magnetism I,” Syokabo. (in Japanese) Kobayashi, K., Nakatani, K. and Fujimura, Y., 1987, “A Study on Thermal Properties and Electrical Conductivity of Iron Carbon Binary System,” Netsu Bussei, Vol. 1, pp. 29–35. (in Japanese) Koiwa, M., 1998, “Diffusion in Materials- Historical and Recent Developments,” Bulletin of Japan Inst. Metals, Vol. 37, pp. 347–355. (in Japanese) Krainer, H. and Swoboda, K., 1944, “Die Wahl des Abschrecköles für das Härten von Baustahl,” Arch. Eisenhüttenwes., Vol 17, pp. 163–176. Kramer, J. K., Pound, G. M. and Mehl, R. F., 1958, “The Free Energy of Formation and The Interfacial Enthalpy in Pearlite,” Acta Meta., Vol. 6, pp. 763–771. Krielaart, G. P., Brakman, C. M., Zwaag, S., 1996, “Analysis of Phase Transformation in Fe–C Alloys Using Differential Scanning Calorimetry,” J. Mater. Sci., Vol.31, pp. 1501–1508. Kumar, A., Metwally, H., Paingankar, S. and MacKenzie, D. S., 2007, “Evaluation of Flow Uniformity Around Automotive Pinion Gears During Quenching,” 5th Int. Conf. on Quenching and Control of Distortion, Euro. Conf. on Heat Treat. 2007, Berlin, pp. 69–76. Lainer, K and Tensi, H. M., 1996, “Heat Transfer Coefficient and Heat Flux Density Distribution, Influenced by the Wetting Behavior of Hollow Cylinders,” 2nd International Conference on Quenching and Control of Distortion, 4–7 Nov., Cleveland, pp. 93–100. Laumen, C., Midea, S., Luebben, T., Hoffman, F. and Mayr, P., 1997, “Measured Heat Transfer Coefficients by Using Hydrogen as Quenchant in Comparison with Helium and Nitrogen,” in “International Symposium on Accelerated Cooling/Direct Quenching of Steels,” ASM. Lienhard, J. H. IV and Lienhard, J. H. V, 2011,“Heat Transfer Textbook,” 4th Edition, Phlogiston Press. Liscic, B., 1992, “Concept and Aims of the Temperature Gradient Method in the Workshop Practice Based on Heat–Flux Density,” in “Theory and Technology of Quenching,” Springer–Verlag, pp. 234–243. Liscic, B. and T. Filetin, 2011, “Global Database of Cooling Intensities of Liquid Quenchants,” Proceedings of the European Conference on Heat Treatment 2011 “Quality in Heat Treatment”, Wels, Austria, pp. 40–49. McAdams, W. H., 1954, “Heat transmission,” 3rd ed., Sponsored by The Committee on Heat Transmission National Research Council, McGraw-Hill, New York. Macchion, O., 2005, “CFD in the design of gas quenching furnace,” PhD Thesis, Department of Mechanics, Royal Institute of Technology, Stockholm, Sweden. Max Planck Institut für Eisenforschung, 1954, “Atlas zur Wärmebehandlung der Stähle,” Band 1, Verlag Stahleisen mbH, Düsseldorf. Ma, S., 2002, “Characterization of the performance of mineral oil based quenchants using CHTE Quench Probe System,” Thesis of Worcester Polytechnic Institute.
45
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto Maniruzzaman, M.and Sisson, Jr., R.D., 2004, “Heat Transfer Coefficients for Quenching Process Simulation,” J. Phys. IV Fraance, Vol. 120, pp. 269–276. Mendelson, A., 1968, “Plasticity: Theory and Application,” The Macmillan Co., New York. Miettinen, J., 1997, “Calculation of Solidification–Related Thermophysical Properties for Steels,” Met. Mat. Trans B, Vol. 28B, pp. 281–297. Mitsutsuka, M. and Fukuda, K., 1974, “Study on the Transition Boiling and Characteristic Temperature in a Cooling Curve during Water Quenching of Heated Metal,” Tetsu- to- Hagane, Vol. 60, pp. 2079–2084. (in Japanese) Mitsutsuka, M. and Fukuda, K., 1977, “Boiling Phenomena and Effects of Water Temperature on Heat Transfer in the Process of Immersion Cooling of a Heated Steel Plate,” Tetsu- to- Hagane, Vol. 63, pp. 1008–1015. (in Japanese) Moreaux, F. and Beck, G.., 1992, “Effect of Workpiece Surface Properties on Cooling Behavior,” in “Theory and Technology of Quenching,” Springer–Verlag, pp. 182–207. Nakagawa, Y. and Yoshida, T., 1950, “Rapid Heat Transfer from Solid to Liquid,” Kagaku Kikai, Vol. 14, pp. 286–291. (in Japanese) Nakagawa, Y. and Yoshida, T., 1952a, “Film Boiling During Rapid Cooling (Part 1) - Effect of Coolant Conditions,” Kagaku Kikai, Vol. 16, pp. 74–82. (in Japanese) Nakagawa, Y. and Yoshida, T., 1952b, “Film Boiling During Rapid Cooling (Part 2) - Effect of Heat Transfer Surface,” Kagaku Kikai, Vol. 16, pp. 104–109. (in Japanese) Nakagawa, Y. and Yoshida, T., 1952c, “Heat Transfer at Quenching Period (2nd Report),” Trans. Japan Soc. Mech. Eng., Vol. 18, pp. 16–23. (in Japanese) Narazaki, M., 1995, “Effects of Work Properties on Cooling Characteristics during Quenching,” Netsushori, Vol. 35, pp. 221–226. (in Japanese) Narazaki, M., Fuchizawa, S. and Takeda, N., 1988, “Effect of Facing Layer on Cooling Characteristics of Quenching Hot Metals into Water,” Netsushori, Vol. 28, pp. 279–285. (in Japanese) Narazaki, M., Fuchizawa, S. and Usuba, M., 1989, “Effects of Specimen Geometry on Characteristic Temperature during Quenching of Heated Metals in Subcooled Water,” Tetsu- to- Hagane, Vol. 75, pp. 634–641. (in Japanese) Narazaki, M., Kogawara, M., Shirayori, A. and Fuchizawa, S., 1997, “Analysis of Quenching Processes Using Lumped–Heat–Capacity Method,” Proceedings of the 6th International Seminar of IFHT, 10, pp. 428–435. Narazaki, M., Totten, G. E., and Webster, G. M., 2002, “Hardening by Reheating and Quenching,” in “Handbook of Residual Stress and Deformation of Steel,” Edited by Totten, G., Howes, M. and Inoue, T., ASM International, pp. 248–295. Nishio, S. and Uemura, M., 1986, “Study on Film Boiling Heat Transfer and Minimum Heat Flux Condition for Subcooled Boiling: 1st Report, Pool Boiling of Water at Atmospheric Pressure from Platinum Sphere,” Trans. Japan Soc. Mech. Eng., B, Vol. 52, pp. 1811–1816. (in Japanese) Nishizawa, T., 1973, “Thermodynamics of Iron Alloy (Part 1),” Bulletin Japan Inst. Metals, Vol. 12, pp. 35–47. (in Japanese) Nishizawa, T., 2008, “Thermodynamics of Microstructures,” ASM International. Nukiyama, S., 1934, “The Maximum and Minimum Values of the Heat Q Transmitted from Metal to Boiling Water under Atmospheric Pressure,” J Soc. Mech. Eng., Vol. 37, pp. 367–374. (in Japanese) (English translation: 1966, Int. J. Heat Mass Transfer, Vol. 9, pp. 1419-1433) Obinata, I, 1930, “A Method of measuring Quenching Velocity,” Kinzoku no Kenkyu, Vol. 7, pp. 161–168. (in Japanese) Ohki, C, 2001, “Application of Computer Simulation to Heat Treatment,” NTN Technical Review，No.69, pp. 61–68. (in Japanese) Ohtsuka, H., Umemoto, M. and Tamura, I., 1987, “The Effect of Transformation Latent Heat on the Progress of Phase Transformation during Continuous Cooling of Steel,” Tetsu- to- Hagane, Vol. 73, pp. 144–151. (in Japanese) Pallister, P. R., 1946, “The Physical Properties of A Series of Steels II – Specific Heat up to about 1300C,” Alloy Steel Research Committee Paper No. 23/1946, JISI, Vol. 154, pp. 90–96. Pallister, P. R., 1949, “The Specific Heat and Resistivity of High–Purity Iron up to 1250C,” JISI, Vol. 161, pp. 87–90. Peter, W., 1949, “Das Abkühlungsvermögen flüssiger Härtemittel,” Arch. Eisenhüttenwes., Vol. 20, pp. 263–274. Peter, W., 1950, “Der Einfluß der Oberflächenbeschaffenheit des Härtungsgutes auf den Abkühlungsvorgang in flüssigen Härtemitteln,” Arch. Eisenhüttenwes., Vol. 21, pp. 395–402. Pilling, N. B. and Lynch, T. D., 1920, “Cooling Properties of Technical Quenching Liquids,” Trans. AIME, Vol. 62, pp. 665–688. Powell, R. W. and Hickman, M. J., 1939, “The Physical Properties of A Series of Steels I – Thermal Conductivity and Electrical Resistivity,” Spec. Report No.24, “Second Report of the Alloy Steels Research Committee,” Iron and Steel Inst., London, pp. 242–251. Powell, R. W. and Hickman, M. J., 1946a, “The Physical Properties of A Series of Steels II – Electrical Resistivities up to 1300C,” Alloy Steel Research Committee Paper No. 23/1946, JISI, Vol. 154, pp. 99–105. Powell, R. W. and Hickman, M. J., 1946b, “The Physical Properties of A Series of Steels II – Thermal Conductivity of a 0.8 % Carbon Steel,” Alloy Steel Research Committee Paper No. 23/1946, JISI, Vol. 154, pp. 112–116.
46
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto Powell, R. W., 1939, “A Survey of Existing Data on the Thermal and Electrical Conductivities of Iron and Steels,” Spec. Report No.24, “Second Report of the Alloy Steels Research Committee,” Iron and Steel Inst., London, pp. 253–268. Powell, R. W., 1946a, “The Physical Properties of A Series of Steels II – Effect of Heat–Treatment on the Electrical Resistivity and Thermal Conductivity of a 13 % Manganese Steel (Steel 13); Thermal Conductivity Values up to 850C,” Alloy Steel Research Committee Paper No. 23/1946, JISI, Vol. 154, pp. 105–111. Powell, R. W., 1946b, “The Physical Properties of A Series of Steels II – Notes on the Estimation of the Thermal Conductivities of Steels,” Alloy Steel Research Committee Paper No. 23/1946, JISI, Vol. 154, pp. 116–121. Ramakrishnan, R. I., 1992, “Quench Analysis of Aerospace Components Using FEM,” Proc. First Int. Conf. Quenching and Distortion Control, Chicago, pp. 155–164. Rose, A., 1940, “Das Abkühlungsvermögen von Stahl–Abschreckmitteln,” Arch. Eisenhüttenwes., Vol. 13, pp. 345–354. Russell, T. F., 1939, “Some Tests on Quenching Oils,” Spec. Report No.24, “Second Report of the Alloy Steels Research Committee,” Iron and Steel Inst., London, pp. 283–298. Saunders, N. and Miodownik, A. P., 1998, “CALPHAD–Calculation of Phase Diagrams– A Comprehensive Guide,” Pergamon. Sato, S., 1933, “On the Effect of Facing on the Cooling Velocity of a Specimen During Quenching,” Kinzoku no Kenkyu, Vol. 10, pp. 63–70. (in Japanese) Schallbroch H., Bieling, W. and Blank, J., 1941, “Das Abschreckvermögen verschiedener Härtemittel,” Technische Zeitschrift für praktische Metallbearbeitung, Vol. 52, pp. 77-82. Scheil, E. and Normann, W., 1959, “Untersuchung Thermdynamisher Grossen der Martensitischen γ–α Umwandlung von Eisen–Nickel–Legierungen,” Arch. Eisenhüttenwes., Vol. 30, pp. 751–754. Schmidt, R. and Fritsching, U., 2007, “Multi-Scale Modelling and Experimental Validation of Gas Quenching Processes,” 5th Int. Conf. on Quenching and Control of Distortion, Euro. Conf. on Heat Treat. 2007, Berlin, pp. 85–93. Schüttenberg, S., Ströbl, M., Lohrmann, M. and Fritsching, U., 2007, “Controlling Flow Conditions During Ring Gear Press Quenching for Distortion Reduction,” 5th Int. Conf. on Quenching and Control of Distortion, Euro. Conf. on Heat Treat. 2007, Berlin, pp. 77–84. Scott, H., 1924, “Quenching Properties of Glycerin and Its Water Solution,” Trans. ASST, Vol. 6, pp. 13–32. Scott, H., 1934a, “Some Problems of Quenching Steel Cylinders,” Trans. ASM, Vol. 22, pp. 68–96. Scott, H., 1934b, “The Problem of Quenching Media for the Hardening of Steel,” Trans. ASM, Vol. 22, pp. 577–604. Segerberg, S. and Bodin, J., 1992, “Variation in the Heat Transfer Coefficient Around Components of Different Shapes During Quenching,” Proc. First Int. Conf. Quenching and Distortion Control, Chicago, pp. 165–170. Shick, D., Chenoweth, D., Palle, N., Mack, C., Copple, W, Lee, W. – T, Elliot, W., Park, J., Ludtka, G. M., Lenarduzzi, R., Walton, H., Howes, M., 1996, “Development of the Carburizing and Quench and Quenching Simulation Tool: Determination of Heat Transfer Boundary Conditions in Salt,” 2nd International Conference on Quenching and Control of Distortion, Cleveland, pp. 357–366. Shimada, M., Takeda, N. and Akabane, S., 1976, “Quenching Abilities of Sprayed Jet Cooling and Air Blast Cooling,” Tetsuto- Hagane, Vol. 62, pp. 372–379. (in Japanese) Shimizu, N. and Tamura I., 1976, “Inverse Quench-hardening of Steel,” Trans. ISIJ, Vol. 16, pp. 655–663. Shoji, M., 1995, “Heat Transfer Engineering,” University of Tokyo Press. Sinnott, M. J. and Shyne, J. C., 1952, “An Investigation of the Quenching Characteristics of a Salt Bath,” Trans. ASM, Vol. 44, pp. 758–774. Smith, C. S., 1940, “A Simple Method of Thermal Analysis Permitting Quantitative Measurements of Specific and Latent Heats,” Trans. AIME, Vol. 137, pp. 236–245. Speith, K. G. and Lange, H., 1935, “Ueber das Abschreckvermögen flüssiger Härtemittel,” Mitt. Kais. –Wilh. –Inst. Eisenforschg., Vol. 17, pp. 175–184. Stanfield, G., 1939, “Quenching Tests in Various Media,” Spec. Report No.24, “Second Report of the Alloy Steels Research Committee,” Iron and Steel Inst., London, pp. 299–342. Stratton, P., 2007, “Validating the Modelling of a Gas-jet Quenched carburized Gear,” 5th Int. Conf. on Quenching and Control of Distortion, Euro. Conf. on Heat Treat. 2007, Berlin, pp. 21–29. Sugianto, A., Narazaki, M., Kogawara, M. and Shirayori, A., 2009, “A Comparative Study on Determination Method of Heat Transfer Coefficient Using Inverse Heat Transfer and Iterative Modification,” J. Mat. Proc. Tech. Vol. 209, pp. 4627–4632 Sykes, C., 1936, “Methods for the Examination of Thermal Effects due to Order–Disorder Transformations,” J. Inst. Metal, Vol. 59, pp. 257–280. Tachibana F. and Enya S., 1972, “Heat Transfer Problems in Quenching,” Trans. Japan Soc. Mech. Eng., Vol. 38, pp. 1056-1064. (in Japanese) Tagaya, M. and Tamura, I., 1951a, “Studies on the Quenching Media (1st Report) The Apparatus and Method of Research,” J. Japan Inst. Metals B, Vol. 15, pp. 535–537. (in Japanese)
47
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto Tagaya, M. and Tamura, I., 1951b, “Studies on the Quenching Media (2nd Report) Observation of the Cooling Process in Quenching by Moving Picture,” J. Japan Inst. Metals B, Vol. 15, pp. 538–541. (in Japanese) Tagaya, M. and Tamura, I., 1951c, “Studies on the Quenching Media (3rd Report) Theoretical Study of Cooling Process during Quenching,” J. Japan Inst. Metals B, Vol. 15, pp.589–594. (in Japanese) Tagaya, M. and Tamura, I., 1952a, “Studies on the Quenching Media (4th Report). The Cooling Abilities of Water and Aqueous Liquids,” J. Japan Inst. Metals, Vol. 16, pp. 107–111. (in Japanese) Tagaya, M. and Tamura, I., 1952b, “Studies on the Quenching Media (5th Report). The Cooling Abilities of Fatty Oils,” J. Japan Inst. Metals, Vol. 16, pp. 342–346. (in Japanese) Tagaya, M. and Tamura, I., 1952c, “Studies on the Quenching Media (6th Report). Cooling Ability of Concentrated Salts Solutions,” J. Japan Inst. Metals, Vol. 16, pp.652–655. (in Japanese) Tagaya, M. and Tamura, I., 1953, “Studies on the Quenching Media (7th Report) The Cooling Ability of Mineral Oil,” J. Japan Inst. Metals, Vol. 17, pp. 217–220. (in Japanese) Tagaya, M. and Tamura, I., 1956a, “Studies on the Quenching Media (8th Report). The Effect of Dimensions of Silver Specimen on the Quenching Process,” J. Japan Inst. Metals, Vol. 20, pp. 124–128. (in Japanese) Tagaya, M. and Tamura, I., 1956b “Studies on the Quenching Media (9th Report) On the Severity of Quench, H, Determined Using Cylindrical Steel Specimens,” J. Japan Inst. Metals, Vol. 20, pp. 336–339. (in Japanese) Tagaya, M. and Tamura, I., 1956c, “Studies on the Quenching Media (10th Report). On the Severity of Quench Derived from Cooling Curves with Silver Specimen,” J. Japan Inst. Metals, Vol. 20, pp. 357–360. (in Japanese) Tagaya, M. and Tamura, I., 1956d, “Studies on the Quenching Media (11th Report). Correlation between Cooling Curves for Silver Specimen and Quenching of Steels,” J. Japan Inst. Metals, Vol. 20, pp. 428–432. (in Japanese) Tagaya, M. and Tamura, I., 1956e, “Studies on the Quenching Media (12 th Report) On the Severity of Quench, H, of Various Quenching Media,” J. Japan Inst. Metals, Vol. 20, pp.477–480. (in Japanese) Tajima, M., Maki, T. and Katayama, K., 1988, “Study of Heat Transfer Phenomena in Quenching of Steel: 3rd Report, Cooling Curves during Quenching and Transient Boiling Heat Transfer,” Trans. Japan Soc. Mech. Eng., Vol. 54, pp. 3491-3496. (in Japanese) Tajima, M., 1998, “The Effect of the Carbon Content on the Latent Heat of Phase Transformation of Iron and Steel,” Tetsuto- Hagane, Vol.84, pp. 7–12. (in Japanese) Takeo, K., Maeda, K., Kamise, T., Iwata, H., Satomi, Y. and Nakata, H., 1974, “The Direct Patenting of High Carbon Steel Wire Rod by Film Boiling,” Tetsu- to- Hagane, Vol. 60, pp. 2135–2144. (in Japanese) Tamura, I. and Tagaya, M., 1962, “On the Mother Cooling Curve for Hardening Steel,” J. Japan Inst. Metals, Vol. 26, pp. 651–656. (in Japanese) Tawara, S., 1941, “Experimental Research on the Cooling Power of Various Quenching Media Report I,” Tetsu- to- Hagane Vol. 27, pp. 583–599. (in Japanese) Tensi, H. M. und Steffen, E., 1985, “Neue Methode zur quantitativen Bestimmung der Abschreckwirkung flüssiger Härtemedien, hier speziell wässrige Kunststofflösungen (New Method for Determining the Quenching Effect of Liquid Hardening Substance, Here Especially Aqueous Synthetic Solutions),” Wärme–und Stoffüber., Vol. 19, pp. 279–286. Tensi, H. M., 1992a, “Wetting Kinetics,” in “Theory and Technology of Quenching,” Springer–Verlag, pp. 93–116. Tensi, H. M., 1992b, “Methods and Standerds for Laboratory Tests of Liquid Quenchants,” in “Theory and Technology of Quenching,” Springer–Verlag, pp. 208-218. Tokihiro, Y. and Tamura, I., 1974, “Unification on Cooling Curves of Quenched Bars at Center,” Tetsu- to- Hagane, Vol. 60, pp. 661–670. (in Japanese) Totten, G. E., Bates, C. E. and Clinton, N. A., 1993, “Handbook of Quenchants and Quenching Technology,” ASM International, Materials Park. Totten, G. E., Tensi, H. M. and Liscic, B., 1997, “Standards for Cooling Curve Analysis of Quenchants,” Heat Treatment of Metals, Vol. 4, pp. 92-94. Trujillo, D. M. and Wallis, R. A., 1989, “Determination of Heat Transfer from Components During Quenching,” Industrial Heating, Vol. 56, No. 7, pp. 22–24. Uehara, T. and Inoue, T., 1995, “Quenching Process Simulation of Japanese Sword Covered with Clay,” J. Soc. Mater. Sci., Japan, Vol. 44, pp. 309–315. (in Japanese) US Steel, 1951, “Atlas of Isothermal Transformation Diagrams,” United States Steel Company, Pittsburgh. Wallis, R. A., 1989, “Using Computer Programs to Calculate Heat Transfer,” Heat Treating, Vol. 21, No. 12, pp. 26, 27, 31. Wever, F., 1932, “Zur Theorie und Praxis der Stahlhärtung,” Arch. Eisenhüttenwes., Vol. 5, pp. 367–376. Wiberg, R. A, 2004, “Study of Heat Transfer from Cylinders in Turbulent Flows by Using Thermochromic Liquid Crystals,” Technical Reports from Royal Institute of Technology, Department of Mechanics/ FaxenLaboratoriet, SE-100 44 Stockholm, Sweden. Yamagami, S., 2003, “VOF Simulation and Its Experimental Verification on Hot Steel Disc Cooling by Water Jet, Master
48
Chapter 5, “Heat Conduction and Heat Transfer”, in “Heat Treating Distortion and Residual Stresses: Part 1”, by Kyozo Arimoto Thesis, University of Tokyo. (in Japanese) Yamaguchi, T., Fukuei, H., Shigematsu, H. and Harada, K., 1969, Study on the Cooling Rate of Steel in Heat Treatment,” Mitsubishi Juko Gihou, Vol. 6, pp. 30–39. (in Japanese) Yamazaki, S. and Okamoto, K., 1967, “Jet Cooling of Induction Surface Hardening,” Netsushori, Vol. 7, pp. 182–190. (in Japanese) Yoshida, T., 1950, “Heat Transmission at Quenching Period: 1st Report,” Trans. Japan Soc. Mech. Eng., Vol. 16, pp. 32–39. (in Japanese)
49
