Focus on wieghing systems: Don't give it away

The challenge of filling to average weight while reducing product giveaway is one manufacturers continue to grapple with. Lou Reade reports

Try for a moment to imagine what 11t of bacon looks like. Think about ordering it, paying for it and shipping it into your factory then simply giving it to customers for nothing. "That's how much one customer was giving away because of over-packing," says Colin Platt, a systems applications specialist at Marco.

Weighing out precise quantities of food into packs, bottles, ready meals or even sandwiches is, quite literally, a delicate balancing act. On the one hand, legislation (or a customer) dictates that a certain minimum quantity is required; but, in satisfying this demand, there is a real danger of going over the target weight and giving away free product.

Platt says that Marco is a 'productivity expert' saving other people money by boosting their return on investment.

"It's all about yield in every key performance location in the factory," he says.

He cites product giveaway as a key issue of manufacturers. In the bacon example, sandwich maker Greencore was producing BLT (bacon, lettuce and tomato) sandwiches that included a 30g portion of bacon. This amount was not legally enforceable but was stipulated by the customer. The sandwiches are put together by hand: a portion of bacon is weighed, and the operator gets a green light if it is within the correct weight range. Either side of this is a 'red zone' for portions that are too light or heavy.

The 30g portion could vary either way by 10%, but Platt says that this quickly led to overfilling and product giveaway."When you have a weight range, you find that the people filling the sandwiches tend to work towards the top end," he says. "They're consumers too, so are often tempted to use a bit extra."

Greencore found that it was putting an average of 32.6g of bacon in each sandwich: an overfilling of 8.3%. The extra 2.6g would be undetectable by the average consumer, but over the course of the year it would have created the 11t bacon mountain.

Marco's answer was to install its A/O software, which forms a part of its yield control module. The system helps to ensure that the average weight of bacon in each sandwich is pulled down as close to 30g as possible.

An internal database will 'police' each weighing unit. It looks at each packer and may notice if one is working at the top end of the scale and automatically adjusts the boundaries of the green zone."The top end could be re-adjusted to 32g, for example, so that anything above that gets a red light," he says. "This forces operators to use a slightly smaller amount."

Taking away specific data on weights, and having a red or green light system and keeping tabs on the average filling for each sandwich helps to keep 'loss of product' to a minimum.

"It's the closest we get to treating the operator as a filling machine," he says.

Filling machine issues

With real filling machines, whether they are dispensing liquids or solids, the issues are different. Here, the important factor is to understand why a machine might be dispensing too much or too little product and taking the right course of action to correct it.

Most food is now sold on an average weight basis, so the onus is on manufacturers to ensure that only a small percentage of their products fall below the stated weight. In the UK, the rules state that the average output of a batch must be at the stated weight with a small number allowed to be underweight (see panel below).

Machines, like human operators, have a natural variation. Some work within very tight tolerances, time after time. Others are less reliable. Line operators, with help from technology, must understand the causes behind why a line is over- or under-dispensing and take the correct action. "The average operator has what I would call a: 'Don't just sit there: do something' attitude, and will go to the filler and tweak it," says Platt. "The opposite attitude which is 'Don't just do anything: sit there' is much better."

He says that assessing standard deviation a measure of the repeatability of a certain machine is necessary to identify the real reason for an incorrect weight. Rather than the dispenser being set incorrectly, it may be that a dispenser head needs cleaning.

Bob Witney, trading law consultant at NSF-CMI has a similar attitude. He says it is important to treat machines as individually as employees and the natural unpredictability of each one must be understood if it is to be as productive as possible. Take the case of two production lines next to one another. There is a temptation to use the same control chart for both of them but this is a mistake, he says.

"Control lines are based on the performance of equipment," says Witney. "You must find out the inherent variation of your machine. If its standard deviation is 5g, you use that to work out where your control and action lines are set but you have to test it."

Accurate, well-kept records are also important, but he says there is still an ongoing legacy that hinders this. He comes from an inspection background, and says that the notion of reporting that a line is 'out of control' is anathema to most quality controllers which requires a change in philosophy.

Historically, he says, if a production was running out of specification, it would be adjusted before a quality control sample was taken. But, in the principle that you learn through your mistakes, he says that 'out of control' incidents actually help to pinpoint underlying problems within a production machine.

"It could be that the 'wrong' results being reported are actually masking the need for equipment maintenance," he says. "The last thing you want is for the production line to stop altogether."

Imperfections in machinery can manifest mistakes in many ways: multi-head fillers, for example, can be put off balance if just one of the heads is set incorrectly.

"You need to check each head to make sure they're all consistent then you can run them all off the same controller," he says.

Another example is bread dividers, which have a blade to cut up dough. This can lead to complications because dough is measured in volume, while bread is measured by weight. If the blade is slightly blunt, the dough will not be cut into equal portions, so some of the bread may end up being under-weight.

"In a case like this, you could use the results to do something about this piece of equipment," Witney says. "It is counter-productive to hide bad results. Perfect records are not perfect."

Pretty average: A guide to weight rules

Since 1980, most foods in the UK have been sold (and packed) according to average weight, rather than minimum weight. Dispensing an average weight depends on three rules:
The average batch must be at the declared quantity or above
2.5% of samples can be below the agreed tolerance (of 1.5%)
None can be below twice the tolerance. For example, for a 1kg pack, one out of every 40 samples can be below 985g, but none may be below 970g.

FIXED WEIGHT TREND

Accurate weighing is critical to supermarkets as they try to keep costs to an absolute minimum so packers that can achieve fast, accurate weighing can be sure of winning their business.

One growing trend in fresh meats has been to move away from 'check weight' where meat is sold close to a certain weight, which means that the pack price will vary towards 'fixed weight', where each pack weighs the same, and costs the same. "Fixed weights are easier for supermarkets, because they only need to print one set of labels," says Torsten Giese, UK-based marketing manager at Ishida Europe. "In this case, you do not need to print an exact weight and price each time."

There is also a perceived quality improvement, as customers are less likely to scrabble through the shelf searching for the cheapest pack.

In common with many other products, these will be weighed by depositing cuts of meat onto various hoppers, then choosing the best combination of three or four of them to make an exact weight for the pack. If there are 14 hoppers, that can mean choosing between 10,000 combinations, he says. He says the balance for a job like this is between speed and accuracy: "The challenge is getting the products into the hoppers at the right speed, and getting a consistent product feed."