Hardness, Alkalinity and pH, and What Each One Does

Why this matters

Three numbers on a water report get collapsed into one verdict all the time. The customer says the water is hard, the tech says the water is hard, and the correction that gets installed treats hardness when the thing actually driving the failure was alkalinity or pH. They are three independent measurements answering three different questions, two of them are reported in the same units, and moving any one of them moves at least one of the others.

The load-bearing point is this: alkalinity, not hardness, decides whether pH will hold still. Which means a water can be soft and still be stable, hard and still swing, and a correction aimed at the wrong one of the three will either do nothing or overshoot.

Before you dose anything to move any of them

Acids and caustics used to shift pH and alkalinity are both corrosive to skin and eyes, and they are different hazards with different gloves. Get the glove class and the eye and face protection from Section 8 of the specific product's safety data sheet rather than reaching for whatever is in the truck; your employer has to make that sheet available (29 CFR 1910.1200). Where the sheet calls for a face shield over goggles, wear both.

Add acid to water, never water to acid. The reverse concentrates the heat of dilution at the surface, boils, and spatters concentrated acid back at your face.

Never introduce an acid into a vessel, sump or drain that holds or recently held a hypochlorite product. That reaction releases chlorine gas in a confined mechanical space within seconds, and the control for an inhalation hazard is ventilation and respiratory protection under a written program (29 CFR 1910.134), not gloves.

A chemical feeder or drum connected to a line that also serves potable water is a cross-connection, and the backflow protection the local plumbing code requires for that chemical class goes in before the line is opened.

Three different questions

pH is an intensity. It says where the acid-base balance sits at the moment of measurement, on a logarithmic scale, so one pH unit is a tenfold change in hydrogen ion activity. It tells you nothing about how much acid or base is present, only how strongly the balance is currently tilted. Because it is an intensity and not a quantity, it is also the number that moves fastest and travels worst: it starts drifting as soon as the sample loses or gains carbon dioxide, which is why it belongs in the field, at the tap.

Alkalinity is a capacity. It measures how much acid the water can absorb before its pH moves, and in ordinary natural water it is mostly bicarbonate. It is reported in mg/L as CaCO3, and it is the buffer. High alkalinity means the pH is anchored. Low alkalinity means the pH is loose and small disturbances move it a long way.

Hardness is a content. It measures the dissolved calcium and magnesium, the two ions whose salts come out of solution on hot surfaces. It is also reported in mg/L as CaCO3, which is the single largest source of confusion between it and alkalinity. They are two different measurements that share a reporting convention.

Why the two "as CaCO3" numbers are not the same number

Both are expressed in calcium-carbonate equivalents so they can be compared on one basis, and the comparison itself is informative.

When alkalinity is at or above hardness, effectively all the hardness has a carbonate partner available, and that is the fraction that precipitates as scale when the water is heated. When hardness exceeds alkalinity, the difference is non-carbonate hardness - calcium and magnesium paired with sulfate or chloride instead. Those salts have their own, quite different solubility behavior and they do not simply drop out on heating the way carbonate does. A water at hardness 224 and alkalinity 180 carries about 44 mg/L as CaCO3 of non-carbonate hardness, and a treatment plan that assumes every hardness ion will precipitate as carbonate is planning against the wrong 20 percent of the load.

The intensity-capacity distinction, both ends

Alkalinity is reported in mg/L as CaCO3, and because the equivalent weight of calcium carbonate is about 50 (its formula weight of roughly 100 divided by 2), 50 mg/L as CaCO3 is one milliequivalent per liter of buffering capacity to the standard methyl-orange endpoint near pH 4.5. That endpoint matters: alkalinity is the capacity down to that point, not a prediction of pH anywhere along the way, and the titration curve between here and there is not a straight line.

Now state the direction at both ends, because half of it gets quoted alone.

• At low alkalinity, a small acid load moves pH a long way. That is the well-known half.
• At high alkalinity, the same water resists your correction as stubbornly as it resists the disturbance. A pH adjustment of one unit on a strongly buffered water takes many times the chemical it would take on a weakly buffered one, and a technician who sizes the dose from experience on soft water will underdose it and conclude the chemical is no good.

Buffering is not a good property or a bad one. It is inertia, and inertia works against whoever is pushing.

One acid load, two waters

Two buildings on different supplies, both reading pH 7.6 at the tap.

• Water A: alkalinity 25 mg/L as CaCO3, hardness 40.
• Water B: alkalinity 240 mg/L as CaCO3, hardness 320.

Convert the capacity: 25 / 50 = 0.5 milliequivalents per liter for A; 240 / 50 = 4.8 milliequivalents per liter for B. B carries 9.6 times A's capacity (4.8 / 0.5).

Now apply the same disturbance to both. Say each system picks up an acid load of 0.4 milliequivalents per liter - the number itself is illustrative, and the point is that it is identical for both. It could be carbon dioxide absorbed from a combustion process, an overshot acid feed, or nitrification in a stagnant leg.

• Water A has consumed 0.4 of its 0.5, which is 80 percent of its buffer (0.4 / 0.5). It has 0.1 milliequivalents per liter left. pH does not drift here, it falls off a shelf, and it keeps falling on the next small disturbance because there is nothing left to absorb it.
• Water B has consumed 0.4 of its 4.8, which is about 8 percent (0.4 / 4.8 = 0.083). It has 4.4 left. Measured pH barely registers the event.

Same starting pH, same insult, opposite outcomes. If you had only the pH reading from before and after in Water B you would conclude nothing happened, and you would be right about the pH and wrong about the process, because the acid load is still arriving and the buffer is being spent 8 percent at a time. That is what makes alkalinity the number to trend on a system with a recurring pH complaint: pH tells you the state, alkalinity tells you how much warning you have left.

What flips this reading: if Water A's low alkalinity is the result of upstream treatment rather than the source, the disturbance may be the treatment itself, and the correction is upstream, not in the loop. And if the acid load in Water B is biological rather than chemical, buffer consumption will accelerate as the population grows rather than staying linear, so a trend line drawn from two points understates it.

The failure mode: the shop dosing Water A chases pH with alkali on every visit, gets a good reading on the way out, and comes back to the same complaint, because the correction raised the intensity without adding capacity. Raising alkalinity is what stops the recurrence, and alkalinity is what nobody measured.

What each one actually does to equipment

Keep the three effects separate; a sibling article works through the material-by-material verdicts in detail and this is the short form.

• Hardness sets the deposit potential at hot surfaces. It is a quantity of scale-forming material, not a rate.
• Alkalinity sets stability, and in steam service high alkalinity contributes to carryover and to carbon dioxide in the condensate, which turns into carbonic acid where the steam condenses. Note the direction: the alkalinity problem in a boiler shows up in the condensate return, downstream of the boiler, not in the boiler.
• pH sets the intensity of attack and, just as importantly, the window in which a treatment chemical works at all. Most corrosion inhibitors have a stated pH range, and outside it the product is inert or worse.

What each correction moves

Two rows deserve reading twice. A calcite neutralizer raises pH by dissolving calcium carbonate, so it necessarily raises hardness, and a customer sent to it for acidic water can end up with a scaling complaint they did not have. Reverse osmosis lowers all three, which produces water that is clean and, because it has almost no alkalinity left, has almost no resistance to the next thing that touches it.

How to verify

Measure pH and temperature at the sample point, in the field, with a meter calibrated that day against two buffers that bracket your expected reading. A pH from a bottle that traveled is a number, not a measurement.

Take alkalinity and hardness on the same sample, at the same point, on the same date, so the comparison between them is real. Then run the arithmetic check: hardness as CaCO3 should equal about 2.5 times the calcium in mg/L plus about 4.1 times the magnesium in mg/L. If the report's hardness and its individual cations do not reconcile, one of the numbers is wrong and neither should be acted on.

Finally, after any correction, re-test at the same point and confirm you moved the parameter you intended and check the two you did not intend to move. That table above says a correction almost never lands on one number alone.

References

• 29 CFR 1910.1200 (Hazard Communication) and Section 8 of each product's safety data sheet for the glove class, eye and face protection for acids and caustics
• 29 CFR 1910.134 for respiratory protection where a chemical reaction can generate chlorine or chloramine gas in an enclosed space
• Standard Methods for the Examination of Water and Wastewater (APHA, AWWA, WEF) for the alkalinity titration endpoints and the hardness calculation from calcium and magnesium
• See related: The Water Chemistry That Attacks a System; What Is Actually in the Water You Put Into a System; What a Scaling Index Tells You and What It Does Not