Model-Based and ANN Control

Multivariable controls (MVCs) are particularly well suited for controlling highly interactive fractionators where several control loops need to be simultaneously decoupled. MVCs can simultaneously consider all the process lags, and apply safety constraints and economic optimization factors in determining the required manipulations to the process. The technique of multivariable control requires the development of dynamic models based on fractionator testing and data collection. Multivariable control applies the dynamic models and historical information to predict future fractionator characteristics. For towers that are subject to many constraints, towers that have severe interactions, and towers with complex configurations, multivariable control can be a valuable tool. 

Dynamic matrix control (DMC) is also an MVC technique, but it uses a set of linear differential equations to describe the process. The DMC method obtains its data from process step responses and calculates the required manipulations utilizing an inverse model. Coefficients for the process dynamics are determined by process testing. During these tests, manipulated and load variables are perturbed, and the dynamic responses of all 

Multiproduct fractionator controls, where, after dynamic correction, the boil-up, side-draw and distillate flows are ratioed to the feed flow (left). On the right, the true boiling points are controlled by throttling the product flows, while heat balance is controlled by manipulating the reflux flows. 

254 Post- Oil Energy Technology After the Age of Fossil Fuels 

Because the neural controller is an empirical model as opposed to a theoretical model, it is susceptible to errors if operated outside the conditions of the training set. Data for the training set need to be continually gathered, and the network retrained whenever novel conditions occur in order to increase the robustness of the neural controller throughout its life of operation. 

---