Accelerate shop floor digitization with edge-to-cloud data integration

Good morning. Thank you for joining. I know it's uh early. I know just a few of you. It's a long distance if you're not staying in this hotel. So thank you for being here. Really appreciate it.

My name is Nicholas Perez. I'm working at AWS specifically leading the industrial IO team. I will be joined today on stage with by Torben who's coming from Siemens. You will talk about a couple of things more towards the end of the presentation.

So today we'll cover mostly three big topics.

The first one will be about building a data foundation and how is that important in the context of digitizing your shop floor environment.

The second will be about how do you scale that foundation and we'll cover also a few customer use cases along the way.

So first, I'd like to start a little bit by setting the stage on what's the state of data in industrial environments and manufacturing environments.

So industrial environments are quite fertile grounds for data. What we see is, for example, on an offshore oil rig, you'll see close to the amount of two terabytes of data generated in a single day. And when you look at manufacturing as another example, you can look at over 1800 terabyte petabytes of data generated over a period of a year. So that's close to five petabytes a day on a site. That's an enormous amount of data.

Unfortunately, most of it is locked down. What I mean by lockdown essentially means that the data is in silos, typically in either systems that are hermetic to external connection, so unable to extract the data or it's simply very expensive and takes a long time to implement a solution that's able to extract that data. And in some extreme cases, there is simply no data so does not contribute to that 1800 petabytes. But we have equipment that's simply too old to be connected. So there's a solution for that as well.

The point being there's a lot of data that resides on shop floors, but that data is not accessible in most cases. So over 90% of it being locked down.

So what's interesting though in that is 20% of the value that could be unlocked has already been unlocked, but there's 80% that hasn't. So what that means is there is still an incredible potential to build additional value out of that data.

So unlocking that data brings even more potential in increased value on shop floor environments, specifically in building digital use cases that go from edge to cloud and what we see with companies that start implementing such use cases, build solutions for that is that they see an increase in productivity by about 50% up to 50%. And in some cases, they see a decrease in cost of quality by up to 70%.

So those are extremely interesting numbers. When when you take a step back and look at the current state of affairs in terms of digital transformation, what's interesting for me and this is a very personal view of how i look at the digital transformation. I like to look at it as a journey.

So as with any journey you mature along the way, this is you may have seen if you were in Mike McKenzie's presentation yesterday, you've seen something very similar, even though we have the two of us a little bit of a different spin on it.

From my point of view, I like to look at this more as where manufacturing has started as traditional manufacturing, where it can go. And I'm not saying it's an end destination for everybody but where it can go and we're really progressing into a gradient.

So not everybody will be at the same state of maturity, not everybody needs to be at the same state of maturity. It's really based on use cases, it's based on the industry and it's based on what makes sense for the business that our customers are in.

So we really see that as a progression. Currently, my personal view is we're still very much in the connected manufacturing space. Um we start to see movement towards software defined automation. We've heard in the news we've heard in a few places that lights out manufacturing is coming, but all of this for the vast majority of people is far away. We're still very much in connecting our manufacturing facilities up to the cloud.

So what does that mean if you're at the beginning of the journey or in the middle of the journey? How do you progress into this? How can you achieve a higher level of maturity to gain more insight about your prediction, how to increase your efficiency?

So what we're proposing today is really twofold. We're going to talk first about data foundation and second, we're going to talk about making that data foundation scale.

So what does it mean to build the data foundation? Well, first, it's about breaking those silos i was talking about. So being able to start extracting the data from different environments, different equipment, different system, take that data and start moving it towards the cloud so that we can have a bigger view of the facility at first at the edge. And second, in the cloud of multiple facilities and start building a better view of how we can optimize and learn across facilities.

Second is to organize that data. So most of you that come from manufacturing or industrial node equipment, data does not contain a whole lot of information, right? That's it's mostly a value that's associated with a tag, a time stamp and a quality and sometimes a little bit more. But most of the time, that's what it is.

So in order to make sense out of that data and transform it into consumable information, we need to add context and that's by organizing the data. And finally, once we have that information, we make it consumable, it can be translated into insight and taking those insights into actions at the facility when it comes to making it scale, i will let torben talk to us more about that later on. But mostly we're talking about taking a small core that relates to a use case and being able to take that small course, that solution that helps with that use case and be able to take it from a proof of concept to a production pilot on a single production line to the entire facility to a global deployment, right?

So what does it take to be able to do that? So we'll cover that a little bit later.

So when it comes to building a foundation, which will be our starting point here, i'm going to start talking to you about what we're doing with AWS I Sitewise to build a data foundation to help you build a data foundation.

So Sitewise has been has been built with the simple concept of saying we need to be able to collect data out of shop floors, be able to contextualize that data add structure to that data, make the data consumable so that it can be queried in a centralized manner. And we can basically visualize that data with components like Sitewise Monitor but also open to third-party products because essentially the data is accessible through either res apis or through sequel queries as i will explain a little bit later.

So take the data contextualize it, move it to the cloud enables our customer to draw insights from it and take action, right? So you can configure rules within Sitewise using events using alarms to make the data actionable.

So those are the the big four components that you will find in Sitewise. Now, one thing that is interesting and probably not well known about Sitewise is that we started in the cloud with a simple gateway at the edge that was collecting data, sending it to built, built all of those services in the cloud. And in 2021 we release Sitewise Edge, Sitewise, Edge extends those capabilities at the edge.

So what we are doing is we're building a ubiquitous model of functionality that resides both on the edge and the cloud.

So let's look at a few customers that have used Sitewise and how they've got benefits out of Sitewise.

So first one, Toyota again, my apologies for those who were in Mike McKenzie's session. I won't go and do a deep dive like he did yesterday with Brad that was on stage. But essentially Toyota implemented a solution for monitoring and maintenance, predictive maintenance using Sitewise IO Sitewise and using Lookout for equipment. And through that, they prevented 16 incidents on the shop floor that saved them 20 hours of downtime and essentially 80 k, roughly 80 k of savings

Marman, which I'll talk in more details in a minute is another great use case, mostly focus on making data visible to the people on the shop floor. And through that, they were actually able to increase shift productivity by roughly 20% and they were able to save in the production.

The pilot facility under k in their first three months of operation with this

Yarra is a great use case. Been a customer with us for quite some time. They've implemented Sitewise across 130,000 assets and across 100 plants or a little bit over 100 plants. Mostly to give self access to data, to information to their plant managers, to supervisors that have created their own dashboard to consume the information that's coming from each of their sites, from those 130,000 assets.

So what's interesting in what they have done is they're able to compare site to site, how the same process is going, how the same assets are performing? What are the the level of quality outputted on each of the production line in the different facilities and be able to learn from that.

So when they see, for example, another facility performing better than theirs. They can reach out understand how that was achieved and apply those learning themselves.

And finally, Hill Corp has implemented Sitewise along with Embassy of thing. One of our partners to provide a single source of truth on the data coming from over 20,000 wells in five different states.

So again, really about first building that strong foundation of bringing data and bringing visibility to that data.

Ok. So i'll talk a little bit about Marman. Marman is a customer we've worked with now for a little bit over a year company that has taken Sitewise zed specifically as something important for them to deploy a solution that would bring them more visibility to their assets.

So a little bit about Marman at first. So it's a 12 billion plus revenue company has operations in 20 different countries or over 20 countries has product and services sold in more than 130 different countries, 400 plus facilities and more than 25,000 employees.

So when you look at this, it's large-scale manufacturing. What's interesting when we start talking about Marman, from my point of view is the diversity that exists within Marman.

So Marman is a holding company has over 100 different business unit, providing industrial components, product services, quite diversified. You look at live, major business groups going from health care, so providing health care devices to things like railways services around railways. So leasing and also managing the railway operation, electrical and looking at the production of industrial products as well as examples of what they do.

Highly innovative company has been around for a long time, but good 70 years of innovation, producing over 100 patents a year. What's interesting is to see the pace at which they're innovating. I've worked with them now, as i said, for a little bit over a year. And it's quite incredible to see how fast they're turning things around and trying them out. So that's, that's really nice to see from a company such as Marman.

So talking a little bit about what they've implemented, their solution is called Marin and the middle industrial mimi and short for internally at Marman.

So their solution is really about providing operational visibility or visibility to operational data to people on the shop floor, to supervisors, shift supervisors and plant managers, but truly focus in the facilities. And all of this is to help everybody in the facility to implement lien lien techniques across the board and focus on operational improvement in terms of solution.

As i mentioned, an early Sitewise edge customer, they've implemented their solution around Sitewise edge for a couple of reasons. The main one was to be able to collect that data and centralize that data collection at the edge. Be able to store some of that data at the edge before sending it to the cloud, also be able to support intermittent connections, right.

So not all of their facilities are located in a place where they can have a resilient internet connection, a connection to the cloud or at times the speed of that connection is not sufficient to transfer all the data. So they were really looking at building a hybrid edge solution on something, a product that allowed them to support disconnected environments or disconnected situation.

So through the building of that solution, they've enable workers on the shop floor to look at the production data through various dashboards, the dashboards as we'll see on the next slide were created using Grafana and build on top of Sitewise edge, using the same APIs that you would find in Sitewise in the cloud.

So when i mentioned hybrid edge, this was also an important criteria is being able to have similar APIs in the cloud in ad edge building once a solution only once that can work in both environments.

So what are the outcomes of this well provided visibility to key metrics. So they have 12 metrics that they've implemented at first, but they've deployed those 12 metrics as standard measures across their facilities, right?

So that's and that's the the key aspect for them is providing that normalization of KPIs across each of their different facilities, independent of the business that they are running, but adapting to the reality of that business.

So when i talk about diversity that makes it even more difficult to scale in those environments, right? So you have highly diversified use cases, highly diversified equipment, but being able to draw similar insights or similar KPIs out of that was quite important for Marman, as i mentioned earlier, that allowed them to save 100 k in the first three months on the production pilot, right.

So this was the pilot facility, good savings and they continued to operate on the on that facility as well as they've scaled to a good number of facilities. At this point, we've seen the increase in shift productivity by around 20%.

And the way they've approached the deployment of the solution was very human centric. So they involved a lot of people on the shop floor specifically at the pilot facility

And through that really got a good response in the deployment of the solution afterwards. So looking quickly at what is the solution, what they have deployed. So essentially the Mimi Edge Server components, what's involved Greengrass, Sitewise Edge data processing, Packet, Sitewise Edge UPS, which is component that's not often used out of Sitewise Edge, but provides visibility to everything that's in movement, all the data that's in movement within Sitewise Edge as well as providing visibility to some key health metrics.

So this was really the foundation for their hybrid edge cloud architecture, focus on normalizing the data before it moves into data warehouses where they have well-defined tables that can be used by the the consuming software. In some cases using Grafana to build dashboards that brings visibility to their shop floor employees.

And as I mentioned, build with third-party components but also built with the involvement of the employees. This was really critical in Marin's point of view on how to execute the project. I'm in complete agreement with that digital transformation is also very human, we say digital, but it's human driven digital transformation. So that built huge amount of collaboration across the different teams that were involved, trust and now it helps with the amplification of the solution, which is a key factor in helping them scale.

So quote from Jeff Dr Jeff Garcia, he's essentially the chief innovation officer at Marman. We've worked closely with him. We've worked with his team very closely to help with the success of this project. We continue to work with them. I really appreciate the effort that uh that they're doing and how we're working in collaboration with them as well.

Ok. On this, we've seen how to go and retrieve data, some use cases to retrieve that data. What does it bring the question now is if you were to have access to all of that data, what would you do? And the answer is through use cases, we see three very common use cases with asset monitoring, which I think I've talked enough about. It's really about visibility on your asset data, predictive maintenance. Talked a little bit about it with the Toyota use case where really it's about detecting events ahead of time and being able to have countermeasures in place using machine learning and then enterprise level visibility. Talked a little bit about that with the Yarra use case, right, where customer Yarra is able to see data from multiple facility and benefit from that visibility to get insights into how to improve, improve in different places.

Those are not obviously the only use cases, the list of use cases could be extremely long. I won't go through all of it. But one of the key thing for us is we cannot achieve or help our customer achieve a solution that will cover all of those use cases. We need help from partners to make that happen.

So we've developed what you see here on screen is what we often call our industrial data fabric, more of an internal name than anything at this point. But it's really how we think about the data foundation Sitewise is at the court providing services, but we see multiple extension points that are required to complement the Sitewise service so that use cases can be delivered as the end value point to industrial customers.

So speaking of partners really happy to mention that we've released about a week ago or two weeks ago, support for the integration of Dometic Easy Edge in Sitewise. What that means is that Dometic through their Easy Edge product allows customer to now use 10 different protocols on top of CU. So very important as I mentioned early on several data silos, trying to break those data silos requires connecting two different types of equipment, PLCs, KAS DCS. So being able to do that requires support for additional protocols and very happy to have worked with Doma de for essentially the last year was a great involvement on their part.

So really happy to announce this as we continue to work with partners and we'll talk potentially a little bit more about that later. We are also continuing to make Sitewise evolve and build more capabilities focused on two major pillars. One is resolving. One of the core challenge we've heard from customers about the cost of implementing a solution with Sitewise specifically a lot around ingestion of data storage of data.

So we've released features that help with the cost optimization, one that we call adaptive ingest that have helped us reduce the cost of ingestion by up to 70% compared to what it was before the release of this feature. We've also worked on the the second pillar which is about helping and simplifying the customer experience to get started.

So, one of the major challenge we heard often from customers was their ability to get started by creating models and Sitewise, this was often referred as a duplication of effort or something that took a long time before getting going. We've done a few things two years ago with supporting undeed ingest, which was not possible prior to that. And we've continued to evolve on this by being able to now import metadata from better systems.

So that means if you have your OSIP PI infrastructure, you you have your entity models into OSIP PI you can now export and reimport into Sitewise. So that's helping customers start much faster with the creation of their models. We've created reusable components as part of our models. That's an important aspect as well. That helps with repeatability and finally storing metadata, time series and time series for the integration with sorry with Lookout for for Equipment.

So really making sure we have good structures to exchange data between Lookout for Equipment and Sitewise. Both of those products were conceived completely separately. So we've established a bridge between those two to help our customers integrate faster.

Ken on this. We'll switch to looking at how we can make the data foundation scale. And for this, I'd like to invite Torben to come on stage and talk to us a little bit about this.

You think about what of that data you want to have really in the cloud so that you can start thinking about building, for example, a ML models using AWS SageMaker, for example. And then the question is how you package that model and deploy it back on the shop floor in a way that you don't disturb running processes on the factory floor.

And also there we have you providing services together that you can package that model, you deploy it on the edge device on an in the running there. And this is also only just extras having the acceleration needed for these for these ML use cases.

So why is that important? Because we really think that this is providing productivity measures. This is really taking us to smart manufacturing. If you think about that, you want to have also have some closed loop use cases. For example, machine vision based quality inspection, they have a lot of pictures you take on the shop floor, you want to provide that seamlessly to the cloud you want to train them. But the model needs to run on the shop floor at the edge, for example, in real time, sorting out wrong and good products with a closed loop integration with a PLC. And that's what we want to realize here.

So next is that we said, ok, now we have the machine connectivity at scale. You can have these data foundations, you have that continuous flow. So how we can now start building these IoT solutions on it? Also at scale for that we are providing so called off the shelf applications. We call them industrial edge applications. They will not be provided only by Siemens but also by AWS. So AI at edge is a industrial edge application and everybody else can provide them as well.

So we are also working with independent software vendors to provide applications for different use cases on the shop floor. But it also means for us as SES that be moving into what we call a software defined automation that some of our core automation functionalities like control motion SCADA MES, they become industrial edge applications and can run on different hardware either at the machine level or at the factory level.

So these industrial edge applications are available on the industrial edge marketplace. You can get different applications for connectivity to the shop floor connectivity to the IT systems. You can have data management applications. You can also have different applications for doing value adding use cases like machine up time improvements. OEE you can do a predictive maintenance and also what we talked about machine based vision and machine vision based quality inspections applications for us.

I said it's very, it's an open approach. So we invite everybody to also provide their off the shelf applications for use cases. They think that will also they will scale across the manufacturing base. So these applications can run either on the machine level for really real time applications close to the closed loop integration is required, but they can also run on the factory level where you want to have more horizontal scalable IT solutions for for example, providing a horizontal scaling data integration level after the cloud and back.

But it's also very important for us is that these applications can actually are inter with each other that you can flexibly exchange them that they are qualified. So we spend also a lot of time to make them work together to have really a proper quality quality measures in place that they work together. Because in this sense, B2B is very different from B2C as you know, for example, on your smartphone, you have different applications. Normally one application constitutes one use case. So if you have your calendar application that helps you manage your time.

But in B2B on the shop, though you need to have different applications for connectivity, data management value, adding stuff that is all integrated and works as a system together and can also be managed as a system, right? So this is how we want to help building IT solution that scales offering or at least off the shelf building blocks. But then also this is an open approach where we also say everybody's invited to build applications for some specific customer needs or manufacturing needs.

So you can use different programming languages to build the industrial edge applications very easy to put them on an industrial edge device. And you can do that as manufacturing yourself as a developer. But you can also ask global system to create a solution partners to do it on your behalf. Or you're taking for example, Mendix as a low code, no code application development platform to do that. And Mendix will help you to develop applications 10 times faster than in a traditional way.

The beauty also with Mendix is that this is an application development platform running on the AWS, you can build lots of applications in the cloud. Now it's also running on edge. So what we did is that we kind of really um made it very simple to develop an application in Mendix and deploy it on industrial edge. Also from a business model approach that you can scale with the number of applications running on SES industrial edge. And with Mendix also using that continuous data flow between machine edge cloud level and back, you have that data integration at all levels.

Meaning in Mendix, once you have built a dashboard, you can have on the same data level, providing insights to the machine operator at the machine level. But you can also provide insights to the factory manager at the factory level on the same data of the same application and the same on the cloud level. If you want to have insights and benchmark across different factories in lines, the results.

So let's recap a bit. What we said is we want to build our connecting machine at scale. We want to build IT solution at scale. Now helping with off the shelf applications but also using local local platforms. Now, how do I get from this one installation? I built up to many installations to many use cases across the factory network. And for that, we we using a software, a orchestration software which we call Siemens Industrial Edge Management and it basically does two things.

So first, it helps you to connect to all your industrial edge devices which are spread across the factory network. So you can remotely connect to these industrial edge devices. You on board them, you can manage the firmware on it. You can do the monitoring of the health and resource status. Of course, you can off board them as well. And the next industry management does the full application life cycle so that you can start mass roll out of applications, you install them on many devices can start configure them for connectivity, you update, upgrade them as you like and then this becomes really repeatable.

Once you have built one use case on one device you configured it, you just need some clicks on a button to repeat it and to spread it to many other industrial edge devices at similar installation, similar machines on the factory level. So and we're doing this with industry's edge management really like in an approach that helps customer depending on their competence profile.

So if you don't want to have all the necessary work to manage and operate in the edge management system, we provide it as software as a service running on AWS. But you can also have the industry's edge management deployed on premise running for example as a VM and a VM cluster or as a container running on coons, right?

So again from a recap perspective, so we said we want to make smart manufacturing real for that. We want to provide data foundations a uniform language for data where we can create and scale IT solutions on top. So we're doing that with connecting to machine data at scale building the solutions with standard applications off the shelf, you can even provide your own applications in a very easy manner. And we also said, ok, we have to the management system for make it to make it repeatable.

So this is the joint offering actually, we have been talking about between AWS and Siemens and this is like how it looks bring it all together in one kind of slide. I personally think this is game changing. I'm very excited about about this. I think it's the first time in the industry that you have the full edge to cloud experience. And this also from a workflow from a user experience in a very, very easy manner.

So first time you can have this consistent data flow, manage your data life cycle from the machine edge to cloud and backwards. You can drive these ML use cases. You can drive other solutions via providing via these industry edge applications where you can create and scale them out for different topics like productivity measures on the shop floor for saying the sustainability measures but also to make it more flexible how you operate on the shop floor, the what AWS will provide as a value as a key value which i think is very important is infrastructure as a service so that you can make it very robust, very secure and simple to scale that out.

Be from Siemens will provide as a key value, some infrastructure components on the industrial edge level like the industrial edge management system, but we will also focus very much on the application value. So we want to provide applications which will help the manufacturers to reap the full benefits of digital transformation. Again, it's if it's quality control, predictive maintenance or OEE improvements, all of this right.

So with that, this is, i think what i wanted to talk about, this is like what we really announcing. I hope you also excited about this as we are. And with that, I want to ask Nicholas back on stage and Nicholas can give us an outlook how we want to progress from here.

Sounds good. Thank you, Keith. Great. I think Torben covered this extremely well. I'm going to move right away into the next slide which is to talk about which we've announced a little bit earlier this week, you may have seen this already if not SiteWise Edge is formally available on the Siemens Industrial Edge marketplace.

Tord. The the great explanation over the last few minutes around this, i share the excitement for me. This is amazing to see how we're connecting the machine level all the way up to the cloud in a very secure manner. And in a way that data can be contextualized can be transformed into consumable information all the way through. So basically attaching skill, attaching the shop floor at skill with the cloud offering and be able to start drawing better insights from that data. So that simplifies drastically. Also the user experience in the sense that now you don't have to deal with multiple disjointed parts to assemble. You have a single pipeline that's coming from Siemens Industrial Edge with SiteWise Edge residing on it all the way to AWS.

So coming back to the initial question at the beginning of this session, how can you achieve higher level of digital maturity? So i hope that through this session, you've been able to get some level of learnings of how you could achieve that. I'll start by saying as a key takeaway that building a strong data foundation is absolutely essential as a starting point. As we've seen with a few customer use cases, starting with building that foundation to monitor data. Just bringing visibility to data itself brings an extreme amount of value to customers, right?

And then we i think i covered read the part of we want to make it scalable, right? Remember this picture of the lighthouse project where you have put so much effort in one installation. In this one use case. Now you want to, we need to make it scale and i think we can only make it scale by bringing edge and cloud together by breaking down the OT and IT's data silos for building the data foundation models. But then having the right applications on the shelf to really try productivity measures on the shop floor. Ok?

And that should lead you to be able to enable more use cases, accelerate time to value with a simpler offering that's already connected, reduce scaling cost because it's highly repeatable. And in the end being able to gain more meaningful insight.

So that brings us to the end of the session, I'll leave you with a few resources that you can scan the QR code and get access to that will give you access to a blog post that was produced to, that explains what we have done with Siemens one on AWS IoT, SiteWise Edge and the press release that was made by Siemens. I'll say Monday, I think was around midnight this time on Monday.

So thank you very much, really appreciate your attention. We have about 10 minutes left so we won't take questions on stage. But if you want to talk to us, if you have any questions, please feel free to come and join us offstage. Thank you very much. Thank you, sir, please.