Let’s Talk Big Data in manufacturing: transforming the industry and operations


Data volumes are exploding. More data has been created
in the last couple of years than in the history of the human race. Why do we need
all these connective devices? What’s the purpose
and the overall benefits of this data? And the big question for industry is: How do we turn big data into small,
smart data and create value? Being connected in an IOT world
is essential today, but that in itself
does not generate value. Let’s get some answers.
Let’s talk. I want to welcome our three panelists.
First, Victoria Van Camp from SKF. She’s Director of Technology and
Solutions for SKF’s industrial market. Then we have professor Johan Stahre, the Chair of Production Systems
here at Chalmers in the department of
Product and Production Development. And Kent Eriksson, an
Internet of Things business consultant at PTC’s Global Services organization. Welcome, all of you. “Big data” is nothing new, but
lots of people probably think of it, along with digitalization,
as a concept that is relatively new. But not within industry – correct? Already in the 50s, some of the
numerically controlled machines saw the day of light. I think that
this is actually digitalization. People think digitalization
started in 2007 when Steve Jobs
launched the first iPhone. But this was not
in the manufacturing industry. This is not disruptive from
a manufacturing point of view, and I know SKF
has been doing this for very long. We started out about 30 years ago. At that time, we didn’t even call it
neural network or anything, but it was taking in data
from our customers… At that time with handheld devices
like thermometers. …and using analysis tools to predict
what was going to happen to a machine and to prolong the service intervals. Kent – smart, connected machines have
played a recurring role in your career. First at McKinsey, then PTC. What are the most significant changes
you see in industry through the years? Today, big data definitely
has better means of being analyzed. Better computer power, and we can
get benefits we couldn’t get before. But it’s true that smart, connected
products have been here since the 50s and produce a lot of data we don’t use. On an oil and gas rig, 1 per cent of the data captured
is used for decision making. 99 per cent gets put on a stack, or the
data lake, as we say in big data. It’s not analyzed. But there could be hidden value
in that 99 per cent of data. If we look at oil and gas, there is a lot of wear and tension from
the fluids going through the pipes, so you can do a data analysis on when
repairs are needed before things break. Johan, would you say we’re in the midst
of an industrial renaissance? How would you define
this transformation? I want to show a picture
I took last year in Mölndal before they tore down the old industry
and put up the new. I think it shows the difference
between the steam engine generation and the 5G of manufacturing
that is coming. And it is actually an
industrial renaissance in that sense. The real Renaissance was also about
information and knowledge. Now, it’s information and knowledge,
but in another way. Victoria, do you agree that there is
a paradigm shift within industry? Yes, there definitely is, but it’s going at a different pace
in different parts of the industry. This can be different sites in Sweden
or different sites in the U.S. where you find customers who already
use predictive maintenance practices with existing amounts of data. They will benefit very much from
the Internet of Things and big data. Then you have users such as
small quarries and mine sites that are still not even using
basic condition monitoring. A guy walks around and kicks the tires,
which perhaps works in their world. But you can see that with these
differences, it won’t help that guy with the tires, if his tires are sending
signals back and forth and texting him. And you won’t get the value out of… What you actually get value out of
is changing your maintenance practices and using big data
to improve your maintenance. What is the reality of companies today
working with data analytics? Can you give some examples of companies
that really are leading the way here? We have Rolls Royce, who do
equipment for aircraft jet engines. If you buy an Airbus or a Boeing,
you actually buy the engine separate. As are Lufthansa and Air France today. The model that you sell is actually not
the product, and it’s not services. You pay power by the hour.
You pay the engine per flight hour. For the airlines, this is great. They know the ticket revenue,
the fuel cost and the service cost for equipment,
which makes calculations easy. For a customer this is better
than trying to estimate what the stochastic costs
will mean per flight. This is what Internet of Things does: turn it to an outcome-based service
rather than a product. There are lots of ways
to use big data. How can big data be used in an
SKF factory at this point, for example? This film was made by
some very skilled colleagues who have scanned the SKF facilities. Every point in space here is
a scanned point with the color of it, so this is actually billions of points
that can be used as a model of reality. This is really, really big data. But when you look at it,
it looks familiar, right? So the human mind is
a very good concentrator of big data and forming it into smart data
that can be useful. For the operational phase, we have also
been looking at augmented reality where you can see how maintenance
in operations can be supported. This is a colleague who is working… You can see the hands
of the remote maintenance support that actually helps her
to choose the right buttons to press. This can be done from Bangalore
or from Mölndal – it doesn’t matter. This is also a very good example. Having the exact data
and the exact information that he needs to run this facility… There are 150,000 operators
in Sweden today. If each and every one of them
saved just one hour per week every week of the year, you can imagine how much money would
be saved in Sweden for the industry. Or used for creative tasks,
things that people are better at. This is a good example, partly of collaboration and development
between academia and industry, but also an industrious example
of what you are talking about. You should use big data
where it creates value. It should create value for the company
and the person using it. Victoria, what’s the latest from your
end on M2M and predictive analytics? The newest example is
what we call the Insight™ bearings, which are bearings
with integrated sensors. You can measure a lot of things,
so we start by using our computer tools,
our data systems, that we use to design bearings
or build them into machines. You can ask: “What do I need to sense?
What’s the most likely failure mode?” And then, that’s what you sense. You also know whether it’s needed
inside or outside the bearing. So, the Insight™ bearing
connected to bearing analysis tools, that’s the new thing we have. Then you have the collection of data either in a cloud or on a server
that SKF has, or it could be as part of
what a customer is doing. One example where we are doing this
is in the rail industry, with Swedish SJ
and a few other rail companies, where we have this kind of
instrumentation on wheel bearings. It is sent to their central computer. Our algorithms are used
to interpret this data so you don’t have a complete train
informing you that everything is fine. And, if needs be,
we or the customer take action, like getting the train off the tracks
and to a depot before anything happens. The combination of analytics
and diagnostics with people is going to be very, very powerful. I’m really happy that you said that. -We still like people to be there.
-Yes, of course. But yes, how does all this affect
the design engineers? The product developers out there,
working with this every day? I can continue. I said we started here,
trying to figure out what to measure. Of course, you then learn
from here, from here and from here what to do and not do
the next time you design a machine. One of the good things
with smart, connected products is to increase the product quality
for product development. There is a good example from Tesla,
who had a quality problem from the benchmarking information
from all the cars running on the road. They found that the reason was certain
speeds and certain types of roads and they could remotely
give the cars a software update that given certain speeds
and road conditions, like the vibrations in the wheels, they increased the suspension height
and reduced the product quality issue by having a smart product that could be
modified even after having been sold. If we talk about new business models, how open and willing should
companies be to share their data? A McKinsey study shows that 40 per cent
of value comes with interoperability, so you need some interoperability,
and not just an internet connection. Different makers
having products working together. In a manufacturing site, you might have
different vendors having the machine, and with an overall performance manage-
ment, you want it all to be connected. Not just the vendors. You need some kind of interoperability. General Electric have made
their Predix platform open, so that everyone can use it. This way, you avoid the lock-in
for the manufacturing site. They give data so their developers can
develop other applications than GE and create an ecosystem to sell things that will actually benefit both GE and
others using that – even competition. They say: “In the future,
we will all be frenemies”. We need to work together even if we
compete, because we share customers. That’s very interesting. In the 1990s, there was a huge effort
very similar to Industry 4.0. It was called computer-integrated
manufacturing. where people tried to connect the kind
of CNC machines that evolved and they had central data systems that would provide access to
all the sensors and all the systems. What happened was
that the interoperability failed. So what came out of it was what
was called “islands of automation”. They couldn’t communicate and then it
took 20 years before it came up again. I think that Industry 4.0
is facing the same problems: Problems of standardization,
problems of interoperability. It’s estimated that German industry
has to spend 2 billion euros per year for the next five years to invest in new
communication equipment and new systems
that will be interoperable. And they have not yet
decided on the standard. So this is a big issue
for the development and use of big data in manufacturing. This is probably an issue of maturity
in different companies and industries. Yes, most companies think about data
as very valuable. That’s a known, so they keep it, even
if they’re not sure what to do with it. For SKF as a component manufacturer,
to sit on a lot of data about bearings isn’t very useful, but becomes useful
in combination with other bearings, with information on how
the electric motor is running, or with something
from even bigger machinery. So data in small pieces
is less valuable. Data as part of an entire puzzle
is more valuable. Some data needs to be closed,
and some can be open. If you look at what SKF is doing,
we are opening up our design software for our users as well so they can start designing advanced
bearing arrangements on their own. That’s a good thing for us. We used to hold it
really close to our chest. But we can’t be everywhere, and if our
customers can start designing this is the way for us
to interact with them. They have to give some information, and
we give some, but to us, it’s worth it. What are the biggest challenges to get
to this kind of ideal business model that we see here in front of us? Over here,
it’s the purchasing department. They don’t usually have
control over the maintenance budget. Here, they look
at price per kilo for steel. “Oh, here’s electronics.
Have some for that.” You talk to different budget parts
when you talk in these different areas. When you start talking about the whole, you have to convince the
management of the company to perhaps change the way
of buying and paying for maintenance. But I think more and more companies
realize the value is more in this side than arguing about the peanuts
on the component side. Let’s talk about predictions.
Kent, let’s start with you. Short-term, mid-term and long-term,
where 10 years is “long-term”. What do we see happening within this? Short-term, we’ll have new standards
on the expected services level. Internet of Things, after-sales
services and maintenance is the easiest way to actually
apply the technology and get value. We see a lot of investments
in this area, and it’s increasing. If we look mid-term… The economic value will be up to 11 per
cent of the global economic value, according to McKinsey Global Institute. Most value will come from
factory and operations settings. That means hospitals,
farming, production… Processes that are repeatable. So the highest value
will be in this setting. Not the fastest, because that will be
maintenance, but the highest. And it would be new eras of LEAN,
better ways of working, as Johan described with the arm, saving
time and using that for something else. And providing outcome-based value where the ecosystem is optimized
and providing services. So you spend more time on providing
value to your customer and ecosystem than fixing things
that have broken internally. And looking at the long term, 10 years,
which is not so far away, I think our imagination and creativity
is the only boundary. I think we may overestimate
the short-term impact of Internet of Things and big data, but I am convinced we are under-
estimating the long-term potential. So, Johan, how much of a key role will
people play in this transformation? A study from some Oxford researchers said that 47 per cent of the jobs
will disappear in 10 years. They looked at different professions
and said certain professions would go. But media multiplied the professions
with the number of employees. That’s not the case. The tasks for different jobs
have been changing since the 1800s. After a while, professions change
and we do other things. I see automation and big data used to
add value for people in their workplace like it is for you and me using
our iPads and telephones to do things. And seeing this as something integrated
which we don’t really see after a while. It becomes invisible to us. And we use the way people are being
creative in their workplace and doing maintenance, which is
very complex for a robot to do. But with the help
of big data, sensors, etc, we can be more productive
and create more value. Victoria, let’s have your predictions.
What do you see in 10 years? If we could at least get
modern maintenance practice in, and perhaps big data and Internet of
Things will speed this up, people will realize:
“Wow, this is all around me now.” “To make use of it,
I have to change my practices.” I really hope for that. Then I’d like to say that
maybe the best way of saving, or conserving our environment
is to do less maintenance. Every time we do
unnecessary maintenance or have an unnecessary production stop we are using resources and money
without needing to. To conserve resources,
keep your machines running as smoothly and as long as possible, that would be the best thing we can do. Thank you, it’s been very insightful. I want to thank my guests Johan Stahre,
Victoria Van Camp and Kent Eriksson. I’m sure people out there still have
a lot of questions about big data, its role, how it’s going to transform
industry and digital manufacturing, but that’s the whole idea
of Let’s Talk. We want to share ideas
and spark conversation. So we want you out there to do just
that by following #letstalkbigdata. Thanks for watching.