Hard water in Los Angeles plants

We are based in Bell, and almost every plant we walk into within ten miles of here has the same problem in a different costume. A heat exchanger that will not hold its outlet temperature. A cooling tower that has to be chased with chemistry. A pump that eats mechanical seals every eight months and nobody can say why. Four different symptoms, one cause, and it comes out of the tap.

What the water here actually is

LADWP publishes hardness by area, and the spread across the city is not subtle:

• San Fernando Valley and western LA — average 8.6 grains per gallon (147 mg/L as CaCO₃), ranging 5.4 to 16.1 gpg
• Silver Lake, downtown and south LA — average 9.8 gpg (168 mg/L), ranging 5.4 to 17.0 gpg
• Eastern LA and the Harbor — average 16.3 gpg (279 mg/L), ranging only 16.1 to 16.4 gpg

The USGS calls anything above 180 mg/L "very hard". Eastern LA and the harbour sit at 279 — better than half again past that line. And that band covers Vernon, Commerce, Santa Fe Springs, Wilmington, Carson and Long Beach, which is to say most of the industrial floor space in this county.

Look at the third figure again, though, because the interesting part is the range, not the average. Out in the Valley hardness swings between 5.4 and 16.1 grains depending on how the aqueduct, State Water Project, Colorado River and local groundwater are being blended that month. A plant out there gets soft water sometimes. In the eastern basin the range is 16.1 to 16.4. There is no soft month. Whatever your equipment does with hard water, it does all year, every year, and it has been doing it since the day the plant opened.

One caveat before you use those numbers: they are LADWP's. If you are in Vernon, Long Beach, Santa Fe Springs or anywhere else with its own utility or its own wells, pull your Consumer Confidence Report. Same basin, different blend.

Why a film you cannot see costs real money

Scale is not a housekeeping problem. It is an insulator, chemically bonded to the exact surface whose entire job is transferring heat. The US Department of Energy puts it plainly: scale has a thermal conductivity roughly an order of magnitude below bare steel. You are wrapping your heat transfer surface in a blanket and then paying fuel to push heat through it.

The DOE's own numbers for fuel loss on a boiler:

• 1/64 inch of normal scale — 1.0% more fuel. With iron present, 1.6%. With iron and silica, 3.5%.
• 1/32 inch — 2.0% normal, 3.1% with iron, 7.0% with iron and silica.
• 1/16 inch — 3.9% normal, 6.2% with iron.

Now hold that against reality. One thirty-second of an inch is 0.8 millimetres — about two business cards stacked. Nobody has ever opened a manway, seen that, and called it a problem. It reads as a light film, the kind you wipe with a rag and forget. Run the fuel arithmetic on a boiler that fires most of the year and that film is a line item.

The same physics applies everywhere heat crosses a wall: exchanger tubes, tower fill, jacket water, chiller barrels. The boiler just happens to be the one somebody published a table for.

Five ways to identify it, cheapest first

1. Read your Consumer Confidence Report

Free, published annually, takes ten minutes. It gives you hardness, total dissolved solids, pH and alkalinity — four of the five numbers you need for anything that follows. The limitation is that it is a distribution-system average across a year, not what came out of your makeup line on Tuesday. Treat it as the starting position, not the measurement.

2. Titrate at the point of use

A hardness test kit costs about what one hour of our time does, and it answers a question the CCR cannot: what is happening in your loop. Test two samples — the makeup water and the recirculating water. On a cooling tower, dividing the recirculating conductivity by the makeup conductivity gives you your cycles of concentration, which is the single most useful number on the whole system and the one most plants cannot tell you.

3. Trend approach temperature — the real answer

This is what separates a maintenance department that catches fouling from one that discovers it. Approach temperature is the gap between the outlet of one stream and the inlet of the other across a heat exchanger. On a clean unit it is a fixed number for a given load. As scale accumulates, that gap widens — long before anything trips, leaks or fails.

Record the approach when the exchanger is known clean. That is your baseline. When it drifts up fifteen or twenty per cent at comparable load, you are fouling, and you now know it months ahead of the day the process starts missing temperature.

Trend pressure drop across the unit alongside it. The two catch different things: approach temperature catches a thin insulating film, pressure drop catches a restriction as the tube bore closes in. A system that is scaling shows both, in that order.

4. Calculate the Langelier Saturation Index

Everything above tells you what already happened. LSI tells you what the water wants to do. It compares actual pH against the pH at which calcium carbonate is saturated: positive means the water will deposit scale, negative means it will be corrosive to metal, near zero means balanced. You need pH, temperature, calcium hardness, total alkalinity and TDS.

The term that catches people is temperature. Water sitting comfortably balanced at 70 °F can be firmly in scaling territory at 140 °F. That is precisely why scale never forms evenly — it forms first and worst on the hottest surface in the system, which is also the surface you least want insulated.

It also explains the trap of "fixing" scale by dropping pH until LSI goes negative. Congratulations: the water no longer scales. It now eats your tubes instead.

5. Look — but look in the right six places

Scale deposits where it is hottest and where flow is slowest. In roughly that order: boiler tubes on the fire side, the hot end of exchanger tubes, cooling tower fill, the seal flush orifice, small solenoid and instrument orifices, spray nozzles, and any dead leg in the piping.

When you find deposit, identify it before you buy chemistry. Calcium carbonate fizzes in dilute acid — household vinegar is enough for a field test. Silica scale does not: it is glassy, far harder, and needs entirely different treatment. Cleaning silica with a carbonate cleaner wastes a shutdown, and shutdowns are the expensive part.

How to fix it

Three separate jobs, and they get confused constantly: stop making new scale, remove what is already there, and change the design so it stops mattering. Most plants do the middle one repeatedly and wonder why it keeps coming back.

Stop making more

Ion exchange softening trades calcium and magnesium for sodium. It is the standard answer for boiler makeup and it is cheap to run. Two things to know: it does not reduce total dissolved solids, it just changes which ones, and it adds sodium — which matters if something downstream cares about that.

Reverse osmosis removes hardness and TDS together. Higher capital, higher operating cost, and it produces a reject stream you have to deal with. The right answer for high-pressure boilers and for process water where quality is part of the product.

Antiscalant chemistry — phosphonates and polymers — holds calcium in solution past the point where it would normally drop out. This is standard practice on cooling towers and it works, but only with dosing control tied to actual system conditions. A drum in the corner with a hand pump is not a treatment programme.

Cycles of concentration is the cheapest lever on any tower and the most commonly mismanaged. Every cycle concentrates the hardness that came in with the makeup. Running high cycles saves water, which is a live concern in this state — but at 279 mg/L makeup you cannot run the cycles a plant in a soft-water region runs. A conductivity controller on the blowdown valve costs less than one exchanger cleaning and settles the argument permanently.

Remove what is already there

Chemical cleaning with inhibited acid — hydrochloric, sulfamic or citric depending on the metallurgy and the deposit. Get the deposit analysed first. The wrong acid on the wrong metal does permanent damage: hydrochloric on stainless risks chloride stress corrosion cracking, which turns a cleaning into a replacement. Spent acid needs neutralising and disposing of properly, and in California that is a regulated waste stream, not a floor drain.

Mechanical cleaning — brushing or rodding exchanger tubes, high pressure water on tower fill. Slower and more labour, but no chemistry risk and sometimes the only safe option on thin-wall tubing.

Design it out

Velocity. Scale forms where water moves slowly. Raising tube-side velocity reduces deposition, and it is often achievable by re-passing an exchanger rather than replacing it. Our pipe velocity calculator will tell you where you currently sit.

Skin temperature. Since LSI worsens with temperature, lowering the hot-surface temperature directly reduces scaling. More surface area at a lower temperature difference scales far less than less surface working harder.

Dead legs. Stagnant water at temperature is where scale starts. Every abandoned branch and capped tee is a scale factory feeding the rest of the loop.

Cleanable orifice unions on seal flush lines instead of a drilled plug you have to break the piping to reach. If the orifice can be inspected in ten minutes it will actually get inspected.

What we build into a PM route here

On preventative maintenance customers in this part of the county, the water items are not optional extras — they are the ones that pay for the programme. Baseline approach temperatures on every exchanger while it is clean. Titrate makeup and recirculating water quarterly. Log tower conductivity and cycles. Put the seal flush orifice on the route as an item with a number attached, not a visual check. And analyse any deposit before anyone quotes a cleaning.

None of that is exotic. It is the difference between replacing an exchanger and cleaning one, and in a basin running 279 mg/L it is the highest-return maintenance you can do. If you want the arithmetic for your own plant, the downtime cost calculator takes about two minutes.