OEE for Keg Filling Lines: Lanes, Rejected Kegs and Starved Time

A keg line is slow by packaging standards. A bottling line counts in tens of thousands an hour, and most keg lines count in tens or hundreds.

That changes how you measure it. One rejected keg is visible in the hourly number, and one lane out of use takes a fixed slice of the output for as long as it lasts.

This page stays on the keg line itself. For bottles, cans and the brewery as a whole, see our page on brewery packaging lines.

Measure the keg line at the racker

The washer and filler sets the pace. It is the slowest machine, and the rest of the line exists to feed it and clear it.

So take the line's stops and speed from the racker. Take the good count further down, after the check weigher and the capper, where a keg is ready for the pallet.

The other machines still matter, but as causes. When the racker waits for kegs it is starved, and when the discharge is full it is blocked.

Your software has to tell those two apart from a fault on the racker itself. Our page on starved and blocked machines covers the logic.

Ideal rate: lanes x cycles per hour

A keg racker is not sped up with a dial the way a bottle filler is. Each lane steps a keg from station to station on a fixed cycle: wash stations first, then steam, then the racking station.

A keg spends several cycles inside the machine, but each lane hands out one keg per cycle. So the ideal rate is the lanes multiplied by the cycles per hour.

Example: 4 lanes with a 60 second cycle give 4 x 60 = 240 kegs an hour. The ideal cycle time for the line is 15 seconds a keg.

On a rotary machine, or one with separate washing and racking stations, the slowest group of stations sets the rate. The rule stays the same: kegs out per cycle x cycles per hour.

Three things to settle before you trust the number:

• Per keg size: a 30 litre keg may fill faster than a 50 litre one, but the wash steps often set the cycle. Use the real cycle for each size.
• Per recipe: if you lengthen a caustic or steam step, the cycle grows. Decide if that is the new ideal or a speed loss, and write it down.
• In kegs, not litres: OEE works on kegs and cycle time. Report hectolitres next to it if the brewery plans in volume.

Use the design cycle, not the best shift you remember. The method is the same as in any OEE calculation.

One lane down is a speed loss, not a stop

This is the loss a keg line hides best. A head fails at one station, the operator switches that lane off, and the line keeps going on the others.

The run signal stays on, so the stop log is empty. On 4 lanes you have lost a quarter of the output for as long as it lasts.

Here is how to see it:

• Look for a step in the rate. A steady 180 kegs an hour on a 240 line is three lanes of four. Jams look ragged, a dead lane looks flat.
• Take a signal per lane if the PLC offers one: a lane fault or lane disabled bit. Then the loss has a name and a start time.
• Give it a reason. "Lane 3 out, fill valve" is a performance loss with a cause, and it belongs in the same Pareto as the stops.

Some plants turn it into time instead: 100 minutes on 3 of 4 lanes is 25 minutes of lost line time. The OEE comes out the same, so pick one way and keep it.

Short jams at the capper or on the conveyors land in the same speed loss. See our page on micro-stops for how to split them out.

Where each keg line loss goes

Most arguments about keg line OEE are about which box a loss belongs in. This is the split we would use.

Two rows need a word. Waiting for empties feels like someone else's problem, but it is still lost time on a line you planned to run.

If you take it out of planned time, the OEE looks fine and the keg float problem disappears from the report. Keep it in, and name it.

The second is planned cleaning. Leave the planned length out of OEE, and count any overrun as a stop.

Rejected kegs: the quality loss you can sort

On most lines a reject is scrap. On a keg line, a keg thrown out in the racker has usually lost no beer, because it was caught before the fill.

It still used a cycle on a lane and produced nothing you can sell. That is why it sits in quality.

The causes are few, and each has a different owner:

• Damaged spear or valve: the head cannot seal or the keg leaks. The keg goes to repair.
• No residual pressure: the keg came back open or tampered with, so the machine will not trust it. It is often sent for a manual check.
• Failed wash: a step did not reach its set point, or the keg did not drain. It can usually go round again.

A keg that goes round again is rework. Count it as a reject the first time and as a good keg the second time, so first pass quality stays honest.

Underfills at the check weigher are different, because beer is already in the keg. If one lane gives most of them, you have found a fill valve problem, so log the lane if you can.

Kegs pulled out before the racker, at the external washer, use no cycle. They only cost OEE if the gap they leave starves a lane.

A worked shift, in kegs

Example data, round numbers, describing no plant. A 4 lane racker with a 60 second cycle on both keg sizes, so 240 kegs an hour.

The shift is 480 minutes. Breaks and a planned sanitation take 80, so planned time is 400 minutes and the ideal output is 1,600 kegs.

Now check it the short way. 1,104 good kegs x 15 seconds is 276 minutes of perfect running.

276 of 400 planned minutes is 69.0%, so the two answers agree.

Look at where the kegs went. Stops took 320, and 140 of those were the line waiting for empty kegs.

Lane 3 took 100 more without a single line in the stop log. That one lane cost more than all 48 rejects.

Small counts: report in kegs too

In this example the line makes 4 kegs a minute. A 5 minute wait for a pallet of empties is 20 kegs, and you can see it in the hourly figure.

That is good news for the shift leader. An hourly target of 240 with the actual next to it says more than a percentage does.

It also means sloppy counting shows fast. If the good count is typed in at the end of the shift, a miscount of 20 kegs moves OEE by more than a point here.

So take the count from a signal: the racker PLC, the check weigher, or a sensor on the discharge conveyor.

Where no usable count signal exists, a camera that counts parts can count kegs on the conveyor after the capper. Read how camera-based OEE works before you plan on it.

What to check in OEE software for a keg line

Run these in the demo with any vendor, and with us. Bring one real shift from your keg line as the test case.

1. Ideal rate per keg size. Ask to set 240 an hour for one size and a different rate for another, and see performance follow the product.
2. One lane out. Describe a shift with 3 of 4 lanes for two hours. Ask where that loss shows, and if someone can give it a reason.
3. Starved, blocked, or own fault. Ask which signal tells "no empties" from "discharge full" from a racker fault, and what the operator has to add.
4. Rejects by cause. Ask if spear, residual pressure, failed wash and underfill can be separate reasons, and if the counts can come from the machine.
5. Count point. Ask where the good count is taken, and what the vendor does when the PLC count cannot be reached.
6. Size changeover. Ask how a changeover is timed, and if it runs to the first good keg or only to the restart.
7. Hourly view in kegs. Ask for target and actual kegs by hour, not only an OEE percent.
8. Your data out. Ask for an export of the shift's stops and counts, and check you can rebuild the OEE by hand.

For a wider view of the market, see our review of OEE software.

Where Fabrico fits

Fabrico is a cloud-based OEE and light MES platform, on the web, iOS and Android. Here is what it does on a line like this.

• Stops: they come from the machine signal, through the PLC or a Fabrico IoT device. The operator adds the stop reason on a tablet or phone.
• OEE and downtime: availability, performance and quality, by line and by shift. Micro-stops are detected from the machine signal.
• Losses in money: losses can be shown in money, not only in minutes.
• Counting: the AI camera is an add-on that counts parts. In the standard setup the video is streamed, not stored.
• Follow-up: a stop reason can be set up to open a maintenance task, for example on a capper fault.
• Your data: REST API and webhooks, and any list exports to Excel or CSV.

Production planning and scheduling are not part of Fabrico today. The keg return plan and the filling schedule stay in your planning tool.

The eight checks apply to us too. Ask us to show the one lane out case on your own racker signals before you believe it.

Book a demo or Contact us.

What people ask next

What is a good OEE for a keg line?

There is no figure that fits all keg lines. Keg mix, the state of the returned kegs and the number of size changes move it a lot.

Compare the line with its own last month, loss by loss, in kegs.

Should waiting for empty kegs count against OEE?

Yes, if you planned to run. Keep it inside planned time as starved time with its own reason, so the size of the keg supply problem stays visible.

Is a keg rejected in the washer a quality loss or a speed loss?

We count it as quality, because a cycle was used and no good keg came out. Some plants put it in performance.

OEE is the same both ways. What matters is that the rejects are counted by cause.

How do I set the ideal rate with two keg sizes?

Set one ideal cycle per size, from lanes x cycles per hour for that size. The software should switch the rate when the product changes.

Can I measure keg line OEE without a PLC connection?

You need a run signal and a count from somewhere. A sensor on the discharge conveyor can give both, and a camera that counts parts is another way to get the count.

Do semi-automatic keg fillers need OEE software?

On a one or two head machine loaded by hand, the operator sets the pace. A count per hour and a short stop log may be enough to start.

Want to see your keg line's losses in kegs and minutes? Book a demo or Contact us.