Home  About us  Contact us  Training  Optimisation Services  Protuner  Educational PDFs  Loop Signatures   Case Histories

Michael Brown Control Engineering CC 

Practical Process Control Training & Loop Optimisation

CASE HISTORY189

POOR FURNACE TEMPERATURE CONTROL

Many process engineers do not appreciate the importance of flow loops in their unit controls.  A senior process engineer once told me that flow loops need not be tuned well as they generally have little effect on the more important and much slower loops like temperatures and pressures.  In actual fact flow loops can be vitally important for ensuring that other controls work properly.  The example given here is an excellent illustration of this.

The process unit used for this example is the steam desuperheater on a boiler in a petro-chemical refinery.  The water is heated using gas as the fuel.  The final desuperheater outlet temperature is a cascade secondary loop with the primary control being the outlet temperature control of the upstream desuperheater.  (Just out of interest, there was a measurement of the outlet steam flow after the second desuperheater, but they had not used this as a further cascade to ensure the temperature control worked properly, which is something I recommended they do.  A secondary flow cascade loop takes care of any valve problems as it would have quickly ensured that the flow of steam keeps to the amount demanded by the temperature controller.  In actual fact there were several problems with steam desuperheater valve, and this may be used for a future article.)
The problem being experienced on the steam temperature was that it was cycling quite badly, and they had not been able to establish the cause.

Investigations were carried out as to what was causing the cycle by checking all the various controls that could have an effect on the desuperheaters outlet temperatures.  It was found quite quickly that the culprit causing the problem was the combustion air flow control loop.

Figure 1.

Figure 1 is a “Closed Loop as Found” test on the airflow with the controller in automatic and using the original tuning parameters.  A series of setpoint changes was made as can be seen in the figure.  It can be seen that the flow was apparently in a continuous slow cycle with a 20% amplitude and a period of roughly 100 seconds.  The other 2 traces shown in the figure are the outlet temperatures of the two desuperheaters in cascade.   It can be clearly seen what a large influence the cycle of the air flow was having on them, particularly on the secondary cascade desuperheater.

Figure 2.