The Valle Kraft standard

It is not a description of what the mill currently makes. It is the condition of entry, written before the mill is assessed, and it is the same document for every mill that wants to supply that grade. That is what lets one grade name mean one thing across the whole group, including our own machine.

The requirements govern a handful of properties per grade, not everything measurable. A specification that demanded the best figure on every property at once would be a specification nobody could hold, and a bar nobody clears is not a standard, it is a marketing document.

Most mills publish typical values, because a single mill can simply describe its own machine. A typical value is an average, and an average is something a given reel is free to miss. We publish the floor instead. It is a worse number and a better promise, and it is the only kind of number that can mean the same thing across machines we do not own.

Where the numbers come from. Paper in these grades ends up in an enormous range of converting operations, and each one finds its own way to fail. A sack that splits on a filling line. A liner that crushes three pallets down in a warehouse. A wrap that picks up moisture and blocks on the reel. We set each requirement at the level that makes the grade work across that whole range of uses, rather than at the level a particular machine happens to reach.

That is why a bar does not move when a mill cannot meet it. The number came from what the paper has to survive, not from what a machine would prefer to produce. We will tell a mill exactly what the requirement is. We will not tell it how close it came, because a bar you can creep up on is a bar that gets aimed at rather than met.

The published minimums for each grade are on the register. The full document, including the properties we track without enforcing, goes to a mill under qualification and to a buyer who asks for it.

See the grade register

• One document per grade. Issued under a code that is quoted in all correspondence and stays fixed across revisions.
• Reviewed, and revised as a whole. A new issue supersedes the previous one in full, so there is never a question about which version is current.
• Never lowered for one supplier. If a mill cannot reach the bar, the answer is that we do not sell that grade from that mill.
• Conditions fixed, not just values. Testing is under ISO 187 conditions, with all tensile work to ISO 1924-3 at 100 mm/min. Results from a different method are not comparable and are not accepted.
• It binds us too. Our own machine near Cali is qualified against the same document as everyone else's.

On any machine running recovered fiber, the variation between reels made on different days is far larger than the variation across a single reel. That one fact decides how a qualification should be sampled, and most sampling gets it backwards. A plan that maximises the number of specimens taken from one reel measures the wrong thing very precisely.

So twenty samples cut from a single reel answer a question nobody asked. They tell you how that one reel came out. The request is small and specific instead: one sample from each of eight to ten different reels, from as many different production dates as possible, spread across recent months, and where the mill can identify them, from different recovered fiber purchase lots.

They come from finished rolls in the warehouse. Nothing is taken off the machine and no production is interrupted for this, which matters when you are asking a favour of a mill that has not yet sold you anything.

Detailed cutting, labeling and packing instructions go out with the request. Where it is easier on the mill, we arrange for a third party inspection company to attend and take the samples itself. That is ordinary practice for this kind of qualification and it takes the work off the mill's people entirely.

• Different days, not different metres. Spread across dates is what makes the result mean anything.
• Fiber lots matter most. On recovered furnish the incoming material is the dominant source of variation, and it moves with whatever the mill bought that week.
• Nothing folded. A crease ruins a tensile and a tear specimen, so samples travel rolled or flat and wrapped against moisture.
• Labeled by hand is fine. Production date, machine, shift, grade, target basis weight and a roll number where one exists.
• We pay. Freight and the full testing cost, at our expense, before the mill has earned anything.

• Three production runs, and we define what one is. Separate calendar days, at least seven days apart, and different recovered fiber purchase lots wherever the mill can identify them.
• A certificate with every delivery. Mill test data accompanies the paper, every time, not on request.
• Written notice before a change. Furnish source, sizing or coating chemistry, and major equipment. A silent change is a breach in its own right, whether or not the paper drifts.
• A retained sample from every delivery, held for two years. It is the only way to settle a complaint that arrives late, or to establish when a drift actually began.
• Our test governs. We test independently whenever we choose to, and where the two disagree the independent result is the one that counts.
• Non conforming deliveries are rejected. Not discounted, not accepted with a note on file.

Why the run definition is written down. A mill with thin records will happily offer three runs that were Tuesday morning, Tuesday afternoon and Wednesday. That satisfies the words and captures none of the variation the requirement exists to catch. The fiber lot condition is the one that matters most on recovered furnish, because the incoming material is the dominant source of variation and it moves with whatever the mill bought that week.

Ten determinations per reel, taken across the full deckle, is the minimum for a figure to count. A single measurement from the middle of a web says very little about the edges, and the edges are where a converter finds the problem.

Test methods are named against each requirement on the grade register. Where a mill's own laboratory uses a different method for the same property, the specification method governs.

Every partner mill has a legitimate reason to wonder whether the testing quietly runs in our favour. The first time we reject one of their loads, that suspicion is the story that gets told about us, and it would be a reasonable story to tell.

So the regime runs on our own machine first, and it runs in the open. The same accredited independent laboratory that tests a candidate mill tests ours. The same sampling, taken across production dates by someone who does not work for us. The same certificate on every delivery, against the same published bars. Mills in the program can see how our own machine scored.

A supplier who can see that we hold our own asset to the identical number has nothing left to argue about. There is also a structural half to this: the more the Yankee runs grades the partner mills do not, the less overlap there is to be suspicious of in the first place.

We do not obscure the mix. Some of what we sell comes off our machine. Most of it does not. All of it ships under the same specification and the same certificate. A buyer who assumes we made everything and then works out the sourcing structure will not forgive it, and it is trivially easy to work out.

• Same laboratory. Not our quality room, and not theirs.
• Same sampling protocol. Across production dates and fiber lots, on exactly the terms we impose on everyone else.
• Same certificate. Every delivery off our own machine carries what every delivery from a partner mill carries.
• Results visible to the program. A mill under qualification can see how our machine performed against the bar it is being held to.
• No exemption when we miss. Our machine goes to tightened inspection on the same trigger as anyone else, and it can lose a grade the same way.

None of this is generosity. A specification enforced by a party that exempts itself is not a specification, and a partner mill would be right to treat it as a negotiating position rather than a standard.