This research investigates the formation mechanisms of two types of hot tearing defects; surface tears and centre-line cracks, during horizontal direct chill (HDC) continuous casting of magnesium.
The heat and fluid flow and solidification behaviour of HDC casting of magnesium and alloy AZ91 have been studied using mathematical modelling and a variety of new experimental methods. The tensile strengths of magnesium and alloy AZ91 were measured at high temperature including the semi-solid region. Ingot microstructures and tear surfaces have been examined. Hot tearing theory has been investigated and developed to explain the effects of microstructure and alloy on hot tearing propensity.
Mould heat flow is shown to be controlled by the formation of the solid shell, which forms an air gap after initial solidification by contracting away from the mould, resulting in reduced cooling and some reheating. The semi-solid AZ91 shell (found to be weaker in tensile tests) maintains contact longer with the mould than stronger pure magnesium, resulting in higher mould heat fluxes in the AZ91 case. Natural convection driven flow in HDC casting is shown to give rise to an asymmetrical temperature distribution. The hot liquid jet feeding the mould rises in the liquid pool, causing remelting of the top shell. This flow meets a stream of colder liquid flowing down the solidification front. The two impacting jets do not exactly engage, resulting in a swirling asymmetrical flow.
Experimental data and modelling results show that surface tear formation occurs primarily due to liquid solidifying onto the refractory orifice plate and grabbing the solidifying shell, resulting in tearing of the shell. In the case of AZ91, the surface tears are hot tears occurring in the semi-solid shell. The higher liquid thermal diffusivity and lower latent heat of magnesium make it more prone to this problem than aluminium. Although shell tearing due to poor lubrication does not appear to have played a role in this study, calculated friction loads under poor mould lubrication conditions appear sufficient to exceed shell strength. Surface tears can be controlled by using higher melt super heats but this may increase the tendency for centreline cracks. Mould designs that reduce the cooling of the refractory by the metal part of the mould should be pursued. Model results indicate that judicious design of the liquid metal inlet to the mould should result in flows that tend to reduce the tendency for cold spots to form against the refractory.
Centre-line cracks in alloy AZ91 are found to be hot tears resulting from the usual mechanism in DC casting, i.e. differential thermal contraction between the centre and surface of the ingot. In cases where the structure was equiaxed no centre-line cracks occurred under normal conditions, whereas when the structure was columnar, centre-line cracks formed. Grain structure modelling showed that the cause of this variation in structure could be explained by changes in nuclei concentration. Nuclei concentrations were probably affected by changes in melt preparation practices between two different sites. Control of centre-line hot tears can be achieved through the usual methods to reduce stresses applied in aluminium DC casting, such as reduced casting speed and delayed water quench. However, it would be more desirable to utilise a suitable grain refiner for magnesium to ensure a more crack-resistant equiaxed structure is formed.
A new hot-tearing model calculates the head, feeding, capillary and dissolved gas pressures acting on a void front within the mushy region and the critical strain rate which will cause hot tearing. The model predicts the equiaxed structure is less crack prone because the greater diameter of the equiaxed grains compared to the primary arm spacing of the columnar structure, results in a higher permeability mush. The higher permeability allows the liquid to feed the imposed strain more readily and propagation of a void is prevented. The strain transmitted into the mush is also proposed to be less for the equiaxed case resulting in insufficient pressure for the crack to propagate at the solidification front velocity. The hot tearing model suggests that as the grain diameter reduces and permeability decreases cracking tendency will increase. Thus, the cases in the literature where small grain sizes (-100-200 |im) have been found to have increased cracking tendency are explained by this model. The well known action of grain refiner in reducing hot tearing is shown to be due to the change in grain morphology rather than refinement of grains. Alloys in which solid bonding occurs at lower fraction solid will be less prone to hot tearing. Results from the semi-solid strength measurements are consistent with the hot tearing model proposed. Unlike previously proposed hot tearing criteria, the new model allows incidence of hot tears to be predicted without any calibration, given input of strain data from a suitable stress model of the casting process.