Optical Sensors for Corrugator Quality Control

Optical sensors on a corrugating line exist to shorten the time between a defect and a decision. Offline lab samples taken every pallet or every hour cannot see a ten-minute crush streak or a glue-line skip that already filled a stack. Cameras, laser profilers, and simple photoelectric arrays each catch a different family of faults. The plant’s job is to choose a defect list first, then a sensor set, then a reaction rule that operators will actually follow at speed.

Start with the defects that cost you claims: warp that fails converting, crush that fails ECT, glue skip, smear, and splice flags. Write the current detection method and the delay in metres or minutes. If warp is only found at the converting hopper, you are already too late for the corrugator crew to correct moisture or tension. An optical or profile sensor near the dry end is useful only if someone is assigned to act on the alarm within a defined length of board.

Warp and twist are geometry problems. Laser or structured-light profilers can map cross-direction height on a moving web when lighting and dust are controlled. A 2–4 mm cup that grows after the stack sits is still a moisture problem, so the sensor should be paired with a hold-and-remeasure rule, not treated as a single absolute number. Log the profile with paper identity and line speed so you can separate a sensor drift from a real climate swing.

Crush and caliper loss show up as a flatter flute profile or a sudden thickness drop. Optical thickness gauges or triangulation sensors at a stable pass-line can flag a pressure-roll or belt event faster than a hand micrometer on a cooling stack. Confirm the pass-line is free of flutter; a vibrating web will look like a thickness defect. After any alarm, check mechanical settings before you change adhesive.

Glue-line and skip defects need contrast. Some plants use cameras after the single facer or before the dry end to see starved tips or flooded valleys on a sample lane. Lighting angle matters more than megapixels: glare from a glossy liner can hide a skip that a side light would show. If you cannot get a stable image at 200–250 m/min, slow the inspection lane or inspect after a controlled sample pull rather than trusting a blurred alarm.

False alarms destroy trust. If the horn sounds twice an hour for dust specks, the night shift will mute it. Set thresholds from a two-week baseline of known-good board, then tighten. Require a reason code when an alarm is bypassed. Review the bypass log weekly the same way you review downtime codes. A sensor programme without a bypass review is just another light on the panel.

Integrate optical events with the encoder and order identity you already trust. A warp spike that cannot be attached to a reel number or a speed change is hard to fix. Use the same event language as corrugator-encoder-sensor-troubleshooting so electrical and quality teams share one timeline. If the encoder jumps, the camera metre stamp is fiction until the pulse train is healthy.

Sampling still matters. Optical systems catch spatial streaks; labs catch bond and moisture that cameras cannot see. Keep the L/C/R pin adhesion and moisture plan in board-quality-sampling-frequency-guide and treat camera alarms as triggers to pull an extra lab set, not as a replacement for the lab. A plant that switches off lab sampling after installing cameras usually rediscovers claims in converting.

Installation environment is half the project. Dust, starch mist, and heat haze will coat lenses and shift calibration. Specify purge air or a cleaning interval—often each shift for cameras near the wet end—and keep a spare window in stores. Mount away from steam leaks and from walkway elbows that operators will use as a rest. A perfect algorithm behind a dirty window is a failed project.

When defects are adhesive-related—skip, washboarding, mottled liner—the sensor is a messenger, not a chemist. Check gap, solids, and viscosity before you open a formulation change. Review application behaviour on /products/alkali-free-adhesive if you are already evaluating a system change, but do not let a camera project become an unplanned glue swap. Lock the reaction tree: mechanical first, kitchen second, chemistry third.

Data storage should be short and useful. Keep images or profiles for the last 24–72 hours at high rate, and keep trend numbers for 30–90 days. Longer video archives that nobody reviews are a disk cost, not a quality system. Export a weekly defect Pareto to the same meeting that reviews waste metres. If optical data never appears in that meeting, the sensors are decoration.

Training closes the loop. Operators need three things: what the alarm means in board language, what to check in the next five minutes, and when to call maintenance versus the kitchen. Post a one-page tree at the dry end. When a new paper grade arrives, retune thresholds with quality present rather than copying last month’s numbers onto a more open recycled liner.

Xuegong New Materials Group can help plants connect optical defect trends with adhesive and process windows when the alarm pattern points at bond or moisture rather than at a single crushed flute. Final sensor brands, safety interlocks, and acceptance tests remain the plant’s and integrator’s responsibility. Use board-quality-sampling-frequency-guide and corrugator-encoder-sensor-troubleshooting as the two checklists you complete before you issue the purchase order.