Filed under: Uncategorized
The purpose of this article is to describe extraction ductwork design approaches that we have seen in practice and try to clarify potential problems with some of these approaches and show that there may be better alternative ways of optimising and extract system.
Large, multi-operational processes (e.g. a sewage works, a chemical manufacturing plant or a Food production facility) will typically have several emission points.
Sometimes the emission sources are of an intermittent nature and there is a need to automatically balance the system to maintain the desired extraction from other sources.
We have often seen the deployment of a pressure control loop to handle this type of application. Here the speed of the main fan is controlled to maintain the system pressure at a particular confluence and in doing so there is some control on the conditions for the extraction system. However, there are, as always, good and bad engineering practices involved in the design of such systems.
Fig 1 below shows a typical manifold for a multiple point extraction system.
We have actually seen this format in operation and needless to say it is not effective. The emission sources are on a manifold and are dependent on the flow from each of the other sources. The static pressure at the end source will vary by the number of combinations of the other sources and hence it is futile balancing that emission source to a single available pressure.
Fig 2 shows a better engineered approach whereby each extract is independent to a collection plenum which is then subject to pressure control. Here each leg is individually balanced to the target static pressure at the plenum. If the fan has the capacity to maintain the static pressure at the plenum then the system can remain automatically balanced. Note that this approach would also favourably respond to variable pressures upstream of the fan. If there was an adsorber as an abatement system, then if that adsorber became blocked then the fan would find it more difficult to reach the target negative pressure in the plenum and would speed up accordingly.
We have actually seen this format in operation and needless to say it is not effective. The emission sources are on a manifold and are dependent on the flow from each of the other sources. The static pressure at the end source will vary by the number of combinations of the other sources and hence it is futile balancing that emission source to a single available pressure.
There is however a risk with this approach. Imagine if one or more of the extracts is critical and it is possible to block the duct with solids or liquids. Any blockage in the extract would cause the fan to slow down thus exacerbating the problem. Fig 2 represents a viable solution where there is no significant risk of blockage or unwanted line closure.
What happens if you move the pressure transducer/ transmitter unit downstream of the fan?
If we were to look at what might be downstream of the fan (e.g. abatement) then it may be desirable to control the fan to respond to pressure variation down-stream of the fan.
In fig 3 the pressure sensor is now up stream of a carbon adsorber and downstream of the fan. If the carbon adsorber became blocked then the fan would find it easier to reach the target pressure and hence would actually slow down, again exacerbating the problem.
Here the transducer/transmitter is measuring resistances which are a function of the full combined flow. This position is no good if the objective is to respond to individual line closures upstream of the fan. In the original case where the objective is to reduce flow in response to a closure of one of the upstream lines, the pressure at the point of measurement would decrease. The fan would then speed up to correct the pressure and would consequently extract more flow from the remaining points.
If the pressure transducer/transmitter is moved downstream of the adsorber then the pressure response would favourably control the fan in response to the adsorber being blocked but again, would not allow control in response to the upstream individual legs.
A means of addressing both scenarios would be to have variable pressure set-points in the controller. This means that there would be 5 pressure set points assuming all 5 of the emission points has the same extract or a larger multitude of set points if the extracts are of differing values. The required static pressure at the point of measurement would have to be determined for each combination in order to capture all the relevant system pressures. The system would also have to receive signals to indicate that one or more of the extracts has been closed (e.g. limit switches).
Obviously, a simpler solution is not to vary the flow at all but to maintain extract at all points even if it is not required. This is ok if the energy consumption is low for the full extract requirement but would be costly for higher flows.
Flow Control
Pressure control can be effective, but pressure is not necessarily proof of flow and direct flow or velocity measurements offer a more robust approach. The placement of a flow transducer/transmitter is not as critical as the pressure scenario however the device must be exposed to a relatively stable velocity profile (ideally 3 diameters downstream of a disturbance or at least 2 diameters upstream of a disturbance). It is not possible to capture the variance in flow with a single set point as with pressure and so the multiple set point approach would be required for a single transducer.
If one or more of the extract lines is critical (e.g. Zoned area required extract rate) then individual flow switches should be used to raise alarms in those extracts. This is because often the reduction or even loss of an individual flow may be less than the alarm setting for the combined flow. As an example, a combined flow may be 25,000 m3/hr and a sensible alarm setting may be 20,000 m3/hr. In this case any extract less than 5000 m3/hr would not trigger the alarm and hence its loss would go unnoticed. Note it may not be possible with the accuracy of flow control and monitoring to streamline the alarm to be smaller gap. In our example if we were to set the alarm at say 24,000 m3/hr then we would more than likely experience numerous nuisance alarms as natural fluctuations would often exceed this bandwidth.
HAZOP
The issues faced with extraction and the nuances of pressure/flow control lead us to conclude that any extraction with any level of sophistication or safety concern needs to be addressed by a HAZOP or other rigorous, structured safety review.
The position of instruments the, way extracts are connected, the consequences of reduced extract or mixed extract are all important considerations and sometimes the optimum solution is counter-intuitive.
We would always recommend a HAZOP study for extract systems, even if there is no accompanying abatement system.
Leave a Comment so far
Leave a comment
ASK Piearcey 