FIELD INTELLIGENCE: Smart Processes, Solutions & Strategies
It has become far too rare for manufacturers’ visions of an IIoT-fueled utopia to survive contact with reality. A Cisco survey finds that nearly 75 percent of Industrial Internet of Things (IIoT) projects are failing. Microsoft research shows 30 percent of IIoT projects don’t even graduate from the proof-of-concept stage. For many manufacturers, expectation hasn’t gelled with experience—even while the competitive stakes of getting it right are higher than ever.
There are three fundamental reasons why these failures are occurring with frequency.
First, actually leveraging machine data at industrial scale requires specific expertise that is rare and expensive to find.
Second, IIoT projects only thrive when leadership can go all-in on supporting IIoT transformation. Siloed manufacturers with cultures or sub-cultures resistant to adaptation will torpedo projects.
Still, the variable most common to manufacturers’ IIoT derailment is a lack of understanding of the very specific data needs of the IIoT. Connected factories produce a tremendous volume of nonstop time-series data. Getting it right requires a data-layer strategy capable of handling sensor inputs at a uniquely intense scale—future scale that must already be baked into the strategy from Day One. Then, that data needs to be delivered with the performance and availability required to derive the operational and business insights manufacturers are seeking from their IIoT investment in the first place.
ALPLA—an Austrian firm founded as Alpenplastik Lehner Alwin that develops and produces plastic packaging solutions—has succeeded where others have struggled.
With nearly 200 plants in 46 countries, ALPLA is showing how important the right data environment and strategy plays in ensuring an IIoT transformation is actually, well, transformative.
ALPLA began its IIoT transition with the goal of enabling data-driven manufacturing. It wanted to move away from standard tasks with time intervals based around six-sigma, 5S, etc., and instead have machine data analyzed and available in real time.
ALPLA wanted to reliably trigger the right information getting to the right person on the plant floor as quickly as possible.
If it could do that, ALPLA could get in front of plant issues before they become big problems.
ALPLA initially deployed SQL Server as a conventional data-storage method before realizing it wouldn’t withstand the data onslaught. The manufacturer had 900 different data tables and each came from 900 different sensor data structures. Then, every production line generated data to the tune of thousands of readings every second. Querying streams of sensor data in real-time or trying to harness complex machine learning analytics on terabytes-wide troves of historical data proved impossible.
Instead of becoming a statistic, ALPLA pivoted—early on—to a revised data strategy. It restructured its stack around data technologies specifically built for smart manufacturing use cases. The results were stark: data queries 250 times faster than with SQL Server.
As one example of what this now yields, production machines are automatically ejecting plastic packaging whenever sensors discover an irregularity. That information is then immediately communicated to the nearest factory floor technician to fine-tune equipment. It’s one part of a larger IIoT data success story that has given ALPLA fewer quality issues, fewer episodes of equipment downtime, reduced complexity for plant workers and lower maintenance costs.