This paper reviews the application of a model predictive controller algorithm to a crude oil unit in Izmit Refinery of Turkish Petroleum Refineries Corporation as a summary of nearly eight months of practical study. The controller is designed to control the crude heating via process flows followed by furnace heating and distillation column with product strippers. After a process overview, the fundamental control loop considerations are discussed. Steps of determining inferred qualities and step test response tests are outlined. The controller design considerations, constraint handling and economic variables are presented. Finally, a comparison of before and after commissioning in the kerosene quality and rate is tabulated as well as simulation results of the system response for different set point changes in the product qualities.