The bag filter is not too small. The flue gas just does not reach all of it 🔍
Design margins cover flow maldistribution. They do not remove it
Bag filter sizing works with design margins on air-to-cloth ratio and pressure drop, and those margins are there for good reason: inlet flow distribution has long been covered by reserve rather than calculated. What a margin cannot do is make the flow uniform. It buys capacity in reserve while the maldistribution it covers stays in the unit.
From the control room that shows up as a throughput limit: the differential pressure climbs back faster than it should after each cleaning pulse, the fan runs into its operating limit, and load is taken off to keep the filter in range. Undersizing is the obvious reading, and sometimes it is the right one. Where it is not, the area is installed and available, and the gas does not reach all of it evenly.
When the usual measures stop helping
Shortening the pulse interval, raising the pulse pressure, reviewing the bag specification. These are the right first steps and they often settle the matter. Where they do not, the reason is usually that they act on the pressure loss that is measured, and on most units that measurement is a single value for the whole filter.
Why the area is not used evenly in the bag filter
The resistance of a chamber is not a fixed property. It rises as filter cake accumulates on its bags, and it drops sharply the moment that chamber is pulsed. Gas follows that change, shifting towards the less resistant neighbour, which then loads faster in turn.
The result is a unit that reaches its cleaning threshold on one chamber while others still have capacity in reserve. Cleaning is triggered early and often, the mean pressure loss sits above the design figure, and the installed area never delivers what it could.
Condensation, temperature spread between chambers and changing dust properties can produce the same symptom, and a single pressure reading does not separate them.
Calculating the distribution settles that one candidate and narrows what is left to look at.
Two stage approach, developed for this bag filter problem
aixprocess models bag filters in two stages, because the system-level and the bag-level questions sit at different scales.
At system level, the bag bundle in each chamber is treated as a porous zone, a volume whose resistance stands in for the bags inside it, with a pressure loss characteristic that depends on the volume flow and the solids flow reaching that chamber. Cake accumulation is carried along in the calculation, as is the cleaning event itself: once a chamber reaches the specified pressure loss it is pulsed, and the pressure loss drops sharply.
At chamber level, a single chamber is then resolved in full, with every bag present and a pressure loss characteristic that varies locally with the approach velocity and the cake built up at that position.
Both characteristics have to be calibrated against the dust and the operating window of the plant in question. A model of this kind is only as good as that calibration, which is where operating data earns its place.
What comes out of it
The system model gives the pressure loss over time, chamber by chamber, including the drop at each cleaning event. That shows the level the unit will actually operate at and which chamber drives it, and it turns pulse threshold, pulse order and frequency into parameters that can be set against the behaviour they produce.
The chamber model gives the velocity and solids distribution across the bags. That answers whether the cake builds uniformly, and where local peaks put enough particle energy onto the fabric to wear it, which is usually where the early failures sit.
Where the distribution itself turns out to be the constraint, the same model is what inlet geometry and deflector changes get assessed against, before anything is cut open.
The take-away
Filter area and flow rate size a unit that separates. Whether it delivers the throughput it was bought for depends on how evenly that area is used, and that follows from a pressure field no single operating point describes.
If your filter is holding plant load down and the usual measures have not settled it, the question worth asking first is whether the chambers are loading evenly. Geometry and the operating data you already record are enough for a first assessment, no additional instrumentation needed to find out whether it is worth looking further. Get in touch here on LinkedIn.
Comments