Modeling the Dynamics of Highly Oscillatory Processes
Optimizing heat rate is a critical success factor for most coal-fired power plants. Too little heat and the plant wastes the energy produced from bituminous coal and other costly production inputs. Too much heat and safety is put at risk both the safety of production staff and of the plants production assets. Although optimizing a power plants injection water systems does not result in obvious financial gains, it plays an important role. Little improvements have a way of adding up to something big. When looking for ways of improving its heat rate, a large mid-American power company put Control Stations LOOP-PRO to work on lots of seemingly insignificant systems including the injection water system.
Invest in Control Payback in Profit quantifies the benefits of improved regulatory control within a typical production facility. In particular, the study which was produced by the UKs Carbon Trust cited a 5% – 15% reduction in energy consumption by regularly tuning PID controllers.
When a Picture Tells a Thousand Words
The injection water system furnishes water for cooling and for sealing of both boiler feed pumps and boiler feed booster pumps. The graphic below shows the Process Variable (top) and Controller Output (bottom) from the injection water system of a coal-fired power generation facility. The graphic also showcases the control loops performance before as well as after tuning was performed using LOOP-PRO.
Due to their requirement for steady-state operation, traditional PID tuning software tools are unable to accurately model the dynamics of oscillatory processes. LOOP-PRO is unique in this regard. LOOP-PRO is equipped with a patent-pending modeling capability that enables it to accurately describe transitional and oscillatory process dynamics.
Using bump test data, LOOP-PRO modeled the process and provided improved tuning parameters for the injection water system. As is shown in the graphic, the new parameters enhanced control dramatically. The process oscillatory behavior and the corresponding variability of the final control element were eliminated.
As with most things, little fixes add up.