For food manufacturers, quality is top of mind for every product. Delivering what they promise in terms of quality and quantity in their packages requires constant diligence. In fact, for many food product manufacturers, the rule of thumb is to err on the side of overfulfilling on those promises, even if it means accepting extra costs.
However, rather than accepting the status quo, a global food product manufacturer wanted to know whether a statistical process control (SPC) or quality intelligence system could make a measurable difference to its bottom line. They decided to analyze the product and process data they were already collecting to find out.
Problem: Extracting Meaning from an Abundance of SPC Data
The company called in InfinityQS® to perform an onsite assessment on one of its many Northern American sites. With just two days to evaluate data from an operation that ran 24 hours a day across three shifts and more than 200 food products, would the company be able to find any way to reduce scrap, waste, defects, or costs?
Would the analysis reveal opportunities to improve efficiency—or profitability?
Proposed Solution: Prioritize the Biggest Win—Net Contents
As weight is a primary cost center for food manufacturers, the company asked InfinityQS to focus our initial assessment on weights. For each of the company’s top 14 product codes, we gathered—
- Means and standard deviations
- Specification limits
- The next year’s projected production rates
- Raw material costs (from the company’s existing ERP system)
InfinityQS SPC solutions like Enact® (in the cloud) and ProFicient™ (on premises) provide quality intelligence by enabling companies to roll up aggregated data like these, for flexible, complex analyses and visual representations. In this case, we looked at a year’s worth of data across all products in a box and whisker chart.
This chart enabled the manufacturer to compare its top products—all of which were running within specifications—to pinpoint opportunities for waste or scrap reduction. Product E, for example, seemed to consistently run heavy over the course of the year: a natural place to begin a waste-reduction effort. That product also displayed extreme variances in production.
Next, we looked at each product, month by month, over the course of the year. By doing so, the company could discern variance patterns that could indicate—
- A need for more regular maintenance of equipment
- Changes related to operational patterns or staff behavior
- A change in raw materials or procedures at a certain point in the year
What happened in months 5, 6, and 7? Could the answer lead to a way to bring the product closer to target for all months? Fortunately, the quality team didn’t need to guess: They could simply look at the data.
Diving into weekly data revealed more granular details. Weeks 2 and 4 showed unusual variations above upper spec limits, with several other weeks remaining in spec but drifting upward.
But another piece of the puzzle was missing. Given the data, could this manufacturer reduce variations? Was such a thing possible? Another chart indicated that it was.
The production line was capable of limited variations during the manufacturing of this product. The question now was, knowing that the equipment and processes were capable of improvement, how could the variation be reduced?
Result: Strategic Process Control Saves Millions
We created several what-if scenarios:
- How much could the company save in raw material costs if mean weights were reduced just 5%?
- What if instead, the company reduced standard deviations by 25%?
- Further, what if standard deviations were reduced 67%?
We determined that a mean weight reduction could produce an annual raw material savings of $160,000. Impressive—but also dangerous. Based on the amount of variation that the quality data revealed, even a 5% reduction could produce some underweight packages, which can cause a great deal of trouble for manufacturers in the food and beverage industry.
Although the causes of the variations we saw were outside the scope of our onsite assessment, we knew that the potential existed to reduce variability by as much as two thirds, without risking underfill. By rooting out the causes of variation and reducing the standard deviation by 25%, the company could save $1,100,000 in annual raw material costs—without the danger of dropping below its lower spec limit.