News

Online control method for wall thickness of extrusion blow molding machine

Online Wall Thickness Control for Extrusion Blow Molding: Methods That Actually Keep Tolerances Tight

Wall thickness variation is the single biggest quality headache in blow molding. Consumers feel it when the bottle feels uneven. Packaging engineers see it when fill levels drift. Quality managers see it when rejection rates climb. And yet most production lines still rely on offline caliper checks or manual parison adjustments that react too late.

The shift toward real-time, in-line wall thickness measurement has changed what's possible. But installing sensors isn't enough — the control strategy behind the data is what separates a system that cuts waste by 15% from one that just generates pretty graphs nobody reads.

Why Wall Thickness Drifts in the First Place

Before talking about control, it helps to understand what's fighting you.

Blow molding is inherently unstable. The parison sags under its own weight between the extruder die and the mold cavity. Air pressure inside the parison pushes it outward unevenly. Cooling rates vary across the mold surface because water channels wear, scale builds up, or ambient temperature shifts. Every one of these factors changes the wall distribution from cycle to cycle.

Then there's material behavior. Polymer melt temperature fluctuates by a few degrees, and viscosity changes exponentially with that small shift. A 5°C drop in melt temp can increase viscosity enough to reduce parison weight by 3 to 4%, which directly thins the wall. Screw speed variations, die swell changes, and even minor moisture in the resin all feed into thickness inconsistency.

The result? A bottle that measures 0.85mm on one side and 1.15mm on the other — well outside most packaging specs. Traditional approaches tried to fix this by over-engineering the parison to be thicker than needed everywhere, then accepting the waste. Real-time control lets you stop doing that.

Measurement Technologies That Work Inside the Production Line

Not all thickness sensors are created equal, and choosing the wrong one will give you noisy data that your control system can't trust.

Beta-ray backscatter sensors sit outside the mold and measure wall thickness by detecting how much radiation passes through the plastic. They're non-contact, fast, and work well for high-density materials like PET and HDPE. The downside is they measure one point at a time as the bottle rotates, so you get a cross-sectional profile rather than a full 3D map. For most packaging applications, that cross-section is enough — but if you need full coverage, you need multiple sensors.

Ultrasonic pulse-echo sensors work similarly but use sound waves instead of radiation. They're safer to operate since there's no radioactive source, and they handle thicker walls better. However, ultrasonic signals scatter in crystalline polymers like HDPE, making readings less reliable on semi-crystalline resins.

Microwave and terahertz sensors are newer and gaining ground. They penetrate plastic well, work on both amorphous and crystalline materials, and can measure through air gaps where other sensors struggle. The technology is still maturing, but early adopters report better accuracy on multi-layer containers where different polymer layers confuse beta sensors.

Laser triangulation and structured light scanning measure the outer surface of the finished bottle. Combined with a known cavity dimension, you calculate wall thickness by subtraction. This indirect method avoids radiation issues entirely but assumes the mold cavity is perfectly consistent — which it never is. Use this for trend monitoring rather than absolute control.

The Parison Programming Strategy That Reduces Thickness Variation at the Source

Sensors alone don't control anything. They tell you what's wrong after the fact. The real lever is the parison.

Most modern extrusion blow molding machines use accumulator-type parison heads with programmable mandrel pins. These pins can move axially and radially to change the gap between the mandrel and the die, which directly controls how much material goes where along the parison length.

If your sensor tells you the bottom of the bottle is thin, the control system pushes the mandrel pin outward at the bottom position, reducing the gap and forcing more material into that zone. If the shoulder is too thick, the pin retracts locally, letting the parison sag and use less material there.

This is called axial parison programming, and it's the foundation of every serious wall thickness control system. Without it, you're just measuring a problem you can't fix.

The timing matters too. The mandrel position needs to be set before the parison drops into the mold. Any delay means you're correcting the previous cycle instead of the current one. Modern systems pre-program the mandrel position based on the last 10 to 20 cycles of sensor data, using a moving average to smooth out random noise while tracking genuine drift.

For non-axisymmetric bottles — ovals, flats, or containers with handles — radial parison programming becomes critical. Rotating the die or using multi-orifice die heads lets you put more material on the wide side and less on the narrow side. Combined with axial control, you can achieve wall uniformity within ±5% on complex shapes that used to require hand-finishing.

Closed-Loop Control Using Feedback from Thickness Sensors

Open-loop parison programming is better than nothing, but it assumes the relationship between mandrel position and wall thickness is constant. It's not. Melt temperature changes, screw wear progresses, and mold cooling shifts over time.

Closed-loop control closes that gap. The sensor measures actual wall thickness on every bottle (or every Nth bottle, depending on cycle time). The controller compares the measurement to the target, calculates the error, and adjusts the mandrel position, air pressure, or extrusion rate in real time.

The trick is tuning the controller. Too aggressive and the system oscillates — the mandrel chases every tiny fluctuation, making the wall thickness worse than if you'd left it alone. Too conservative and it takes 50 cycles to correct a drift that should have been fixed in 5.

A PID controller works for simple cases, but wall thickness in blow molding is a multivariable problem. Mandrel position affects thickness, but so does blow pressure, mold temperature, and even the delay between parison drop and air injection. A model-predictive controller that accounts for these interactions gives better results, though it requires more setup and computational power.

Start with proportional-only control for the first few hours of production. Let the system learn the baseline relationship between mandrel position and thickness. Once that map is stable, introduce integral action to eliminate steady-state offset. Add derivative action only if you see rapid oscillations — which usually means your sensor data is too noisy, not that you need more derivative gain.