Swabbing Is a Job-Change Problem, Not a Shift Problem

It's 02:30 on a Tuesday and bird-swing is climbing on section 4. The operator re-swabs, the rate drops, and forty minutes later it's back. You call it a swabbing problem. It isn't. It's a job-change problem wearing a swabbing problem's face, and the line will keep telling you that story at every run until you fix the protocol.

The first hour after a job change is where your monthly numbers go

Most container glass plants I've audited treat swabbing optimisation as a frequency question. How often does the operator swab each cavity? Is the interval consistent across shifts? Those are useful questions for steady-state production. They're the wrong questions for the window that actually drives your Hot-End Rejection Rate.

In the first 45 to 60 minutes after a job change, mould temperature is still hunting, gob weight has been reset, and nobody has formally updated the swabbing interval from the outgoing job. HERR at steady state runs 2-5% on a well-managed line. In that first post-job-change hour, when swabbing is treated as a general shift task rather than a job-change-specific procedure, I consistently see 8-14%. On a 200 t/day furnace at peak job-change frequency, that's 30-80 tonnes of ware rejected and culletted per day. Not a swabbing discipline problem. A governance gap.

In 2016, I was running a three-line plant with a mixed fleet, one line a newer Emhart 10-section IS machine the forming team was proud of. The operators had swabbing sorted, or so everyone believed. What they had was a steady-state habit that looked clean at hour three and fell apart in the first thirty minutes of every job change. We fixed it by treating the swabbing protocol as a job-change document with named owners, not a shift instruction. First-hour HERR on that line dropped by more than six percentage points inside eight weeks.

Three variables reset simultaneously at every job change, and most plants document none of them

Blank mould temperature is the starting point. The target for grey iron moulds in Blow-and-Blow is 480-520°C. For stainless steel NNPB work, it's 440-480°C. Deviate more than ±30°C from those targets at first-gob contact and you'll see check defects or split-neck failures within the first 50 containers. Not 500. Fifty.

The swabbing interval is the second variable. If the outgoing job ran on a 10-gob cycle and the incoming article is a lighter NNPB bottle, say a 250 g wine vessel replacing a 400 g spirits container, the interval must be shortened to 6-8 gobs per cavity until mould temperature stabilises. That window is typically 20-35 minutes into the run. Carrying the prior job's interval into the first hour of a new run is the single most common root cause of bird-swing spikes I find during a hot end audit. Bird-swing rate climbs +3-7 defects per thousand containers in the first 15 minutes of a job change when compound viscosity grade isn't matched to the new gob weight.

The third variable is compound grade. Heavier gobs need higher-viscosity graphite-oil compound. Lighter NNPB jobs need lower-viscosity, faster-volatilising formulations to prevent parison adhesion and carbon smear on the shoulder and heel of the ware. Rush the swabbing before mould temperature reaches equilibrium and you get a 0.3-0.8 percentage-point HERR uplift through the first production hour. All three, temperature, interval, and compound grade, reset simultaneously at every job change. On most lines I audit, exactly zero of them are documented in a job-change-specific swabbing procedure.

And here's the equipment trap that catches plants off guard. Auto-swab systems, whether Emhart Glass AutoSwab or Heye SprayLub, are programmed to deliver compound at a set interval per section per cavity. At every job change, the interval register in the HMI must be manually updated before speed ramp-up (and yes, I know your auto-swab log shows green, check the register anyway). Across every plant audit I've run, this is consistently the most commonly skipped step in IS machine job-change checklists. When nobody updates it, the system runs the prior job's interval on new mould geometry indefinitely. The supervisor reads a clean log while the ware goes off-quality.

The interval the operator swabs at hour three tells you almost nothing about your swabbing performance. The first 20 minutes after a job change is where the ware quality is made or lost.

Europe's cullet push is rewriting the swabbing calculation

Germany's VerpackG mandates a minimum of 60% post-consumer recycled cullet in single-use glass containers from 2025. Every additional 10% of cullet in the batch lowers feeder-exit glass temperature by approximately 3°C and alters viscosity at the blank. FEVE reported approximately 21 million tonnes of container glass produced across European member plants in 2023, with average cullet input of 52% and Germany already above 80%. Plants running different cullet ratios across successive jobs, which is most of them, will see a measurably different mould-temperature equilibrium from one job change to the next. The swabbing interval that worked on a 40% cullet job doesn't transfer to an 80% cullet job without recalculation.

Verallia is one of the few operators that has built the infrastructure to track this properly. Their 2023 Annual Report disclosed IS machine OEE of 87.3% across their European fleet and confirmed that swabbing-related stoppages are coded as a distinct downtime category in their MES. That data-capture discipline makes job-change-specific swabbing-loss attribution possible. Most Middle Eastern and North American plants are still running paper swabbing logs that nobody correlates with the job-change record. The loss stays invisible right up until a VP of Operations asks why first-hour HERR is running at double steady-state.

Under EU ETS Phase IV, container glass plants emitting above their product benchmark are purchasing EUA allowances at rates that reached EUR 60-65/tCO2 in early 2025. Uncharacterised swabbing-compound combustion is a contributing input to that exposure. It's not the biggest line item, but it's a variable with a documented compound-grade and interval basis. A vendor-neutral container glass consultant with hot-end audit capability can quantify it in a single site visit and put a number against it that holds up in an ETS compliance review.

What a job-change swabbing protocol actually contains

Look, the compound grade conversation sounds like a chemistry problem, but it's really a job-change governance problem. The spec exists. The grade selection logic exists. What's missing is the document that makes both mandatory at the moment they're needed, every job change, every shift, regardless of who's on the floor.

A job-change-specific swabbing protocol is not a paragraph in the shift log. Two pages, at most. But those two pages contain:

• Blank-side and blow-side target mould temperatures verified by calibrated contact pyrometer before first-gob contact, with a hard reject threshold for any cavity outside ±30°C of target
• Compound grade selection tied to incoming gob weight and mould alloy, documented in the job recipe alongside temperature and timing specs
• Swabbing interval for the first 30 minutes of run, explicitly distinct from the steady-state interval
• Auto-swab HMI interval register update confirmed by the Section Technician before speed ramp-up
• Dual sign-off from the Section Technician and the hot-end superintendent before any production ware is released to the lehr

The hot-end superintendent owns recipe lock in this sequence. The operator does not change set points or compound grade without sign-off. That governance structure is exactly what the Job Change Tool is built around, specifically stages five through seven of the 9-stage Job Change Lifecycle: recipe load, ignition, and first ware. Each stage has a named owner and a gate the next stage can't open until the current one closes. There is no version of this that works as a shift-level habit. It works as a job-change governance document or it doesn't work.

If you want to benchmark where your current swabbing performance sits relative to what's achievable, a forming audit covers compound selection, interval discipline, and mould-temperature management across every section over a five-day observation window. That's enough data to see the cross-shift variance pattern clearly and to build a job-change swabbing protocol that holds across all shifts.

Swabbing isn't where container glass plants lose the most ware at job change. It's where they lose ware most consistently, run after run, shift after shift, because nobody built a protocol around the transition. Fix the protocol. The shift discipline follows.