Even on a modern manufacturing line equipped to collect and analyze process data, something can always slip through the cracks to derail quality and efficiency.
This is only to be expected. Take the modern automobile. A vehicle on the road is the end result of a vast and complex supply chain. The automaker may be operating its own stamping, engine and transmission assembly plants in addition to working with suppliers of everything from the airbags to the seats and the infotainment systems. Such complexity breeds a wide margin for error.
These hundreds of parts, systems and subsystems come together at the assembly plant. Regardless of whatever quality assurance processes these various components underwent where they were made, additional testing and inspection during this final assembly is crucial.
Process monitoring of specific operations like welding, crimping, torqueing and dispense can ensure each cycle of a specific process is completed within acceptable parameters. Machine vision inspection systems, leak testing, and end-of-line functional testing provide additional assurance.
The risks of human error remain
But what if a key assembly process still rests entirely with a human being manually completing a task? There is no obvious data trail, no process monitoring in real-time, to shout “error” when this task is done wrong or not at all.
It may be hard to believe this could still be the case on a modern manufacturing line equipped with digital sensors and instrumentation, with all that collected process data constantly correlated and analyzed to catch production problems as they occur.
And yet, it can and it does. This is why we so often talk about continuous improvement with our customers. Industry 4.0 isn’t a sprint, but a marathon. It comes back to the old Pareto Principle — first focus your effort, and your budget, on the 20 percent of the trouble spots that will fix 80 percent of the problems. Once you have that working well, triage how best to tackle the remaining 80 percent. Grease the squeakiest cogs in the machine, first.
One of these problematic cogs that one of our automotive customers had not yet tackled related to head gasket installation at its powertrain plant. The head gasket provides the essential seal between the block and the cylinder head of an internal combustion engine.
Missing head gaskets – a true story
If an engine doesn’t have a head gasket, it will ooze oil. At the customer’s powertrain plant, a human was responsible for placing the head gasket on each block, without the benefit of any kind of quality system to catch an omission. The initial seal of metal on metal between the block and cylinder head was tight enough for the assembled engine to pass the subsequent leak test and engine cold test.
The problem didn’t begin until the vehicle was on the road. When the engine heated up for a sufficiently sustained period, oil started oozing from the crack.
Such an incident happened a few years ago. A consumer brought their recently purchased vehicle back to the dealership, complaining of an oil leak. Tearing down the engine revealed the absence of a head gasket.
This left the automaker facing some uncomfortable questions: Why is this gasket missing? How many other engines have the same flaw?
Without a usable data trail, the automaker would have had no way to know. The only option would have been to err on the side of caution and issue a recall that could affect tens of thousands of consumers. This would most certainly have left the company’s reputation with a black eye, in addition to a financial impact that could easily be in the range of $5,000 to $10,000 per vehicle.
Millions in recall costs avoided
This automaker, however, had deployed modern data management and analytics software in its powertrain plant. It did have all the data related to those leak and cold tests, neatly serialized. That made it easy to pull up the birth history record for that first engine that had come back to the dealership with the missing head gasket.
Upon reviewing the engine’s leak test data, quality engineers did spot an anomaly in the digital process signatures from the test. Not enough for a fail at the leak test’s current parameters, but substantial enough that a new algorithm could be scripted to review several months of engine production to see if any others exhibited the same tell-tale sign of a missing head gasket.
The outcome? Eight engines in total displayed the same anomalous leak test signature. Most were in vehicles that had not yet made it to consumers. Doing a focused recall on just these vehicles and tearing down these engines revealed that they too, were missing head gaskets.
How much time and manpower did it take to trace this problem to eight specific engines? One person in four hours.
The lessons learned
A large-scale recall that could have cost tens of millions of dollars was averted. The automaker also learned some important lessons in support of the continuous improvement of its operations.
First, how to tweak leak test parameters to better highlight the anomaly that pointed to a missing head gasket.
Second, the need for a visual indicator — the automaker worked with the gasket manufacturer to add a tab to the gasket that would be visible outside the joint of the engine block and cylinder head after assembly.
Lastly, the importance of secondary verification that could be saved for future reference in the digital birth history records for those engine components — the automaker installed a camera verification system to ensure correct gasket installation.
Let’s also look at this another way. While being able to trace the problem to only eight engines did save the automaker millions, it doesn’t take that scale of averted disaster to justify the investment in a modern data management and analytics platform. The time, expense and general fuss and bother of dealing with just eight faulty engines can still add up. Averting those costs by investing in a data management and analytics platform is a better investment than dealing reactively with issues after the fact.
Cost versus value: It’s a no-brainer
Industry 4.0 isn’t just about collecting data but also making effective use of that data on a continuous basis to systematically identify and address the blind spots that can lead to costly production issues down the road. It does not matter if you are producing automobiles, agricultural equipment, household appliances or medical devices.
Before focusing on the price tag that may be attached to a data management and analytics platform that can improve traceability, root cause analysis and defect avoidance on your line, consider the cost of the common quality issues this kind of system can prevent or, at the very least, mitigate. Typically, these systems should pay for themselves quickly through yield improvements and of course, in being able to use narrow focus recalls to deal with spills.
Patrick Chabot is Manufacturing IT Manager at Sciemetric Instruments.