In order to improve yield within a manufacturing facility, one must first create the belief that there is more opportunity to realize. The following two case studies show two different ways we created this belief in organizations that needed to realize significant opportunity:
Discovering previously unknown sources of opportunity
Showing that previously perceived “impossible” problems can be realized
Discovering Previously Unknown Sources of Opportunity
In yield improvement, as well as life in general, we don’t deal well with unknowns. We have found that simply shining a light on various hidden opportunities and their magnitudes can shift an entire organization away from “analysis paralysis” and towards meaningful action.
In the following example, we performed a yield analysis at a food processing plant where the overall size of the yield loss was well known: 15MM lbs of ingredients per year were not making it into the final product. The makeup of 60% of these losses was known and measured. However, 40% of the losses, or about 6MM lbs, were not being directly measured. When asked about where this 40% came from, leaders within the organization shared multiple different beliefs including:
Fill Weights – “The missing pounds could be in our overfill, because we know there is some error in the measurements which we have proven with data”
Water Loss – “We cook our product… we must be cooking off some of the water but I’m not sure how much”
Batching Accuracy – “Our batching system may not always be weighing ingredients perfect and could be a source of error”
Delivery of ingredients – “We may be getting shorted when ingredients are delivered”
Without a known cause for the missing opportunity, the organization confined its realization efforts to the 60% that was known. As several previous efforts to identify the missing opportunity had failed, there was significant organizational resistance to even looking for it again.
To get over this mental hurdle, we applied some quick math and logic to the perceived sources to show that it was not likely that the majority of the opportunity sat within any one of them. Looking at the fill weights, they would have to be four times as big to account for the missing pounds. For the water loss, they would have to be losing five percent more than expected (which is a huge gap).
The only source large enough to make up for the unknown pounds was the difficult to measure and almost impossible to see waste stream leaving the plant.
Figure 1: Overall mass balance of plant
To uncover the magnitude of the missing pounds, we performed a plant-wide mass balance. This meant determining everywhere the product could enter and leave the plant. Knowing where to look was only half of the battle, the other half was being able to measure it accurately. The plant leadership had frequently floated the idea of adding expensive flowmeters to the waste piping. That way the plant would be able to accurately measure all of the waste streams. Unfortunately, this never came to fruition because of the cost and complexity to do so.
We challenged the assumption that we needed this intense, expensive level of measurement to gain insights on the missing waste. To make an accurate approximation of the waste streams, we used a combination of various waste reports from numerous sources along with spot measurements throughout the week. This was a meticulous exercise in data analysis that rapidly and accurately mapped out the various magnitudes of waste. A diagram of this analysis is shown below.
Figure 2: Detailed mass balance of streams leaving the plant
What mattered most in completing the plant-wide mass balance was the percentage of each of these waste streams that was “solids.” Hunting down the solid’s percentage and volume of each of the waste streams (often piecing numerous data sources together), we were able to determine the average amount of solid product going through each of the waste streams.
This uncovered that the amount of product flowing through the ECF processing plant and to the city almost exactly matched up to the missing yield pounds. This then allowed the plant to have the full picture of where their wasted product was going.
Figure 3: (Left) Previous mass balance of product with 40% unknown (Right) Updated mass balance of product with unknown streams identified as “Solids to LCL” and “Drains/Silo”
Once the plant understood this, they were able to focus on waste realization with the full confidence that they were working on the largest priorities. Over the following 18 months, the plant leadership drove waste from 10% to less than 4%, saving millions of dollars. Beyond the financial results, the combination of realizing an opportunity that had persisted for years and freeing up the resources that had been focused on it allowed the plant to vigorously tackle their next largest opportunities.
Highlighting Opportunities as Realizable
The second common result of our approach to yield analysis is proving that “impossible” problems are actually possible.
In this case, the yield loss of a facility totaled about $10MM. With significant cost pressure on the facility and the company as a whole, quickly realizing some of this opportunity was a high priority.
Figure 4: Yield loss opportunities by line (J3, L5, G2, etc) and material (Gel, Flex WIP, Swe, etc)
The challenge for this facility was the perception that most of the losses were an inevitable part of the process. Yes, there was $10MM in total opportunity and its location was quite well known, but most of this $10MM had stopped being viewed as an opportunity.
Many factors can build an emotional wall in front of realizing opportunities: trying and failing at realization in the past, previous prioritization towards other areas, and even the routine of it “always being that way.” In our analysis, we realized that it was primarily necessary to tackle this emotional barrier over refining the valuation and location of specific opportunities.
You can’t always just show the data. Often times you have to speak to the heart about realizing opportunities.
In our yield analysis we came up against multiple problems that had been deemed “impossible,” To drive a wedge between the impossible and possible, we focused on showing insights that would sow doubt on an opportunity’s “impossibility.”
An example best illustrates this situation.
On the filling line, the first 15 packages leaving the filler would be automatically thrown away whenever the line restarted from a stop. It was a part of the process, and was even put into the machine’s coding.
Figure 5: Filler line rejecting the first 15 products after every stop
Nobody was sure exactly why the packages were being automatically rejected, but there were a couple main beliefs:
The packages starting underneath the top heaters on a restart weren’t getting reheated properly and thus weren’t fully sealing.
The packages stopping next to the side sealers were overheating and shrinking, causing underfilled packages.
Our first step was to test each of the 15 packages that were being rejected and understand if they would fail a quality check, and why. We discovered that only the first two bags in the top sealer section and the three directly under the filler nozzles were consistently failing the quality checks.
Figure 6: Rejected (red) product based on quality tests when restarting the filling line
The ones in the top sealer had an incomplete seal, while the ones under the filler nozzles weren’t completely filled. This led to two immediate actions:
Changing the programming to only reject the packages that were consistently failing the quality test. This immediately reduced the waste by 66%.
Breaking the previous “15 packages wasted at the filler” waste type into two, actionable problems to solve:
Ensuring packages from under the top sealer had a complete seal after line restart
Ensuring packages from under the filler were totally filled after line restart.
In this case, without doing any detailed problem solving, we proved that a previously “impossible” problem could be improved. This was the first of several “impossible” problems we tackled to show insight that created doubt in the impossibility. Some of the others included: improving recipe target consistency between products, reusing non-destructive quality checks, and reducing changeover times between high value products.
These improvements changed the attitude towards impossible problems in the plant. Most importantly, leadership was reinvigorated to lead the charge in realizing various yield opportunities previously taken off the table.
Like this article? You might like: