Modeling of Cement-Bonded Sand Mould System: An Artificial Intelligence Approach

Introduction

By using the cement as binder, moulds/cores of excellent strength and castings of high dimensional accuracy can be produced. These moulds/cores do not produce any toxic gases during moulding sand preparation and on metal pouring. In the cement bonded sand moulding system, silica sand, cement and water are the important ingredients. The major constituents of cement are tricalcium silicate (3CaO.SiO₂), dicalcium silicate (2CaO.SiO₂), tricalcium aluminate (3CaO.Al₂O₃) and tetracalcium aluminoferrite (4CaO.Al₂O₃.Fe₂O₃). The major drawback associated with these moulds is that, they need considerably long time to attain the required strength. As a result, it will increase the total time of making casting. However, the said problem can be resolved by using accelerator. The use of accelerator increases the rate of hydration and there by reduces the time required to obtain the strength. In the present study, calcium formate (Ca [OOOCH]₂) is used as an accelerator. It is to be noted that this accelerator was already used in the construction of civil works. Moreover, hydration leads to the formation of thin film of hydrated cement for the cement, and prevents further chemical reaction (Sleicher & Providence, 1943). This helps in preventing any gas generation, when the molten metal comes in contact with the cement.

Limited work had been reported on the cement bonded sand moulding system. This might be due to the drawbacks associated with cement-bonded moulding sand system, and availability of other bonding materials. It is to be noted that, the study of input-output relationship will help to increase the commercial use of cement as a binder in cement-bonded moulding sand system. Moreover, there are a number of independent variables, such as the amount of Portland cement, amount of accelerator, water quantity and setting time etc., which influences the properties of sand mix. However, the past investigations were helpful in identifying the variables, affecting cement-bonded moulding sand system and operating range. In Schleede and Ill (1944), the authors had identified the uses of Gypsum cements in pattern and mould making. A classical approach was presented by Uchikawa and Uchida (1976) to study the effects of amount of jet cement, testing time, presence of an accelerator and the amount of water on compression strength. They varied one variable at a time, keeping the others fixed. In addition to these, statistical design of experiments were used by some of the researchers, and proved to be an effective tool for studying the effects of number of independent variables on various responses. Both the linear as well as non-linear regression models based on design of experiments (Chakraborty & Dhindaw, 1977; Cochran & Cox, 1963) and response surface methodology were used to establish accurate input-output relationships in different moulding sand systems. Moreover, Parappagoudar, Pratihar, and Datta (2008a) used full factorial and central composite designs to model cement bonded sand moulding system. It is to be noted that the amount of Portland cement, accelerator, water and testing time were treated as inputs, and compression strength and hardness were considered as outputs.