General Electric (GE 0.01%) has invested a lot of resources in becoming "the digital industrial company." Its flagship digital industrial project is Predix, a cloud-based industrial operating system that can analyze and manage data and operations of GE equipment, facilities, and systems. That sounds great, but what can Predix actually do?
Predix as crystal ball
Industrial machinery is incredibly complex. Even a single steam turbine -- one of GE's major products -- has thousands of variables like heat, pressure, friction, temperature, and speed that all need to be programmed, monitored, and adjusted, often in real time while the machine is running. Small wonder, then, that such machines can generate an incredible number of data points. According to GE, a single plant can generate two terabytes of data in a single day.
As a cloud-based system, Predix can collect and analyze that data to provide critical information about how a machine is operating. This is good, since it can reveal inefficiencies in the machine and detect problems as they occur. But Predix can go one step further and actually detect potential problems before they occur -- sometimes months in advance.
Exelon, the largest utility in the U.S., is seeing this value first-hand. At GE's recent Minds + Machines conference in San Francisco, Exelon demonstrated how it used the Predix Asset Performance Management app to identify increased temperatures on a pair of bearings inside a machine, which led to diagnosis of a faulty cooling system valve elsewhere in the machine. Without Predix, the problem might not have been discovered until the bearings failed, or until the faulty valve caused other, more serious problems.
This can add up to big savings for customers like Exelon. In fact, Ganesh Bell, GE Power's Chief Digital Officer, estimates that deploying this one APM app could save GE Power's customers $390 billion over 10 years.
Predix as guru
Predix has capabilities beyond simply identifying problems. It can even recommend solutions.
To do this, Predix creates a "digital twin" of a real-life machine. By comparing the actual data coming from the machine as it operates with the theoretical data being generated by the digital version, Predix can catch discrepancies between how the machine is actually operating and how it should be operating. It can also simulate how different potential solutions might affect the machine's future operations.
For example, at Minds + Machines, GE's Vice President of Software Research Dr. Colin J. Parris demonstrated a situation in which a single D11 turbine rotor in southern California was experiencing a small amount of damage. Predix was able to identify the problem and diagnose that, if left uncorrected, the machine's useful lifespan would be cut by 69%.
Predix also utilized the machine's historical data, data from other D11 turbines around the world, and nearly 60,000 digital simulations to propose the optimal solution, which involved digitally aligning ramp rates and thermal pressure, and it also had the side effect of burning less fuel and, of course, avoiding a $12 million equipment failure.
Predix as army and commander
So, Predix can optimize the lifespan and operations of individual machines. That's valuable, of course, but through GE's partnership with Microsoft, Predix can also revolutionize an industry.
Microsoft's Corporate Vice President of Intelligent Cloud and Internet of Things Business Development Kevin Dallas demonstrated one way in which this is occurring at Minds + Machines. Microsoft and GE teamed up to use Predix to help an (unnamed) power transmission company client. The client has more than 50,000 miles of "transmission assets," including lines, towers, clamps, and transformers. Up until now, the company needed to maintain a crew of more than 2,000 people to inspect the lines, a costly process involving helicopters in remote areas.
However, through a Predix app developed by Microsoft, the company is now able to use a fleet of drones to travel autonomously along the lines, photographing each asset from multiple angles along the way. The app then analyzes the photos and other transmission data to identify problems such as a missing bolt pin. When a problem is identified, the system flags it so a live operator can check for a false positive. If the problem is confirmed, the system automatically creates a work order and sends it to a repair crew.
This sounds revolutionary, and it is. But the most amazing part is that the app took just six weeks for Microsoft to develop. Yes, you read that right: just six weeks to create an entire autonomous system that can save a company millions of dollars a year. Microsoft -- which used to be an "as-is" software provider -- has now found an additional revenue stream by working with its clients to create these tailored apps.
Predix's capabilities are impressive, there's no doubt about it. Better yet, they create value for GE clients like Exelon. Given the amount of cost savings Predix can offer, it's clear there is a big market opportunity here waiting to be tapped by GE and its partner, Microsoft. GE predicts that in just three years, the industrial internet will be nearly 50 times larger than the consumer internet.
Of course, a big untapped market usually means competitors will emerge, and both GE and Microsoft have competition aplenty in their respective fields. And certainly, there are many Internet of Things software platforms in various stages of development. But with its unique ability to manufacture huge industrial machinery with a software component, and its proven real-world applications of that technology, GE looks to have a clear lead among its peers in this lucrative space.