Journal of Information & Computational Science 12:2 (2015) 641–656 Available at http://www.joics.com January 20, 2015 Simulation on the High Frequency Induction Cladding and Experimental Validation ? Yancong Liu a,b,∗, Chengkai Li b , Jianghao Xie b , Yongjun Shi b , Peng Yi b Guidong Sun b , Xianghua Zhan b , Xiaoli Ma b , Lanfang Cui c a Shengli b College College, China University of Petroleum, Dongying 257097, China of Mech. and Elec. Engineering, China University of Petroleum, Qingdao 266580, China c National Center of Quality Supervision and Inspection for Automobile Fittings of China, Yantai Products Quality Supervision and Testing Institute, Yantai 264001, China Abstract The simulation model of the High Frequency Induction Cladding (HFIC) is built using three-dimensional (3D) and its dependability is validated by the experimental measurement result. Firstly, a mathematical 3D model with respect to the HFIC process is established with corresponding assumptions and border condition. Secondly, the 3D model of the FEM simulation is validated by comparing the calculation results and the experimental results. Thirdly, the thermal distributions at vary time are investigated by the 3D model simulation. The results show that the 3D model simulation has significant effect on the analysis of the thermal distribution, optimal cladding process, and crystal phase transition. It is possible to obtain an ideal temperature distribution for the HFIC process using the 3D simulation calculation. Keywords: Induction Cladding; Experimental Measurement; Three-dimensional Model; FEM; Thermal Field 1 Introduction Induction heating has become quite popular now in industry due to its fast heating rate, good reproducibility and low energy consumption [1, 2], It is a technique of heating electrically conductive materials such as metals and alloys, and is commonly used in process heating prior to metalworking, heat treatment, welding, and melting [3], Recently, a newly-emerging application of the induction heating technique, known as the high-frequency induction cladding, has attracted much attention from process engineers [4]-[13]. ? Project supported by the Fundamental Research Funds for the Central Universities (14CX06061A), Fundamental Research Funds for the Central Universities, China (No. 13CX02076A), and Shandong Province Science and Technology Development Plans, China (No. 2011GGX10329). ∗ Corresponding author. Email address: liuyc@upc.edu.cn (Yancong Liu). 1548–7741 / Copyright © 2015 Binary Information Press DOI: 10.12733/jics20105297 642 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 The induction cladding process is a representative induction heat process [4, 5], including electromagnetic, thermal and metallurgic phenomena. In this process, an alternating electric current induces electromagnetic field, which in turn induces eddy currents in the substrate of the conductor. Making use of inductive skin effect is helpful to utilize heat source efficiently and to reduce heat influence on the bulk material. Then, the released heat from the induced eddy currents melts the surface self-fused alloy coat, and the alloy is coated on the surface of the substrate. The surface alloy coatings usually play an important role in resisting high temperature wear and corrosion for a variety of applications [6]. In the particular case where wear and corrosion resistance at low and moderate temperature are required, it can be expected that the induction cladding will be widely used in the future [7]. The high frequency is the first choice for induction cladding process duo to getting the best thermal distribution using the skin effect. Besides, the application of the high frequency has many merits comparing the medium and low frequency, such as more high efficiency and more saving energy [8]. Therefore, it is indispensable for the study on the induction cladding to adopting the high frequency as the inductive heat source. Many studies have been carried out on the induction heat model to investigate the effects of the process parameters in recent years. Naar and Bay [9] achieved an accurate control of temperature distribution and evolution in the heat affected zone by formulating a classical optimization problem. Shokouhm and Ghaffari [10] analyzed the effects of process parameters on moving induction heat from a hollow cylinder. Luozzo et al. [11] made use of the finite element method to conduct the numerical simulations of the heating stage of the bonding process. Yi Han et al. [12] analyzed the influences of the current frequency, the current density and the distance between the coil and the weld seam on the heating efficiency and the temperature difference across the thickness of weld seam using the finite element method. S. Hansson and Fisk [13] simulated the manufacturing process chain at glass-lubricated extrusion of stainless steel tubing by the finite element method. Besides, many investigations have been done with respect to the induction heat process parameters using the numerical model. However, by far, the simulation of the high frequency induction cladding has not been investigated. As there is no single universal computational method that can optimally fit all cases and be optimum for modeling all induction heating applications [14], it is necessary to study an appropriative computational model of the high frequency induction cladding (FHIC) process. The objective of this study is to build three-dimensional (3D) model of the induction cladding process and to validate the dependability of the 3D model. Firstly, the 3D geometry numerical model of the shaft workpiece of 45 Steel is established for the HFIC process. Secondly, the validation of the 3D Finite Element Method (FEM) model is carried out by using the experiment measurement results. Thirdly, based on the results of the simulation calculation, the thermal distribution is investigated in detail. 2 Numerical Modeling Generally, for the axisymmetrical geometry, the 3D geometry models with FEM method are simplified into two-dimensional (2D) geometry model or even one-dimensional geometry model. This can be attributed to the following reasons [15]. Firstly, the skin effect of thermal distribution in the low and medium frequency induction heating process is not remarkable, so the simplification of axisymmetric geometry models is applicable on that condition. Secondly, the precision of Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 643 controlling thermal distribution with two-dimensional or one-dimensional geometry model is not demanded rigorously for some induction heating applications. Thirdly, it takes much more calculation time using the 3D geometry model than using the two-dimensional or the one-dimensional geometry model. Therefore, the accurate 3D geometry model with FEM method is rarely adopted in induction heating numerical simulation. In our case, because the high-frequency current is employed in the induction heating cladding process, the contribution of skin effect is remarkable. Besides, technical requirements for the induction cladding are needed that metallic coating on the surface is melted by the induction Joule heat and the temperature from the inner part of the shaft workpiece is less than 600◦ C as far as possible [16]. Thus, the more accurate calculation requires the adoption of the 3D geometry model. Since the state-of-the-art computer technology provides more efficient ability in numerical calculation, the calculation time when the 3D geometry model is used can be readily neglected. Therefore, in this case, the 3D geometry model is adopted to the HFIC process for achieving more accurate numerical solution. 2.1 Assumptions The generating heat of the coating needs no consideration for the reason that HFIC coating thickness is about one to two millimeters, so the generating heat of the coating is ignored on the FHIC process and the heat conduction is only considerate into the process. Meanwhile, the assumptions made for the numerical modeling in the process are as follows: (1) The steel material is homogeneous and isotropic; (2) The heat conduction and eddy currents effect for the helix coil are not considered in the process; (3) Hysteresis loss of the high-frequency induction process is ignored; (4) The magnetic saturation for the steel material is neglected. 2.2 Domain Dividing The whole domain for the HFIC process is shown in Fig. 1 (a). Considering the symmetry of revolution and the symmetry shaft workpiece, in this case, the whole domain is divided into three model domain zones, including shaft substrate, coating, inductor coil and air. The model domains zones are shown in Fig. 1 (b) and provided in detail as follow: Ω1: Shaft substrate zone. It is a heat-generating region by electric conduction and induction, where induced magnetic field and induced electric field can be stimulated. Free electricity exist duo to the conductor of shaft workpiece, so Joule heat of induction process can be generated by induced current in this region. Ω2: Coating zone. It is a magnetic conduction and heat conduction region. In this case, it mainly receives the heat by the heat conduction of the shaft substrate. Ω3: Induction coil zone. It is an inducing eddy current drive region, where driving current AC is enforced. As Joule heat of the induction coil conducting is taken away by the cooling water inside coil, Free electricity is not taken into account. Ω4: Air zone. It is a heat convection and heat radiation region. Phenomenon of convection 644 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 Axis of symmetry Γ0e Ω1 shaft substrate Ω2 coating Ω3 induction coil Ω4 air Artificial surrounding box e Γe−Γ0 (a) Whole domain (b) Model domains Fig. 1: Domain dividing in the HFIC process and radiation takes place in this zone. For using the standard finite element method, it is necessary that a closed domain with an artificial border is taken into account. Considering that shaft workpiece is axis-symmetrical geometric profile, a null Dirichlet boundary condition on the symmetry axis Γe0 is prescribed for the electrical field. Avoiding artificial reflections or external borders Γe − Γe0 an absorbing-type Robin-like condition is prescribed [17, 18], as shown in Fig. 1 (b). The numerical models are summarized below, which of proper continuous equations are written with proper applied boundary conditions. 2.3 Electromagnetic Model The global system in the electromagnetic fields can be described by the following four Maxwell equations: ∇·D=ρ ∇·B=0 ∂B ∇×E=− ∂t ∂D ∇×H=J+ ∂t (1) (2) (3) (4) where D is the electric flux density, B is the magnetic flux density, E is the electric field intensity, H is the magnetic field intensity, J is the electric current density associated with free charges, and is the electric charge density. ∇× and ∇· denote the curl operator and divergence operator, respectively. Joule heat of conductor is the focus of analysis, which is produced by free charge directional migration in induction heating process. Besides, since displacement current density ∂D/∂t of conductor is no reason of generating the Joule heat, and when the frequency of the current is less than 10 MHz, the induced conduction current J is much greater than the displacement current density ∂D/∂t, the displacement current density ∂D/∂t is negligible compared with the conduction current density [19]. Thus, the Maxwell equations can be written as: ∇·B=0 (5) Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 ∂B ∂t ∇×H=J 645 (6) ∇×E=− (7) There are constitutive relations for the intrinsic material properties: B = µH (8) J = σE(Ohm0 slaw) (9) where µ and σ are the magnetic permeability and electrical conductivity, respectively. They are temperature independent parameters. Since the magnetic flux density B satisfies the zero divergence condition Eq. (5), a magnetic vector potential A is given in accordance with Helmholtz theorem. B=∇×A (10) Substituting Eq. (10) into Eq. (6), it can be obtained: ³ ∂A ´ ∇× E+ ∂t (11) Electric scalar potential ϕ is introduced due to the reason that ∇ × (∇ϕ) = 0 satisfies Helmholtz theorem. Then, Eq. (11) can be further derived as: E=− ∂A − ∇ϕ or ∂t E= ∂A + ∇ϕ ∂t (12) Eq. (10) and Eq. (12) ensure that Eq. (5) and Eq. (6) can be satisfied. The constitutive relations Eq. (8) and Eq. (9) are taken account in conjunction with Eq. (7). Thus, substituting Eq. (5) to Eq. (9) into Eq. (7), it can be obtained: ∇× 1 ∂A ∇×A+σ + σ∇ϕ = 0 µ ∂t (13) Electric scalar potential ϕ and magnetic vector potential A are not unique, for different gauges have various values. Furthermore, Lorentz gauge and Coulomb gauge are usually employed to solve the electromagnet analysis formula. However, in our case, on magneto-quasi-static approximation condition, the utilization of Coulomb gauge can simplify the model equations compared with that of Lorentz gauge, since scalar potential and vector potential are controlled by electric scalar potential ϕ and magnetic vector potential A, respectively. ThereforeCoulomb gauge expressed as ∇ · A = 0 is put into use. The penalty function ∇ µ1 (∇ · A) is added to Eq. (13) and the following expression can be acquired: ∇× 1 ∂A 1 ∇ × A − ∇ (∇ · A) + σ + σ∇ϕ = 0 µ µ ∂t (14) Introducing the formula of curl ∇ × ∇ × A = ∇(∇ · A) − ∇2 A, and substituting it into Eq. (14), it can be obtained: 1 2 ∂A ∇ A+σ + σ∇ϕ = 0 (15) µ ∂t 646 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 Substituting Eq. (9) into Eq. (11), it can be obtained: J = −σ ∂A − σ∇ϕ = Je + Js ∂t (16) where Je and Js are eddy induction current density and source current density, respectively. Therefore, equation Js = σ∇ϕ can be described. For time harmonic analysis, electromagnetic field quantities are oscillating functions of a single frequency. Thus, A can be expressed as: A = A0 ejωt (17) Js = J0 ejωt (18) where J0 and A0 are the amplitude of source current density and magnetic vector potential, respectively, and ω is the angular frequency expressed as ω = 2πf . Substituting A and Js into Eq. (15), it can be obtained: 1 2 ∇ A0 + J0 + jωσA0 = 0 µ (19) When dealing with axi-symmetrical configurations, Eq. (19) can be transformed into the equation of cylindrical coordinate (R, ϕ, Z). Then, Eq. (19) is rewritten as: 1 ∂A0 ∂ 2 A0 A0 i 1 h ∂ 2 A0 + + − 2 = −J0 − jωσA0 µ ∂R2 R ∂R ∂Z2 R (20) The above presented equations are employed to describe different mediums in electromagnetic fields, including dielectric, conductor and magnetic medium. In our case, three different regions with various mediums, i.e. the workpiece, the coil, and the surrounding air, can be analyzed as follows. For the workpiece Ω1 region, there is no source term. The equation for this region can be written as: ∂A 1 2 ∇ A+σ =0 (21) µ ∂t For the coil Ω2 region, because the coil is linked to an external source current, the current density of the induction coil is composed of two components: induced part and impressed part. The induced part is stimulated by the time harmonic magnetic field B, while the impressed part is defined by the gradient of the electric scalar potential Js = σE = σ∇ϕ and is connected to an external source. Thus, for the coil, it can be obtained: ∂A 1 2 ∇ A+σ + Js = 0 µ ∂t (22) For the air Ω3 region, there is no current density and electrical conductivity is zero. Then, Eq. (15) can be simplified as: 1 2 ∇ A=0 (23) µ Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 647 The source current density Js must be input into Eq. (22). When the induction coil loads an alternating voltage source, Js is unknown. In this case, it is necessary to compute the equivalent impedance of the coil-workpiece by circuit analysis. Finally, the boundary condition is either Dirichlet or Neumanntype along with a given surface depending on whether B · n = 0 or B × n = 0 has to be satisfied. Continuity conditions between regions of different properties are also provided as follows [20]: Ω1: Normal direction: n · (D2 − D1 ) = σeo (24) n · (B2 − B1 ) = 0 (25) n · (J2 − J1 ) = − ∂σeo ∂t (26) Tangent direction: n × Hout n × (E2 − E1 ) = 0 = i0 or n × (H2 − H1 ) = 0 (27) (28) Ω2: Normal direction: n × (J2 − J1 ) = 0 (29) n · (B2 − B1 ) = 0 (30) n × (H2 − H1 ) = 0 (31) Ω3: Normal direction: Ω3: Tangent direction: where Di , Bi , Ji and Ei (i = 1, 2) are the electric flux density, the magnetic flux density and eddy induction current density on both sides of the boundary, respectively, n is unit normal vector, σe0 is free charge surface density, i0 is surface current density on the surface of the conductor boundary, and Hout is magnetic field intensity, of which direction is point to the outside along the interface. Especially, under the condition of the Ultrahigh Frequency (UHF), n × Hout = i0 is used as the substitution for n × (H2 − H1 ) = 0 as tangent direction boundary condition. In our study, the high frequency induction eddy is laid upon the workpiece, so n × (H2 − H1 ) = 0 is employed to the boundary condition. 2.4 Heat Transfer Model The induction heating processes involve conservation of energy and Fourier law. Energy transferring in the induction heating process is governed by the classical heat transfer equation, which can be expressed as: ∂T ρC − ∇ · (k∇T ) = Q (32) ∂t 648 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 where ρ is the material density, C is the specific heat, T is temperature time function, k is the thermal conductivity, and Q is the heat source term due to eddy currents [21]. In our induction heating process, the specific heat C and the thermal conductivity k are assumed to be isotropic with temperature dependence. In this case, with regard to the axi-symmetrical workpiece, the classical heat transfer equation Eq. (25) in the cylindrical coordinate (R, ϕ, Z) can be rewritten as: ∂T ∂T ³ ∂T ´ 1 ∂ ³ ∂T ´ ρC − k − kR =Q (33) ∂t ∂Z ∂t R ∂R ∂R The heat source of the workpiece is generated by the eddy current. Connecting to the Joule heat theorem, by substituting Eq. (18), it can be obtained: ³ ∂A ´2 h ∂(A ejωt ) i2 0 2 Q = J · E = σE = σ =σ = σ(jωσA0 ejωt )2 (34) ∂t ∂t In our case, boundary conditions of the workpiece for temperature can be described through its normal derivative at interfaces. Therefore, modeling function with respect to heat flux, convection (− T ) k = k0 + (k0 − k∞ )e T0 and radiation between the part and the air, and it can be expressed as: −k ∂T 4 = h(T − Tair ) + εemi σb (T 4 − Tair ) ∂n (35) where n is the outward unit normal vector, Tair is the environment temperature of the workpiece, and h, εemi and σb are the convection coefficient, the material emissivity, and the Stephan constant, respectively. 2.5 Material Properties Model Material properties, including the thermal conductivity, specific heat, magnetic permeability and electric resistivity, are nonlinearly correlated with temperature in the heating process. The models of material properties can be obtained through the mathematical methods. The temperature-dependent thermal conductivity and specific heat are expressed as follows [22]: (− T ) (36) k = k0 + (k0 − k∞ )e T0 where k0 , k∞ and T0 is thermal conductivity value for zero degree centigrade, thermal conductivity value for ∞ degree centigrade and initial temperature constant, respectively. h 1 ³ T − T ´2 i Eb b (− T ) ρCP = √ e − + (V0 − V∞ )e T0 + V∞ (37) 2 S0 S0 2π where V0 , V∞ , s0 , Eb and Tb is ρCp value for zero degree centigrade, ρCp value for ∞ degree centigrade, ferrite to austenite temperature transition, transition energy for the ferrite to austenite transition and standard deviation of the Gaussian part of equation in temperature, respectively. The studies of workpieces are made of carbon steel, which is the ferromagnetic material. The temperature-dependent magnetic permeability and electric resistivity are described as follows [23]: h (µ − 1)πµ |H| i³ T´ 2Ms r0 0 arctan 1− (38) µ= π 2Ms Tc 649 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 ρe = ρ0 + ρ1 arctan ³T − T ´ c Tr (39) where Ms is the saturation magnetization at zero degree centigrade, µr0 is the initial relative permeability, µr0 is vacuum magnetic permeability(4π×107 H/m) and Tc is the Curie temperature, and ρe is electric resistivity. 3 FEM Simulation Based on the electromagnetic model, heat transfer model, and material properties model, the Finite Element Method (FEM) is adopted to carry out the different stages of the numerical simulation and ANSYS software is employed to establish the 3D simulation model. 3.1 Temperature-dependent Material In this study, the shaft parts of the carbon 45 steel are used to perform FEM simulation of the HFIC process. The chemical composition of the carbon 45 steel is shown in Table 1. In consideration of the chemical composition of the carbon 45 steel, the Curie temperature Tc = 768◦ C is employed in the induction heating process [25]. Based on an initial relative permeability µro = 700 and a saturation magnetization Ms = 2.2T , by taking Eq. (36) to Eq. (39) into calculation for the process, we can obtain the temperature-dependent material properties data including electrical resistivity, thermal conductivity, specific heat, and relative permeability. The value of the temperature-dependent material properties are illustrated in Fig. 2. Relative permeability gradually reduces with temperature incrementing, and when the temperature is up to the Curie temperature, relative permeability is down to zero. Electrical resistivity steadily increases and thermal conductivity slowly reduces along with the temperature varying. Specific heat fiercely enlarges with the temperature changing. With the temperature up to 756◦ C, the specific heat rapidly reduces. Therefore, the material properties are non-linear parameters, and it is necessary to take into account the temperature-dependent properties for the further precise study on the induction heating process. Table 1: Chemical composition of carbon 45 steel (mass%) [24] C Si Mn S P Cr Mo Ni Al Cu Fe 0.45 0.25 0.65 0.025 0.008 0.4 0.1 0.4 0.01 0.17 Bal. Table 2: Chemical composition of Ni60 power (mass%) [26] C Cr B Si Fe Ni 1.0 ∼ 0.6 17 ∼ 14 1.5 ∼ 2.5 4.5 ∼ 3 ≤ 15 Bal. A kind of the nickel base alloy Ni60 was used as the surface self-fused alloy coating due to the most widely used coating with accurately determined physical properties. Its chemical composition is shown in Table 2. From the above assumptions of the mention, the generating heat of the coating is not taken into account on the HFIC process. Therefore, the temperature-dependent 650 225 200 175 150 125 100 75 50 25 0 100 Electric resistivity (Ω·m) 225 200 175 150 125 100 75 50 25 0 100 1.2×10−6 1.0×10−6 8.0×10−7 6.0×10−7 4.0×10−7 2.0×10−7 0 100 300 500 700 900 1000 Temperature (°C) (a) 300 500 700 900 Temperature (°C) (b) 1100 300 500 700 900 Temperature (°C) (d) 1100 1100 Specific heat (J/Kg·K) Thermal conductivity (W/m·K) Relative permeability Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 1000 900 800 700 600 500 400 100 300 500 700 900 1000 Temperature (°C) (c) 100 540 90 Specific heat (J/Kg·K) Thermal conductivity (W/m·K) Fig. 2: Temperature-dependent material properties of the 45 steel 80 70 60 50 300 500 700 900 1100 Temperature (°C) (a) 1300 520 500 480 460 440 420 400 300 500 700 900 1100 Temperature (°C) (b) 1300 Fig. 3: Temperature-dependent material properties of the Ni60 material properties of the coating Ni60 only include the thermal conductivity and the specific heat and shown in Fig. 3. 3.2 Three-dimensional Model and Meshing The induction heating is usually applied to the axial symmetry shaft part. Thus, the previous researches about induction heating often used the 2D model to analyze the temperature distribution and other properties since the 2D model can simplify the simulation process and reduce the simulation time under certain conditions. In contrast, the 3D model is used relatively less due Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 651 to the difficulty in establishing the model and the more time for calculation. However, with the computer technology developing, the simulation time of the 3D model is remarkably shortened and thus induction heat process can be simulated with the 3D model without difficulty. It is well known that, the 3D model of induction heating process has many merits comparing with the 2D model, such as more fine mesh, more accurate results, more conspicuous temperature gradient, and more explicit normal direction of the heat conduction. Therefore, in this study work, we establish the 3D model of induction heating process and take advantage of the 3D model’s preponderance to analyze the parameters sensitivity in the process. The outline of 3D model for the HFIC process is shown in Fig. 4 (a). The solid elements of whole domain are divided into various groups in accordance with different material properties and thermal distribution. For the surface groups of the axial symmetry shaft part, dense meshes are set along with the coating line due to the significant skin effect of the highfrequency induction magnetic field intensity distribution. Otherwise, for the remaining groups of the axial symmetry shaft part, coarser meshes are set along with the radius direction. Then, with regard to the groups of the air and the coil, homogeneous meshes are set in the region. The mesh division of the 3D simulation model is shown in Fig. 4 (b). Therefore, through the measures of the meshing, the results of the simulated the 3D model are more accurate than the simulated the 2D model. Air Coil Surface Core (a) (b) Fig. 4: 3D simulation model and 3D mesh division 3.3 Numerical Result Coupling the electromagnetic with thermal fields is a characteristic for simulation of induction cladding at each time step. With the thermal field by the numerical simulation of the induction cladding process obtained, the outward contour and inward contour of temperature distribution of the shaft workpiece are shown in Fig. 5. From the figures of thermal distribution, it is observed that the skin effect is prominent in the high-frequency range, and the core temperature of the model thermal field is more lower than that of the model surface. 4 Experiment In order to validate the above 3D numerical model, a measurement experiment is designed. The detailed structure of the experiment equipments in the High Frequency Induction Cladding (HFIC) is illustrated in Fig. 6 (a). The cooling system provides circulating cooling water to take 652 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 (a) The outward contour of temperature distribution (b) The inward contour of temperature distribution Fig. 5: Temperature distribution of numerical simulation Thermocouple Infrared sensor Coating Induction coil A/D transverter Substrate Induction heating equipment Cooling system PC (a) Schematic model of the HFIC equipment (b) Eddy coil heating in the experiment process (c) Thermocouple measurement Fig. 6: Experimental measurement process the heat of the induction coil away from the induction heating equipment. Its aim is to sustain the induction coil temperature. The induction heating equipment supplies the high frequency Alternating Current (AC) to the induction coil. The AC of the induction coil stimulates the eddy current on the substrate. To measure the core and the surface temperature in this experiment, a thermocouple and an infrared sensor are installed. The gathering analog data of the temperature from the thermocouple and an infrared sensor is transfer to the digital analog (A/D) converter. Finally, the Personal Computer (PC) receives and saves the experiment measurement data from the A/D converter. The induction heating equipment has the rated power of 25 kW and the frequency ranging from 20 to 80 kHz. The recording temperature of the experiment environment is 22◦ C. The current density of showing on the induction equipment is 554A in the heating cladding process. From the Fig. 6 (b), we can note that the induction coil heats the shaft workpieces of carbon 45 steel and then the surface temperatures at different positions is measured by the infrared sensor and recorded by the PC. As shown in Fig. 6 (c), a core temperature of the shaft workpiece is measured by the thermocouple that is inserted into cylindrical bore of the shaft workpiece. Geometry data of the shaft workpiece and the coil are shown in Table 3, respectively. Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 653 Table 3: Geometry data for the workpiece (mass%) Size mm Outside diameter 20 Number of turns 4 Gap between coil and workpiece (on the radius) 10 Coil length 38.5 Workpiece length 5 55 Validation The temperature on the surface of the shaft workpiece is obtained by the infrared temperature measurement instrument, while the temperature at the core of shaft workpiece is gained by the thermocouple measurement. In the case, the measured and simulated time is 43 s. From Fig. 6, we can obtain that the maximal temperature is 1312◦ C, which is below the melting point of 45 steel (1350◦ C). When the heating process reaches Curie point, the rising rate of temperature decreases because material properties change. It could be observed that the veracity of temperature-dependent material properties plays such a significant role in predicting temperature. Temperature (°C) The temperatures of both the surface and the core are acquired for the same locations from the above FEM numerical simulation results. Therefore, the reliability of the numerical model for the induction cladding process could be validated by comparing the numerical simulation with the measurement results. The comparison of temperature distributions between the experimental measurement and the numerical simulation is shown in Fig. 7. The evolution of the temperature values on both the surface and the core of the shaft workpiece obtained by the numerical simulation shows a very good matching with that obtained by the experimental measurement. The discrepancy is within the interval [−25◦ C, 25◦ C]. Thus, referring to the comparison results, it can be concluded that the 3D model for the induction cladding process using FEM is reliable, and it can be employed for further study on the induction cladding process. 1500 1400 1300 1200 1100 1000 900 800 700 600 500 400 300 200 100 0 0 Surface Core Finite-element solution Experimental measurment 5 10 15 20 25 Time (s) 30 35 40 45 Fig. 7: Comparison of experimental measurement and numerical simulation of induction temperature 654 6 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 Discussion It is well known that the Austenite Start (AS) temperature is approximately 723◦ C for 45 steel [27], which means that when the temperature at any point of the component exceeds AS during the heating stage, the material at this point starts to become Austenite [28]. For this case, the occurrence of phase transformation of Austenite should be avoided so that the inner temperature of the workpiece is below AS. On the other hand, the ideal induction coating temperature field is that the temperature of the surface of the shaft is above 1300◦ C and the inner temperature of the shaft workpiece is below 600◦ C. Thus, to some extent, the layer location temperature 600◦ C can reflect the quality of the coating in this case, and the bigger thickness ratio is, the better the quality of induction heating process becomes. With the 3D simulation model established, the analysis of numerical simulation results with respect to thermal field can be investigated for the HFIC process. The analysis of thermal field under the condition above in detail is shown in Fig. 8. The thermal distributions on the threequarter volume are in different induction heat time, including 1 second, 3 second, 8 second, 20 second, 30 second, and 43 second. From Fig. 8 (a) and Fig. 8 (b), we can easy to note that the surface temperature of the coating is increase and the core temperature for the substrate remains the initial temperature duo to the skin effect of the induction. When the time is up to 8 second, the core temperature of the substrate raised to 125◦ C. At the time of 20 second, the core temperature of the substrate is heat up below 600◦ C and the temperature of the coating is 805◦ C. However, at the time of 30 second and 43 second, the core temperatures of the substrate seriously exceed 600◦ C and the surface temperature is heat up to the ideal coating melting point of 1200◦ C. The layer location temperature 600◦ C for this case is not exist duo to inductive heating time too long. Therefore, we can cut down the time of the inductive heat and increase the other value of the process parameters, such as the electric power. (a) 1 s (b) 3 s (c) 8 s (d) 20 s (e) 30 s (f) 43 s Fig. 8: Thermal distribution on the three-quarter volume Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 655 Numerical modeling of the HFIC normally involves three main physical phenomena related to electromagnetism, heat transfer and solid mechanics. As far as this study is concerned, electromagnetism and heat transfer are described above in detail. Solid mechanics can be indirectly researched based on the relation with heat transfer. The complete finite elements approach is chosen to carry out the coupling process in this study. We can take advantage of the 3D simulation model to analyze the thermal distribution, optimal cladding process, and crystal phase transition in the future research. 7 Conclusions This study has developed the 3D simulation model to analyze the HFIC process. Firstly, we have established a mathematical 3D model for the HFIC process by coupling electromagnetism with heat transfer computation. Secondly, the validation of the constructed model is accomplished by using the experiment measurement results. Finally, the thermal distribution for the HFIC is discussed to investigate the quality of induction heating process. The construction of the 3D geometry model in this case is of significance for the prediction of thermal field in the HFIC process. The constructed 3D model can be used as a powerful tool to simulate the induction process for better analyzing the process parameters. Moreover, our future research will focus on the further analysis of the process parameters by the 3D simulation model. References [1] Y. Favennec, V. Labb´e, F. Bay, Induction heating processes optimization a general optimal control approach, J. Comput. Phys., 187 (2003), 68-94 [2] J. Acero, C. Carretero, R. Alonso, J. M. Burd´ıo, Quantitative evaluation of induction efficiency in domestic induction heating applications, IEEE. Trans. Magn., 49 (2013), 1382-1389 [3] M. Kranjc, A. Zupanic, D. Miklavcic, T. Jarm, Numerical analysis and thermographic investigation of induction heating, Int. J. Heat. Mass. Tran., 53 (2010), 3585-3591 [4] Z. Z. Zhang, B. Ai, A cold adhesion for self-fused alloy coat by high frequency induction, Transactions of Materials and Heat Treatment, Proceedings of the 14th IFHTSE Congress, 25 (2004), 1316-1320 [5] B. Xiong, C. Cai, H. Wan, B. Lu, Fabrication of high chromium cast iron and medium carbon steel bimetal by liquid-solid casting in electromagnetic induction field, Mater. Design., 32 (2011), 2978-2982 [6] H. J. Kim, S. Y. Hwang, C. H. Lee, P. Juvanonc, Assessment of wear performance of flame sprayed and fused Ni-based coatings, Surf. Coat. Tech., 172 (2003), 262-269 [7] J. H. Chang, J. M. Chou, R. I. Hsieh, J. L. Lee, Microstructure and corrosion resistance of induction melted Fe-based alloy coating, Surf. Coat. Tech., 251 (2014), 300-306 [8] X. B. Zhang, Y. L. Yang, Y. J. Liu, Feasibility research on application of a high frequency induction heat to line heating technology, J. Marine Sci. Appl., 10 (2011), 456-464 [9] R. Naar, F. Bay, Numerical optimisation for induction heat treatment processes, Appl. Math. Model., 37 (2013), 2074-2085 656 Y. Liu et al. / Journal of Information & Computational Science 12:2 (2015) 641–656 [10] H. Shokouhmand, S. Ghaffari, Thermal analysis of moving induction heating of a hollow cylinder with subsequent spray cooling: Effect of velocity, initial position of coil, and geometry, Appl. Math. Model., 36 (2012), 4304-4323 [11] N. Di Luozzo, M. Fontana, B. Arcondo, Modeling of induction heating of carbon steel tubes: Mathematical analysis, numerical simulation and validation, J. Alloy. Compd., 536 (2012), S564S568 [12] Y. Han, E. Yu, H. L. Zhang, D. C. Huang, Numerical analysis on the medium-frequency induction heat treatment of welded pipe, Appl. Therm. Eng., 51 (2013), 212-217 [13] S. Hansson, M. Fisk, Simulations and measurements of combined induction heating and extrusion processes, Finite. Elem. Anal. Des., 46 (2010), 905-915 [14] V. Rudnev, Unique computer modeling approaches for simulation of induction heating and heattreating processes, J. Mater. Eng. Perform., 22 (2013), 1899-1906 [15] Kim Hyun Jung, Youn Sung Kie, Three dimensional analysis of high frequency induction welding of steel pipeswith impeder, J. Manuf. Sci. E-T Asme., 3(10), 2008, 51-57 [16] K. F. Wang, S. Shandrasekar, H. T. V. Yang, Finite-element simulation of moving induction heat treatment, J. Mater. Eng. Performance, 4(4), 1995, 460-473 [17] T. H. Pham, S. R. H. Hooles, Unconstrained optimization of coupled magneto-thermal problems, IEEE Trans. Magn., 31(3), 1995, 1988-1991 [18] M. Malinen, T. Huttunen, J. P. Kaipio, Optimal control for the ultrasound induced heating of a tumor, 4th International Conference on Inverse Problems in Engineering, Rio de Janeiro, Brasil, 2002 [19] Hong-Seok Park, Xuan-Phuong Dang, Optimization of the In-line Induction Heating Process for Hot Forging in Terms of Saving Operating Energy, Int. J. Precis. Eng. Man., 13(7), 2012, 1085-1093 [20] Dominique Coupard, Thierry Palin-luc, Philippe Bristiel, Vincent Ji, Christian Dumas, Residual stresses in surface induction hardening of steels: Comparison between experiment and simulation, Materials Science and Engineering A, 487 (2008), 328-339 [21] S. Zinn, S. L. Semiatin, Elements of Induction Heating: Design, Control and Applications, ASM International, Electronic Power Research Institute, 1988 ´ [22] P. Bristiel, Modelisation mag´ netothermique, metallurgique ´ et mecanique ´ de la trempe superficielle ap`res chauffage par induction appliq´ uee aux vilebrequins, PhD Thesis, ENSAM CER de Bordeaux, France, 2001 [23] D. Coupard, T. Palin-luc, P. Bristiel, V. Ji, C. Dumas, Residual stresses in surface induction hardening of steels: Comparison between experiment and simulation, Mater. Sci. Eng. A, 487 (2008), 328-339 [24] I. Magnabosco, P. Ferro, A. Tiziani, F. Bonollo, Induction heat treatment of a ISO C45 steel bar: Experimental and numerical analysis, Computational Materials Science, 35 (2006), 98-106 [25] V. I. Rudnev, D. Loveless, R. Cook, M. Black, Handbook of Induction Heating, Marcel Dekker, New York, 2003 [26] Guojian Xua, Muneharu Kutsuna, Zhongjie Liu, Hong Zhang, Characteristics of Ni-based coating layer formed by laser and plasma cladding processes, Materials Science and Engineering A, 417 (2006), 63-72 [27] K. F. Wang, S. Shandrasekar, H. T. V. Yang, Finite-element simulation of moving induction heat treatment, J. Mater. Eng. Perform., 4 (1995), 460-473 [28] P. Liua, Y. Y. Wang, J. Li, Parametric study of a sprocket system during heat-treatment process, Finite. Elem. Anal. Des., 40 (2003), 25-40
© Copyright 2024