Recent advances in the area of computer-aided robust (i.e., uncertainty tolerant) multivariable feedback control design based on linear time invariant nominal plant models have so far had minimal impact on process control practice. This is partly because defining the uncertainty and performance requirements for a complex nonlinear model realistically and non-conservatively is difficult, and substantial expertise is required to set up a problem. In this paper, we focus on constructing a framework for generating suitable nominal model-uncertainty set pairs from nonlinear dynamic plant models with associated constraints and operating cost functions. We then demonstrate the effectiveness of the linear robust control approaches by performing robust analysis and synthesis for a fluidized catalytic cracking unit (FCCU), an economically important complex nonlinear, time varying exothermic process with multiple manipulated and regulated variables, and time domain constraints which is a prime candidate for advanced control design.