As stated in "Pattern Making," it is very desirable that a casting should be made as nearly of a uniform thickness as possible. Where the thickness of the metal varies, the thin portions cool more quickly than the thick. Internal stresses are thus set up in the metal which may cause a fracture. If the casting is not actually broken it is strained so that it is not strong enough to stand the load which it was calculated to bear. This is shown in Fig. 38 in an exaggerated form. The thin rim of the pulley cools first. It becomes hard, solid and fixed while the heavy spokes are yet at a high temperature. As the spokes cool they contract. This contraction of the spokes pulls the rim toward the center and prob-ablv results in a fracture.
It is not always possible or desirable, however, to make castings of a uniform thickness. Where the thickness does vary the cooling may be made uniform. This is done by knocking the sand off from the casting and exposing the thick parts to the air while the thin parts are kept covered by the sand. The heat is, therefore, retained in the latter and its radiation checked, while the radiation is facilitated in the thick portions.
In "Pattern Making," the necessity of fillets in all corners was stated. The reason for this is that any sudden variation of form tends to weaken a casting whether the thickness varies at that point or not. When cast iron changes from the liquid to the solid state it crystallizes. The crystals thus formed are either of a right square-based octahedral form as in Fig. 39'or that of a regular octahedral as in Fig. 40. These crystals always arrange themselves with their axes perpendicular to the cooling surface. When the casting cools slowly the crystals are larger than when it cools rapidly. The larger the crystals the softer the metal. Fig. 41 shows the effect of cooling when the corners are rounded, also when they are sharp. It will be seen from this representation of the arrangement of the crystals when cooling, that there is a line connecting the two sharp corners a and b where their axes lie at right angles to each other. The casting will be weak along this line. On the other side where the two changes of direction, r and d, are rounded then; is no sudden change of direction in the line of axes of the crystals and the casting retains nearly its full strength. A fillet in the corner at a would tend to relieve the stress. Fillets should therefore always be used in such places. If it is necessary that the finished work should have a square corner or a sharp angle, let it be made in the machine shop by cutting away the excess of metal left on the casting.