The Carbon Fixed the Taste. Nobody Asked What It Was Doing About the Chlorine.
A water treatment specification came to us with a line on it that read, near enough, activated carbon filter โ one (1) unit, FRP vessel, complete with media. Everything else on that sheet carried a number. The pump had a head and a duty flow. The vessel had a diameter. The membranes had a model and a count. The carbon had a vessel size and a word.
That is not a careless sheet. That is how carbon is bought in this market, by us as much as by anybody. You order a vessel, you order the volume of media that fills it, and the media arrives in bags. It is true across the whole water treatment side of what we do, and it is true in the sewage plants as well. The specification that decides whether the carbon actually does its job is not on the purchase order, and in most cases nobody on either side of the order has asked for it.
We have written before about when a building actually needs reverse osmosis and when it does not, and about the four vessels that sit in front of the membranes and what the feed flow does to their size. That second note says, in one line, that the carbon filter has two jobs and only one of them is visible to a customer. This note is about the number that decides whether it can do the invisible one. It is not about the vessel, the flow or the cartridge โ that ground is covered โ and it is not about the tertiary filter bed on a sewage plant, which is sized by area and washes itself clean. Carbon does neither of those things.
Carbon is sold by volume. Its capacity is a different number entirely
Two vessels, same diameter, same bed depth, same quantity of granular activated carbon by the cubic foot. Same weight on the delivery note, give or take. Put them side by side on a plant room floor and there is nothing to choose between them. One of them may be carrying something close to twice the adsorption capacity of the other, on the one figure that measures it.
The figure that separates them is the iodine number. It is the result of a standard laboratory test, reported as milligrams of iodine adsorbed per gram of carbon, and it stands in for how much pore surface the carbon has available for adsorption. A sand or glass bed is a mechanical object and you can reason about it from its dimensions, which is the whole basis of sizing a filter by bed area and filtration velocity. Carbon is not a mechanical object. It is a quantity of surface, and the volume in the vessel tells you almost nothing about how much surface you bought.
Our own written process description for a refilling station states the figure we specify and the figure we see in the market: we supply activated carbon with an iodine content above 1000, where the market is, in the words of that page, littered with carbon below 500. Those are the two numbers worth carrying out of this note. They describe two products sold under the same four words, delivered in the same bags, installed in the same vessel, and in a quotation comparison they look like the same line item at different prices โ which is exactly the situation we described when two quotations for the same plant are not comparable, arriving from the media side rather than the equipment side. It is also why, when we put three water treatment options in front of a client, the media grade belongs in the comparison rather than under the vessel price.
The two jobs, and only one of them ever complains
Carbon in a drinking water or process train is doing two things at once. It takes out taste, odour and the yellowish or brownish discolouration that organics and some inorganic contaminants put in the water. That is the job the customer can detect, and it is the reason most owners think the vessel is there at all. We keep the equipment side of that job on our refilling station page, where the taste and the clarity are what a customer thinks he is buying. The second job is chlorine. On any plant fed from a chlorinated municipal supply, the carbon is what stops free chlorine reaching a thin-film membrane, and chlorine damages those membranes, as that reverse osmosis note sets out โ it is not a contaminant the membrane rejects, it is a contaminant that eats it.
Here is the part that makes a weak carbon dangerous rather than merely disappointing. The customer only ever audits the first job. He drinks the water. If it tastes clean, the carbon is working as far as anyone in the building is concerned, and nobody asks a second question. The membrane, meanwhile, has no sense of taste. It is being oxidised quietly on a chlorine residual nobody is measuring, and the result does not appear as a carbon problem. It appears months later as a membrane replacement, on a plant where the owner believes the pre-treatment is fine, because the only part of the pre-treatment he can perceive is fine. The same blindness has a sewage equivalent, and we have written it up: two parameters failing for one cause, where fixing what you can see leaves what you cannot see exactly as it was.
That is a shape we keep running into and it is worth naming on its own, because it is the same shape as a fire code seat count used as a design population and a specification quoting a regulation that had been dead for ten years and a flow rating on a grease trap read as if it were a size. In every one of them there is a figure, or an observation, that is real and authoritative and is measuring something other than the thing the decision depends on. Clean-tasting water is a real observation. It is not a statement about chlorine. The habit that protects you from all of them is the same: find out what the number is actually measuring before you let it size anything, which is what reading a discharge standard properly comes down to as well.
Contact time is the other half, and it does not come in the bag
The iodine number tells you how much surface you bought per litre of media. It does not tell you whether the water stays in contact with that surface long enough to use it. That second half is geometry and flow: bed depth divided by the velocity through it. Our own description puts it plainly โ the longer the water is in contact with the carbon, the more adsorption and surface reaction take place.
So there are two ways to end up with a carbon filter that does not remove chlorine, and they are independent. You can buy too little surface, or you can run good carbon too fast. The second is the commoner failure on packaged plants, and it has a specific cause we have written out at length: the vessels in front of a membrane sit on the feed side, so they pass the feed flow, not the product flow, and on a plant recovering roughly a third of its feed that is close to three times the number printed on the brochure, which that refilling-station note works through in full. A bed sized against the product figure is being asked to work at three times the velocity it was chosen for. The media is fine. The contact time is gone. It is the same mistake as sizing a sewage chamber off a daily average and discovering the peak, which is the reason an equalisation tank exists at all. The same arithmetic ruins a softener on the same skid, for the same reason, which is why a softener is sized by service flow rate and not by the salt it can hold, and why on that side too no amount of extra regenerant fixes a bed that is too small in area.
Two carbons at the same iodine number in beds of different depth behave differently. Two beds of identical depth carrying carbon at 1000 and at something under 500 behave differently. Neither figure is on a one-line specification, and the one-line specification is what gets priced. If you want to see what a written basis looks like when the numbers are on it, our design generator prints one for a sewage plant, and the computation sheet shows the arithmetic underneath it.
Exhaustion is invisible, and there is no gauge for it
A filter bed blinds and the differential pressure rises, which is a reading an operator can act on โ the kind of number a daily log exists to capture. Carbon does not do that. Adsorption consumes a finite capacity, and when the capacity is gone the bed still passes water at the same pressure drop it did on day one. There is no differential, no flow change, no visible appearance, and no cheap field test anybody keeps on site. It simply stops removing what it was removing.
Backwashing does not restore it. A backwash redistributes the bed and flushes out the fines and the biological growth; it does not empty the pores. That is a real difference between carbon and the other beds in the same plant room, and it is the difference people get wrong most often โ on the treatment train we set out step by step, the multimedia filter is washed and the carbon is replaced, and those two words describe genuinely different kinds of maintenance obligation โ the difference between a service visit and a purchase order, which is the whole of how we think about operating a plant. We keep a separate note on what a gravity filter bed needs for the same reason: a bed that cleans itself and a bed that gets used up are not the same asset.
Because exhaustion is silent, there are only three honest defences, and all three are decisions made before the plant is bought. Specify the iodine number. Give the bed the contact time its duty requires at the feed flow. Replace the media on a schedule you set in advance, rather than on a symptom that will arrive downstream of the carbon and look like something else. That last one is a maintenance obligation written into the design, which is the same argument as choosing equipment for what it does to you after the sale rather than for what it costs on the day. On the sewage side the equivalent discipline is a cone reading taken on a schedule โ a cheap measurement made routinely beats an expensive one made after a failure.
Where I will not overstate the case
The iodine number is a proxy. It measures adsorption of iodine from solution, and it correlates with the pore surface available, which is why it is the number the trade uses. It is not a direct measurement of how many kilograms of chlorine a bed will destroy before it stops. Chlorine removal on carbon is not pure adsorption either โ a good part of it happens as a surface reaction, chlorine being destroyed at the carbon rather than merely held there, which is one reason a bed can still be dealing with chlorine after it has stopped doing much for taste, and the reverse can also be true. So the two jobs do not report on each other in either direction. That is the useful conclusion, and it is more useful than a single number: nothing you can taste tells you anything about the chlorine, and nothing about the chlorine tells you the water will taste right.
The figures above are also somebody else's figures. Above 1000 is what we specify and what the supplier's datasheet carries; below 500 is what we have found in the market. They are not measurements made on our bench in Cordova, and we publish them the way we publish any manufacturer's number โ as the supplier's claim, which is the same care we took when we set out the published figures for a filtration medium we supply. What makes a datasheet figure mean anything on a delivery is the only practical step available to a buyer: ask for the batch certificate for the carbon that is actually being installed, not the brochure for the grade that was quoted. A number on a brochure and a number on a truck are two different facts, and we take the same view of a rate in a costing. We keep notes on the other media we supply the same way โ iron and manganese, the oxidation window those metals need before any filter can hold them, and what happens to a membrane when the filter in front of it channels โ and in each case the figures on those pages are the supplier's.
The same problem, one step up: media is the part nobody specifies
Carbon is the clearest case but it is not the only one. Nearly everything sold by the cubic metre in this industry has a performance specification that does not appear on the order. Ion exchange resin is sold by the litre and differs in capacity and in how long a run it will give. Filter media of the same nominal grade differ in grain shape and in how they foul. Fixed-film media in a sewage plant is sold by volume and what actually matters is protected surface area per cubic metre and whether it holds its biomass through an upset โ which is the argument behind comparing one treatment technology against another rather than comparing tank volumes. We keep a side-by-side comparison of the systems we supply for the water treatment equivalent.
In every one of those cases the vessel is visible, the media is invisible, and the money is in the media's specification. That is why a written design basis matters more than a bill of quantities, and why we say the design is the part that decides what the plant costs to own rather than what it costs to build. If the basis names the media specification, it can be checked โ including checked by somebody who is not us. If it says "complete with media", there is nothing there to check, and the cheapest bid is the one that reads that line most generously. That is the honest reason we will put a design basis in writing before anybody has signed anything, and where we draw the line between what is free and what is paid.
Where this turns up on the wastewater side
This is a drinking water and process water argument at first sight, and it reaches the sewage side in three places.
The first is reuse. Once a plant is treating effluent for anything beyond discharge, carbon appears in the train, and it appears after a filter. We have set out how far to treat for the reuse the client actually wants and when reuse pays at all, and the carbon in those trains is carrying the dissolved organic load the filter cannot touch โ which is a far harder duty than municipal water, so it is the last place to put a weak media. What level of treatment the reuse actually calls for is the first question, and treating above the rung the client needs is money spent on quality nobody uses.
The second is the high-end trains. The one we have described publicly is a hospital effluent polished for a dialysis duty, where the train needed ozone with venturi injection and catalytic granular carbon with manganese-impregnated media, not plain carbon by the cubic foot. When the carbon is specified for a reaction rather than for adsorption, the grade is the design, and the iodine number is only one of the figures that has to be right. That train is listed among the projects we have finished, and there are more of the water treatment ones on the systems page.
The third is chlorine itself, which is where the two halves of this business meet. On the wastewater side chlorine is dosed deliberately, in a contact chamber, and we have written about what a sludge bed in that chamber does to a phosphate result and about the two chlorine-and-ammonia routes we will not sell. On the water side chlorine is the thing being removed. The same molecule is a tool in one plant and a threat in the next one along, and in both cases the quantity is the whole argument. A plant that reuses its own effluent has both conversations on the same site, often in the same plant room. Which parameters matter at all depends on what the building does, and we keep a table of that by industry; the nutrient end of it is set out in the ammonia, nitrate and phosphate field guide.
What to ask before the vessel is ordered
In the order they are useful, and none of them needs a laboratory.
- Ask for the iodine number of the carbon being supplied, in writing, before the order is placed. A supplier who sells carbon by grade will have the figure immediately. A supplier who does not quote it is quoting volume.
- Ask whether the feed is chlorinated at all, and at what residual. A deep well with no chlorine changes the whole argument; a municipal line in a city changes it back. On a commercial kitchen there is usually a softener in the same room and the two beds are often sized off the same wrong flow figure.
- Ask for the batch certificate on delivery, and keep it with the plant records. It is the only thing that connects the grade you bought to the bags that arrived.
- Work out the velocity through the bed at the feed flow, not the product flow, and check the bed depth against it. If the plant recovers about a third of its feed, the vessels in front of the membrane are passing roughly three times the headline figure.
- Ask what the replacement interval is and what it costs, and put it in the operating budget before you sign. This is the number that belongs beside the capital figure, and on the sewage side we argue the same way about operating cost per cubic metre.
- Ask who will notice. Carbon exhaustion is silent, so the honest answer is a date in a maintenance schedule and a named person, which is the same conclusion we reached about designing for the operator who leaves and about what an operator can and cannot fix.
None of those six questions is about the brand on the vessel. Most of them are answered on paper, before anybody digs or pours. That is the general pattern of the questions worth asking, and it is why we keep a list of the ones to ask before signing. If a plant is already being quoted, tell us what the building is or send the proposal request through and we will read the media lines with you. We do this across Cebu and nationally โ the Cebu work is described here.
Where this does not transfer
If the plant has no membrane and no chlorinated feed, the chlorine half of the argument falls away and the carbon is doing a taste and odour job only. On that duty the customer's own palate is a reasonable, if crude, instrument, and a lower grade may be a perfectly rational purchase. The reason to care about the iodine number on a refilling station is the membrane sitting behind the carbon, and the reason to care on a reuse plant is the organic load reaching it. Take the component out of the argument and the argument changes.
Feed water varies more than the media does. A carbon bed on a clean deep well with no chlorine will outlast the same bed on a chlorinated municipal line by a wide margin, and neither experience transfers to the other. We assume influent and feed quality constantly because almost nobody has an analysis โ we have written about how often that is the case on the sewage side โ and every assumption has to be made safely, which in practice means a longer contact time than the calculation strictly needs. That costs money in a vessel, and it is a cost we would rather not spend on uncertainty. The limits the water is being judged against differ too, and the current ones are the ones to read โ the 2021 amendment changed six parameters and left the rest alone, which is worth knowing before anybody designs to a figure from memory.
And the two figures in this note came off our own written process description for one class of plant. The 1000 we specify is our specification for that duty, not a universal threshold, and the sub-500 material we describe is what we have found in this market rather than a survey of it. Ask for the number on your own project, do the contact-time sum for your own flows, and treat what we found as a reason to ask rather than as your answer โ and if you would rather we sat in that conversation with the supplier, that is what our consultations are for. Every site is different, and this has worked for us often enough that it is worth passing on.
If you want a second pair of eyes on a water or wastewater train before it is ordered โ the media specifications included โ book a free 15-minute consultation. We will tell you what we would ask the supplier, and whether the vessel in front of your membrane is big enough for the flow it will actually see.