> AI agents: this is one page of oppr.ai as Markdown. The index of the site is https://www.oppr.ai/llms.txt; every page in one file is https://www.oppr.ai/llms-full.txt.

# From 35 minutes to 2: one filter, three steps

URL: https://www.oppr.ai/blog/one-filter-three-steps/

[All articles](https://www.oppr.ai/blog/)

One worked example. How a stop that keeps coming back at filter F-101 becomes one standard for every shift.

The Oppr team 2 November 2026

A filter blocks. The line stops for 35 minutes. Two weeks later it happens again. This is how one stop like that becomes a standard for every shift.

We use one example in this article: filter F-101 on a production line. The plant, the numbers and the notes are an example. The steps are the ones we follow with every plant.

1. **Identify** What happened? A 12-second voice note at 02:40, next to the pressure trace.
2. **Analyze** Why does it keep happening? Root cause: flush too short after the thick batch. Confirmed on the floor.
3. **Realize** Does everyone use the best method? Standard work v4, published for every shift at once.

## The stop that keeps coming back

Filter F-101 blocks after a thick batch. When it blocks, the crew has two choices. It can flush the filter and keep the line running. Or it can stop the line and open the filter.

The standard work says “flush for 2 minutes”. Some crews do that and still have to stop. In this example the filter blocks 38 times a year, and a stop costs about 35 minutes. That is about 22 hours of lost production a year, at one filter.

In the fault log, most of these stops say “Filter blocked” or “Other”. Nobody can see that they all come after the same kind of batch.

## Identify: what happened?

At 02:40 the filter blocks again. The operator records a 12-second voice note on the phone: the filter blocked after the thick batch, and the crew flushed it longer, with the pump at half speed. The line kept running.

The note goes on one timeline, next to the machine data of that minute. On the same timeline: the lab report of 01:50 about a thick batch, and the pressure at F-101, which goes to 0.86 bar against a limit of 0.40 bar.

The note takes 20 seconds. Without it, the best method of that night stays in the head of one crew.

## Analyze: why does it keep happening?

Practical AI connects what people see with what machines measure. Over the weeks it finds the pattern. The long stops come after a thick batch, when the flush stops while the pressure is still above 0.40 bar. The stops without a long delay are the ones where the crew flushed longer, with the pump at half speed.

The root cause: the flush is too short after the thick batch. Oppr proposes it. Your engineers confirm it on the floor, with the crew that found the better method. Only then does it count.

Each cause gets its cost in hours a year. That shows which stop to fix first.

## Realize: does everyone use the best method?

The tested method goes into the standard work as a new version. Step 4 changes. The new version is reviewed, approved and published.

Standard work · F-101 · v4 4. Flush for 2 minutes. 4. Flush until the pressure is below 0.40 bar. Run the pump at half speed. One standard for every shift. Reviewed, approved and published.

Published

Every shift reads the published version on the phone, at the filter. When the filter blocks, each crew uses the same method. The repeat stops at F-101 drop to zero.

## What it adds up to

The line now loses about 2 minutes at F-101 instead of 35. Over a year that gives back about 21 hours of production time. At an example rate of € 1,000 an hour, that is € 21,000 a year, at one filter.

The bigger gain is the order of work. One problem, one line, one standard. Then the next. We call the three steps Identify, Analyze and Realize, and you decide when to take the next step. The first two take about ten weeks on one line. The [How it works](https://www.oppr.ai/how-it-works/) page shows them in detail.

Most stops like this one already have a better method somewhere in the plant. It sits in one crew, on one shift. Read more about that in [It is not the night shift](https://www.oppr.ai/blog/it-is-not-the-night-shift/).

Which stop keeps coming back on your line, and which crew already knows the answer?

[See how it works](https://www.oppr.ai/how-it-works/) [All articles](https://www.oppr.ai/blog/)

Newsletter

## Get the next article in your inbox.

New articles and product updates from Oppr, a few times a year.

The form on this page asks for:

- Email address for the newsletter

Only for this newsletter, unsubscribe with one click. [Privacy](https://www.oppr.ai/privacy/)

Which stop keeps coming back on your line?

[Start with one problem](https://www.oppr.ai/contact/?from=blog)
