Reduction of Liquid Clad Formation Due to Solid State Diffusion in Clad Brazing Sheet

Numerical Simulation of Brazing Aluminium Alloys with Al–Si Alloys

“…The predicted values are significantly larger than the experimental ones at low heating rates. However, the same conclusion as drawn by Benoit et al [6] can be made, namely that a minimum heating rate of 10-20 K/min will be necessary to avoid consumption of the clad during the heating stage of the brazing process.…”

“…The intention in the present study is rather to use the above-mentioned assessed value of the diffusion coefficient in order to be able to draw conclusions on the appropriateness of the assumptions made. Quite recently, Benoit et al [6] carried out a metallographic estimate of the thickness of residual clad after holding the brazing sheet at 525-550 °C for up to 120 min. At the end of the holding period, the samples were heated to 574 °C-which corresponds to their experimental onset temperature for melting of the clad-and then cooled to room temperature.…”

“…In addition to clad thickness measurements, comparison of calculated and experimental silicon profiles in the core could possibly give some support to the present numerical analysis. In the literature, two studies were found reporting silicon profiles after one thermal cycle below the eutectic temperature of the clad [5,6]. It was decided to use the study by Gao et al [5] who heated a coupon rapidly to 520°C, held it for 5 min at this temperature and finally cooled it quickly to room temperature.…”

“…Fig. 2 also shows the calculations corresponding to shorter holding times as carried out by Benoit et al [6], namely 5, 10 and 30 min long. The change in position at the end of holding shows the same characteristics as for the 120-min holding.…”

“…To complement their investigation on the effect of solidstate diffusion on the amount of available clad at the brazing temperature, Benoit et al [6] also performed where N dof is the number of degrees of freedom at each time step, (U i ) is the solution vector corresponding to the solution at a certain time step, E i is the estimated error in the latest approximation to the ith component of the true solution vector, A i is the absolute tolerance for the ith degree of freedom, and f is the relative tolerance; for the computation, f = 0.01 and A i = 0.001 for i =1,…, N dof . Also, a uniform mesh was used and a mesh independence check was performed; for the results shown, a mesh consisting of 240 elements, corresponding to N dof = 481, was employed.…”