I am implementing a 3-element boiler drum controller.

Has anybody had any success using a lead/lag block with a negative lead time constant? My thinking is that, by effectively "cancelling out" the inverse response of shrink/swell, this could be used to keep the level closer to 50% and decrease the chance of hitting the hi or lo trips.

There are some that say that the steam feedforward should be heavily lagged to prevent hitting the trips, but this just seems bad to me. It totally kills the dynamic response of the system -- which is the whole point of having the feedforward. Definitely not good if you need to deal with rapid swings in load.

With the lead/lag, on an increase in demand, the steam feedforward will cause an initial reduction in the feedwater flow. This would combat the swell phenomenon. The feedwater would then filter up to its final value under the lag component. Because you have the lead action, you do not need to lag it to death, thus retaining the dynamic response and ability to really swing the demand.

Very interested in any experience or thoughts on such a technique. Why won't this work?

 

Sam,

There are 4 main influences on drum level dynamics: steam flow and pressure, fuel flow and feedwater flow. Holding back feedwater flow intuitively reduces drum swell caused by falling steam pressure and increasing fuel, but actually it also has an inverse transient. If feedwater flow is reduced or held from increasing with steam flow, the additional heating time in the tubes causes void expansion and may swell the drum level initially.

So using a negative lead may help reduce drum swell but not in a simple way. Also forced and natural circulation boilers can have different swell responses from feedwater changes.

In general, by simply lagging the steam flow feedforward, a reasonable compromise is reached. The lag is not usually long - between 10 and 30s in my experience. This doesn't cause problems during fast changes as the dynamic influence is similar under all conditions. If you want to try compensating further, add a pressure rate of change influence to the level set point. This impact however varies with boiler pressure.