How to Treat Mold Cooling Water: Scale, Corrosion and Algae

An eight-cavity PP bumper bracket mold came off the press for six days after a teardown nobody had scheduled. The tool ran 38 second cycles at trial and 44 second cycles by month fourteen. Delta-T across the cavity face had widened from 2 °C to 9 °C, sink marks appeared on cavity 6 and scrap reached 4 percent. One borescope look through a plug port showed 0.3 to 0.5 mm of hard scale inside the 12 mm gun-drilled circuits. The steel was sound. The cooling water had quietly taken 16 percent of press capacity before anyone dipped a test strip in the tank.

This guide treats the water itself: a six-number chemistry spec, the corrosion families, glycol practice, a weekly and monthly test program, dosing and equipment options, and the decision tree for a loop that is already plugging.

The Snapshot

• Hardness below 100 ppm as CaCO3, pH 7.5 to 9.0. Above 150 ppm the loop needs a softener or continuous dosing, because calcium carbonate drops out of solution faster as the wall heats.
• Chlorides below 50 ppm for steel loops, below 15 ppm with aluminium in the water path. That single number decides whether an aluminium insert lasts 5 years or 18 months.
• Velocity 0.6 to 1.5 m/s per circuit. Below 0.3 m/s solids settle and stick; above 2.5 m/s erosion starts at the bends.
• 50 to 100 micron filtration at every branch, inhibitor residual verified monthly by test kit. Zero residual is the same as no inhibitor, even if the drum is full.

Why Scale Is a Heat-Transfer Crime

Cooling water does almost all of the work in an injection mold. The melt arrives at 220 to 300 °C, and roughly 90 percent of the extracted heat leaves through the channels, not through the part or the platens. Cooling occupies 50 to 70 percent of a typical cycle, so any heat-transfer loss at the channel wall shows up directly as added cycle time.

Scale is calcium carbonate, and it precipitates at the wall, not in the bulk water. Its solubility falls as temperature rises, so the hottest surface in the loop is exactly where the solid appears. Deposition climbs sharply once the wall passes 60 °C, and a hot spot behind a gate takes the first layer.

Calcium carbonate conducts at roughly 2 W/m·K against 45 W/m·K for tool steel. A 0.1 mm scale film on a channel wall adds about 0.00005 m²·K/W of resistance, which cuts local heat flux by 10 to 20 percent against a clean wall. Cycle time then pays for that loss in full.

Scale also shrinks the bore, and that part is self-reinforcing. A 0.5 mm layer inside a 10 mm channel leaves a 9 mm bore: the flow area drops 19 percent, so pressure drop rises 20 to 30 percent and the circuit receives less flow at the same pump setting. Slower flow deposits more scale. That is the death spiral that takes one circuit from clean to half-plugged in eight months while the mold looks healthy from outside.

Corrosion products add the second fouling layer. Iron and copper oxides, typically 0.05 to 0.2 mm, travel from steel surfaces and plate out wherever velocity drops. Biofilm adds a slimy third layer that insulates like scale and corrodes like a battery at the same time.

← swipe to scroll →

Table note: conductivity figures and loss bands are typical published values for cooling-water deposits. Confirm against your own loop with a teardown inspection before sizing a cleaning interval.

The Water Spec: Six Numbers That Define Good Cooling Water

"Clean enough" is not a spec. Write these numbers on the tank lid and test against them, with an action limit for each one.

Total hardness: under 100 ppm as CaCO3, action limit 150 ppm. Below 100 ppm you can run a closed loop with inhibitor only. Above 150 ppm you are dosing or softening, and the choice is equipment versus chemistry, not luck.

pH: 7.5 to 9.0. Below 7.5 the water attacks steel and copper as acid corrosion. Above 9.0 aluminium plates start to dissolve, and calcium precipitates faster. If the loop carries aluminium alloy inserts or baffle plates, hold 7.5 to 8.5 and treat the top of the band as a limit, not a target.

Chlorides: under 50 ppm for steel circuits, under 15 ppm whenever aluminium alloy sits in the water path. Chloride drives pitting, and pits are where MIC sets up later. Well water and marine air routinely push a raw-water loop past both numbers.

Conductivity: under 800 µS/cm for a treated closed loop, alarm at 1,200 µS/cm or at 1.5 times the commissioning baseline. Conductivity is the cheapest evaporation and leak detector that exists, which is why it gets tested weekly.

Inhibitor residual: held inside the supplier label band, verified by test kit monthly. A full drum with zero residual protects nothing.

Filtration: 50 to 100 micron at the branch to each mold, plus a 25 micron polishing element on the main return. A plugged 10 mm bore starts as one 200 micron chip wedged at a drill cross.

Two numbers belong on the log sheet even though they are not water chemistry: iron in the return water under 1 ppm, because rising iron means active corrosion somewhere in the loop; and mold surface temperature kept at least 3 °C above the measured shop dew point, which is the subject of its own section below.

Where the feed water will not meet the spec, fix the supply before the additive. A carbon filter at 5 micron ahead of the softener removes the silt and chlorine that destroy resin beads. A 200 litre buffer tank between softener and loop damps the hardness spike between regenerations, which is when most dosing programs quietly fail. Retain the commissioning numbers in the tool file, because a supplier dispute two years later is argued with that baseline, not with memory.

Flow Velocity Is Part of the Treatment

Chemistry buys you clean water. Velocity keeps that water doing its job, and it decides whether the solids stay in suspension or settle into a deposit. The working rule: hold every circuit at 0.6 to 1.5 m/s, turbulent, all the time the mold runs.

Below 0.3 m/s, particles drop out of the stream and plate onto the bottom of horizontal runs. Dead legs, capped tees and drill-cross cavities become sediment traps at any velocity, which is why plumbing design and water treatment are the same argument.

In a 10 mm bore, 0.6 m/s equals about 28 L/min and reaches a Reynolds number near 6,000. The stagnant boundary layer where deposits anchor simply does not exist in that flow. Above 2.5 m/s the water starts cutting: bends, baffle edges and bubbler openings erode, and pressure drop rises with the square of velocity for no heat-transfer gain.

Balance the parallel circuits and measure each one. A manifold feeding eight circuits without branch throttles pushes roughly double the flow through the shortest loop and starves the longest one. Log per-circuit flow at commissioning, then read the same flowmeter positions monthly: that series is the real fouling trend chart, long before the part shows a defect.

Glycol breaks the velocity math, so treat it as a different fluid rather than a colour change. A 25 percent propylene glycol mix is about three times more viscous than water at 20 °C, so the same L/min gives a lower Reynolds number and a weaker film coefficient. If a circuit already runs laminar, glycol makes it worse.

Corrosion Family One: Galvanic Attack

A mold cooling loop is a battery accident waiting for an electrolyte. Steel plates, brass fittings, copper tube heat exchangers and aluminium baffle plates sit millimetres apart in the same water. Once the water conducts, the more active metal corrodes to protect the nobler one, and nothing about the mold looks wrong while it happens.

Aluminium is the most active common metal in the loop. Pair it with brass or steel and aluminium gives up its ions at the joint. The visible signature is grey-white powder at fitting interfaces and pits under brass seat washers, typically 0.1 to 0.5 mm deep within a year on raw water. Copper released from a heat exchanger plates out as a thin film on aluminium surfaces even metres downstream, and every copper speck starts a fresh local cell.

Three defences, in order of preference. Keep one metal family per loop: all stainless or all steel wetted parts, with the heat exchanger separated by a plate-and-frame unit. Where dissimilar metals cannot be avoided, isolate them with dielectric unions and dielectric bushings so the electrical path runs through plastic, not through the fitting. Then raise the water resistance: conductivity under 800 µS/cm slows the cell, and inhibitor film covers both metals.

Galvanic attack also explains a rule that surprises shops: sacrificial anodes belong in open atmospheric tanks, not in closed mold loops. A magnesium or zinc rod inside a sealed loop will consume itself in months and can push local pH above 9.0, which starts attacking the aluminium it was meant to protect. If a brass-cored anode cartridge came pre-installed in a controller tank, pull it and log what the chemistry actually needs.

Corrosion Family Two: Pitting and MIC

Pitting is the quiet second family. General corrosion thins a wall evenly and shows on a gauge; a pit eats 0.5 mm through a 3 mm wall and announces itself as a drip onto the part or a pressure loss between shifts. Chloride is the main driver, which is why the spec caps it at 50 ppm for steel and 15 ppm with aluminium present. Stagnant water with low flow and high chloride pits fastest at the waterline of an open tank and in dead legs.

Microbiologically influenced corrosion finishes the job that pitting starts. Bacteria attach to the channel wall, build a biofilm, and set up an oxygen differential cell under the deposit. Sulfate-reducing bacteria in low-oxygen pockets produce hydrogen sulfide, which blackens steel and eats pits at rates several times higher than sterile water. MIC pits run 0.2 to 1 mm deep within a year in a warm, stagnant, slightly organic loop.

Algae is the surface symptom of the same problem. Algae needs light, so green growth inside a mold means light is reaching water: a translucent tank, a clear sight glass in sunlight, or an open fill port. Algae itself does not attack steel, but every strand that dies feeds the bacteria underneath it and sheds debris that plugs a 50 micron screen in a week.

Control the biology with four cheap moves. Run a closed, opaque loop: metal or black poly tanks, lidded, with the fill port screened. Keep disinfectant residual per the supplier program, typically a non-oxidiser dosed continuously or a shock of oxidiser at 25 to 50 ppm for a few hours with the loop bypassed around any sensitive internals. Keep velocity above 0.6 m/s so biofilm cannot anchor. And drain and dry any loop idle longer than two weeks, because a dead warm loop grows more bacteria in fourteen days than a running loop grows in a year.

Watch elastomers while you dose. Some biocide and inhibitor chemistries swell EPDM and nitrile O-rings, baffle seals and quick-disconnect gaskets. Check every seal compound against the supplier compatibility chart before the first shock dose, and re-check the baffle and bubbler seals during the quarterly teardown. A 0.3 mm swollen seal is a flow restriction and a future plug.

The Dew-Point Rust Loop

Half of the rust found on mold plates never touched liquid water inside a channel. It grew from air. When a water or oil circuit holds a plate below the shop dew point, moisture condenses straight onto the steel face, and a polished cavity can flash-rust overnight between shifts.

The physics is simple to manage and easy to ignore. Condensation starts when the surface sits at or below the dew point of the surrounding air, and a shop at 28 °C with 70 percent relative humidity has a dew point near 22 °C. A chiller loop set at 15 °C runs its whole plate below that line. Measure the dew point with a pocket meter every shift in humid months; it takes ten seconds and it decides your controller setpoints.

Two rules follow. Hold coolant at least 3 °C above measured dew point whenever the process allows, and where a cold setpoint is genuinely required, protect the steel instead of the air: rust preventive on every ferrous surface between runs, and plates stored below 50 percent relative humidity. This is the same protection logic as the full rust and corrosion prevention guide, applied to the water program rather than to storage.

The loop closes back on water chemistry because condensation is not clean water. Shop air carries carbon dioxide, ammonia and chlorides, so the condensate film is slightly acidic and salt-laden. It concentrates as it dries at the edge of the wet patch, and rust rings form exactly where the film thins. If a plate shows rust halos around cooling-line traces after a humid week, the diagnosis is dew point, not channel corrosion, and the fix lives on the schedule below rather than in the acid tank.

Glycol and Freeze Protection Practice

Glycol earns its place in winter, in rooftop loops, and in any mold parked where heat is off. Use inhibited propylene glycol, never automotive engine coolant, and hold the mix at 20 to 30 percent by volume for freeze protection down to roughly minus 10 to minus 20 °C. Above 30 percent the heat-transfer penalty outruns the freeze benefit, since pure glycol conducts about 10 times worse than water.

Inhibitor is the part shops forget. Plain glycol degrades to organic acids over 12 to 24 months at loop temperatures, and those acids drop pH and attack copper and solder. Buy inhibitor-package glycol, then re-test the reserve alkalinity and pH once a year. A glycol loop that has been running three winters without a test is an acid loop with a pretty colour.

Never mix inhibitor chemistries. Silicate, OAT and phosphate packages gel when combined, and the gel plugs exactly the 10 mm circuits you are trying to protect. When changing brands, flush with clean water first and verify with a visual check of the return line. And label the loop concentration permanently: mixing a 25 percent stock straight into a full-strength top-up silently makes an unprotected blend.

Glycol also changes the treatment budget. Propylene glycol feeds bacteria worse than water does, so the biocide program matters more, not less. And viscosity at 10 °C can be double the 20 °C figure, so verify pump head against the winter curve before a cold snap, not after. Keep the same 50 to 100 micron filtration; glycol carries sludge particles that water would have flushed out.

The Test Program You Can Actually Run

Before any dosing, baseline the loop. Fill with the water you intend to run, record hardness, pH, chlorides, conductivity and iron from two sample points, and log per-circuit flow at the controllers. Those numbers are the reference for every trend later. A shop that starts dosing on month five with no day-one record cannot tell recovered flow from luck.

The cadence below runs on two 15-minute dips per week plus a monthly half hour, and it catches almost every failure before the part shows it.

• Weekly: conductivity and pH at two sample points, supply side and return side. Rising conductivity with rising delta-T means evaporation concentration or a leak; a pH drift of 0.5 in one week means chemistry ingress.
• Monthly: total hardness and inhibitor residual by test kit, plus visual check of filter differential pressure across the 50 to 100 micron branch elements.
• Quarterly: borescope inspection through a removable teardown plug in the worst circuit, one photo per port logged against the previous quarter, plus per-circuit flow against the commissioning sheet.
• Annually: full passivation and rust preventive on any steel exposed by an open cleaning, glycol reserve-alkalinity re-test if glycol is in the loop, and a blind sample sent to a lab for iron, copper and bacteria counts as a check on your own kit.

← swipe to scroll →

Table note: limits are typical industry bands for closed mold cooling loops. Confirm against the inhibitor supplier datasheet and your commissioning baseline, and keep the log sheet in the tool file.

Dosing and Equipment Options

Four setups cover almost every shop, chosen by makeup-water hardness and by what the loop protects. For a 10 ton-per-hour loop fed by 200 to 300 ppm hard municipal water, a salt-softened makeup line plus a continuous inhibitor blend is the standard economic answer: softener capex is moderate, salt is a consumable measured in bags per month, and one inhibitor drum covers the whole system.

Without a softener, a phosphate or polymer antiscalant dosed at supplier rate still works, but it converts the program from removal to stabilisation. Hardness stays in the water, held in solution by 2 to 10 ppm of polymer. If a controller runs hot and the wall passes 80 °C, stabilisers lose the argument and the scale drops anyway.

Deionised or reverse-osmosis water belongs where optics and medical parts demand it, typically below 10 µS/cm. Treat DI as a specialist tool, not a default: near-zero conductivity water is aggressive and leaches ions from whatever it touches, so a DI loop still needs a noble-metal-safe inhibitor, or it trades scale for uniform corrosion. Full DI plumbing with non-ferrous fittings is a real project, priced as hardware plus annual resin or membrane replacement, not as a jug.

Bleed-and-feed or a plate heat exchanger handles the evaporation-concentration problem on open cooling towers, which is the loop type that sneaks chloride into the spec every hot week. Whatever the box, the controller itself matters: a unit with ±1 °C control, a sealed tank and a pump sized for real circuit pressure drop is covered in the mold temperature controller guide, and the same site covers the internals at mold heating and cooling components.

Decision Tree: When Channels Are Already Plugging

A mold that keeps losing temperature control is a triage problem, and 60 percent plugged is the line where opinions split. Work the tree from the cheapest test up, and do not start with acid.

Step one, separate the loop from the tool. Measure flow and delta-T at each circuit with the mold off the press and at the press. A single bad circuit with good tank chemistry points inside the mold; several bad circuits point at the controller, filtration or the water program.

Step two, quantify the deposit. Borescope through a plug port. Soft sludge that moves with a flush is corrosion product and biofilm. Hard white scale that rings under a pick is calcium. The deposit tells you the cleaning chemistry before you buy it.

Step three, clean in place if under 60 percent of channels show restriction and the bore still passes a sized pig or brush. An acid circulate on a tool with open plugs, neutralised and flushed to pH within 0.5 of the feed water, recovers most of the flow; the full procedure lives in the cooling channel cleaning guide, and re-plumbing options for stubborn circuits are in the guide to improving cooling on an existing mold.

Step four, at 60 percent plugged or worse, in-place cleaning is a downgrade. The channels have narrowed enough that acid contact time and flow distribution cannot be guaranteed around bends, and partial cleaning leaves ridge-scale that re-plugs in weeks. Pull the plates, blow out, and re-drill or re-bore with the drill-pattern and plug rules from the circuit-drilling guide, then baseline the water program on the same day. Route the teardown through mold repair and maintenance and record the plug percentage by circuit as the reason code.

Whatever branch you take, the water decision follows the metal, not the calendar. If the borescope shows aluminium with pits over 0.3 mm, chlorides are already above the 15 ppm limit, and cleaning without fixing the chemistry rebuilds the problem inside a year.

Symptom-to-Cause Mapping

← swipe to scroll →

Table note: cause bands assume a closed loop on treated water. Confirm each row against your own log sheet before opening an acid cleaning ticket.

Mold Cooling Water Treatment FAQ

Q1. Can I just run distilled or DI water and skip the chemicals?

Only with a corrosion inhibitor. DI water under 10 µS/cm has no buffering and no passivating film, so it leaches ions from steel, copper and aluminium faster than hard water does on the scale side. Pure water is a good antiscalant and a poor metal protector. Below 15 ppm chlorides it slows pitting, but the noble-metal-safe inhibitor is still what keeps a mixed-metal loop alive.

Q2. My tap water is 250 ppm hard. Is softening worth it or do I just clean more often?

At 250 ppm expect 0.1 mm of scale per year on hot walls with no treatment. A salt softener drops makeup under 5 ppm and removes the deposition driver at the source; dosing a polymer stabiliser is cheaper capex but fails above 80 °C wall temperature. Cleaning every 12 to 18 months is the no-treatment path, and each acid clean carries plug-handling labour and elastomer inspection.

Q3. How do I know the inhibitor is actually working?

Three independent checks: residual by test kit monthly, iron in the return water under 1 ppm quarterly, and the borescope photo at the teardown plug each quarter. A loop can carry label-rate dosing and still show zero residual if the tank is being dumped and refilled weekly; iron trend catches what the residual test misses.

Q4. Algae keeps growing in our controller tanks. Biocide or black-out the shop?

Both, mostly the cheap one. Algae needs light and nutrients; opaque tanks and closed loops remove the light and cut nutrient ingress by 90 percent. Keep a disinfectant residual for the bacteria that follow, and shock-dose at 25 to 50 ppm oxidiser with sensitive internals bypassed when a film already shows. Translucent tanks in sunlight will lose the chemical argument every summer.

Q5. When does a plugging problem become a re-drill problem?

At roughly 60 percent of circuits restricted, or any single circuit under 30 percent of its commissioning flow after a proper acid clean. Past that line in-place cleaning cannot guarantee contact and flow around every bend, and ridge-scale left behind re-plugs within weeks. Pull the plates, restore or re-drill the bores to the 12 mm and turbulence rules, and fix the water program the same month or the new channels will age exactly as fast.

The Bottom Line

Cooling water is a process parameter, not a utility. Six numbers define it: hardness under 100 ppm as CaCO3, pH 7.5 to 9.0, chlorides under 50 ppm with steel and under 15 ppm with aluminium, conductivity under 800 µS/cm with an alarm at 1,200, inhibitor residual inside the label band, and 50 to 100 micron filtration at every branch. Velocity carries the rest of the load at 0.6 to 1.5 m/s, because deposits anchor below 0.3 m/s and erosion starts above 2.5 m/s.

The failure arithmetic is unforgiving. A 0.1 mm scale film costs 10 to 20 percent of local heat transfer and the cycle pays it; galvanic pairs and MIC open pits that chloride started; dew point rusts plate faces that the channels never touched. The program that prevents all three fits on one sheet: weekly conductivity and pH at two points, monthly hardness and residual, quarterly borescope and flow against the commissioning baseline, annual passivation after any open clean.

DieStrike builds and restores precision molds and dies to IATF 16949 with ±0.005 mm mold tolerance, ±0.002 mm on EDM and jig grinding, HRC 62 maximum heat treatment, and 120+ machines behind standard components in 3 to 7 days and tool lead times of 2 to 5 weeks. TE, Amphenol, Luxshare and Dongshan programs run with cooling circuits drilled, plugged and commissioned against a documented water spec, because a mold that is cooled properly from shot one is a mold that never needs the acid tank.