If an engineer spends five minutes making a phone call to report the issue to the maintenance team, the plant is already down $108,300. That’s in five very quick minutes – and that’s before you consider the cost of the repair.
In a smart manufacturing environment, sensors on the machinery can accurately predict when parts will wear out or even break down – which means that replacement parts can be ordered so that they are on hand for a maintenance crew to make the necessary repairs during scheduled downtime.
If it can be measured, it can be analysed
At its very basic level, smart manufacturing is a process whereby internet-connected machinery is monitored and the data is analysed to improve the overall performance of the manufacturing process. The management consultants at McKinsey predicted that the advent of the smart manufacturing era will save factories around the world $3.7 billion by 2025.
But when it comes to enterprise, machine learning isn’t just about repairing faulty machinery. It also focuses on all the other processes in the business including rostering and staffing, marketing, the energy your manufacturing plant uses and much more. If it can be measured, it can be analysed.
Most manufacturing plants are concerned with the day-to-day running of the machinery. But how many plants employ someone whose sole task is to look at how things are running and to optimise the process? Not many, we’ll bet. But that’s what smart manufacturing offers – a way to analyse the data in real time so that improvements can be made almost instantly.
Machine learning finds the problem
Analysing the data can lead to surprising results – as General Electric (GE) recently found out. In 2016, some of GE’s aviation clients – which include most of the biggest airlines in the world – noticed that the aircraft engines they had bought from GE (at a cost in the tens, if not hundreds, of millions) were degrading faster than they expected.
Initially, a flaw in the manufacturing process was suspected. GE launched an investigation and the data showed that the manufacturing process wasn’t to blame.
Firstly, the machine learning software ruled out any mechanical faults in the manufacturing plants. Then the software diagnosed that the affected planes shared many characteristics in common – the most relevant being that they were frequently being flown in the Middle East and Far East.
Why did that matter?
Because the Far East has far higher levels of air pollution than anywhere else on earth. The Middle East has far higher levels of sand than any other location. Both the pollution and the sand created friction which in turn caused the fan blades in the engines to wear out far faster than anyone had predicted.
Two fairly simple solutions saved the day. The airlines changed their take-off techniques in the Far and Middle East. Instead of the energy efficient gradual take-offs, the pilots now do a full-power lift off to get out of the sand or pollution as fast as they can. And secondly, the engines on the planes that fly in those areas are washed more frequently.
The machine-learning analysis was carried out in the space of seven days. It stopped a costly and embarrassing recall for GE and its customers were happy as they hadn’t wasted millions on faulty technology. The simple solutions of faster take-offs and more frequent washing of the engines? That came from human engineers.
That’s the future of smart manufacturing: machine learning does the number crunching – and human ingenuity does the rest.
Want to see how machine learning can impact your business?
If you are thinking about harnessing the power of machine learning in your business, then Statwolf is your first port of call.
We have used a range of custom services to suit our customers’ needs including advanced data analysis and modelling, custom algorithm creation and implementation as well as predictive maintenance.
Our advanced data science consultancy can team up with you to interpret your data and make your business work more efficiently so get in touch today to unleash your business’ potential.