Chemicals & Process Industry
From Gut Feeling to a Real Number
How a filling operation found out where its flushing volume really goes.
Starting Point
An industrial company fills different products on several lines. At every product changeover the line is flushed so the products do not mix. A matrix defines how much is flushed. What comes out during flushing can only be resold cheaply.
There were plenty of ideas for reducing the flushing volume. What was missing was proof. Is the matrix even correct? Where does the volume really arise? And which measure pays off first? Before putting money into technology, the company wanted these questions answered. That is where we started.
What We Did
1. FIRST UNDERSTAND THE PROCESS: We focused on one loading bay and documented the filling process step by step: from planning through the flushing operation to lab release. That made clear where which volume arises and who is involved.
2. MEASURE INSTEAD OF ESTIMATE: We evaluated the filling history and determined the flushing volumes retroactively from weighings. Result: the real flushing volume is 18 to 48 percent above what the matrix plans. On top of that comes the lead volume from the filter line, which no system had recorded so far. Because the weighings were made retroactively, we show a range instead of a seemingly exact figure.
3. TAKE THE CAUSES APART: Together with the team on site we collected eight factors that influence the flushing volume. The matrix takes exactly one of them into account: which products follow each other. Fill volume, line length, filter volume, pigging or shift change play no role in the planning. The flushing volume can only be set precisely on one line.
4. CALCULATE THE MINIMUM: From the pipe geometry and a dispersion equation we calculated how much flushing is technically necessary. The gap between the measured volume and the minimum shows the potential in euros. In systems engineering we derived three requirements for a digital product changeover system: detect product changeovers automatically, log every changeover, and let staff at the ramp perform the changeover in two steps.
5. SET THE ORDER: Not every measure can be assessed right away. Extraction of the filter line comes first, because it already works at another bay. Automating the flushing operation follows. Optimizing the matrix only makes sense once the automation delivers real data.
Result
Reliable flushing volumes for one loading bay, as a range with a disclosed data basis.
Eight influencing factors, cleanly separated into considered and not considered.
Three prioritized measures with savings potential in euros and a clear order.
Requirements and a roadmap for a digital product changeover system, all the way to a digital twin.
The biggest open question is stated clearly in the document: measured values are missing to extrapolate to all lines. That also makes clear where data has to come from first.
What You Can Take Away
Do you rely on planning values for a process that nobody has ever measured? Then start there, not with the technology. In the Accelerator Sprint we put the real numbers on the table with you and turn them into a document you can use to decide which measure pays off first.
18–48 %
more flushing volume in reality than the matrix plans
1 of 8
factors influencing the flushing volume is considered by the matrix
3
prioritized measures with savings potential in euros