Case Study: Air Knife Drying Retrofit on a Two-Filler Canning Hall

Water that leaves the pasteuriser on the can body is cheap until it reaches the labeller. Then it becomes rejected cases, seam corrosion returns, slipping pallet film, and a stopped line while the crew wipes changeparts. On this class of canning hall the label reject rate alone can run at several thousand cans per shift. The raw compressed air used to fix it can cost three times the power it should.

This case is a composite drawn from European beverage and cooking-oil canning halls with two fillers in the 400 to 900 cans per minute class. Every figure below is the generic shape of this class of installation, not a record of any single plant. What transfers between plants is the arithmetic of the air knife drying retrofit, not the plant name.

Case Snapshot

The table below describes the generic shape of this class of installation, not a record of any single plant.

Item	Reference value
Operation	Beverage and cooking-oil canning hall, Europe, two fillers
Speed classes	Filler A near 400 cans per min, filler B near 900 cans per min
Can format	Steel can about 66.2 mm diameter, 117 mm tall, 0.031 m2 surface
Wet station	Tunnel pasteuriser and spray cooler discharge into labeller infeed
Old setup	Compressed-air drying nozzles on a manifold, about 4 m3/min at 6 bar
New setup	Blower-fed air knife system: one top knife plus side knives per lane group
Residual water before	About 90 to 150 g/m2, with 120 g/m2 as the working figure
Residual water target	Under 5 g/m2, held near 4 g/m2 at commissioning
Blow-off force	About 25 N from 8 m3/min of free air at 0.5 bar g
Drive power	About 9 kW blower shaft against about 25 kW for the same force at 6 bar

A two-filler hall feeding one pasteuriser, one labeller group, and one palletiser is the common shape in European canning. The sizing logic below transfers to that whole class.

The Challenge

The hot-filled beverage line ran a steam-tunnel pasteuriser, then a spray cooling section. Cans left the discharge carrying a continuous water film plus droplet holdup in the double seams and the neck. The cooking-oil line was wetter, because its caustic wash left detergent film that holds water longer than clean steel. Nothing between the cooler and the labeller removed that water except gravity and time.

Three defects followed the water. Paper labels lifted at the wrap seam, because cold-glue tack needs a dry body and a water film breaks the bond. Corrosion appeared at the double seam within weeks of packing, because capillary water in the groove keeps the bare tin edge wet. Pallet film slipped on the night shift, because shrink film over wet cans does not pin the load.

The old fix was a compressed-air manifold over the conveyor with a row of drying nozzles. It wasted power. It sprayed mist across the aisle. It left a dry stripe where the nozzle row had been set for an older lane count, and sound at the station sat near 100 dBA. Our air nozzle selection guide covers why a loose nozzle row is the wrong tool for a sheet-peel job like can drying.

The cost sat in two places. The plant paid compressor electricity for a blow-off that was loud and shallow. It also paid label rejects and seam returns, the expensive item in any canning loss review.

The Moisture Budget on the Can Web

The retrofit started by writing the water budget on one sheet. How much water arrives, how much the adhesive needs removed, and what each removal option costs in power. Air knife drying systems are sized against that budget, not against a feeling about damp cans.

The can is a 66.2 mm by 117 mm cylinder with two ends. Surface area is pi x d x h plus two circles, about 0.031 m2 per can. A film at 120 g/m2 therefore carries 120 x 0.031, roughly 3.7 g of water per can. Worked at the two speed classes:

Duty	Residual film	Water per can	Water per hour
Filler A, 400 cans per min	120 g/m2	3.7 g	3.7 x 400 x 60, about 90 kg/h
Filler B, 900 cans per min	120 g/m2	3.7 g	3.7 x 900 x 60, about 200 kg/h
Filler B after knives	4 g/m2	0.13 g	0.13 x 900 x 60, about 7 kg/h
Labeller requirement	1 to 5 g/m2	0.03 to 0.16 g	Cold-glue tack window on paper

Two readings matter. 200 kg/h of carried water at full rate is a physical load, not a cosmetic one. That is a filled bathtub every four hours walking past the labeller. And the adhesive does not need bone-dry steel. Published cold-glue practice on paper labels asks for a residual film in the low single g/m2. That is a blow-off duty, not an oven duty. A warm-air tunnel would over-build the job. A well-set air knife strips the film to 3 to 4 g/m2 in a fraction of a second, with no heat at all. A 120 g/m2 film is 0.12 mm of water depth, and that thickness sets the knife gap.

Air Knife Geometry: Gap, Angle, Standoff

An air knife is a plenum with a continuous slot that extrudes a thin sheet of air. Three geometry numbers decide whether the sheet removes the film or slides over it. The slot gap, the exit angle, and the standoff to the moving can.

Gap. The slot height was 0.6 mm on the top knife and 0.5 mm on the side knives. Published blow-off practice sizes the gap at several times the film thickness, and the 0.12 mm film sits about five times under the 0.6 mm slot. A gap much narrower than that chokes the flow and raises back pressure for nothing. A wider gap wastes air and lets the sheet break into turbulence. Narrow slots are debris-sensitive, so the blower intake carries a filter and each knife has a blow-down port.

Angle. The slot lip is bevelled so the sheet leaves at a shallow included angle to the blade face, commonly in the 15 to 30 degree region. That wedge keeps the sheet coherent across the knife length. The side knives sat roughly 30 degrees to the can surface, a sweeping angle that peels the film downward into the drain trough. The top knife sat near 90 degrees over the lid so the sheet pushes standing water off the crown and out of the seaming groove. Same equipment, two jobs, set by angle alone.

Standoff. Side knives held 5 to 10 mm from the can wall. The coherent sheet stays effective within roughly twenty slot heights, so a 0.5 mm slot carries useful momentum at 10 mm and loses it past about 25 mm. The cans wobble a millimetre on the infeed table, so the setting is the mean gap plus a compliance allowance, checked with a feeler card while the line creeps.

The exit velocity follows from flow and slot area. The blower delivered 8 m3/min of free air, which at 1.2 kg per m3 is 0.16 kg/s. Inside the knife at 0.5 bar g the density rises to about 1.8 kg per m3, so the slot sees 0.16 / 1.8, roughly 0.089 m3/s. The total slot area is 450 mm x 0.6 mm on top plus two side knives at 300 mm x 0.5 mm, so 570 mm2 or 5.7e-4 m2. Velocity is 0.089 / 5.7e-4, about 155 m/s. The momentum at the sheet is 0.16 x 155, about 25 N of blow-off force. That 25 N is the number the compressed-air comparison hangs on. The flat fan air nozzle blow-off guide applies the same momentum logic to header nozzles, the cheaper option for short coverage strips.

Dwell completes it. At filler B, 900 cans per min across six lanes at a 72 mm pitch moves each lane at about 180 mm/s. A 40 mm contact arc gives 40 / 180, about 0.2 s of blade time per can. The stripping happens inside that window, which is why knife drying replaces a minute of drip time without a minute of drip time.

Knife Length, Line Width, and the Seam Trap

Knife length is set by the wetted width of the can web, not by the can diameter. Six lanes at 72 mm pitch make a 432 mm belt, so the top knife is a 450 mm bar with a few millimetres of overhang per side. A knife cut to the exact lane count leaves a dry stripe at the belt edge, the same failure the old nozzle row had at the centreline. The rejects arrive in the outer lanes nobody inspects. Where a hall runs mixed lane counts, use a sectional knife with shim-off blanks over empty lanes instead of a second knife size.

Side knives only see the body. They cannot reach the two places water hides on a can. One hiding place is the double seam, where the groove holds 0.03 to 0.08 mL of capillary water against surface tension. That is a tiny mass beside the 3.7 g body film. It is also the exact metal edge where corrosion staining starts, so mass is the wrong metric and location is the right one. The other hiding place is the neck under the lid curl, where a horizontal side sheet passes straight over the shoulder.

The retrofit attacked both with geometry instead of power, because can drying is really seam drying plus body drying. The top knife swept the crown and pushed groove water off the lid edge. A short angled blade over the shoulder, set at 45 degrees, broke the surface tension at the neck. A drip lip downstream of the knives caught what the seams released late. This is why air knife drying systems for cans are quoted as a knife package with drain plan, not as one bar over one belt. The air nozzle for air compressor guide covers the single-nozzle equivalent for spot duties like the shoulder purge, where a full knife is oversized.

Blower Duty Against Compressed Air

The power arithmetic is why the retrofit was approved. Both options deliver the same 25 N. They pay for it very differently.

Blower side: 8 m3/min of free air raised from 1 to 1.5 bar absolute. Adiabatic specific work for that ratio is about 44 kJ per m3, so ideal air power is 44 x 8 / 60, roughly 5.8 kW. A regenerative blower at 60 to 70 percent overall efficiency lands the shaft draw near 9 kW.

Compressed-air side: the same 25 N from a knife fed at 6 bar g means an exit velocity choked near 310 m/s, so the mass flow needed is 25 / 310, about 0.081 kg/s. That is 4.8 kg per min, roughly 4 m3/min of free air at 1.2 kg per m3. Compressing to 7 bar absolute costs about 264 kJ per m3 ideal, and at 70 percent compressor efficiency the shaft figure lands near 25 kW.

Same force, nearly three times the power. The blower-fed knife delivers about 2.8 N per shaft kW, and the 6 bar knife about 1 N per kW. Per can at 900 per min, the blower spends about 0.6 kJ and the compressed-air option about 1.7 kJ. Over 6000 running hours at 0.20 EUR per kWh, that is roughly 11000 EUR per year against 30000 EUR. The old manifold was leaking from the airline anyway, and that is pure waste on top of the 25 kW.

Noise moves the same direction. The air knife blower package in its housing sits at 75 to 80 dBA at one metre. An open compressed-air knife at 6 bar lands at 95 to 105 dBA, and the station came out of hearing-protection territory. One honest cost: the blower inlet and silencer add footprint, and the discharge runs warm. The spray nozzle flow rate calculation guide walks the same flow-times-pressure discipline for liquid duties. The cooling nozzle selection guide covers pre-cool duties upstream of the knives.

Controlling Re-Entrained Mist

Blow-off does not destroy water. It moves water off the can and into the room, and water in the room is the next defect. The 25 N sheet strips about 3.6 g per can, a mist load near 200 kg per hour at full rate. Uncontrolled, part of it rains back on downstream cans, the label magazine, and the film reel.

Three containment measures were built into the frame. A drain trough under the side knives with drip lips collects sheet-borne water as a liquid stream before it atomises. About four fifths of the removed water leaves as trough flow. A capture hood over the top knife draws the fine fraction down, and a face velocity of 0.5 to 1 m/s across a hood opening of 0.5 by 0.8 m needs an exhaust near 1000 to 1500 m3/h. Between stations, a short enclosed transition ends the mist zone before the labeller glue head opens. Knife orientation does quiet work too. The side sheets angle downward so gravity takes the peel water into the trough instead of throwing it upward. Re-entrainment control is mostly the knife set itself, and the hood is the insurance. Housekeeping is part of the same spec, because mist on a walkway is a safety item before it is a quality item.

The Solution

The install went in on a weekend changeover at each filler, over the pasteuriser discharge table and the labeller infeed. Each lane group received one top knife bar, two side knives, and a shoulder purge blade. All stainless 316 with machined slot lips, bolted to a welded frame with feeler-gauge shim sets. The old compressed-air manifold stayed valved off as standby.

One blower per filler, not one plant-wide blower. The two lines peak and idle on different schedules, and a shared header would run the fast line’s knives at the slow line’s pressure. Pressure trim came from a variable-speed drive on the blower motor, so the operator dials blow-off strength at changeover instead of throttling at the knife. The BSP and NPT thread guide is the reference the maintenance crew received, because a mixed thread inventory is how a Friday changeover becomes a Monday leak hunt.

Material followed the washdown regime. The wet zone is all stainless, and polymers were limited to slot gaskets and wear strips rated for hot humid contact. The plastic versus stainless corrosion guide covers where plastics earn their place, which is on the chemical dosing side and not at the pasteuriser discharge. Nothing downstream of the pasteuriser that is not smooth, drainable, and hot-water flushable. The spray applications overview carries the same framing for the wider hall. Two spot jets on the standby air stayed for manual clean-down of the drip lips. The air knife system takes the continuous duty, and compressed air retreats to the manual corner.

Commissioning: Weighing, Tape, and Pattern Checks

Verification was three tests against the moisture budget, run at both speed classes and recorded per lane.

The weighing test. The residual number is measured, not judged. A 100 by 100 mm aluminium witness plate taped to the belt carries 0.01 m2, so a 120 g/m2 film shows as 1.2 g and a 4 g/m2 film as 0.04 g, both readable on a 0.1 mg balance. The crew tared dried plates, ran them past the knives, blotted the rim, and re-weighed within a minute so evaporation could not steal the result. The cross-check was can weighing on a 0.1 g balance: before the retrofit the body film read about 3.7 g per can, after it read about 0.1 g. Four plates per lane group, at 400 and 900 cans per min, and the g/m2 number became a shift record instead of an opinion.

The tape test. Low-tack tape pressed on the body at the label wrap point and peeled back tells the adhesive story directly. Dry steel leaves clean tape. A residual film leaves a dull watermark on the adhesive side. Run it at every changeover, because it samples exactly where the glue lands.

The sheet check. A ragged peel line tells you the slot is fouled before the rejects do. Commissioning ran a dry flour dusting through each knife and photographed the peel on a catch card. A clean straight peel means a coherent sheet. A ragged one means a nicked lip, a stone, and a re-shim before sign-off. Slot fouling is the silent drift mechanism here, the same family as a worn orifice silently over-applying water. The nozzle wear guide is the companion read for that failure mode. Sign-off numbers: 3 to 4 g/m2 on every lane at both rates, trough and hood capturing, blower holding 0.5 bar g at the fast line.

The Results

No audited plant data appears in this table. Every figure is the generic shape of this class of retrofit, anchored to published ranges for blow-off films, specific power, and label adhesive practice.

Metric	Before (nozzle manifold)	After (blower-fed knives)	Basis
Residual film after pasteuriser	90 to 150 g/m2, about 3.7 g per can	3 to 4 g/m2, about 0.1 g per can	Weighed witness plates
Label lift rejects at the wrap seam	0.3 to 0.5 percent of cases	Under 0.05 percent	Adhesive tack needs a dry body
Double-seam corrosion returns	Seasonal complaint entries	No entries in the following campaign	Groove water removed at source
Pallet film slipping	Night-shift rewrap calls	Stable loads	Film pins a dry surface
Blow-off station sound	100 to 105 dBA at 1 m	75 to 82 dBA at 1 m	Housing against open knife exit
Drive power for the same 25 N	About 25 kW compressor shaft	About 9 kW blower shaft	Adiabatic work per m3 of free air
Energy per can at 900 per min	About 1.7 kJ	About 0.6 kJ	Shaft power over throughput
Stripped water destination	Mist across the aisle	About four fifths to trough	Drain lips plus hood face velocity

The honest readings. The film figure is a band, and the band widens as the slot fouls. The reject row assumes a healthy glue head, because an air knife cannot negotiate a wrong glue film weight. The corrosion row covers one campaign window, which is how tin staining reports arrive.

Reading the Numbers Against Published Ranges

Residual film. Published blow-off practice treats 1 to 5 g/m2 as the landed band for a well-set knife against a water film, and reserves oven drying for requirements below that. The 4 g/m2 commissioning figure sits mid-band. That is where can drying should sit when the demand is adhesive tack rather than coating adhesion.

Exit velocity and force. Knife sheets in this duty class commonly run 50 to over 150 m/s at the exit, so the computed 155 m/s lands at the working end from an ordinary regenerative blower. Carry the 25 N figure into any supplier conversation, because force is the removal capacity and geometry decides whether it reaches the film.

Standoff. Published knife practice keeps the working distance inside roughly twenty slot heights, under about 10 to 25 mm for a 0.5 mm slot. A setting sheet claiming a hand-span standoff from a narrow slot is quoting open-nozzle spray distance, not a coherent sheet. The distinction matters on a wobbling can web.

Specific power. The comparison here, 2.8 N per shaft kW against about 1 N per kW at 6 bar, follows from the adiabatic work ratio and the exit-velocity difference. It matches the published position that low-pressure high-volume air wins momentum duties. Can drying is that duty. Blasting baked residue under pressure is not. The air nozzle selection guide lays out the split.

Noise. The 95 to 105 dBA open-knife band and the 75 to 80 dBA housed-blower band are stock figures in compressor and blower literature. A quote for air knife drying systems that omits a sound figure at one metre is incomplete, because the acoustic result was half of this approval.

Why This Matters

Canning halls own both halves of this problem and connect them only after an audit. The wet half is physics: a pasteuriser exports a known g/m2 load, and the budget above shows its kilograms per hour at each speed class. The dry half is money: compressed-air blow-off buys momentum at nearly three times the blower cost and rents the difference back as hearing protection and leak schedules. An air knife drying retrofit puts both halves on one sheet, which is why it reads well in a capital request and better in next year’s utility cost per thousand cans.

The failure mode changes too. A nozzle manifold degrades invisibly. Knife slot lips are inspectable geometry, checked with a feeler card and a flour sheet in minutes. Once the residual film is a recorded number per lane, the labeller stops guessing and the rejects become a control chart. That is the practical value of air knife drying systems: not dryness as a feeling, but dryness as a measurement.

And the seam lesson generalises. The mass that mattered least, the drops in a seaming groove, carried the costliest defect. A blow-off design judged only on the big film number will miss it. Judge the package on where the sheets point, not only on how hard they blow.

Size Your Can Drying Duty Against the Numbers

If your line carries a film into the labeller and a hiss into the hearing zone, start with three measurements instead of a quote request. Weigh the film per can. Count the lane width at the pitch. Note the free air the manifold actually burns. We can size a knife package against that budget. Gap, angle, standoff, knife length against belt width, blower duty against the momentum needed, and the verification plan to prove it. Bring the g/m2 figure, the cans per min, and the can format. Talk to us through the contact page, and see the air nozzle range plus the flat fan range for the header option when a knife is oversized for the duty.