Direct-chill (DC) casting is one of the most common methods to produce ingots of high-strength aluminum alloys such as an AA7050. Despite of its superior mechanical properties, this alloy is prone to both hot tearing (HT) and cold cracking (CC) during DC casting. HT form above the solidus while CC occurs below the solidus temperature. CC occurrence appears to be random and thus it is difficult to foresee. Moreover, the propagation of CC is usually catastrophic. Thus, not only it decreases the yield of the production and may damage the casting equipments, but it also poses safety hazard to the casting personnel. A previous study on CC indicated this type of crack could start from a solidification defect, for example an inclusion. The same study also suggested that undeveloped or micro hot tear (HT) may also act as CC initiation point but this assumption has not yet being proven. In addition to trials and errors methods, numerical models are more and more considered as relevant tools to find process conditions minimizing the risk of both HT and CC. At the moment, we have a numerical process simulator to simulate aluminium DC-casting process, called ALSIM. Although at its current version (ALSIM 6), ALSIM has criteria to estimate the susceptibility for both HT and CC, it was reported that ALSIM’s crack prediction capability has not yet reached its full potential. Therefore, the main goal of this study is four-fold: (1) Verify the connection between HT and CC experimentally. (2) Complete ALSIM materials database for AA7050 for a better crack prediction accuracy. (3) Analyze ALSIM model sensitivity with respect to different DC casting parameters and materials database. (4) Propose a concept of a new crack prediction criterion so that ALSIM could capture the connection between HT and CC quantitatively. Before developing the connection between HT and CC within ALSIM, we need to make sure that the current version of ALSIM is able to properly estimate the susceptibility of different types of cracks during DC casting. For this validation, we performed benchmark tests between trial casts that produce different cast output (healthy, HT and CC) and ALSIM simulations that mimic those trial casts. For obtaining different cast outputs, we vary two casting parameters; casting speed (in combination with cooling rate) and melt inlet geometry. The analysis shows the output of the ALSIM simulation is in qualitative agreement with the experimental results of the trial DC casting. This verifies that ALSIM is sensitive with respect to different casting parameters. Moreover, the results also indicate that ALSIM has the potential to capture the connection between HT and CC. ALSIM uses a set of material properties database to simulate the DC casting process. One of the ALSIM drawbacks in the beginning of the project was the incompleteness of the AA7050 mechanical properties database in the vicinity of the solidus point (T = 465 °C). To complete this database, we performed isothermal tensile tests in both the sub-solidus temperature regime (between 400 and 465 °C) and in the super-solidus temperature regime (when the material is in its semi-solid state but already possesses mechanical strength), i.e. between solid fraction of 0.85 and 1.0 (fully solid). We fitted the result of the tensile tests in respective temperature regimes with the constitutive models implemented in ALSIM to obtain the material property parameters. The focus was on the tensile behavior of the alloy as this loading mode is the main one for the HT formation in the billet. The tensile test results also reveal the mechanical behavior and the failure mechanisms of AA7050 around the solidus point and shed some light on the possible mechanisms of the connection between HT and CC. In the sub-solidus temperature range, we performed tensile tests at different temperature points (400, 420, 440, 450, 455, 460, 465 °C) and three strain rates (0.05, 0.005, 0.0005 s-1). For each combination of temperature and strain rate, we repeated the tests three times to capture the statistical behavior of the alloy at the test condition. From the test result, we found that the ductility of the alloy decreases as temperature and strain-rate increases. Meanwhile, the strength of the alloy increases with strain rate but it decreases as temperature increases. From the fracture surface assessment, we observed that the main mode of failure gradually changes from ductile transgranular (at 400 °C) to ductile intergranular (at 465°C). In the sub-solidus temperature range, we fitted the tensile constitutive behavior data with both extended-Ludwik equation and creep-law. For the former fit, we found that the obtained constitutive parameters are continuous with the data from the lower temperature regime (room temperature up to 400°C) while for the latter fit, we found that the AA7050 parameters are relatively comparable to the value of other 7XXX series alloys. In the super-solidus temperature range, we performed tensile test at different solid-fractions (fs); 0.85, 0.88, 0.9, 0.94, 0.97, 0.99 and 1.0 (fully solid) or in terms of temperature at 550, 520, 485, 475, 473, 470 and 465 respectively. To observe the displacement-rate sensitivity, we did tests at two different displacement rates (0.2 and 2.0 mm/min) for some solid fractions. To assess the statistical behavior of the alloy, for each test conditions, we repeated the test three times. The results show that grain coalescence starts between solid fraction of 0.94 and 0.97. This occurrence is signified by the steep increase in the peak stress as solid fraction increases. The alloy undergoes a typical brittle temperature range behavior – transition from ductile-brittle-ductile as the solid fraction decreases from the fully solid state, with the lowest ductility observed at fs = 0.94. The fracture mode observed from fracture surface analysis is generally mixed between transgranular and interdendritic. We extracted the constitutive parameters by comparing the tensile force-displacement curves obtained from the experimental tensile tests and the result from numerical thermo-mechanical tensile tests that were built using ALSIM. Using this method, we selected the constitutive parameters that provide a good fit of the force value between numerical and experimental tensile tests in this temperature range. From the fitting, we found that the semi-solid behavior of AA7050 is different compared to the other alloys that were already available in ALSIM semi-solid mechanical database; Al-2% Cu and AA5182. We observe that AA7050 is stronger than Al-2% Cu but weaker than AA5182 in the semi-solid temperature regime. Using the newly obtained AA7050 materials database, we carried out the sensitivity analysis of ALSIM with respect to different materials database at both temperature regimes (semi-solid and fully-solid). We ran ALSIM simulations using the same set-up as the benchmarking cases but we varied the materials database at different temperature ranges. We assessed the results of both cracking criteria. We conclude that ALSIM model is sensitive with respect to the materials database in both temperature ranges. However, we also found that ALSIM HT criterion has a weak dependency with respect to different materials databases, which is unexpected and shows this criterion needs to be improved. To experimentally validate the connection between HT and CC, we need to produce samples containing undeveloped HT and then quantify the dimension of these HT. To produce these samples, we used a solidification experiment with controlled conditions. During the solidification, we varied two process parameters namely cooling-rate and constrain conditions. We made sure only HT were formed in the sample by relieving the constraining force as soon as the material is fully solidified and let the sample cool down unconstrained to room temperature. Subsequently, we performed three dimensional imaging on the re-solidified and most deformed part of the sample by using X-Ray microtomography technique and analyzed the image data by image analysis method. From the result of the image analysis, we found that by changing these two process parameters, four regimes of HT can be distinguished based on its severity; compressive, micro-HT, macro-HT and complete fracture. Only the samples with micro-HT conditions were selected for the subsequent tests. The produced samples with micro-HT were then tensile tested at room temperature, which is the temperature where CC commonly occurs. The result of tensile tests shows significant drop in strength and ductility of the alloy if micro-HT are present in the sample. This verifies the connection between HT and CC. The result from the quantitative image analysis shows that usually HT do not form individually. Moreover, the geometry of the formed HT is not simple (i.e. branching crack) and the approximate ratio between the longest and shortest crack axis is finite (unlike the penny-shaped geometry). These HT features are not taken into account in the crack initiator geometry of the current ALSIM CC criterion. Therefore, the criterion needs to be improved in order to increase the accuracy of the crack prediction. One of the most important drawbacks in the current ALSIM HT criterion in the framework of connection between HT and CC is that it cannot provide the dimension of the formed HT. This information is crucial for the CC criterion to decide whether crack will occur. Since HT is a non-linear and complex phenomenon, we propose to capture its formation by using a stochastic (probabilistic) model. The new HT model considers the relationship between different physical phenomena during solidification and utilizes ALSIM simulation result (physical fields) as the model input. From the simple example presented in this study, we illustrate the potential of this new approach to provide probable dimension and location of the formed HT. Although further development is necessary to make this model fully operational in ALSIM, the implementation of this new HT model in ALSIM may offer the possibility of quantitative crack prediction of the connection between HT and CC.