Give the same 1.16 MMBtu/h milk heating duty to the three sanitary constructions and you get three answers that differ by a factor of four in surface, a factor of ten in footprint, and a factor of two in price, and any of the three can be the right one depending on what else is on the sheet. The plate wins on coefficient and approach. The shell-and-tube wins on pressure, steam and pharma documentation. The tube-in-tube wins the moment the product has pieces in it or will not flow through a plate channel.
The six questions below settle most selections. Answer them in order and the construction is usually clear by the third.
The milk duty in tube-in-tubeRunning near 300 Btu/h-ft2-F, a tube-in-tube needs about 60 square feet in a rack of modules for the same milk heater, which is why it waits for cream with fruit in it.
Ten thousand lb/h of milk from 40 F to 165 F, hot water at 190 F returning at 175 F, LMTD about 65 F. In a sanitary gasketed plate at a service coefficient of 900 Btu/h-ft2-F the unit is about 20 square feet: a small frame, a few dozen plates, tri-clamp ports of 2 inches, and a pressure drop around 12 psi. In a sanitary shell-and-tube at 250 it is about 72 square feet: a 6 inch shell around 8 feet long with two tube passes, or an 8 inch shell around 5 feet, at 6 to 8 psi on the tube side. In a tube-in-tube on this product it would run near 300 and need about 60 square feet, which is a rack of several modules; it is the wrong choice here and the right one the moment the product becomes cream with fruit.
Change the utility to 150 psig steam and the plate is out because of wall temperature, the shell-and-tube shrinks to a 4 inch shell, and the wall has to be managed with a throttled valve.
The gasketed plate is the default for thin sanitary product because nothing else comes close on coefficient per square foot or on approach. Corrugated plates force turbulence at velocities that would be laminar in a tube, so the coefficient on milk, juice, beer or water sits at 800 to 1200 Btu/h-ft2-F. True counterflow means a 2 to 3 F approach and a temperature cross, which is what makes 90% regeneration a modest frame rather than a room full of shells. The unit opens by backing off the tie bolts, so every square inch of product-contact surface can be seen, which is why plants that inspect between CIP cycles prefer it.
The plate's limits are the gasket and the channel. Elastomer gaskets cap the temperature, typically 250 F for EPDM and less for nitrile, and cap the pressure, usually 150 to 230 psig depending on frame. Each gasket is one barrier between product and utility, and although the vented double gasket around each port makes a port leak visible, a plate crack is not visible until the product tests positive. The channels are narrow, so viscosity above about 500 cP and pieces above an eighth of an inch are out, and wide-gap plates only stretch that so far. Plates are also the construction most sensitive to fouling, because a deposit that is a nuisance in a tube is a blockage in a channel.
The sanitary shell-and-tube is the construction for pressure, temperature and documentation. Tubes are seal-welded or expanded and welded into the tubesheet, the shell takes steam at any plant pressure, and the whole vessel is code-stamped, hydrotested and documented in a way that a validation engineer can file. A double tubesheet, a sloped shell with low-point drains, a 20 Ra electropolished tube side and a full ASME BPE package are all available on the same basic geometry. This is why WFI coolers, clean steam condensers, point-of-use coolers and CIP heaters are almost always tubular.
The shell-and-tube's cost is thermal. Coefficients of 150 to 350 mean three to four times the surface of a plate for the same thin-product duty. A single shell with two or more tube passes is partly parallel flow, carries a correction factor, and cannot deliver a temperature cross, so a 10 F approach is economic and a 3 F approach is not. It takes more floor length, and its product-contact surface is inspected with a borescope rather then by eye. On thin product with a tight approach it is the wrong choice; on steam, pressure, pharma water or anything where the second barrier matters, it is the right one.
The tube-in-tube is a shell-and-tube reduced to its simplest form: one product tube, or a few, inside a jacket, in modules connected by return bends. Because the product path is a single open tube it passes anything that fits through it, and because the modules are in series it is true counterflow and can run a tight approach. It is drainable by design, cleans end to end with no dead legs, and the corrugated version raises the coefficient on products that can take the mixing. It also condenses and evaporates, which is useful in extraction and solvent recovery.
Its cost is length. Surface per foot of a single 2 inch tube is under a square foot, so a 60 square foot duty is a rack of modules, and on thin product it will never be the cheapest answer. But on yogurt with fruit, sauce with pieces, egg, tomato, and anything that would either block a plate or need a positive displacement pump to get through one, it is not competing with plates at all; it is competing with a scraped-surface unit at several times the price.
The sensible procedure is to size the thin-product duties in plates and check the pressure and hygiene questions, size the pressure, steam and pharma duties in shell-and-tube and check the approach, and send everything with pieces or above a few hundred cP to tube-in-tube. Where a duty sits near a boundary, between 200 and 500 cP, or at a 5 to 10 F approach, it is priced both ways. Send the sheet as it is and the comparison comes back with it.
Call and talk it through with an engineer: 1-805-484-2992
The pressure drop at the cold end and the pump that has to supply it, priced on the actual pump curve rather than a nominal limit. In that band a plate still gives a better coefficient than a tube, but the gain shrinks fast: a plate running 1,000 Btu/h-ft2-F on milk gives about 300 at 200 cP and 150 at 500 cP, and holding even that costs several times the pressure drop. A sanitary centrifugal pump runs out of head long before the plate runs out of surface, so a selection that works on paper arrives in the plant as a positive displacement pump nobody budgeted. A product that gels as it cools, or whose viscosity triples between inlet and outlet, takes plates off the table regardless. Send both viscosities and the pump data and both constructions are priced.
Plates are the most sensitive, tubes the least, and the difference is geometry rather than metallurgy. A protein deposit a fraction of a millimeter thick barely changes a 1 inch tube, but in a plate channel only a few millimeters wide it narrows the passage, so pressure drop climbs while the coefficient falls, and the two effects compound toward the end of a run. On raw milk the 0.001 h-ft2-F/Btu allowance cuts a plate's coefficient roughly in half; the same layer costs a tube a smaller fraction because its clean coefficient was lower to start with. Compare fouled run length, not clean size.
The gasket, and the plate itself: each is a single barrier between product and utility that can fail without an outside sign. EPDM gaskets are limited to about 250 F and age with every CIP cycle, and a hairline crack in a plate passes fluid across the pack with no leak to the floor. The vented double gasket around each port shows a port leak, but not a cracked plate. Plants with a strict leak policy budget for regasketing and for periodic dye penetrant or pressure decay testing; both costs belong in the comparison beside the purchase price.
Each needs its service envelope, which is larger than its footprint, and the plate's one-fifth-to-one-half advantage shrinks once that is drawn. A plate frame needs clear length behind the movable plate to slide it back and lift plates out; a shell-and-tube needs bundle-pulling room at one end; a tube-in-tube rack needs access to every return bend. Draw the opening clearance before deciding on footprint.
Yes, and some pharma and allergen-sensitive food plants do exactly that, buying a longer unit to get a gasket-free product path that drains completely. The price is length: a 2 inch inner tube carries about half a square foot of surface per foot, so a 60 square foot thin-product duty is roughly 115 feet of tube in a rack of modules, against a plate frame a person can step around. On thin product the coefficient is well below a plate's, but series counterflow keeps the approach close and regeneration is still possible. It is worth pricing wherever full drainability or no elastomer in the product path is written as a requirement rather than a preference.
A gasketed plate on thin product and a tube-in-tube in series on thick product, because both are true counterflow and the utility only needs to run a few degrees above the product outlet. A single shell with two tube passes cannot do that: its economic approach is 10 F or more, so for the same 165 F milk outlet the hot water has to run hotter, and the wall runs hotter with it. Since the wall temperature, not the bulk temperature, is what denatures protein and burns sugar on, the wider approach becomes a fouling problem rather than just a thermal one. The plant symptom is a tubular pasteurizer that cleans well on Monday and shows protein film by Wednesday, with operators raising the loop set point to hold temperature, which makes the wall hotter still. Where the shell-and-tube is wanted for pressure or documentation, two shells in series recover most of the counterflow behavior; where it is not, plates or tubes in series are the straightforward answer.
Where a tube-to-tubesheet leak would go, what it would cost, and whether the second sheet is required.
+ Learn More
The utility sets the wall temperature, and the wall temperature decides whether the product survives the pass.
+ Learn More
Plate, shell-and-tube or tube-in-tube, single or double tubesheet, and the utility that suits the product.
+ Learn More