The Design of a Gamma-Ray for Detecting Mushy Zone in Tomographic Sensor Continuous Casting

The y-ray tomographic sensor invented by the CastScan consortium at MIT for the non-invasive real time monitoring of the solidification front in the continuous casting of metals is a new industrial computed tomography (CT) tool. The sensor potentially offers great economic benefits to the continuous casting industry by preventing break-outs, and increasing productivity and efficiency through optimal process control. In this thesis, the proposed concept was experimentally verified, based on radiation attenuation measurements, to prove its viability for determining the thicknesses of the liquid and solid sections of a solidifying metal, with a density difference as low as 7%. In addition, the preliminary experiments identified several fundamental factors that affect the operation of the sensor and which must be considered in the prototype design. The y-ray sensor integrates numerous system components: the radiation source, detector array, actuators, and tomographic algorithms. Therefore, in developing an effective sensor that meets various industrial specifications, component design and system integration must be optimized. A CT sensor was designed specifically for detecting the mushy zone in continuous casting. Through mathematical modeling of radiation attenuation through the liquid and solid phases, quantitative relationships among source strength, exposure time, and other critical system parameters were established. These relationships serve as a quantitative framework for the design of an industrial tomographic system. In addition, optimal spatial resolution and tomography techniques for identifying the mushy zone were investigated and confirmed by simulations using several critical parameters, such as the number of translational and angular steps. From these considerations, the approach of generating 3D tomographic image of density profile across the liquid-solid interface from radiation attenuation data offers a superior and more direct method than the conventional approaches. In conclusion, the tomography system geared to mushy zone identification can serve as a standard model for the CT sensor. Furthermore, the guidelines developed in this thesis can be used to design the future generations of sensors for different applications. Thesis Advisor: Jung-Hoon Chun Title: Esther and Harold E. Edgerton Associate Professor of Mechanical Engineering Thesis Advisor: Nannaji Saka Title: Principal Research Scientist, Mechanical Engineering