Sack 173 of 200: Picking Twelve Bags From Eight Pallets

Two hundred sacks arrived on eight pallets. The plant sampled four, from the top layer of the first pallet, and all four passed. Sack 173, on the bottom of the last pallet, had been wetted in transit and held a solid cake. It reached the mixer eleven days later and blocked the feed screw for two hours. The sampling plan was not wrong about the four sacks it tested. It was wrong about which sacks it tested.

Sampling is the cheapest quality control a plant can buy, and it is usually the first thing to be reduced when receiving is busy. The fix is not more samples; it is a plan that covers the pallet, the sack and the lot, in that order.

How many sacks

For a lot of 200 sacks, twelve is a reasonable number, taken from at least four different pallets. That is roughly the square root of the lot size plus a few, which is the usual starting point for attribute sampling when you have no history with the supplier.

If the supplier has a good record over twenty lots, six will usually do, provided they come from different pallets and different heights. If the material is new or the supplier is new, go to twenty and treat the first three lots as a qualification exercise rather than a purchase.

The number matters less than the spread. Twelve sacks from one pallet tells you about that pallet. Twelve sacks from six pallets tells you about the delivery.

Where on the pallet, and where in the sack

Water enters a pallet at the bottom and at the edges. Sample the bottom layer of at least two pallets, the top layer of at least one, and a sack from the middle of the stack. If any pallet has a torn or damp outer wrap, sample that one specifically; it is not a random sack any more, it is a suspect.

Within the sack, sample the top, the middle and the bottom with a thief probe, and combine them in equal parts. Fines migrate downward, so a top-only sample reads coarser than the sack. Moisture also stratifies, especially if the sack has been near a cold wall.

Do not sample from the sack that was dropped, the one with a stain, or the one already open at the stitching. Those are the sacks you reject on sight, not the ones you use to judge the lot.

Composite or individual

A composite tells you the average. Individual samples tell you the variation, and variation is what causes plant problems. Keep both: test the composite for the certificate parameters, and test each individual sack for the two or three numbers that matter most to your process.

Split every sample into three. One goes to the lab, one is retained sealed for twelve months, and one is sent to the supplier if there is a dispute. Record the sack number, pallet number and position on the label, because a retained sample without provenance is a bag of white powder.

Sample mass should be at least 500 g per sack, and a full kilogram if you will run sieve analysis, bulk density and absorption on the same material. Splitting a 200 g sample three ways produces results that disagree for reasons that have nothing to do with the polymer.

What to test, in what order

Start with the cheap tests. Visual check for colour, contamination and caking. Moisture by balance. Sieve distribution on three screens. Bulk density, loose and tapped. Those four take about forty minutes for a full set and catch most of the problems that reach the plant.

Then absorption: free swell in 0.9 percent saline at 30 minutes, and absorbency under load if your process applies pressure. Run them on the composite, and on individual sacks only if the composite is out of specification.

Residual monomer is a send-out test unless you have the equipment, and it usually only needs to be run on the first lot from a supplier and then annually, unless the application is regulated.

Acceptance limits that survive an audit

Write the limit as a range with a method, not a single number. “Free swell in 0.9 percent saline at 30 minutes: 240 to 320 g/g” is auditable. “Good absorption” is not. Add the sampling plan itself to the quality manual, with the number of sacks, the positions and the splitting routine.

Decide in advance what happens when one sack fails. The usual rule is: re-test from twice as many sacks in the same area. If the re-test passes, accept the lot and note the anomaly. If it fails, reject the pallet, not the delivery, and ask the supplier what happened on that pallet.

The sack that started this story should have been caught by the visual check before anyone opened it. A wet pallet has a stain, a soft corner, or a wrap that has been re-taped. The incoming inspection of SAP lots checklist covers the receiving step, and it is worth pinning to the wall of the goods-in office rather than filing it.