Cutler takes on DMC...

Charlie Cutler is widely recognized as one of the earliest developers of advanced process control and model predictive control algorithms and controllers. His DMC and DMCplus are the most widely used multivariable controller technologies in the world. Cutler, a member of the Control Process Automation Hall of Fame, announced at the Invensys Customer Conference a global agreement for Invensys to market Cutler Technology Corporation's new Adaptive Dynamic Matrix Control technology. This is a non-exclusive agreement, but Cutler said, "I am not going to let just any engineering company have this. Invensys has demonstrated the ability to sell these products." With nearly 300 consultants worldwide, Invensys is now the largest authorized CTC implementer and re-seller, according to Doug Kelly, vice president of global solutions services for Invensys. Cutler described ADMC as the first major advance in advanced process control in nearly thirty years, as a next generation controller, and as a strong solution in the process industries, with step out performance against the competition. "We thought we were in a saturated market, but when we tested the ADMC products, we found that we could improve performance dramatically, even where existing multivariable controller software was in place and operational," Cutler said. Cutler described a test at a ConocoPhillips FCCU (Fluid Catalytic Cracking Unit) as showing "distinctively different performance" than the DMC controller that was already on the unit and used specifically for comparison. ADMC is a suite of two products: the Operator Advisor simulator (OA), which interfaces with the DCS and updates its basis every five seconds, and the Adaptive Dynamic Matrix Controller itself, which is a multivariable controller with PID dynamics removed from the model. Both products use the same model, and this synergy elimiates maintenance problems with the simulator model. In the ConocoPhillips FCCU test in July 2006, the fact that ADMC was able to permit all the valves to be driven to 100% without invalidating the model permitted very large performance gains. "And when you open the valves all the way up," Cutler noted, "what you find is that the system runs a lot more stable. Production variability significantly decreased." ADMC removes the dynamics of the PID equation from the hierarcy of the model, and without the PID controllers, the model is now a true dynamic model of the process. Basic process dynamics rarely change, since they are vessel sizes, inventories, lengths of pipe, and so forth. So, since model dynamics don't change, maintenance is greatly reduced with a corresponding increase in the controller stream factor. Since the OA simulator runs 150 times real time, and updates the state of the process every five seconds, the operators can model "what ifs" in near real time, any time. Cutler and Invensys believe that this can be a replacement for every existing multivariable model predictive controller currently installed. "And there are lots of sticky valves," Cutler noted. "Those valves are what make traditional controllers not work. We can deal with them quite well, and since they aren't going to go away, we are a good choice to replace existing controllers."