Liu Jiang， Li Yuanji， Wang Shuai
a.School of Energy Science and Engineering， Harbin 150001， China； b.Zhengzhou Research Institute， Harbin Institute of Technology， Zhengzhou 450001， China
Abstract
Objective Melt infiltration is a favorable manufacturing route for fabricating aluminum matrix dispersed fuel elements. Using preform skeletons and melt infiltration densification， this method effectively overcomes the limitations of conventional manufacturing routes， such as low fuel volume fraction， uneven fuel distribution， and poor densification. Nevertheless， the transient filling mechanism of the infiltration front in porous media and the quantitative correlation between infiltration pressure and saturation remain insufficiently understood. Therefore， clarifying the intrinsic relationship between aluminum melt infiltration behavior and its key influencing factors， and revealing the effects of infiltration processes and melt physical properties on infiltration characteristics are of great significance for developing advanced fabrication technologies for high-performance aluminum matrix dispersed fuels.
Methods In this study， the phase-field method was adopted to characterize two-phase flow behaviors. The Cahn-Hilliard equation was applied to capture the evolution of the interface position， where the chemical potential was defined as a function of mixing energy density and interface thickness control parameters. A porous medium model consisting of fuel particles was established， and pore-scale simulations of aluminum melt infiltration were carried out. The feasibility of the proposed model was verified against experimental data.
Results and Discussion Under external pressure， the melt preferentially advanced toward both sides. After the displacement fronts on both sides reached a certain depth， the melt gradually infiltrated and filled the intermediate region. When the melt flowed into dead-end pores， the liquid meniscus ruptured， and newly formed liquid-gas interfaces encapsulated residual gas within pores to form discrete isolated bubbles. When two opposing menisci converged and coalesced， the convex liquid surface could not fully spread， and residual gas was trapped by pinned liquid interfaces to form permanent pore void defects. For porous architectures featuring large cavities confined by narrow pore throats， interfacial pinning readily occurred， which suppressed effective gas evacuation from internal cavities and ultimately generated isolated bubbles. In the late infiltration stage， pressure fluctuations exhibited a lower frequency. A plateau appeared on the flow velocity curve in the late stage， indicating that the melt was trapped in narrow pores and flowed steadily along dominant flow channels at a nearly constant flow velocity. Sufficient driving pressure in the early infiltration stage yielded high infiltration saturation. Saturation fluctuated in the middle and late stages， mainly caused by Haines jumps induced by driving pressure attenuation. In the final stage， the discrepancy in infiltration saturation between samples under different inlet pressures became pronounced， and higher inlet driving pressure consistently resulted in greater infiltration saturation. Low-viscosity melts could fully occupy pore voids within an extremely short period， yielding high infiltration saturation. However， the absence of flow stabilization from viscous shearing readily triggered disordered channeling during melt infiltration. This trapped a greater amount of gas inside pore channels that could not easily escape， thereby deteriorating infiltration densification. When aluminum melt exhibited poor wettability with pore wall substrates， the liquid front became pinned at narrow pore throats， and a large external pressure was required to overcome pinning. In contrast， systems with favorable melt wettability against pore walls required much lower external pressure for depinning， and the depinning process proceeded spontaneously， facilitating uniform full pore infiltration. Poor wettability resulted in larger differences in displacement front depth， which continuously expanded as the overall infiltration depth increased. For strongly wetting interfaces， the differences in displacement front depth increased in the early infiltration stage. In the later stage， the average throat radius of unfilled pores decreased， the capillary driving force increased， and the differences in displacement front depth declined.
Conclusion In this study， pore-scale numerical simulations are conducted to characterize aluminum melt infiltration within porous preforms using a constructed packed-particle porous medium model. The results show that isolated bubbles during melt infiltration originate from meniscus evolution induced by pore geometric configurations and interface pinning effects. Under low inlet infiltration pressure， the dominant seepage channel is preferentially formed， which reduces overall infiltration saturation and increases the occurrence probability of Haines jumps. Low-viscosity melts can rapidly fill porous voids and reach high saturation. Nevertheless， the lack of flow stabilization from viscous shear induces chaotic channeling. A large amount of residual gas remains trapped inside pore channels and cannot escape， impairing the uniformity of pore filling. When capillary force acts as flow resistance in melt-solid systems， the liquid front is prone to pinning at narrow pore throats， and high external pressure is required to achieve depinning， resulting in non-uniform infiltration. At strongly wetting interfaces， capillary force cooperates with external pressure to promote infiltration， and less external force is needed for depinning. Smaller contact angles correspond to improved infiltration densification. However， isolated gas bubbles easily form when melt flows through large cavities surrounded by tiny pore throats.
Keywords： aluminum melt； infiltration； saturation； pore scale
Get Citation:Liu Jiang， Li Yuanji， Wang Shuai. Numerical simulation of infiltration behavior of aluminum melt in porous preforms［J］. China Powder Science and Technology， 2026， 32（5）： 1-12.
Received：2026-05-16， Revised： 2026-08-06， Online： 2026-08-26。
Funding： The research was supported by the National Natural Science Foundation of China (Grant No. 52076060).
CL C No.：TB333
Type Code：A
Serial No.：1008-5548（2026）05-0001-12
