There are a number of goals for the use of advanced model-based control in rotary kilns:

ü       Improve on currently used schemes (i.e. cascade PID controllers), which rarely provide satisfactory process control.

ü       Maintain the kiln operating temperatures in a narrow band, thus maximizing the product reaction occurring within the kiln.

ü       Yield direct economic payback through a satisfactorily holding of an oxygen target with safety objectives in mind.

ü       Address faults in the temperature or oxygen sensors in a different manner that an unproductive shutdown.

ü       Improve through estimation on the erroneous results that radiation pyrometers that are exposed to interference from dust and the kiln flame are giving.

ü       Address at the supervisory level the issues of lime quality, which will define set-points for the draft or fuel correcting for the product test results.

ü       Decrease the emissions of particulate matters (PM), CO, CO₂, SO_2, NO_x

We believe that rotary kilns are inherently Multi Input Multi Output (MIMO) systems that exhibit strong coupling and separation between slow and fast loops. At present Universal Dynamics Technologies Inc. is developing a multivariable adaptive predictive controller which already has a development version coded in Matlab, a mathematical simulation and development software package which is in the possession of U.D.Tech. All this creates the need and also the opportunity for high fidelity modeling of rotary kilns. The high fidelity model will have a modular structure allowing interchangeability of its components such as the principal kiln, pre-heaters or coolers, all of which would be treated independently in Matlab.

It is proposed that this model is to be achieved through the integration of Computational Fluid Dynamics (CFD) flame modeling and heat balance equations, as well as the raw material and its chemical reactions. These areas are to be addressed independently, but integrated into an overall Matlab simulation that would approximate the infinite dimensional system of the rotary kiln. This model gives the opportunity for:

• More rapid development of controller design through a more accurate theoretical model.
• A controllability and observability analysis of the models used for control.
• Obtaining a Linear Parameter Varying (LPV) model through a series of linearizations around several operating points or the use of function substitution or quasi-LPV techniques.

Such a project would require a student with strong engineering mathematics and computer science background that can easily integrate knowledge from fields such as process control and chemical engineering together with fluid dynamics. The final goal is to provide a universal solution for the simulation, control, and hence operation of rotary lime kilns.