Dispensing automation: putting adhesive, sealant and grease down the same way every time

Material behaviour sets the process window

Dispensing equipment is specified around a material, and the material moves. Viscosity falls as temperature rises, so a bead metered by time and pressure grows through the morning unless the material is conditioned or the system meters by volume. Two-part chemistries add a working life: once mixed the clock runs, and any stoppage longer than that clock means a purge and a fresh static mixer. Filled materials abrade wet ends and settle in reservoirs. None of this is exotic, but it explains why a cell that laid perfect beads during a supplier trial lays variable ones on a floor with no temperature control and a different material lot.

The bead lands where the part is, not where the program says

A dispense path is programmed against a nominal part in a nominal position. Real parts arrive with moulding variation, fixtures wear, and an operator loads with a slight rotation the clamp tolerates. The bead then sits half on the flange and half on the radius, and the joint fails a pressure test weeks later. Two remedies exist: tighten presentation until the programmed path is valid, or add vision that finds a feature and shifts the path. The second is more forgiving and more expensive, and it still needs a datum the camera can see reliably. Surface condition matters as much as position, since release agent, coolant residue or handling oil defeats any bead geometry.

Blockages, entrained air and the end of a drum

What actually stops a dispensing cell is unglamorous. Nozzles skin over during idle periods, so cells need a park station and a purge shot before the first part after a break. Air drawn in at a drum change appears later as voids in the bead and gets blamed on the applicator. Follower plates that do not seal leave a crust the pump eventually pulls through as a lump. Filters load. Assign these to a named routine with an interval the maintenance plan genuinely contains, because dispensing equipment degrades gradually rather than failing outright, and gradual degradation is the sort nobody raises until scrap appears at final inspection.

Proving the bead without opening the assembly

Once parts are joined the bead is invisible, so verification has to happen before closure. Inline inspection measuring bead position, width and continuity as it is laid catches skips, wander and a nozzle that stopped mid-path. Dispensed quantity can be checked independently by monitoring shot weight against the program. Neither proves adhesion. Bond strength is confirmed by destructive audit on sampled assemblies plus control of what adhesion depends on: surface preparation, open time before mating, clamp pressure and achieved cure. Treat geometry monitoring and bond verification as two separate obligations, because passing the first is routinely reported as passing both.

Purge waste, extraction and the operator's new job

Automating the applicator does not remove people, it changes what they do. Someone replenishes material, swaps mixers and nozzles, runs the opening bead check and answers faults. Purge waste is a genuine consumable line: every material change, every restart after a long stoppage and every mixer swap discards usable product, and on a high-mix schedule that discard can rival what reaches the parts. Solvent cleaning and uncured chemistry bring skin and inhalation exposure into a job that previously involved a cartridge gun, so extraction, gloves and a written handling routine belong in the cell design rather than being bolted on after the first complaint.

Frequently asked questions

Why does a dispensing cell that ran perfectly at the supplier drift once it is on our floor?

Three things usually differ. The supplier ran conditioned material at a stable temperature and your building is not stable. The supplier used hand-selected parts and yours come off a production tool with real variation. And the supplier ran continuously, while your schedule contains breaks long enough for the nozzle to skin. Insist the acceptance run uses your parts, your material lot and a deliberate pause, because those conditions expose almost every fault later blamed on the equipment.

Can one cell dispense more than one adhesive?

It can, but each change costs a purge and usually a wet-end clean, and incompatible chemistries can cure inside the equipment. Where the schedule alternates materials often, dedicate reservoirs and swap the module rather than flushing a shared path. Where materials are similar and changes are rare, a documented purge with a verified opening bead works. Settle this before purchase, since fitting a second material path into a compact cell afterwards is awkward and sometimes impossible.

What evidence shows the joint is sound rather than just that a bead is present?

Bead monitoring evidences geometry and continuity only. Sound joints need the whole chain controlled: substrate cleanliness and preparation, material within its stated shelf and working life, mating inside the open time, correct fixture pressure, and cure temperature and duration achieved on the part rather than at the oven setpoint. Add periodic destructive testing on production assemblies, and record which material lot and which cure cycle each batch saw so a failure can be bounded instead of guessed at.