† Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant No. 51901148), the Fund of the State Key Laboratory of Solidification Processing (Northwestern Polytechnical University), China (Grant No. SKLSP202006), and the State Key Lab of Advanced Metals and Materials (University of Science and Technology Beijing), China (Grant No. 2019-Z15).
We simulate the evolution of hydrogen concentration and gas pore formation as equiaxed dendrites grow during solidification of a hypoeutectic aluminum–silicon (Al–Si) alloy. The applied lattice Boltzmann-cellular automaton-finite difference model incorporates the physical mechanisms of solute and hydrogen partitioning on the solid/liquid interface, as well as the transports of solute and hydrogen. After the quantitative validation by the simulation of capillary intrusion, the model is utilized to investigate the growth of the equiaxed dendrites and hydrogen porosity formation for an Al–(5 wt.%)Si alloy under different solidification conditions. The simulation data reveal that the gas pores favorably nucleate in the corners surrounded by the nearby dendrite arms. Then, the gas pores grow in a competitive mode. With the cooling rate increasing, the competition among different growing gas pores is found to be hindered, which accordingly increases the pore number density in the final solidification microstructure. In the late solidification stage, even though the solid fraction is increasing, the mean concentration of hydrogen in the residue melt tends to be constant, corresponding to a dynamic equilibrium state of hydrogen concentration in liquid. As the cooling rate increases or the initial hydrogen concentration decreases, the temperature of gas pore nucleation, the porosity fraction, and the mean porosity size decrease, whilst the mean hydrogen concentration in liquid increases in the late solidification stage. The simulated data present identical trends with the experimental results reported in literature.
With the demands for light weight in automobiles and aircrafts, the application of aluminum alloys in these industries increases greatly. For aluminum alloy castings, the formation of microporosity is one of the most detrimental defects occurred in the solidification process.[1–4] The mechanical properties of the castings, particularly the fatigue resistance, could be reduced significantly due to the presence of the porosity defects.[4] Accordingly, the research of microporosity generation in aluminum alloy castings has collected extensive attention from the industrial production and academic study. Owing to the different porosity formation mechanisms, the microporosity defects can be categorized to shrinkage porosity and gas porosity. The former is caused by insufficient feeding of the melt because the permeability of the mush zone decreases with the solid fraction increasing, while the latter is produced due to the evolution of insoluble hydrogen gas in liquid during solidification.[4]
Numerous studies have been carried out to explore the mechanisms of gas pore formation through experimental observation and theoretical analyses.[5–12] The techniques of in situ observation for transparent organic materials using optical microscopy and for aluminum alloys using x-ray radiography provided a great deal of meaningful information, including the images of gas bubble emergence, bubble–dendrite interaction, bubble morphological evolution, and bubble movement.[5–7] Since the formation of gas bubbles in aluminum alloys is dependent on the variations of the local hydrogen concentrations and hydrogen solubility in melt, it is vital to obtain temporal and spatial evolution of hydrogen distributions in melts as solidification occurs. However, real-time quantitative determination of hydrogen concentration in aluminum alloy melts during solidification is so far unavailable. In the aspects of theoretical study, mathematical models were proposed and focused on the prediction of quantitative data under different conditions, such as porosity percentage and porosity size.[8–12] Yet, those models are incapable of providing the graphical morphological output of the microporosity and dendritic microstructure.
Nowadays, owing to the rapid development in the area of computing techniques, numerical modeling has emerged as an important complementary approach for the study of microstructural evolution in the process of alloy solidification. Numerical modeling not only reproduces the dynamic evolution of the solid fraction and concentration fields, but also fills the technical gap between the data obtained from experimental observations and theoretical analyses.[13–15] Various computational models based on cellular automaton (CA) and phase-field methods have been developed for simulating the formation and development of microporosity as the dendrites grow.[16–21] The simulation results have revealed the irregular shape of gas porosity affected by the nearby dendrites,[16] morphological variation of the growing bubbles and columnar dendrites after confronting each other,[17] formation of multiple gas pores in the process of welding of an Al–Cu alloy,[18] and gas porosity formation together with the equiaxed grains.[19,20] We previously developed a CA model for simulating hydrogen porosity formation in the dendritic and eutectic solidification stages for an Al–Si alloy.[21] The regimes of the hydrogen gas pores in the entire solidification processes were reasonably presented. In the above-mentioned simulations, however, the gas pore nuclei were either placed arbitrarily in the domain or generated by the stochastic model. As a result, some levels of artificial effects were inevitably introduced in the simulations, such as the number and position of the gas pore nuclei.
In past decades, the lattice Boltzmann (LB) models have been widely applied for simulating various multi-phase phenomena involving the gas/liquid/solid (G/L/S) phases.[22,23] By utilizing the nature of the multi-phase LB models for describing the interactions between fluid particles, the nucleation of a fluid phase in another phase owing to the evolution of the surrounding environment can be simulated.[24,25] The multi-phase LB models were coupled with the CA approach for modeling gas bubble formation as aluminum alloy solidification takes place,[26–28] in which the nucleation, movement, and merging of the growing bubbles during dendritic solidification could be visualized. However, these models have not encompassed the mechanism of hydrogen partitioning as dendrites grow. Recently, we developed a coupled model by combining the LB, CA, and finite difference (FD) methods to investigate the emergence of gas bubbles as the equiaxed and columnar dendrites grow for an Al–Cu alloy.[29] The LB-CA-FD model takes the effect of hydrogen partitioning into consideration, and this model has the capability of reasonably describing the evolution of hydrogen concentration as bubbles and dendrites grow. However, the influences of solidification conditions, such as cooling rate and initial hydrogen concentration, on gas pore formation and dendrite growth have not been investigated exhaustively.
In the present study, the LB-CA-FD model is utilized to simulate the nucleation and growth of hydrogen gas pores during dendritic solidification of an Al–(5 wt.%)Si alloy. This model could reproduce the hydrogen concentration variations, nucleation of the gas pores, and the morphological evolution of the growing hydrogen porosities and dendrites. The effects of cooling rate and the initial concentration of hydrogen in the Al–(5 wt.%)Si melt on gas pore nucleation and growth are investigated. Comparisons between the simulation results and the experimental data are performed for better understanding the effects of the various factors on microporosity formation.
In the recently proposed LB-CA-FD model, essentially the LB model describes the hydrogen concentration evolution and the features of bubble flows, while the CA-FD model calculates the dendrite growth and solute diffusion.[29] In the present study, the solidification shrinkage is not considered, and the effect of convection is neglected.
For the multi-relaxation-time (MRT) multi-phase LB model, the transport of hydrogen atoms is described by the evolution of the distribution functions, which is written as[23]
The distribution functions in the moment space are obtained by m = M f and meq = M feq, in which the distribution functions are written in vector forms. We can write meq as
After the linear transformation of Eq. (
S in Eq. (
The CA-FD model is briefly described as follows. Within each time step, the variation of solid fraction in an S/L interface cell is evaluated by[13]
In Eq. (
The effects of gas pore-dendrite interaction and hydrogen partitioning provide the connection between the multi-phase LB model and the CA-FD model. The hydrogen amount rejected at the S/L interface is incorporated into the multi-phase LB model through the source term in Eq. (
The physical and thermodynamic parameters of Al–Si alloys used in the simulations are displayed in Table
In our previous study, the proposed LB-CA-FD model was verified by the Laplace law test and contact angle simulation.[29] In order to further validate its capability for modeling the fluid dynamics in a complex system involving the G/L/S phases, the dynamic process of capillary intrusion is simulated. In this process, the wetting fluid phase migrates in a capillary tube since a pressure difference exists on two sides of the curved fluid-fluid interface. The simulated data are compared with a theoretical model applied in the work of Liu et al.[32] In the analytical model, the position of the migrating curved fluid-fluid interface in a capillary tube, ξ, is a function of time t,
In the simulation of capillary intrusion, the computation domain consists of 400 × 35 lattice units (l.u.). A horizontal capillary tube with a width of r = 21 l.u. and a length of L = 200 l.u. is set in the middle portion of the computation domain. As shown in Fig.
The simulations are carried out to investigate the growth of equiaxed dendrites and porosity formation as an Al–(5 wt.%)Si alloy solidifies under different cooling rates. The computational domain is composed of a 300 × 300 mesh, and the lattice unit and the mesh size are arranged as Δx = 1 l.u. = 3 μm. At the start, several initial equiaxed dendrites are set in the computational domain. The undercooling is arranged to be 3 °C, and the crystallographic orientations of the equiaxed dendrites are randomly generated. The initial concentration of hydrogen in the Al–(5 wt.%)Si melt is assigned as CH0 = 0.022. Figure
Figure
Figure
The influence of initial concentration of hydrogen in the Al–Si melt on the solidification porosity formation for Al–(5 wt.%)Si is investigated using the present LB-CA-FD model. Figure
Figure
Figure
The equiaxed dendrite growth and gas pore formation during hypoeutectic Al–Si alloy solidification is simulated using a previously proposed LB-CA-FD model. In this model, the hydrogen concentration variation as well as the gas pore nucleation and gas pore growth are described by the multi-phase LB model, whilst the growth of dendrites and solute transport are described by the CA-FD model.
The present model is validated by simulating the dynamic process of capillary intrusion of the wetting liquid phase. The simulated profile of the curved fluid-fluid interface position varying with time is in good agreement with the analytical prediction, revealing the capability of the present model for reasonably describing the interactions between different phases.
The phenomena of gas bubble formation and dendrite growth with different cooling rates and initial hydrogen concentrations are simulated using the LB-CA-FD model. The hydrogen concentrations increase as the solidification proceeds. When the hydrogen concentrations exceed a certain level, the gas pores favorably appear in the corners surrounded by dendrite arms. Then, the gas pores start to grow, and a competitive growth mode is observed among the gas pores with various radii. The simulations visualize the evolution and distribution of hydrogen concentrations during gas pore and dendrite growth, which is coherently affected by hydrogen rejection as dendrites grow and the hydrogen consumption by gas pore nucleation and growth. It is found that the mean hydrogen concentrations in the residue melt become relatively stable as the equiaxed dendrites and gas pores grow, leading to a dynamic equilibrium state for the amount of hydrogen atoms in the residue melt. In this stage, the mean hydrogen concentration remains slightly higher when a higher cooling rate or a lower initial hydrogen concentration is applied. With the increase of cooling rate, the gas pores nucleate at a higher temperature. Moreover, as the cooling rate increases, the final percentage of porosity and mean pore radius decrease, whereas the porosity density increases. As the initial hydrogen concentration increases, the gas pores emerge earlier, and both the final porosity percentage and mean pore radius increase. The tendencies of the simulated porosity percentages varying with cooling rate and initial hydrogen concentration are identical with the experimental results. The simulated data reveal that applying a higher cooling rate and decreasing the initial concentration of hydrogen in the aluminum alloy melt through the degassing process are beneficial for reducing the final porosity percentage of the aluminum alloy castings.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] |