Activated carbon injection for flue gas: what it removes
By the time the gas has been cooled and its acid gases neutralised, what is left is a short list of pollutants that are neither soluble enough to be scrubbed nor solid enough to be filtered out as dust. Mercury leaves the furnace as a vapour and stays one at the temperatures a fabric filter runs at. Dioxins and furans survive the chamber in small quantities, and form again on the fly ash as the gas cools. Neither dissolves in a scrubbing liquid at a useful rate, and neither settles out of the gas on its own.
Both are captured on carbon, and activated carbon injection is the usual way of putting the two in contact. Powdered carbon is dosed into the duct, and the gas carries it forward to the particulate filter. A gram of the powder presents an internal surface measured in hundreds of square metres — a material property rather than a design figure — and that surface is where an organic molecule or a mercury atom stops.
That also fixes the outside of the duty. Carbon is dosed for mercury and for dioxins and furans. It is not the answer to the acid gases, which the scrubber or a dry sorbent upstream has already removed, and it is not the answer to nitrogen oxides, which do not adsorb on carbon in any useful quantity at these temperatures.
- Mercury — the pollutant that most often decides whether activated carbon injection for flue gas is fitted at all, and the one that cannot be caught as dust.
- Dioxins and furans — the fraction that survived the chamber together with the fraction that formed again on the fly ash downstream.
- Other trace organics — chlorinated aromatics and similar compounds, taken up on the same surface without a stage of their own.
- Heavy metals other than mercury — carried on the fly ash as solids, so they are the duty of the dust collector rather than of the carbon.
- Acid gases and nitrogen oxides — outside this stage; they are answered upstream or in the furnace.
Mercury removal from incinerator flue gas
Mercury behaves differently from every other metal in the gas, and the difference is what makes it an adsorption problem. The other metals leave the furnace on the fly ash, so a fabric filter takes them out with the dust. Mercury leaves as a vapour, and part of it stays a vapour all the way down the train, where no filter can intercept it.
It also arrives in more than one form. Oxidised mercury is water-soluble and a wet scrubber removes part of it; elemental mercury is close to insoluble and passes through a scrubber almost untouched. The split between the two depends on the chlorine in the waste, the oxygen in the gas and the temperature history of the gas, none of which is fixed, so the fraction that has to be caught on carbon is not a constant either.
Two properties of that capture matter later in the design. Adsorption weakens as the gas gets hotter, so the injection point sits as far downstream as the temperature allows. And mercury that has been adsorbed can be released again from the collected ash if the ash is warmed, so the filter inlet temperature and the ash handling route belong to the mercury answer and not only to the dioxin one.
Dioxin removal in flue gas: the adsorption stage
The formation half of this subject is treated elsewhere. The article on why incinerators run at 850 °C for two seconds covers destruction inside the chamber, and the one on de novo synthesis and the quench window covers what forms downstream on the fly ash. What is left for the carbon is the material present at the filter inlet — the compounds the chamber did not destroy and the cooling section could not prevent.
Adsorption is weaker when the gas is hot and stronger when it is cold, which puts the injection point at the cold end of the train by preference. It is also reversible. Dioxins taken up on the carbon earlier can be released again when conditions move: a temperature rise, a load change, a stop and a restart. A plant can therefore emit more after an upset than its steady-state figures predict, and the answer is to keep the carbon dosing running through transients rather than to switch it off with the waste feed.
How the carbon injection rate for waste incinerators is set
No table gives the rate. The carbon demand of a gas depends on the concentrations of mercury and dioxins in it, the temperature and moisture it arrives at, how long the powder has to mix with it before the filter, and how much carbon is already held on the bags — none of which is a property of the equipment being bought. A rate copied from a plant burning a different waste is a starting guess, not a design.
What is specified is the method. A starting rate is chosen at the conservative end, the plant is run at it, and the result is read from the stack: mercury and dioxins from a sampling run over a defined period, and the continuously measured parameters from the monitoring system. The rate is then adjusted until the figures hold with margin at the lowest load the plant runs, where the gas is slowest and the contact poorest, and rechecked whenever the waste stream changes materially.
The rate itself is usually quoted as a mass of carbon against a volume of gas, or as a proportion of the ash the filter collects, and either form has to be converted into kilograms per hour of powder for the dosing equipment. Which figures are published for a given unit, and the starting point that suits a particular waste, are confirmed with the factory for your waste composition.
- The mercury and dioxin load of the raw gas, which follows from the waste rather than from the plant.
- The gas temperature at the injection point, because adsorption falls away as the gas gets hotter.
- Moisture and the acid dew point there, because damp powder sticks to duct walls and to the bags.
- The mixing length between the lance and the filter, which sets how much contact the powder gets before it is collected.
- The cleaning cycle of the filter, which sets how much carbon is left on the bags between pulses.
Activated carbon injection and the bag filter
The arrangement works because of what happens on the bags. The carbon arrives as a dispersed powder and the fabric collects it; the cake that builds on the bags then becomes the bed the gas is forced through. The gas passes through carbon rather than alongside it, and the contact time that the injection lance alone cannot provide is supplied by the filter. That is why the injection point sits upstream of the dust collector, and why a plant that sends its gas straight to the stack after injection has not installed an adsorption stage at all.
Three consequences follow. The filter has to run below the temperature limit of its bag material, which keeps the cake cool and the adsorption working. A bag that has failed, or a compartment that is bypassed, sends carbon — and whatever it has already picked up — to the stack. And the collected ash leaves the plant carrying the mercury and the dioxins that were taken out of the gas, so its disposal route is part of the same duty rather than a downstream detail.
The alternative to injection is a fixed carbon bed: the gas is passed through a vessel of carbon, and the carbon is replaced or regenerated on a schedule. A bed is easier to monitor and easier to hold at a steady rate, and it needs more space and a heavier vessel than a lance. Both arrangements depend on the gas arriving within the temperature range where adsorption works, and a gas that is too cold carries wet and sticky dust into the filter.
Common mistakes
- Specifying an injection rate as a fixed number without stating the waste it was set for. The rate follows the gas, and the gas follows the feed.
- Injecting upstream of the quench or the scrubber, where the powder is washed out of the gas before it reaches the filter and never forms a bed.
- Running the filter above the temperature limit of its bags to improve adsorption, which shortens bag life and moves the collected ash back towards the range where dioxins form again.
- Treating the dust collector as a dust collector only, and not checking that the cake on its bags is the contact bed the adsorption design assumes.
- Stopping the carbon feed during a stop or a load change, when adsorbed material can be released again and the release is what the stack sees.
- Assuming the ash is clean because the gas is. The mercury and the dioxins are in the ash, and the ash has a disposal route of its own.
Self-check before the injection system is specified
- Are the mercury and dioxin concentrations in the raw gas known, or at least the waste characteristics they follow from?
- Is the injection point downstream of the cooling and scrubbing stages, and upstream of the particulate filter?
- Is the gas temperature stated at the injection point and at the filter inlet, and is the filter inlet below the limit of the bag material?
- Is the carbon rate expressed against the load the plant actually runs, with a figure for the lowest load as well as for the design point?
- Has the ash from the filter been characterised for disposal, given what it now contains?
- Is the injection a continuous duty, including through start-up, shutdown and load change?
- Is the stack test that closes the loop scheduled for a representative load and a stated sampling period?
From a stack limit to a carbon dosing rate
Activated carbon injection is a small stage in footprint and a decisive one in result: it is what takes the two pollutants the rest of the train cannot reach. The stages it sits between are set out in the article on flue gas treatment in an incinerator, and the limits it has to hold on mercury, dioxins and particulates are in the incinerator emissions chart. The injection rate, the point it is dosed at and the filter duty for your plant are confirmed with the factory for your waste composition and the emission standard that applies.