Product-to-product heat recovery
The regeneration section is a plate pack with raw milk on one side and pasteurized milk on the other. Cold product going in picks up heat from hot product going out, so the heating section supplies only the difference. On a 40,000 lb/h line at 90 percent the section moves about 4.4 million Btu/h between the streams, heat the plant neither buys as hot water nor removes as refrigeration.
It is also the one place where raw and pasteurized milk are separated by a single 0.5 mm plate, which is why it has its own pressure rule, its own pump and a plate count decided by economics.
Regeneration efficiency is the raw milk rise across the section divided by the total rise from raw inlet to pasteurization temperature. Raw at 40 F, pasteurized at 161 F, raw outlet at 149 F: 90 percent. Because both streams are the same fluid at the same flow, the temperature difference is constant along the section, and that difference is the approach: 121 degrees times 10 percent is 12 degrees at 90 percent, 6 at 95.
Plate area scales inversely with approach, so 95 percent needs twice the plates of 90 while the heat saved rises by only 5 percent of the total. That is the whole economic argument: the first 85 percent is nearly free, 90 is the normal design point, 95 pays back in two to four years on two shifts, and anything above is bought for a reason other than money.
If a regeneration plate cracks, product crosses. The PMO requires the crossing to go from pasteurized to raw, so the pasteurized side runs at least 1 psi higher at every point, in forward flow, in divert and during CIP. A plate pack does not do that by itself; the layout has to. In the standard arrangement the timing pump draws raw milk through the raw regenerator, so the raw side runs under suction while the pasteurized side leaves the pump under pressure.
A booster pump, where the line needs one for capacity, is the part that can upset that. It sits on the raw side between the balance tank and the raw regenerator, interlocked so it cannot run unless the timing pump runs, the diversion valve is forward and the pasteurized side is at least 1 psi above raw. A differential pressure controller pens on the same chart as the temperature. The detail that gets missed is shutdown: the raw side must lose head first, so HeatX lays the section out with the raw side draining before the pasteurized side and a booster pump that does not backflow when stopped.
The pressure rule makes a single-wall regenerator legal, but pasteurized product crossing into raw is a yield loss the plant notices late. A double-wall plate is two plates pressed together with a vented gap; either can fail and the product weeps out the gap edge, never into the other stream. On a tubular regenerator a double tubesheet does the same job.
Double-wall plates transfer about 20 percent less heat per plate, so the section needs more of them. HeatX quotes them on infant formula, where a leak means a recall, and wherever QA wants physical rather than pressure-based separation. On fluid milk the single-wall section with a correctly interlocked raw-side booster is the standard answer.
Buying 95 percent and giving it back to fouling. At 95 percent the approach is 6 degrees. Let 2 degrees of fouling build over a week and the raw outlet drops 2 degrees, the hot water valve opens, the heating section wall climbs and fouls faster. The line still pasteurizes; it just loses the efficiency it paid for. A 95 percent section needs a CIP interval and a pressure drop alarm a 90 percent section does not.
Tell HeatX the flow, the temperatures, the shifts and the steam cost, and the regeneration point falls out.
Call and talk it through with an engineer: 1-805-484-2992
Each row gives a design point for the raw and pasteurized streams, from approach to the pressure rule. Compare them with your line and send your own temperatures and flows so the plate count starts from your duty.
| Condition | Detail |
| Product | Raw milk, cream, mix or juice; the same product pasteurized on the other side |
| Flow range | 5,000 to 100,000 lb/h, equal mass flow both sides |
| Temperatures in / out | Raw 40 F in, 145 to 155 F out; pasteurized 161 F in, 46 to 56 F out |
| Hold | None; the section sits either side of the holding tube |
| Utility | None; product against product |
| Approach | 6 degrees at 95 percent, 12 at 90 percent, 18 at 85 percent |
| Construction | Gasketed plate, single or double-wall; double tubesheet tube on tubular lines |
| Finish / class | 3-A 11-10 plates, 316L, 32 Ra; EPDM or nitrile gaskets |
| Pressure | Pasteurized side at least 1 psi above raw at the section connections |
| CIP | Both sides in the circuit; raw side drains first on shutdown |
The interlock stops the booster pump, which puts the raw side back under timing-pump suction and restores the safe pressure direction; the booster restarts only when the differential recovers. The diversion valve does not move on this signal and the timing pump keeps running, so the line carries on at reduced capacity. One trip at start-up while pressures settle is normal. Repeated trips mid-run are hydraulic: the pasteurized side is fouling and losing pressure at the regenerator outlet, a pump speed has changed, a gasket is starting to weep, or the differential switch itself has drifted. Check the switch first with a hand gauge at the two section connections, because a switch reading 1 psi when the real margin is 3 psi costs capacity, and one reading 3 psi when the margin is gone costs you the public health argument. Log every trip with time, product and flow; after a week the pattern usually names the cause.
Test on a schedule with salt water, before a mass balance drifts, because the pressure rule decides which way product crosses a hole, not whether it does. Circulate water with a few percent salt on the pasteurized side at its normal higher pressure and plain water on the raw side, and watch the raw outlet with a conductivity meter: a rise means a hole. A pinhole leaks pasteurized into raw, so it never shows as a quality complaint; it shows as pasteurized milk going around twice, lost capacity and a hot water bill that climbs for no reason. On a 40,000 lb/h line a leak of even 1 percent is 400 lb/h reheated all day. Fluorescent dye on the opened plates finds the crack once the salt test says there is one, and the double-wall section is the answer where QA will not accept a test interval at all.
Gaskets by section on a schedule, plates by inspection result, and the tightening dimension written down every time the frame closes. A pack of mixed old and new gaskets compresses unevenly and leaks at the old ones first, so replacing a whole section at once is cheaper than chasing weepers one at a time. EPDM on a regenerator that sees 161 F milk and 180 F caustic every day lasts a few years, and the frame tells you when: the closing dimension has to go tighter each time to seal as the rubber takes a set. Plates are checked for cracks and thinning at the ports and contact points whenever the frame is open and replaced on what is found. A frame tightened below the plate maker's minimum dimension crushes gaskets and bows plates, so the stamped record matters.
Two legal routes for fluid milk, 161 F for 15 seconds or 145 F for 30 minutes, and the exchanger each needs.
+ Learn More
Tube length from velocity and hold time, a valve that diverts in under a second, and a chart that proves both.
+ Learn More
Plates for thin milk and short payback; tubes for viscosity, fiber, fouling and 20-hour runs.
+ Learn More