Procedure: Mold cooling water (b-cooling-water). Revision 10 — October 18, 2026. Plant sizing, the looped header, and towers and outdoor units added from Mike's notes; hydraulic and feed throat allowances from published chiller-maker guidance. Index: https://ancreationoffice.com/sop.txt Source: Mike's practice from years of building and running injection molds (Mike's notes, kept private). Lines marked "Mike" are his practice; lines marked "physics" are standard heat transfer; anything else the office adds is marked "office". Tier: deep. Read this when a mold is being designed or quoted, a mold runs hot, cycles long, warps, sticks, binds or galls, or a plant is sizing chillers, towers or pumps. Calculation: https://github.com/mminwi/an-creation-office/blob/main/specialists/calcs/cooling_water.py (input format: example-cooling-input.json). WHY IT MATTERS Cooling water is usually the biggest factor in molded part cost, and often in part quality (Mike). Most of the cycle is cooling time, and uneven cooling shows up as warp, sink, sticking and dimensional spread. The goal is not "more water"; it is balanced cooling: every area of the mold giving up its heat at the same rate. WHAT TO ASK FOR (a cooling review needs these; unknowns become questions) - The part: resin and grade, wall thickness, shot weight, the corners and deep areas. - The mold: steel type, any inserts (beryllium copper such as MoldMax, aluminum), every water circuit with its line size, how circuits are hooked up (single or looped), bubblers and baffles, slides, lifters, cores. - The process: cycle time and how much of it is cooling, mold temperature setting. - The measurements: flow per circuit (gpm) and water temperature in and out of each circuit. If the plant has no flow meters, that is the first finding. - The resin supplier's heat content from melt to ejection (BTU/lb), for the heat load. Never filled from memory. 1. WHERE THE HEAT GOES (mold design) - The core side does most of the cooling (Mike). Plastic shrinks onto the core and pulls away from the cavity, so cavity-side cooling is partly cut off. Put the cooling effort, and the conductive materials, on the core. - Balance by conductivity, not only by water (Mike). Count how many hot plastic faces feed heat into each area of steel: - One face: steel with normal water. - Two faces meeting (an edge or a corner of a box): a more conductive insert, such as MoldMax (beryllium copper), on the core. - Three faces meeting (the bottom corner of a box or pan): three units of heat into one spot. An aluminum core cap there pulls the most heat out. The gate is often at that spot too, and aluminum's poor wear resistance does not matter on a core cap. - Office: beryllium copper dust from grinding or machining is a health hazard; the tool shop follows its beryllium handling rules. Aluminum wears; keep it off shut-offs and sliding faces. 2. WATER INTO EVERYTHING THAT MOVES - Get water into every component, especially small cores, slides, lifters and the plates behind them (Mike). If it can't reach the front, water the back end anyway: even a bubbler hitting the back of a lifter, or a line through the back plate, gives the heat a way out. - Why: heat does not cross a sliding fit (ways, gibs, slides) well, and a little corrosion on those faces stops it entirely. An uncooled moving part heats up, grows, binds and galls; you see grinding in the ways and splay in the parts (Mike). - Ejector pins are the accepted exception: they usually can't be cooled. 3. CIRCUITS - Keep circuits short (Mike). Short lines pass water well. - Loop where it makes sense; don't single-circuit everything (Mike). A straight-through line, especially in aluminum, moves a lot of heat and can share its water with a second circuit. Watering every line on its own wastes flow. - Find the restrictive circuit and give it its own water (Mike). It is usually the one with a bubbler or baffle, and it is the one people loop. That line may only pass about 2 gpm on its own, and the tool needs all of it. - Judge each loop by flow through it against heat it must remove (calculation below). 4. FLOW - A standard line, on a 200 to 300 ton mold, is drilled 7/16 in with 3/8 in pipe fittings; some molds use 1/2 in holes (Mike). - Target about 3 to 4 gpm through a standard line (Mike). That is about as good as cooling gets; more flow adds turbulence but little cooling. - Don't swing to the other extreme (Mike). Plants that are starved for water often add big pumps and big hoses; at about 9 to 10 gpm through a standard line the water starts eroding the line, aluminum worst of all. - Physics: the flow must be turbulent (Reynolds number above about 4,000) to carry heat well; the calculation reports it for every circuit, with the velocity. - Physics, for a 7/16 in line: 3.5 gpm is about 7.5 ft/s and well turbulent (Reynolds about 22,000 in 60 F water); 9.5 gpm is about 20 ft/s. So 3 to 4 gpm is already well past the turbulence a line needs, which fits Mike's point that more flow buys little cooling. 5. MEASURE - Buy flow meters, ideally with temperature sensors, on every circuit (Mike). Put the temperature on the water coming out of the mold. - Water out vs water in, per circuit (Mike; degrees F): - More than about 5 degrees: take a serious look at that circuit. - Within a degree or two on each line: perfect, and usually reachable. - The entire mold running within a degree or two: world class. - Physics: heat removed by a circuit = 500 x gpm x (out - in) in BTU/hr. A circuit with a big rise is working hard and is short of water; a circuit with almost no rise may have more water than it needs, which could go to a starved one. 6. THE CALCULATION (https://github.com/mminwi/an-creation-office/blob/main/specialists/calcs/cooling_water.py) Input per mold: shot weight, cycle, the resin's heat content from melt to ejection (from the supplier), and per circuit: line size, gpm, water in and out, insert material. Output per circuit: velocity, Reynolds number, temperature rise, heat removed, and flags (below 3-4 gpm, 9+ gpm erosion, not turbulent, over 5 degrees, within 2 degrees). Per mold: heat from the plastic (from an enthalpy number, or specific heat with melt and ejection temperatures), heat carried by the water, the total gpm needed for a target rise, and tons (BTU/hr / 12,000). Plant mode: see section 7. Every number is marked proposed. 7. PLANT WATER: SIZING CHILLERS AND PUMPS (Mike) Mold heat load, by pounds: - For each press, take its worst job: the heaviest pounds per hour (parts per hour x part weight). A packaging plant with a few materials can do this straight from throughput, for example 500 lb/hr. - Heat per pound = from melt temperature down to ejection, for example from about 350 F to under 150 F. Office: the heat per pound comes from the resin supplier (specific heat over that range, plus the heat of crystallization for semi-crystalline resins, or one enthalpy number); never from memory. - Worst-case plant load = every press running its worst job at once. Size for that worst case (Mike). Three units, two running: - Split the worst case over three chillers and three pump sets, each about a third (Mike). - Typical load is about two thirds, so two units run and the third is the backup. Running everything at once is rare (Mike recalls about two weeks near 100% in his career). - Rotate them: when a pump or chiller cycles off and comes back, the next one in line starts (lead rotation), so all three run regularly and you know each one works (Mike). - Maintenance: one unit can come off line without shutting the plant down (Mike). - Chillers run best with about a 10 F rise across the chiller, even while the molds run at 2 to 5 F rise; the return water is a bit warmer and the chiller runs more efficiently. Two chillers at part load run better than units pushed to their limit (Mike). Header piping: run it in a loop (Mike, highly recommended): - From the pump tank, take the supply out to the floor, tee it, and run one branch each way around the plant so the two meet on the far side. The split can be right after the pumps or where the header reaches the production floor; the loop around the floor is the part that matters. - Why: demand is never even. Some areas are shut down, one side pulls more than the other. A dead-end header has to be oversized to get enough water to the far end; a loop feeds every press from both directions, so the pressure stays steady and the last press in the row does not see the drop. - It also lets you close off one section of the loop to add a press, clean a line or repair it, while the rest of the plant keeps running. Office: that needs isolation valves between sections of the loop; place them so any one section can be closed with both directions still feeding the rest. Towers and outdoor units (Mike): - In Mike's plants the cooling tower cooled the chillers (it takes away the chillers' heat). Air-cooled chillers work too, but they also need regular cleaning; they pick up whatever is in the air. - Get towers and air-cooled chillers off the ground. Below about 8 feet the air carries dust, dirt and, in farm country at planting and harvest, a haze of organics. A tower or an air-cooled coil acts like an air filter: it fills with dirt and organics, and then it is cleaning and water treatment from then on. - Best: on the roof. If not, on a stand about 10 feet tall. Two or three feet up is not enough. - Height also keeps people from fiddling with them: at 10 feet it takes a lift to reach them. - Filter, filter, filter: everything should be filtered as much as possible (Mike). On a tower, add a bag filter system on a separate side set. At certain times of year the bags need cleaning daily, so the filters need something that shows clearly when they are loading up. Office: a pressure gauge on each side of the filter (differential pressure) is the usual way to see that. - Office: plan safe access for the service people (lift, platform, fall protection), and keep the tower on its water treatment and Legionella program (ASHRAE 188, Guideline 12). Tower size and climate (Mike): - Size the tower at about 1.25 to 1.5 times the chiller capacity (Mike). - Units, so it isn't counted twice (Perplexity answer Mike pasted, consistent with the physics): the tower rejects the chillers' load plus the compressors' heat, about 1.25 times the chiller load. A chiller ton is 12,000 BTU/hr; a nominal cooling-tower ton is 15,000 BTU/hr, which already includes that 1.25. So Mike's 1.25-1.5x in chiller tons equals about 1.0-1.2 nominal tower tons per chiller ton. Never multiply a tower-ton rating by 1.25 again. - Nominal tower ratings assume set conditions (reported as 95 F water in, 85 F out, 78 F outdoor wet-bulb; via Perplexity, confirm on the tower maker's data). Size from the chiller's stated condenser heat rejection at the site's design wet-bulb, water temperatures and flow. - Towers are evaporative coolers ("swamp coolers"). They work very well in dry climates (Arizona, the desert West); the hard part there is keeping them clean. - In places that run near 100% humidity most of the time (Georgia, Florida, the East Coast), a tower struggles: it can't get the water cold without the temperature climbing, and the extra plant life puts more organics in the air and the water. There, an air-cooled chiller system sized for the local warm temperatures is often the better choice. Air-cooled chillers have gotten very good, and Mike leans that way. - Office: the physics behind it: an evaporative tower can only cool water toward the outdoor wet-bulb temperature, and in humid air the wet-bulb is close to the air temperature. Use the site's design wet-bulb (for a tower) or dry-bulb (for an air-cooled chiller) when sizing. Hydraulic oil and feed throats (published rules of thumb, found 2026-10-12; Mike confirmed 0.1 ton/hp is the rule he remembered): - Hydraulic injection molding machines: about 0.1 ton of cooling per hydraulic motor horsepower, 1 ton per 10 hp. Sources: Advantage Engineering's sizing sheet (https://advantageengineering.com/fyi/073/advantageFYI073.php), Conair's chiller sizing guidance, and a Plastics Technology data card (https://www.ptonline.com/cdn/cms/PlasticsCard03152022.pdf). Read through search results and Perplexity; the office did not open the pages itself. - Feed throat: about 0.5 ton per throat on machines under 500 tons, about 1 ton at 500 tons and up (Plastics Technology card; Advantage lists 0.5 ton per throat). - General hydraulic systems, as a cross-check: heat is a fraction of pump input power, about 25% for simple circuits up to 60-75% for servo systems (Flodraulic, via Perplexity). Physics: 1 hp = 2,545 BTU/hr = 0.212 ton, so 25-33% heat is 0.053-0.070 ton per hp. Injection molding hydraulics run hotter than a simple circuit, which is why the molding rule is 0.1. - Use running power and duty cycle where known, not just nameplate; and size the oil cooler from its maker's heat rejection at the real oil flow and temperatures. - Put the hydraulics on chilled water (Mike). His plants started them on tower water and moved them to the chillers over time: tower water is too dirty for servo-driven hydraulics and the heat exchangers on them, which need cleaner water. Size the chillers for mold + hydraulic + feed throat loads together. The calculation (plant mode) adds up each press's worst-job pounds and heat per pound, adds the hydraulic (0.1 ton/hp) and feed throat allowances when given, then gives the total load in BTU/hr and tons, the size of each of three units, the typical two-thirds load, and the chiller loop flow at a 10 F rise (gpm = BTU/hr / (500 x 10)). Library: ASHRAE Handbook (HVAC Systems and Equipment), CTI codes, ASHRAE 188 and Guideline 12 (Legionella; they apply to cooling towers), chiller and controller makers' guides. HANDS OFF A mechanical contractor for plant piping, chillers and towers; a water treatment vendor for chemistry and the Legionella program on towers; the tool shop for inserts and circuit changes. The office prepares the review and the numbers; it does not specify plant equipment for purchase. WHAT THE REVIEW PRODUCES One cooling review per mold, in its own dated folder: the circuit table (flow, in, out, rise, heat, flags), the heat map by face count (one, two, three faces), the list of moving parts with and without water, recommended changes in order (single-circuit the restrictive line, add water to uncooled moving parts, conductive inserts where three faces meet, re-balance loops), and what to measure next. Recommendations are proposals; Mike or the tool shop confirms.