Process & Water Chemistry

The mole is chemistry's counting unit — a fixed number of particles large enough to weigh on a bench balance. Dividing a measured mass by the molar mass is the bridge between the two worlds: weigh out 36.0 g of water, divide by its molar mass of 18.02 g/mol, and you know you have 2.00 mol — about 1.2 × 10²⁴ molecules. Every stoichiometry problem starts or ends with this conversion, because balanced equations speak in moles while balances speak in grams.

The word mole was coined by Wilhelm Ostwald in the 1890s, but the idea goes back to Avogadro's 1811 hypothesis that equal gas volumes hold equal numbers of molecules. Since the 2019 SI redefinition, the mole is defined by an exact count — 6.02214076 × 10²³ particles — so molar masses in g/mol are now measured quantities rather than definitions.

Molarity answers the practical question "how much stuff is in this bottle?" by counting moles of solute per liter of solution. Dissolve 58.44 g of table salt — exactly one mole of NaCl — in water and top up to the 1.00 L mark of a volumetric flask, and you have a 1.00 M solution. Note the fine print: it is per liter of solution, not per liter of water added, which is why chemists fill to a calibrated mark instead of adding a measured liter of solvent.

Molarity is the workhorse concentration unit because reactions are mole-to-mole affairs: multiplying C by a dispensed volume immediately gives the moles delivered, which is exactly what a titration calculation needs. Physiological saline is about 0.154 M NaCl, ocean water roughly 0.5 M, and concentrated hydrochloric acid around 12 M — a span that dilution calculations cross daily in every lab.

Mass percent is the concentration measure that needs nothing but a balance: the solute's mass divided by the mass of the whole solution — solute plus solvent — times 100. Dissolve 25 g of salt in 225 g of water and the solution is 25/250 × 100 = 10% salt by mass. The classic mistake is dividing by the solvent mass alone, which would wrongly give 11.1%.

Because it is temperature-proof and instrument-free, mass percent dominates industrial and household labels: household vinegar is about 5% acetic acid, physiological saline 0.9% NaCl, seawater roughly 3.5% dissolved salts, and concentrated sulfuric acid ships at 98%. Converting to molarity requires the solution's density, which is why bottle labels for concentrated acids list both figures side by side.

A mole is a count and nothing more mysterious — the chemist's dozen, scaled up to something useful. Multiply an amount in moles by the Avogadro constant and you have the literal number of particles in front of you. The only real question is why the constant is that particular size, and the answer is that it was chosen to make the bridge between two worlds land cleanly: $N_A$ is the number of atoms that makes a mole of carbon-12 weigh exactly 12 grams. That choice is what lets a mass in grams read off a balance be converted into a count of atoms, which is the single most useful trick in chemistry.

Scale is the thing worth feeling here. A 250 g glass of water is 250/18.015 = 13.9 mol, so it holds $13.9 \times 6.022\times10^{23} = 8.4\times10^{24}$ molecules. Now count the other way: all the water on Earth, about $1.4\times10^{21}$ litres, divided into 250 g glasses, comes to roughly $5.4\times10^{21}$ glasses. There are about fifteen hundred times more molecules in one glass of water than there are glasses of water in every ocean on the planet. That ratio is why chemists never think about individual molecules and why statistical behaviour is so reliable at this scale.

The constant carries Avogadro's name but not his arithmetic — he proposed in 1811 that equal gas volumes hold equal numbers of particles, and never estimated the number. Jean Perrin did, from Brownian motion, and named it for Avogadro in 1909, work that took him the 1926 Nobel Prize. The 2019 SI redefinition then reversed the logic entirely. $N_A$ is now exact by decree at 6.02214076 × 10²³ per mole, and the mole is defined as that many entities. Carbon-12's role is retired: a mole of it now weighs 12 grams only to within experimental uncertainty, rather than by definition. So 2.00 mol contains exactly 1.204428152 × 10²⁴ particles, with no uncertainty in the constant at all — though your measured 2.00 mol still has its own.

The error that swallows this page whole is leaving "particles of what" unstated. A mole of O₂ is 6.022 × 10²³ molecules but 1.204 × 10²⁴ atoms. A mole of NaCl is 6.022 × 10²³ formula units, which is 6.022 × 10²³ sodium ions plus the same number of chloride ions — 1.204 × 10²⁴ ions in total. A mole of Al₂(SO₄)₃ contains three moles of sulfate. Almost every wrong answer here is a correct calculation attached to the wrong noun, so write the noun down before you multiply: molecules, atoms, ions, or formula units.

Two smaller slips. Dividing a molar mass by $N_A$ gives the mass of one particle in grams — water comes out at $2.99\times10^{-23}$ g — and people routinely lose a factor of 1000 by mixing kilograms into that step, or confuse it with the mass in unified atomic mass units, which is just the molar mass number again with different units. And because $N_A$ is exact, it never limits your significant figures; only the amount you measured does.

Adding solvent to a solution spreads the same solute through a larger volume — the moles do not change, only their crowding. Since moles equal concentration times volume, $C_1V_1$ must equal $C_2V_2$ before and after any dilution. To prepare 250 mL of 1.0 M hydrochloric acid from a 12.1 M concentrated stock, solve for $V_1$: (1.0 × 250)/12.1 ≈ 20.7 mL of stock, made up to the 250 mL mark with water. (And always add acid to water, never the reverse.)

The law is not limited to molarity — any concentration measure proportional to moles per volume works, as long as both sides use the same one. Water-treatment operators lean on it daily when dosing inhibitor or biocide from concentrated drums into recirculating loops, and biologists use the identical arithmetic for serial dilutions, where each step divides concentration by a fixed factor.

$PV = nRT$ says that for a gas, four quantities are not independent: fix any three and the fourth is decided. Squeeze it and the pressure rises; warm it and it pushes harder or swells; add more of it and both go up. What makes the equation remarkable is not that those things are true — anyone with a bicycle pump knows them — but that one constant serves every gas. Helium, nitrogen and steam all obey it with the same $R = 8.314$ J/(mol·K), which is a strong hint that pressure has nothing to do with what the molecules are and everything to do with how many there are and how fast they are moving.

A worked case in units you would actually read off a gauge. A 20 L cylinder sits at 150 kPa absolute on a 20 °C morning. Rearranged for amount, $n = PV/RT = (150\,000 \times 0.020)/(8.314 \times 293.15) = 3000/2437 \approx 1.23\ \text{mol}$. Note what had to happen before the arithmetic: pascals not kilopascals, cubic metres not litres, and kelvin not Celsius. This page converts your entries for you, but the discipline is worth keeping in your head, because a scrap of paper will not.

The law arrived in pieces. Boyle established $PV$ constant at fixed temperature in 1662; Charles and Gay-Lussac tied volume and pressure to temperature around 1800; Avogadro proposed in 1811 that equal volumes of gases hold equal numbers of particles. Émile Clapeyron folded them into a single expression in 1834. Kinetic theory later derived the whole thing from mechanics: treat molecules as point masses that bounce elastically and never attract one another, average over their collisions with the walls, and $PV = nRT$ falls out — with $RT$ revealed as a measure of the average kinetic energy per mole.

Those two assumptions are also the fine print, and here a common textbook line deserves correcting. The ideal gas law is a limit, not a fact about gases. Real molecules do occupy volume and do attract each other, so the equation is exact only as pressure approaches zero and the gas gets out of its own way. Near condensation it fails plainly: at 100 atm, or anywhere close to the boiling point, the error runs to tens of percent and you want van der Waals or a compressibility factor. Under ordinary room conditions the error is well under 1%, which is why the approximation earns its keep.

Two errors account for most wrong answers on this page, and both are unit errors rather than physics errors. The first is feeding in Celsius. Doubling a gas from 20 °C to 40 °C does not double anything — in kelvin that is 293 to 313, a rise of 7%, and a calculation that used 20 and 40 would be wrong by a factor of nearly two. The second is feeding in a gauge pressure. A tire gauge reading 220 kPa means 321 kPa absolute; $P$ here is absolute pressure, measured from vacuum, because the equation counts molecular impacts and vacuum is where there are none. A third, quieter trap: the familiar 22.4 L per mole belongs to 0 °C and 1 atm. IUPAC redefined standard pressure to 100 kPa in 1982, and at that pressure the molar volume is 22.71 L. Both numbers circulate, and quoting one against the other's conditions is a 1.3% error hiding inside a memorised constant.

Fix the temperature and pressure and a mole of any ideal gas occupies the same volume — so converting between moles and litres needs one constant and no information about the substance. This is Avogadro's principle, and it deserves a moment of surprise before it becomes routine. A mole of hydrogen weighs 2 g and a mole of sulfur hexafluoride weighs 146 g, seventy-three times more, yet at the same conditions they fill the same flask. The reason is that pressure comes from the number of impacts and their momentum, and at a common temperature the heavier molecules move proportionally slower. The mass cancels out of everything the container can feel.

This page uses classic STP, 0 °C and 1 atm, where $V_m = 22.414$ L/mol. Burning one mole of methane, $\text{CH}_4 + 2\text{O}_2 \rightarrow \text{CO}_2 + 2\text{H}_2\text{O}$, consumes 2 mol of oxygen and produces 1 mol of carbon dioxide — which is 44.83 L of O₂ in and 22.41 L of CO₂ out at STP. Note that the two moles of water do not appear in that volume tally, because at 0 °C the water is a liquid and the equation only counts gases.

The constant is not independent; it is $PV = nRT$ evaluated once. $V_m = RT/P = (8.314 \times 273.15)/101\,325 = 0.022414$ m³/mol, and every other molar volume in circulation is the same calculation at different conditions. Historically the logic ran the other way. Gay-Lussac reported in 1808 that gases combine in simple whole-number volume ratios — two volumes of hydrogen to one of oxygen — and Avogadro's 1811 hypothesis explained why: equal volumes hold equal counts, so the volume ratios are the mole ratios of the balanced equation. That insight was ignored for half a century before Cannizzaro revived it at Karlsruhe in 1860.

The error that dominates this page is using 22.4 L/mol for conditions that are not STP. Room temperature is not 0 °C. At 25 °C and 1 atm the molar volume is 24.47 L/mol, so applying 22.4 to a bench-top measurement understates the volume by 9%. And "standard conditions" is not one definition but several: classic STP gives 22.414 L/mol, IUPAC's post-1982 STP (0 °C, 100 kPa) gives 22.711, and SATP (25 °C, 100 kPa) gives 24.79. The spread between the smallest and largest is over 10%, which is far more than the precision most people think they are carrying. Check which convention a source assumes before you borrow its number, and when the conditions are anything other than a listed standard, abandon the shortcut and use the ideal gas law directly.

Two smaller cautions. This is an ideal-gas result, so it degrades for gases near their condensation point — ammonia and sulfur dioxide at STP already deviate by a percent or two, and water vapour at 0 °C is not a gas at all. And when you use volume ratios as mole ratios in a reaction, they apply only to the species that are actually gaseous at the stated conditions. A dissolved or condensed product contributes no volume, and counting it is the quiet way to get a stoichiometry problem wrong.

Start from $PV = nRT$, substitute $n = m/M$, and rearrange for mass over volume: $\rho = PM/RT$. What the result says is that a gas has no density of its own. Unlike a solid or a liquid, whose density is close enough to a fixed property to tabulate, a gas takes whatever density its pressure and temperature impose, and the only thing the substance itself contributes is $M$. Squeeze it and it gets denser in exact proportion; warm it and it thins in inverse proportion.

Air at 101.325 kPa and 20 °C, using $M = 0.028964$ kg/mol: $\rho = (101\,325 \times 0.028964)/(8.314 \times 293.15) = 1.204\ \text{kg/m}^3$ — the figure every ventilation calculation starts from. Helium at the same conditions, with $M = 0.0040026$, comes to 0.166 kg/m³. Subtract, and a cubic metre of helium lifts about 1.04 kg. That is the entire physics of a party balloon, and it explains why a balloon large enough to lift a person has to be the size of a house.

Rearranged for molar mass the equation becomes a measurement rather than a prediction, and a historically important one. Weigh a bulb of known volume empty, fill it with a vapour at measured temperature and pressure, weigh it again, and $M = \rho RT/P$ hands you the molar mass of an unknown. This is the Dumas method, and through the middle of the nineteenth century it was one of the few routes to a molecular formula. It also settled arguments: measured vapour densities are what showed that many elemental gases travel as diatomic molecules rather than lone atoms.

The dominant error here is the molar mass unit, and it is a clean factor of a thousand. With $R = 8.314$ J/(mol·K) the equation demands $M$ in kilograms per mole. Air is 0.029 kg/mol. Enter 29 and the answer comes back as 1204 kg/m³ — air denser than water — which at least announces itself. Enter 0.029 when the calculation wanted grams and you get the mirror error. The usual companions apply too: $P$ must be absolute, not a gauge reading, and $T$ must be in kelvin.

Two conceptual notes. There is no such thing as "the molar mass of air" in the strict sense — air is a mixture, and 28.96 g/mol is a mole-weighted average of nitrogen, oxygen and argon. It works precisely because an ideal gas is indifferent to what its neighbours are; only the total count matters. That same indifference produces a result most people find backwards: humid air is lighter than dry air. Water is 18 g/mol against air's 29, so at a given pressure and temperature every water molecule that joins the mixture has displaced a heavier one. Muggy days are low-density days, which is why aircraft performance charts include humidity and why a hot, humid runway is a long takeoff.

A balanced equation promises a certain mass of product; the flask rarely delivers all of it. Side reactions consume reactant, some product stays dissolved in the mother liquor, and a little is always lost on filter paper and glassware. Percent yield is the honest scorecard: the mass you actually isolated divided by the stoichiometric maximum, times 100. In a classic teaching lab, synthesizing aspirin from salicylic acid might predict 5.00 g of product; if 4.21 g of dry crystals come off the funnel, the yield is 84.2%.

The number matters far beyond the classroom. Process chemists judge manufacturing routes largely by yield, because in a multi-step synthesis losses multiply — five steps at 80% each deliver only 33% overall. A reported yield above 100% is a red flag, not a triumph: it usually means the product is still wet or carries impurities.

A chemical formula is a recipe by count, and this equation translates it into a recipe by mass. Molar mass is additive — the compound weighs exactly what its atoms weigh, summed — so an element's share of the total is its atoms' contribution divided by the whole. In water, hydrogen brings 2 × 1.008 = 2.016 g/mol out of 18.02, which is 11.2% by mass; oxygen carries the other 88.8%. Two atoms out of three, and barely a ninth of the weight.

A case with real stakes. Ammonium sulfate, (NH₄)₂SO₄, has a molar mass of 132.14 g/mol. Nitrogen appears twice, so $\%N = (2 \times 14.007)/132.14 \times 100 = 21.2\%$. That is not a coincidence of the classroom — it is why the fertilizer is sold as 21-0-0, and why a 25 kg bag delivers 5.3 kg of actual nitrogen to a field.

Run the arithmetic backwards and it becomes analysis rather than description. Burn an unknown compound, weigh the CO₂ and H₂O produced, convert those to masses of carbon and hydrogen, take oxygen by difference, then divide each element's mass percent by its atomic mass and normalise to the smallest result. What comes out is the empirical formula. This is combustion analysis, and it was the central technique of organic chemistry from Liebig's refinement of it in the 1830s until spectroscopy arrived. Behind all of it sits Proust's law of definite proportions: a given compound always holds the same elements in the same mass ratio, which is exactly the claim this equation encodes.

The subscript $a$ is the number of atoms of that element per formula unit, and parentheses are where people lose it. In (NH₄)₂SO₄ the hydrogen count is 8, not 4 — the subscript outside the bracket multiplies everything inside. Hydrates catch people the same way: copper(II) sulfate pentahydrate, CuSO₄·5H₂O, has a molar mass of 249.7 g/mol including its water, so the copper is 25.4%, not the 39.8% you would get from the anhydrous salt. If the bottle says pentahydrate, the water is part of what you weighed.

Two further traps. Percent composition can only ever give you an empirical formula, never a molecular one: formaldehyde CH₂O, acetic acid C₂H₄O₂ and glucose C₆H₁₂O₆ all analyse to identical percentages, and separating them needs an independent molar mass. And on fertilizer bags, only the first number is what it appears to be. The N in N-P-K is genuinely percent nitrogen, but P is reported as P₂O₅ equivalent and K as K₂O — a nineteenth-century convention that never died. A bag marked 0-46-0 is 46% P₂O₅, which works out to about 20% actual phosphorus. Reading it as elemental phosphorus overstates the dose by well over a factor of two.

At the equivalence point the moles of titrant delivered, $C_b V_b$, exactly match the moles of analyte present, $C_a V_a$, scaled by the balanced equation's mole ratio n. Everything else in a titration — the burette, the indicator, the swirling — exists only to find that point precisely. Titrate a 25.00 mL aliquot of hydrochloric acid with 0.1000 M sodium hydroxide and take 23.45 mL to reach the endpoint: with n = 1, $C_a$ = (1 × 0.1000 × 23.45)/25.00 = 0.09380 M, good to four figures from nothing but glassware.

The mole ratio is where marks are lost. For a diprotic acid such as H₂SO₄ titrated with NaOH, one mole of acid consumes two of base, so n = 0.5: a 25.00 mL aliquot needing 30.00 mL of 0.100 M NaOH is 0.5 × 0.100 × 30.00/25.00 = 0.0600 M. Karl Friedrich Mohr systematised the whole technique in his 1855 Lehrbuch der chemisch-analytischen Titrirmethode, introducing the burette clamp and the pinchcock that made reproducible volumetric analysis possible; his methods still underpin water-hardness and chlorine testing today. Note the difference between the endpoint (where the indicator changes) and the equivalence point (where the stoichiometry balances) — the gap between them is the indicator error, which is why the indicator is chosen to change colour on the steep part of the titration curve.

Normality counts reactive capacity rather than molecules. One mole of sulfuric acid delivers two protons, so 0.500 M H₂SO₄ is 1.00 N — and 1.00 N of any acid neutralises 1.00 N of any base volume for volume, which is exactly why the unit survived so long in analytical labs. For redox work the equivalents are electrons: potassium permanganate in acid picks up five, making 0.0200 M KMnO₄ a 0.100 N oxidant.

The catch, and the reason IUPAC deprecated normality decades ago, is that the equivalent count depends on the reaction, not the substance. Phosphoric acid is 3 N when fully neutralised to phosphate but only 1 N in a titration stopped at the first endpoint, so a bottle labelled "1 N H₃PO₄" is ambiguous without knowing the intended reaction. You will still meet it constantly in water-treatment and clinical chemistry, where hardness, alkalinity, and electrolyte balances are quoted in equivalents per litre or milliequivalents per litre — blood sodium at 140 mEq/L being the most familiar example.

Reactions in water happen equivalent-for-equivalent, not gram-for-gram: one charge neutralizes one charge, one H⁺ neutralizes one OH⁻. The equivalent weight is simply the mass of a species that carries one mole of that reacting capacity — molar mass divided by valence. Calcium carbonate, M = 100.087 with a divalent cation, has EW = 50.04 g/eq, which is why 50.04 turns up in every "as CaCO₃" conversion in this trade. Sulphuric acid, M = 98.08 with two replaceable protons, has EW = 49.04 g/eq, so a pound of pure H₂SO₄ neutralizes almost exactly a pound of alkalinity expressed as CaCO₃ — the near-1:1 coincidence that lets operators do acid-feed arithmetic in their heads.

The valence you divide by is the one that applies to the reaction in question, and that is where people go wrong. Sodium carbonate reacting as a base uses z = 2 (EW 53.0), but if you are counting sodium ions it behaves as z = 2 for a different reason. Bicarbonate is z = 1 (EW 61.0) in the alkalinity titration but the carbonate it came from is z = 2 (EW 30.0). Write down which reaction you mean before you pick z, and the rest of the water chemistry falls into line.

Chlorine added to water does not stay chlorine. It is consumed by iron, manganese, sulphide, ammonia, and every scrap of organic matter it meets — that consumption is the demand — and only what survives shows up on a DPD test as residual. The bookkeeping is a straight sum: dose equals demand plus residual. Feed 3.3 mg/L into a water with 2.5 mg/L of demand and you keep 0.8 mg/L residual, which is a typical distribution target. There is no way around the demand; it has to be satisfied before a single tenth of a milligram of residual appears, which is why a well with high iron can swallow 4 mg/L and still test zero at the tap.

The practical procedure is a chlorine demand study: dose a set of bottles of the actual water at increasing rates, wait the real contact time — usually 30 minutes, but overnight for a distribution main — and test each. The plot of residual against dose is flat until demand is satisfied, then rises with unit slope. Two field cautions. Demand is not a fixed property: it climbs in warm weather, after a main break, and with every degree of surface-water turbidity, so the study needs repeating seasonally. And test at the far end of the system, not at the plant — the residual that matters legally and microbiologically is the one still standing at the last service connection.

Every North American operator learns this one as a chant: pounds per day equals milligrams per litre times million gallons per day times 8.34. The 8.34 is nothing mysterious — it is the weight of a US gallon of water, 8.345 lb at 60 °F, rounded. Dosing 2 mg/L of chlorine into a 1 MGD plant needs 2 × 1 × 8.34 = 16.7 lb/day. Underneath the trade constant it is simply concentration times flow, mass balance in its plainest form, and this page will do it in kilograms per hour and cubic metres per hour just as happily.

Three warnings from the field. First, this gives pounds of active chemical; if you are feeding 12.5% sodium hypochlorite you still have to divide by the strength to get pounds of product, and again by the solution's density to get gallons the pump must deliver. Second, the 8.34 quietly assumes the water is water — feeding into a brine, a slurry or a glycol loop needs the actual density. Third, calibrate the pump against a drawdown cylinder rather than trusting the dial: a diaphragm pump's stroke setting drifts with discharge pressure, and the difference between a calculated feed rate and a measured one is where half of all chemical overspend hides.

A gauge on a chemical tank reads volume, but inventory, dosing and safety paperwork all want mass of active ingredient. Three multiplications get you there: volume, then specific gravity to account for the solution being heavier or lighter than water, then the percent active. In trade units, gallons × 8.345 lb/gal × SG × %/100. A 500 gallon tank of 50% caustic soda at SG 1.53 holds 500 × 8.345 × 1.53 × 0.50 = 3,192 lb of NaOH — nearly a ton and a half, in a tank that only reads "500 gallons."

Specific gravity is where the money and the mistakes are. Concentrated chemicals are dense: 50% caustic is 1.53, 93% sulphuric acid is 1.835, 12.5% sodium hypochlorite is about 1.16, and 40% ferric chloride is 1.42. Ignoring SG on the caustic tank understates your inventory by a third. Two field notes: 50% caustic freezes at 12 °C (54 °F) and will plug an outdoor line solid, which is why northern plants buy 25% and pay to ship the water; and SG drifts with temperature, so a hydrometer reading taken on a hot summer afternoon will underread concentration unless you correct it back to the reference temperature on the supplier's table.

Chlorine chemicals are sold by product weight but dosed by chlorine content, and the gap between the two is large. Sodium hypochlorite bleach ships at 10–15% available chlorine (12.5% is the standard water-plant grade), calcium hypochlorite granules at 65–70%, and chlorine gas is the only 100% product. To deliver 10 lb of available chlorine from 12.5% bleach you need 10 × 100/12.5 = 80 lb of solution. From 65% cal-hypo you would need only 15.4 lb of granules. The phrase "available chlorine" is itself a convention — it measures oxidizing power relative to elemental Cl₂, which is why calcium hypochlorite can test above 100% on a molar basis and still be quoted at 65%.

Liquid hypochlorite decays, and this is the number one cause of failed disinfection in small systems. A 12.5% solution loses roughly half its strength in 30–60 days at room temperature, faster if it is warm, exposed to light, or contaminated with trace metals. A drum bought at 12.5% and used three months later may genuinely be 8%, so the pump set from the label under-doses by a third. Titrate incoming shipments, store cool and dark, rotate stock, and never dilute hypochlorite with hard water — the calcium precipitates and plugs the injection quill.

This is the shock-chlorination and batch-dosing formula: you have a bucket of product, a tank or a well of known volume, and you need the resulting parts per million. One pound of 65% calcium hypochlorite in 1,000 gallons gives 0.65 lb of chlorine in 8,340 lb of water, which is 0.65/8,340 = 78 parts per million. Well disinfection specifications typically call for 50–200 mg/L held for 12–24 hours; a new water main is usually chlorinated to 25 mg/L for 24 hours or 50 mg/L for 3 hours before flushing and bacteriological clearance.

Getting the volume right is more than half the job. A drilled well's water column is π/4 × d² × depth-of-water, which for a 6-inch casing is about 1.5 gallons per foot of standing water — and you must add the volume of the pressure tank, the plumbing and any storage, because the chlorine has to reach everywhere the bacteria are. The other trap is where you pour it. Dumping cal-hypo granules down a well casing without pre-dissolving drops undissolved pellets onto the pump and can pit stainless screens; make up a slurry in a clean pail with soft water first, and circulate by running each fixture until it smells of chlorine, then shut everything and let it stand.

Ammonia in the source water hijacks chlorine. The first chlorine you add makes monochloramine, a weak combined residual; keep adding and you make dichloramine, then nitrogen trichloride, and eventually the chlorine oxidizes the ammonia all the way to nitrogen gas. The residual actually falls as you increase the dose over that middle stretch — the famous dip in the breakpoint curve — and only past the low point, the breakpoint, does free chlorine start to accumulate. Stoichiometry puts breakpoint at 7.6 parts chlorine to 1 part ammonia nitrogen by weight (from 3Cl₂ + 2NH₃ → N₂ + 6HCl), so 0.5 mg/L of ammonia nitrogen needs about 3.8 mg/L of chlorine before you see any free residual at all.

Practice runs richer than theory: real waters demand 8:1 to 10:1 because organics and other reduced species compete, and plants commonly target 10:1 to be safely past the hump. The failure mode is stopping halfway. An operator sees the residual dropping as they increase the feed, assumes something is wrong, and backs off — landing squarely in the dichloramine valley where the water tastes and smells worst and the disinfection is weakest. That swampy "swimming pool" odour is chloramines, not excess chlorine; the cure is almost always more chlorine, not less. The mirror image is deliberate chloramination, where a plant holds a 4:1 to 5:1 ratio on purpose to keep a long-lived combined residual in a sprawling distribution system with minimal trihalomethane formation.

Disinfection obeys a dose law, and the dose is concentration multiplied by time. That is the whole of it. A regulator writing a drinking-water rule needs one number that says "this water has been disinfected enough", and $C \times T$ is the number: the residual a pathogen swam through, times how long it swam through it, in milligrams per litre times minutes. Hold 1.2 mg/L of free chlorine for 45 minutes of real contact and you have delivered 54 mg·min/L.

This page computes the CT you achieved. It does not tell you whether you passed. I want to be very plain about that, because it is the thing people most want a calculator to do and the thing a calculator has no business doing. The required CT depends on four things at once — the pathogen, the disinfectant, the water temperature and the pH — and it is a table published by your regulator, revised on their schedule, keyed to conditions you have to measure rather than assume. Look it up. There is no table on this site and there never will be, because a stale table is worse than no table: it looks authoritative and it is wrong, and the person who trusts it is running a plant.

Three things sink real plants, and I have watched all three.

The C is the residual at the END of the contact zone, not the dose you fed. Demand eats the difference on the way through, and on a coloured surface water in the spring it can eat most of it. Feeding 2.5 mg/L and measuring 0.6 at the outlet means your CT is built on 0.6. Where the residual is falling noticeably along the contact chamber, the honest practice is to break the chamber into segments and add up the CT of each, using the residual measured at the end of each one — that is what the EPA guidance describes, and it is more work and it is more nearly true.

The T is T₁₀, not the detention time. This is the big one and it gets its own page. Volume over flow gives you θ, the time an average drop spends in the tank if the tank behaves; T₁₀ is the time by which only a tenth of a tracer has come out the other end, and real basins short-circuit badly enough that T₁₀ can be a tenth of θ. Using θ where the rule asks for T₁₀ overstates your disinfection by up to a factor of ten, and it does it silently, on a spreadsheet that adds up correctly.

And the credit was earned at a temperature. The same 3-log Giardia credit costs roughly twice the CT at 5 °C as at 15 °C, because everything about the reaction slows in cold water. A plant that passes comfortably every August can be failing every February on identical operation, and the way it usually gets found is an inspection rather than a violation. Take your credit at the worst temperature of the day, not the average, and certainly not the design value.

The disinfectant matters as much as the temperature. Ozone and chlorine dioxide need far less CT than free chlorine for the same credit; chloramine needs vastly more, which is exactly why it is a distribution-system residual and not a primary disinfectant. And free chlorine loses potency as pH rises, because the species doing the killing is hypochlorous acid, and above about pH 7.5 an increasing share of it has dissociated into the far weaker hypochlorite ion. That is why the CT tables have a pH axis at all.

Cryptosporidium is the reason the whole framework looks the way it does. Its oocysts are effectively immune to free chlorine — the CT for a single log runs into thousands of mg·min/L, which is to say hours at residuals no plant would ever feed. You cannot chlorinate your way past it. That single fact is why UV and ozone appear in treatment trains, why filtration performance is regulated so hard, and why the rules that came after the Surface Water Treatment Rule were written the way they were. If Cryptosporidium is your concern, nothing on this page addresses it.

One note on units, because water operators will notice. The residual here is typed as a mass concentration, and this site deliberately will not convert mg/L to ppm. In fresh water they are numerically identical, and operators use them interchangeably every day and are entirely right to — a litre of water weighs a kilogram, so a milligram of solute in it is a part per million by mass. But that equality is a property of the solvent, not a conversion. Put the same measurement in brine, in a glycol loop, or in anything denser than water and the two numbers separate. The site keeps them apart so that the one time it matters, it does not quietly get it wrong.

This is the most important page in this shard, and if you read only one of them, read this one.

The T in CT is not the detention time of the tank. Divide the working volume by the flow and you get $\theta$, the theoretical detention time — the time an average drop spends inside if the basin behaves like a well-ordered queue. Basins do not behave like queues. A jet from the inlet pipe drives a stream of water straight across to the outlet in a fraction of the calculated time; a corner behind a wall holds water for hours that never joins the flow at all. What comes out of the far end at any moment is a mixture of water that has been in there for very different lengths of time, and the disinfection credit has to be earned by the water that got the least.

So the rule takes credit on $T_{10}$: the time by which only 10% of an injected tracer has appeared at the outlet, which means 90% of the water has had at least that much contact. The relation to the theoretical time is a single multiplication, $T_{10} = \theta \times \mathrm{BF}$, where BF is the baffling factor — and everything hangs on how small that factor is.

The published defaults run roughly like this. About 0.1 for an unbaffled tank with no inlet or outlet structure worth the name — a bare clearwell with a pipe in one end and a pipe in the other. About 0.3 for poor-to-average baffling, an inlet baffle and an outlet weir. About 0.5 for superior baffling, with intra-basin baffles. About 0.7 for a proper serpentine basin with a perforated inlet diffuser. And 1.0 for perfect plug flow, which is a limit and not a design target, because no real basin reaches it.

Look at what those numbers do to a compliance calculation. Take a clearwell with a 45-minute detention time at a 1.0 mg/L free chlorine residual. Multiply straight through and you report CT = 45 mg·min/L. Now apply an average baffling factor of 0.30: the effective contact time is 13.5 minutes and the CT you may actually claim is 13.5 mg·min/L. Same water, same chlorine, same tank — a third of the credit. At the unbaffled end, with a factor of 0.1, it is a tenth. That is the gap, and the reason it is dangerous is that the wrong version is not obviously wrong: the arithmetic is correct, the spreadsheet adds up, and the number is off by a factor of ten.

Two things follow for anyone designing rather than reporting. First, run the equation backwards and it sizes concrete: the detention time you must build is $T_{10}/\mathrm{BF}$, so a basin at 0.1 has to be seven times the size of one at 0.7 to deliver the same credit. Interior baffle walls are dramatically cheaper than seven times the tank. Second, notice the alternative you are choosing against. A poorly baffled basin can be made to pass by carrying a higher chlorine residual, and that residual is then carried for the life of the plant, showing up as taste and odour complaints and as trihalomethanes and haloacetic acids in the distribution system. Fixing the hydraulics is a one-time cost; chasing them with chemistry is a permanent one.

Measure it rather than assuming it. The published factors are conservative defaults assigned by description — "average baffling" — and your basin is not a description. A tracer study with salt or fluoride gives you the real number, and it can go either way: a basin measuring 0.40 when it has been claiming 0.30 is free credit the plant already owns and is not taking, and a basin measuring 0.18 when it has been claiming 0.30 is a compliance problem nobody knew about. Run the study at the highest flow the plant sees, because short-circuiting gets worse as flow rises, and that is the condition the credit has to survive. And when you compute $\theta$, use the working volume — the sludge blanket and the dead corners are not detention, whatever the tank drawing says.

This is the measurement the whole framework is trying to predict: count the organisms going in, count them coming out, and the base-10 logarithm of the ratio is the log reduction. Two hundred and fifty thousand per 100 mL entering, twenty-five leaving, and the ratio is ten thousand — four logs, flat.

Logs rather than percentages, and the reason is that percentages crowd together where the interesting behaviour is. Ninety-nine percent and 99.99% differ by two characters on the page and by a factor of a hundred in survivors, and no one's intuition handles that gap. The log scale spreads it out, and — more usefully — logs add across barriers. Two-log removal in the filters followed by two-log inactivation in the contact chamber is four logs overall, and that additivity is the entire architecture of multi-barrier treatment. Percentages do not add: the same pair is 99% then 99%, which is 99.99% overall, not 198% and not 99%.

A non-detect is not a zero. This is the mistake that inflates more log-reduction claims than any other, and it is easy to make in good faith. A plate that grows nothing does not mean the count was zero; it means the count was below the detection limit. Divide by zero and the log reduction is infinite, which is a number no disinfection process has ever achieved. The honest substitution is the detection limit itself, and what you get is the minimum log reduction your data can support — say "at least 4.4 logs", not "complete inactivation". If you need to demonstrate more, you need a bigger sample volume or a more sensitive method, not a different arithmetic.

The counts themselves are noisier than they look. Microbial counts are Poisson-distributed, which means a plate reading of 4 carries a standard deviation of 2 — the uncertainty is the square root of the count. Any log reduction built on small numbers inherits that scatter in full, and the difference between 3.6 and 4.1 logs on a single pair of plates is frequently nothing at all. This is why performance is demonstrated over many samples and many days rather than announced from one good result, and why a seeded challenge study uses enormous spikes: a starting count of $10^7$ leaves room to measure six logs of removal with counts still large enough to be stable at both ends.

And the reduction you measured is not the reduction you are credited. Regulatory credit for inactivation comes from a demonstrated CT under approved conditions, and credit for removal comes from an approved and properly operated process — not from a favourable pair of samples. The measurement is how you find out whether your process is doing what it is credited for. It is monitoring, and it is valuable precisely as monitoring: a plant whose measured reduction drifts down over a season has a real problem, whatever its paperwork says.

One last framing that surprises people. When the arithmetic returns a survivor count below one — say 0.02 organisms per 100 mL — that is not an error and it is not zero. Read it as a probability: roughly a 2% chance that any given 100 mL sample contains an organism. A utility producing many millions of such volumes a day is still delivering a real number of them, which is why drinking-water targets are framed as an acceptable annual risk of infection rather than as an absence, and why the log framework does not stop at "none detected".

Three-log is 99.9 percent. Four-log is 99.99 percent. They are the same statement said two ways, and this page exists because people mix them up constantly — in reports, in tender documents, in conversations with regulators, and occasionally in the same sentence.

The arithmetic is one line. A log reduction of $\mathrm{LR}$ leaves a surviving fraction of $10^{-\mathrm{LR}}$, so the fraction killed is $P = 1 - 10^{-\mathrm{LR}}$, and running it backwards, $\mathrm{LR} = -\log_{10}(1-P)$. One log leaves a tenth and kills 90%. Two logs leave a hundredth and kill 99%. Six logs leave a millionth and kill 99.9999%.

The percent form hides everything interesting in its last decimal places. Moving a claim from 99.9% to 99.99% looks like a typographical refinement and is a whole extra log — a tenfold cut in survivors, which in treatment terms is often another barrier, another vessel, another capital project. That asymmetry is exactly why every disinfection rule in the world is written in logs, and why a percent kill quoted to several decimals should always be converted before it is believed.

The confusion that costs money is double-counting. I have read reports that say a process achieves "3-log inactivation, that is 99.9% removal", and then add a further "99.9% removal" from a second barrier, and conclude something enormous. Those are two barriers of three logs, which is six logs total — 99.9999% — and the correct way to get there is to add the logs. Percentages cannot be added or multiplied intuitively, which is the practical case for never working in them at all.

One hundred percent is not a number this equation accepts. Its log form is infinite, and no disinfection process delivers it — 100% is always a rounding, or a non-detect being reported as an absence. If a supplier's literature says 100%, ask what the detection limit was and what the starting count was, and convert those two into the minimum log reduction the data actually supports. That number is usually respectable and always defensible; the 100% is neither.

Worth keeping the survivor form in your head as well as the percentage. A log reduction of $\mathrm{LR}$ means one organism in $10^{\mathrm{LR}}$ survives: 3-log is one in a thousand, 6-log is one in a million. Said that way, the difference between 99.9% and 99.9999% stops being three extra nines and becomes what it is — a thousandfold difference in what reaches the tap.

In 1908 Harriet Chick published in the Journal of Hygiene the observation that disinfection behaves like a chemical reaction: plot the surviving fraction on a logarithmic axis against time and you get a straight line. Organisms do not die all at once at some threshold; they die at a constant proportional rate, so each equal interval of time removes the same fraction of whatever is left. In the same volume H. E. Watson added the piece Chick's law was missing — that concentration does not simply multiply the rate, but enters raised to an exponent — and the pair of them has carried disinfection theory ever since.

Written for the log reduction, it is $\mathrm{LR} = k\,C^{n}\,t$. Three of those symbols are ordinary. $C$ is the disinfectant concentration, $t$ is the contact time, and $\mathrm{LR}$ is the base-10 log reduction achieved. The interesting one is $n$, Watson's dilution coefficient, and what it measures is which of your two knobs is worth turning. When $n = 1$, concentration and time trade one for one — halve the residual, double the time, get the same kill — and the model collapses to exactly the CT product the regulations use. Free chlorine sits close enough to 1 that the regulatory shortcut works, which is not a coincidence: the framework was built on the disinfectant it works for. When $n$ is well above 1 the disinfectant is concentration-driven, and a stronger residual buys more than a longer basin. When it is below 1, the reverse, and the concrete is the better purchase.

$k$ is unit-bound, and this is the honesty problem of the page. It carries the units $(\text{mg/L})^{-n}$ per minute, which is not a rate you can convert, because the exponent on the concentration changes the dimensions. A $k$ fitted at $n = 1$ cannot be used at $n = 0.8$; a $k$ taken from a paper that worked in different concentration units is a different number entirely, and papers very often do not say. This page takes $k$ on the trade basis — milligrams per litre and minutes — and no calculator can guess what basis a bare number came from.

Which is a good reason to fit your own. A $k$ measured on your water beats any published figure, because it silently absorbs everything the model leaves out: the organism, the pH, the temperature, the particle load, the mixing in your jar. But fit it honestly. One pair of numbers cannot recover both $k$ and $n$ — that is two unknowns and one equation, and putting one point through a two-parameter model is a coincidence, not a fit. Run several concentrations at several times, plot log survival against time to get the slopes, then plot those slopes against log concentration to get $n$ from the second slope.

And then be sceptical of the line itself. Chick's law assumes every organism is equally susceptible and equally exposed, and real survival curves say otherwise in two visible ways. At the start there is often a shoulder, a lag while the disinfectant works through clumps, through cell walls, through whatever is shielding the target. At the end there is a tail: the line flattens and further contact buys almost nothing.

The tail is the half that matters, and it is worth being blunt about why. The tail is the resistant fraction — organisms shielded inside particles, aggregated into clumps, or simply tougher than their siblings — and it is precisely the fraction that survives to reach a customer's tap. Extrapolating a fitted straight line past the data you actually collected is optimistic in exactly the direction you cannot afford to be optimistic in. If your bench work stops at 4 logs, you know about 4 logs. The model will happily print 8 and it does not know anything about them.

Various refinements exist to bend the line — Hom's model puts an exponent on time as well, and delayed-Chick–Watson forms build in a shoulder — and all of them cost you another fitted parameter. For plant work the honest position is usually the simple model plus a healthy distrust of its extremes, rather than a more elaborate model fitted to the same thin data.

One organism is outside all of this. Cryptosporidium resists free chlorine at every CT a drinking-water plant can physically deliver, and no value of $k$ worth writing down will change that. Its control comes from filtration, from UV, or from ozone. That is the reason the modern treatment train looks the way it does, and it is why the log-credit framework separates the pathogens rather than treating disinfection as one number.

Free chlorine does not sit still. It reacts — with natural organic matter it did not finish with at the plant, with biofilm on the pipe wall, with corrosion products, with whatever else the distribution system offers — and the residual falls exponentially as it goes. $C = C_0 e^{-kt}$ is the standard first-order description, and what "first order" means in practice is that the residual loses the same fraction in every equal interval: if it halves in twelve hours, it halves again in the next twelve, whatever it started at.

Take a main carrying 1.4 mg/L out of the plant with a bulk decay coefficient of 0.15 per hour. Eight hours of travel gives $kt = 1.2$, and $e^{-1.2} = 0.301$, so 0.42 mg/L reaches the sample point. The implied half-life is $\ln 2 / k = 4.6$ hours — a useful figure to carry in your head, because it converts a decay coefficient into something you can reason about without a calculator.

The coefficient belongs to your water and should be measured on it. The standard method is a bottle test: fill headspace-free bottles at the plant, hold them at distribution temperature in the dark, and read the residual at intervals over a day or two. Plot the natural log of the residual against time and $k$ is the negative of the slope. Three cautions about that fit. Decay is strongly temperature dependent, so a coefficient measured in February badly understates August loss. The first hour or so is usually much faster than everything after it, because the fast-reacting organics go first, and a $k$ fitted through that initial plunge will overpredict decay at the ages your system actually runs — fit the long tail. And a bottle test measures bulk decay only.

That last point matters more than it sounds. A glass bottle has no biofilm, no tuberculation and no corroding iron. A real main has all three, and wall demand can dominate bulk demand entirely in old unlined cast iron. So expect the field residual to come in below what this equation predicts, and treat the difference as information rather than as error: a large gap between bottle and field is a measurement of your pipe's condition.

Travel time is not a single number either. A distribution system is not one pipe. The trunk main might deliver in three hours while a dead-end leg holds water for four days and an oversized storage tank stratifies and turns over a fraction of its volume a day. It is the old water that loses its residual, and losing the residual is where nitrification starts in a chloraminated system and where a coliform positive eventually appears. Running the equation backwards to estimate water age from a residual drop is a legitimate blunt instrument for finding those places, but it assumes bulk decay and a single path, and a sample point fed by a blend of fresh trunk water and old tank water returns a meaningless average of the two. Use it to decide where to point a hydraulic model or a fluoride tracer, not as a substitute for either.

The design direction is the honest one to end on. Solve for $C_0$ and you get the plant setpoint needed to hold a target residual at the far end — and the exponential is unforgiving, because every additional half-life of water age doubles what you must feed. Which is why the answer to a distant dead-end is almost never "raise the plant residual". That dulls the whole system with taste and odour complaints and drives disinfection by-products up everywhere, not just where the problem is. Flushing the dead end, or re-operating the tank so it actually turns over, treats the cause.

In a first-order reaction each molecule decomposes independently with a fixed probability per second, so the rate is proportional to how much is left and the concentration falls exponentially. Taking logarithms gives ln[A] = ln[A]₀ − kt, a straight line — the diagnostic test that distinguishes first order from every other order. The defining property is a constant half-life, t½ = ln2/k = 0.693/k, independent of where you start.

The maths is identical to radioactive decay, which is why the same equation covers the elimination of most drugs from the bloodstream, the fading of a chemiluminescent glow stick, and the isomerisation of cyclopropane to propene. Worked case: with k = 0.0231 s⁻¹, a 0.100 M solution falls to 0.100 × $e^{-0.0231 \times 30}$ = 0.0500 M in 30 s — the half-life, since 0.693/0.0231 = 30.0 s. The usual arithmetic slip is mixing k in s⁻¹ with a time in minutes; the calculator converts both to SI first, so enter each with its real unit rather than pre-converting.

Hardness is a property, not a substance: it is the combined effect of every divalent cation in the water, and in practice calcium and magnesium account for essentially all of it. Because they have different atomic weights, the trade converts both to a single reference salt — calcium carbonate — so that numbers from different labs, different waters and different treatment steps can be added and subtracted. The factors come straight from equivalent weights: 50.04/20.04 = 2.497 for calcium and 50.04/12.15 = 4.118 for magnesium. A well water reporting 80 mg/L Ca and 25 mg/L Mg therefore has a total hardness of 2.497 × 80 + 4.118 × 25 = 199.8 + 103.0 = 303 mg/L as CaCO₃, or about 17.7 grains per gallon.

The single most common field error is reading a lab sheet that already says "as CaCO₃" and multiplying by 2.497 anyway — instant 2.5× overstatement, an oversized softener and a customer paying for salt they do not need. Check the units column before you touch the calculator. The second trap is the magnesium factor: because Mg²⁺ is light, a modest 25 mg/L of it contributes as much hardness as 41 mg/L of calcium, which is why waters from dolomitic aquifers feel far harder than their calcium number suggests and why magnesium-heavy water scales heat exchangers with a stubborn magnesium silicate rather than easily-acid-cleaned calcite.

A water analysis lists a dozen ions with a dozen different atomic weights, and you cannot add or subtract them directly — 20 mg/L of calcium and 20 mg/L of sodium are not the same amount of chemistry. Expressing everything "as CaCO₃" fixes that by converting each ion to the mass of calcium carbonate that carries the same number of charge equivalents. Calcium carbonate has a molar mass of 100.09 and a valence of 2, so its equivalent weight is 50.04 g/eq; divide that by the ion's own equivalent weight and you have the factor. Calcium (EW 20.04) gets 2.497, magnesium (12.15) gets 4.118, sodium (23.0) gets 2.18, bicarbonate (61.0) gets 0.82, and sulphate (48.03) gets 1.04. Forty mg/L of Ca²⁺ becomes 40 × 2.497 = 99.9 mg/L as CaCO₃.

Once every ion is on the CaCO₃ scale you can do the things that make an analysis useful: check that cations balance anions to within a few percent (the classic sanity test on any lab report), split total hardness into carbonate and non-carbonate fractions by comparing hardness to alkalinity, and size a softener or a dealkalizer directly from the numbers. The trap is the word "equivalent" — the divisor is the equivalent weight, molar mass divided by charge, not the molar mass. Using 40.08 for calcium instead of 20.04 halves every hardness figure you produce.

The grain is a survivor from the old apothecaries' system — 1/7000 of a pound, originally the weight of a single barleycorn — and North American water treatment never let it go. One grain of CaCO₃ dissolved in one US gallon is 64.79891 mg in 3.785412 L, which is 17.118 mg/L. That is the whole conversion: 10 gpg is 171 mg/L, and a residential softener rated "30,000 grains" will remove 30,000 grains of hardness before it needs salt. Softener control valves, resin capacity tables and salt-dose settings are all published in grains, while every laboratory report and every regulatory limit is in mg/L, so the technician lives in both worlds at once.

Watch out for the imperial gallon. A British or Canadian "grain per gallon" historically meant grains per imperial gallon (4.546 L), which works out to 14.25 mg/L — a 20% difference that will undersize a softener if you take a UK spec sheet at face value. And remember the informal hardness bands are quoted in grains: under 1 gpg is soft, 1–3.5 slightly hard, 3.5–7 moderately hard, 7–10.5 hard, and anything over 10.5 gpg (180 mg/L) is very hard and will scale a water heater within a few years untreated.

Alkalinity is the water's capacity to absorb acid, and it lives in three species: bicarbonate below about pH 8.3, carbonate between roughly 8.3 and 11, and free hydroxide above that. An ICP or ion-chromatography report gives them as their own ions, but every alkalinity limit, every dealkalizer sizing chart and every LSI calculation wants a single number as CaCO₃. The factors are equivalent-weight ratios: 50.04/61.02 = 0.8202 for bicarbonate, 50.04/30.00 = 1.6679 for carbonate, and 50.04/17.01 = 2.9425 for hydroxide. A groundwater with 122 mg/L HCO₃⁻ and 30 mg/L CO₃²⁻ has TA = 0.8202 × 122 + 1.6679 × 30 = 100.1 + 50.0 = 150 mg/L as CaCO₃.

In the field you rarely get the species — you get a two-stage titration. Titrating to pH 8.3 with phenolphthalein gives P alkalinity, then on to about pH 4.5 with methyl orange gives total M alkalinity, and the classic relations back out the split: if P = 0 it is all bicarbonate, if 2P < M you have bicarbonate plus carbonate, if 2P = M it is pure carbonate, and if 2P > M you have carbonate plus hydroxide. Note that carbonate's 1.6679 is roughly double bicarbonate's 0.8202 — a water whose pH has drifted up into the carbonate range gains alkalinity as CaCO₃ without gaining a single milligram of carbon, which routinely confuses operators chasing a runaway boiler.

Acid feed is how a cooling tower runs at high cycles without cementing itself shut. Alkalinity is what drives the LSI upward, so knocking it down from, say, 200 to 150 mg/L as CaCO₃ buys back most of a scaling index unit. The arithmetic is equivalent-for-equivalent: every equivalent of alkalinity needs one equivalent of acid. Because CaCO₃ has an equivalent weight of 50.04 and sulphuric acid 49.04, pure H₂SO₄ neutralizes alkalinity at almost exactly 0.98 pounds per pound — then divide by the commercial strength. Removing 50 mg/L from 1 MGD with 93% sulphuric acid takes 50 × 1 × 8.34 × 0.98 / 0.93 ≈ 440 lb/day, and the same job with 100% pure acid would take 409.

Two hard-won cautions. Acid feed and pH control are not the same knob: you are removing buffer capacity, so as alkalinity falls the water's pH becomes progressively twitchier and a pump that was fine yesterday can overshoot into the low 6s and start dissolving copper. Never run a tower below about 50 mg/L residual alkalinity for that reason. And plumb it properly — feed into a turbulent, well-mixed return line, never into the tower basin or a dead leg, use a corrosion-resistant quill that discharges into the middle of the pipe, and interlock the pump to the recirculation flow switch. A concentrated acid slug sitting in a stagnant carbon-steel header will eat a hole through schedule 40 pipe in a single afternoon.

Evaporating a litre of water to dryness and weighing the residue is the only true TDS measurement, and it takes a drying oven and several hours. A conductivity meter gives an answer in two seconds, and because dissolved salts are what carry the current the two track each other closely enough for daily work: TDS in mg/L ≈ k × conductivity in µS/cm. The factor k is not a constant of nature — it depends on which ions dominate. Chloride-rich waters run near 0.55, typical fresh surface and well waters around 0.65, and sulphate- or bicarbonate-heavy waters climb toward 0.70. A tower reading 500 µS/cm at k = 0.65 is carrying roughly 325 mg/L of dissolved solids.

Two things will bite you. First, conductivity is strongly temperature-dependent — about 2% per °C — so every meaningful reading is temperature-compensated to 25 °C; a probe without compensation reading warm blowdown will overstate TDS by 15% or more. Second, non-ionic dissolved solids are invisible to the meter. Silica, dissolved organics and sugars contribute real gravimetric TDS but carry almost no current, which is why a high-silica cooling water or a food-plant effluent can gravimetrically test far above what its conductivity predicts. Calibrate k against one oven-dried sample of your own water and it becomes a genuinely reliable daily tool.

Above about 10 µS/cm nobody in water treatment talks about resistivity; below it, nobody talks about anything else. The two are exact reciprocals of the same physical property, but the resistivity scale spreads out the ultrapure end where the interesting decisions live. Theoretically perfect water at 25 °C, ionized only by its own dissociation into H⁺ and OH⁻, has a conductivity of 0.055 µS/cm and therefore a resistivity of 18.2 MΩ·cm — the number stamped on every laboratory polisher. Drop to 1 MΩ·cm and you are at 1 µS/cm, roughly 0.5 mg/L of dissolved solids; at 100 kΩ·cm (10 µS/cm) a mixed-bed cartridge is exhausted and needs changing.

The trap is the centimetre. Instrument resistivity is quoted in ohm-centimetres and conductivity in siemens per centimetre, so 18.2 MΩ·cm is 1.82 × 10⁵ Ω·m in SI, and mixing the two scales throws a factor of 100 into your answer. Temperature matters even more here than in ordinary conductivity work: water's self-ionization roughly doubles for every 10 °C, so uncompensated ultrapure water reads dramatically "dirtier" when warm. And resistivity says nothing about dissolved gases or organics — an 18.2 MΩ·cm loop can still be carrying enough TOC or dissolved CO₂ to ruin a semiconductor rinse.

Every softener sizing calculation starts here: how much hardness, by weight, has to come out of the water between one regeneration and the next. Concentration times volume, and the only trick is keeping the units honest. In North American trade practice the arithmetic is done entirely in grains — hardness in grains per gallon times gallons treated gives grains of hardness — so 20 gpg water and 1,000 gallons between regenerations is 20,000 grains, which is 20,000/7,000 = 2.86 lb of calcium carbonate, or 1.30 kg. That figure is what you compare against the resin bed's rated capacity.

The classic field mistake is sizing on average daily use and forgetting the peak. A family of four using 300 gallons a day looks like a 6,000 grain/day load on 20 gpg water, so a 30,000 grain softener seems to give five comfortable days — until laundry day, a filled hot tub, or an irrigation zone doubles the volume and the bed exhausts mid-shower with untreated hard water going straight to the heater. The other trap is forgetting to add clear-water iron, which the resin also removes: the rule of thumb is to add about 4 gpg of equivalent load per mg/L of dissolved iron before you size anything.

Ion-exchange resin is sold and rated by volume, so once you know the hardness load per cycle the bed size follows immediately. The industry rating is grains of hardness per cubic foot at a stated salt dose: about 20,000 gr/ft³ at 6 lb of salt per cubic foot, 24,000 at 9 lb, and 32,000 at a saturating 15 lb. In mass terms 30,000 grains per cubic foot is 4.29 lb/ft³, or about 68.7 kg/m³ as CaCO₃. A load of 30,000 grains per regeneration at a 30,000 gr/ft³ rating therefore needs exactly one cubic foot — the standard residential "one cube" tank, roughly 28 litres of resin in a 9 × 48 inch vessel.

Volume is not the only constraint, and this is where undersized systems come from. The bed also has to satisfy service flow (about 7–10 gpm per cubic foot before hardness leakage climbs), backwash flow (enough to fluidize and expand the bed 50%, which sets the tank diameter), and minimum contact time. A tall thin tank meets the volume target and fails the backwash; a short fat one backwashes beautifully and channels in service. Pick the resin volume from this formula first, then check that the tank geometry that holds it can also pass the peak flow — and add 15–20% if the water carries iron or the customer is on a well with variable quality.

This is the number the customer actually feels: how often the softener wakes up at 2 a.m. and how much salt the bag loses each month. Divide the bed's rated capacity by the hardness load per unit of water and you get the run length. Working it in trade units, a 30,000 grain softener on 10 gpg water fed at 1 gallon per minute exhausts after 3,000 gallons, which is 3,000 minutes or about 50 hours of continuous flow; on typical residential intermittent use at 300 gallons a day that same bed runs 10 days. Metered control valves do exactly this arithmetic internally, which is why programming them requires an accurate hardness number and an honest household size.

The published capacity is the trap. A cubic foot of standard 8% cross-linked resin holds about 32,000 grains at a heavy 15 lb/ft³ salt dose, but only about 20,000 grains at an economical 6 lb/ft³ — and the rated number on the box is almost always the high-salt figure. Size the interval on the salt setting you actually intend to run. Two more field rules: never let a bed go longer than about 14 days between regenerations regardless of capacity, because stagnant resin fouls biologically, and set the reserve capacity so the valve regenerates before the day's demand rather than during it.

The salt setting is the single most consequential number in a softener's programming, and it is a straight multiplication: dosage per cubic foot times cubic feet of resin. Two cubic feet at 8 lb/ft³ draws 16 lb of salt per regeneration. What makes the choice interesting is that capacity and salt efficiency pull in opposite directions. At 6 lb/ft³ you get about 20,000 grains per cubic foot — roughly 3,300 grains per pound of salt, excellent efficiency. Push to 15 lb/ft³ and capacity climbs to about 32,000 grains, but efficiency collapses to around 2,100 grains per pound. You buy the last third of the capacity at nearly double the salt cost.

That trade-off drives real regulation. Several US states and California in particular set minimum salt efficiency standards, and high-efficiency programming means running low salt doses with proportionally more frequent regenerations. The counter-argument is hardness leakage: a lightly regenerated bed keeps more residual calcium on the bottom of the column and leaks 1–3% of inlet hardness into service, which matters if the softener is feeding a boiler or an RO. Field practice on a well with 25 gpg and an RO downstream is to run 10–12 lb/ft³ and accept the salt; on a municipal supply at 8 gpg feeding a house, 6 lb/ft³ is plenty. Also confirm the brine tank actually makes saturated brine — a bridged salt crust means the valve draws water, not brine, and the bed never regenerates at all.

Removal efficiency is the honest way to grade a softener, because "soft water" is a marketing word and 1 gpg out of 20 gpg is not the same thing as 0.1 out of 20. Subtract the outlet from the inlet, divide by the inlet, and you have the fraction removed; whatever is left is leakage. A bed taking 20 gpg down to 1 gpg is running 95% removal and 5% leakage — acceptable for a house, unacceptable ahead of a low-pressure steam boiler, where the specification is typically under 1 mg/L as CaCO₃, or better than 99.7% on the same water.

Leakage is not random; it comes from the bottom of the resin column. During regeneration the brine flows counter-current or co-current through the bed, and whatever calcium is left on the last few inches of resin is what the service water meets on its way out. Low salt doses leave more of it, high flow rates give less contact time to strip it, and high inlet TDS makes sodium a less effective competitor for the exchange sites. If a customer complains that their new softener "doesn't feel soft," measure the outlet hardness before touching anything: 3 gpg leaking through a system that should deliver 0.2 usually means a low salt setting, a bridged brine tank, or a bed that has channelled and needs a manual backwash.

An evaporative cooling tower throws away pure water vapour and keeps every dissolved mineral behind, so the water in the basin steadily gets saltier. Cycles of concentration is simply how many times more concentrated that circulating water has become than the makeup that feeds it — and because every litre of makeup either evaporates or leaves as blowdown, the ratio of makeup to blowdown gives the same answer as any chemical ratio. A tower taking 100 gpm of makeup while bleeding 20 gpm to drain is running at 100/20 = 5 cycles.

Cycles is the single number that decides whether a tower saves water or eats itself. Before the 1970s most towers ran at two or three cycles because nobody trusted the chemistry to hold scale off; modern phosphonate and polymer programs routinely hold six to ten, and every cycle you add cuts blowdown sharply — going from 3 to 6 cycles halves the bleed. The trap is chasing cycles past what the makeup can support: once calcium × alkalinity crosses the saturation line the tower fills with carbonate scale, and a fouled condenser costs far more in compressor power than the water ever saved.

Nobody meters makeup and blowdown on a routine service call — they dip a conductivity probe in the makeup line and again in the basin, and divide. Dissolved salts carry current, they do not evaporate, and they concentrate in lockstep, so 2400 μS/cm in the tower against 400 μS/cm in the city water means the system is running at exactly 6 cycles. The same probe is what the blowdown controller uses: set a 2400 μS/cm setpoint and the solenoid opens whenever the basin drifts above it.

Two traps catch people. First, conductivity meters are temperature-compensated to 25 °C — a hot basin and a cold makeup tap read on different bases unless compensation is on, and a 10 °C error is worth roughly 20% on the reading. Second, if you dose an acid or a conductive inhibitor, that chemical shows up in the tower reading but not in the makeup, inflating the apparent cycles; that is precisely why many programs cross-check against a non-reactive tracer such as chloride.

Chloride is the referee. Calcium precipitates as scale, alkalinity boils off as CO₂, silica drops out at high pH and conductivity gets muddied by whatever you are feeding — but chloride stays dissolved through all of it and appears in no cooling-water inhibitor. So when the conductivity ratio says nine cycles and the chloride ratio says five, the conductivity number is lying: 250 ppm Cl in the basin over 50 ppm Cl in the makeup is five cycles, full stop.

Field technique matters more than the arithmetic here. A silver nitrate (Mohr) titration on a 50 mL sample resolves chloride to a few ppm, which is plenty when the makeup carries 30–80 ppm, but if the makeup is soft well water with 5 ppm Cl the ratio becomes noise and you should switch tracers. Watch also for chloride sneaking in from elsewhere — a leaking heat exchanger on a brine loop, or hypochlorite biocide slugs, both add chloride to the tower only, and both will make a well-controlled system look badly over-cycled.

Rejecting heat by evaporation costs about one percent of the circulating water for every ten Fahrenheit degrees of range — the arithmetic behind the constant 0.001. It falls straight out of an energy balance: to drop a pound of water by 10 °F you must remove 10 BTU, and boiling off a pound of water at tower temperatures absorbs roughly 1000 BTU, so one pound in a hundred leaves as vapour. A 1000 gpm tower on a standard 10 °F range therefore evaporates about 10 gpm, or 14,400 gallons a day. The solver applies the rule per Fahrenheit degree even when you enter the range in Celsius, converting for you.

The rule is deliberately blunt and it is honest about it: real evaporation varies with ambient wet bulb, because in cold, dry weather part of the heat leaves as sensible warming of the air rather than as latent heat. Field factors of 0.00085 in winter and 0.00110 on a humid summer afternoon bracket the reality, so treat the 0.001 answer as a design-day figure. The common mistake is applying it to a closed-circuit or dry cooler — those lose no water at all, and a technician who bills a dry cooler for makeup has some explaining to do.

Every mineral entering with the makeup leaves either in the blowdown or as scale on a tube. Write the salt balance — makeup in equals blowdown out, at COC times the concentration — and with makeup being evaporation plus blowdown the algebra collapses to B = E/(COC − 1). A tower evaporating 20 gpm and asked to hold 5 cycles must bleed 20/4 = 5 gpm; ask it to hold 10 cycles and the bleed drops to 2.2 gpm.

The shape of that curve is the whole economic argument for a chemical program. Going from 2 to 4 cycles cuts blowdown by two thirds; going from 8 to 10 barely moves it. Beyond about six cycles you are paying scale-inhibitor money for very little additional water saving, which is why most well-run comfort-cooling towers settle in the 4–7 range. And note what COC = 1 means physically: makeup with no concentration at all, which would require infinite bleed — the solver refuses it rather than dividing by zero.

Water leaves an open cooling tower by exactly three doors — up the stack as vapour, out the bleed line as blowdown, and over the drift eliminators as entrained droplets — and the makeup valve has to replace all three. A 1000 gpm tower on a 10 °F range with 5 cycles typically needs about 20 gpm evaporation, 5 gpm blowdown and a fraction of a gpm of drift, so roughly 25.5 gpm of makeup, which is where the "makeup is about 2.5% of recirculation" shorthand comes from.

Utilities care about this equation for money reasons. Sewer charges are usually billed on metered water in, but only the blowdown and drift actually reach the sewer — the evaporated portion never does. In most North American jurisdictions you can install a second meter on the makeup line, or a bleed meter, and claim an evaporation credit worth a third or more of the water bill. The trap is signing up for the credit and then losing track of drift: modern eliminators cut it to 0.001–0.005% of recirculation, but a tower with damaged eliminators can throw ten times that, and those droplets carry full basin chemistry — and, in the wrong circumstances, Legionella — onto whatever is downwind.

Drift is water that never evaporates — actual basin water, at full basin chemistry, flung off the fill and carried out of the stack as droplets. It is quoted as a percentage of the circulating flow, so a 10,000 gpm tower rated at 0.005% drift loses 0.5 gpm, about 720 gallons a day, that no one ever sees. Because drift leaves at tower concentration it counts as blowdown for the cycles balance, which is why the honest form of the blowdown equation subtracts it: an uncontrolled tower with heavy drift can hold low cycles without ever opening the bleed valve.

The numbers have moved by two orders of magnitude in a lifetime. Towers built in the 1950s with wooden slat eliminators drifted 0.1–0.2%; cellular PVC eliminators introduced from the 1970s brought new towers to 0.005%, and current low-drift designs claim 0.0005%. That progress is public-health engineering, not just water conservation — the 1976 Philadelphia Legionnaires' outbreak was traced to a hotel cooling tower, and drift is the vector by which aerosolised basin water reaches people. If you find scale spotting on cars in the parking lot or white dusting on the roof deck, your eliminators are damaged, and the drift figure you are using in the water balance is fiction.

Range is what the load does, not what the tower does. It is set entirely by how much heat the plant dumps into the water and how fast the water circulates — a chiller condenser rejecting a fixed duty at a fixed gpm will show the same range in January as in July. The classic North American design point is 95 °F in, 85 °F out: a 10 °F range at 3 gpm per ton, which is exactly the 15,000 BTU/h of condenser heat a ton of refrigeration rejects.

Because range depends on flow, a falling range is a diagnostic, not a compliment. If the range on a constant-load condenser drifts from 10 °F down to 6 °F, the circulating flow has gone up — usually a fouled or bypassed control valve. If it climbs to 15 °F, flow has dropped: a plugged strainer, an air-bound pump, or a partially closed isolation valve. Watch the range and the approach together and you can separate a hydraulic problem from a heat-transfer problem without opening anything.

Wet bulb is the coldest temperature evaporation can ever reach, so it is the floor a cooling tower is pushing against — and approach is how far above that floor the tower actually delivers. Unlike range, approach genuinely measures the tower: same load, same flow, and a fouled, scaled or air-starved tower will show a wider approach on the same day. The 1950s design convention of 7 °F approach persists; 5 °F is aggressive and expensive in fill and fan power, while nothing on earth reaches 0 °F.

An example a service tech runs weekly: on a 78 °F wet-bulb afternoon a tower delivering 85 °F basin water is at a 7 °F approach and is performing to design. If that same tower delivers 92 °F on the same wet bulb, the approach has doubled and something is wrong — plugged fill, a fan belt slipping, a bird screen matted with cottonwood, or scale on the fill sheets from a program running too many cycles. The trap is comparing basin temperatures between days; without the wet bulb the number means nothing, which is why a sling psychrometer or a hygrometer belongs in the truck.

This is the water-side sensible-heat equation wearing trade clothing. Q = ṁ·c·ΔT with mass flow in pounds per hour and c = 1 BTU/(lb·°F) becomes Q = 500 × gpm × ΔT°F, because a gallon of water weighs 8.34 lb and there are 60 minutes in an hour: 8.34 × 60 = 500.4, rounded to 500 forever. A 1000 gpm tower on a 10 °F range is therefore rejecting 500 × 1000 × 10 = 5,000,000 BTU/h — about 1465 kW, or 417 tons of refrigeration.

Two habits keep this honest. First, the constant assumes plain water near ambient temperature; a 30% propylene glycol loop has a specific heat near 0.9 BTU/(lb·°F) and a different density, so the real constant is closer to 460 and using 500 overstates the duty by roughly 8%. Second, on a chiller the tower rejects the evaporator load plus the compressor work, which is why condenser water flow is sized at 3 gpm/ton against 2.4 gpm/ton on the chilled-water side. The solver applies the 500 constant per gallon per minute per Fahrenheit degree regardless of the units you type, converting first.