Furnace control systems for reheating furnaces celFCS-rht
In contrary to old approaches, which only observe single elements in the furnace (e.g. critical material), celano GmbH has developed a new three step procedure for their furnace control system celFCS-rht (celano Furnace ControlSystem Reheating) which observes the whole material inside the furnace, thus optimizing the operation mode while abstaining from the usual simplifications.
1. step observer
The current material temperature is determined (energy bilance in the furnace, 3-dimensional thermal conduction equation). Heat carriage through convection and radiation as well as heat conduction and added energy through combustion are incorporated during this step.
2. step prediction
The expected fure temperature distribution for all materials inside the furnace is calculated with the help of prediction. The observer is called multiple times to simulate the future material movement and the planned changes of furnace temperatur. The prediction then returns the expected chronological sequence of the local furnace temperature for every slab as a result.
3. step optimization
The setpoint assignation results from the optimization with the aid of an evolutionary algorithm. Changed setpoints (so called scenarios) are simulated based on the current situation. With a so called fitness function the chronological sequence of material temperature (results of the prediction) for every scenario is evaluated. By the use of mechancial selection, heredity, mutation and consideration of fitness results the evolutionary algorithm creates new scenarios, which in return are simulated by the prediction. The determination of the scenario with the highest fitness result is goal of the optimization. The flexibility of the fitness function, which can easily be modified for customer specific focal points is a huge advantage. The different (potentially contradictional) goals can also be weighted.
To ensure the online capabillity of the system, even though complex and computing time intensive operations are used, the software is designed with parallel algorithm for multi processor systems.
Observation of thermoelements
A special function inside the modell is the observation of thermoelemts. Since the model is recreating the furnace situation at any point in time, an additional algorithm can identify a difference in the meassured temperature of thermoelements and offer corresponding messages.