Select a language
Davie Street Enterprises (DSE), our fictional case study company with real-world problems, has made significant progress in its modernization efforts in this past year. The company has automated a large part of its IT infrastructure and has completely revamped its development processes using DevSecOps.
It has also rebuilt the Parts and Supply (PAS) system to allow for more efficient supply chain management. With these successes behind it, DSE is feeling a little more confident in its plans to transform the company.
However, it looks ahead soberly and sees some very big challenges in modernizing its warehouse operations and is not sure how to address them. In this post, we’ll share how DSE plans to add predictive maintenance capabilities with Red Hat and an ecosystem of vendors, and begin its path toward Industry 4.0.
What is Industry 4.0?
Industry 4.0, also known as the Fourth Industrial Revolution, refers to the increase in automation, communication, and self-monitoring in traditional manufacturing and industrial practices.
From embedded sensors to robotics, manufacturers are integrating internet of things (IoT) and artificial intelligence/machine learning (AI/ML) into their factories to gain new efficiencies, achieve higher quality and better adapt to customer demands.
With the introduction of the new PAS, new DevSecOps processes and automation, the plant operations are more efficient than ever. This higher demand is putting stress on plant operations to keep up.
When Stephanie Wilson, Director of Plant Operations, looks at the weekly global production report, she notices something that bothers her. The Chinese plant was down for half of the week because of equipment failure. This is a perfect example of what she has been complaining about for some time.
First of all, she wasn’t aware of this failure in real time. She had to wait for the weekly report to find out what happened. Secondly, this will put a serious strain on order fulfillment. This is unacceptable, and Wilson is frustrated. Through her user group of manufacturing professionals, she’s heard a lot about Industry 4.0 and the ability to get real-time information about all aspects of the manufacturing process.
Utilizing this outage as motivation, she calls a meeting with Daniel Mitchell, DSE Chief Architect, and Monique Wallace, DSE CIO. In the meeting, she explains the outage and problems that it caused with the production schedule. She also does a brief introduction about Industry 4.0 and how it could help to solve the problems that outages cause. As part of the meeting, they also discuss the changes required to support the new Smart Widgets that are being designed.
As the modernization efforts progress, one of the great side effects is an increase in demand for Widgets. The Smart Widgets will be a game changer for the widget industry. Smart Widgets are intelligent versions of typical widgets.
They are network-attached, have a unique identity and provide information about themselves, such as current description and status. They can also provide lineage information that will allow DSE to track failing Widgets from their creation date, plant, plant line number and run.
Exciting times are ahead for DSE. The meeting was a great starting point. Wilson got the nod to move forward with implementing Industry 4.0 technology into operations of all the plants worldwide. She is excited by that outcome but it leaves her with a lingering question, now what?
Ready, Set…Industry 4.0
Wallace, DSE CIO, puts together a meeting with Mitchell, Chief Architect, and a group of lead architects in her organization.
The goal of these meetings is to put together an architecture and to evaluate the current tools, vendors and capabilities already in place. Also, they want to examine the new requirements defined by Wilson and determine if the established tools and vendors collectively could address the new Industry 4.0 requirements.
As they learn more about Industry 4.0, they have come to understand that the only way to implement a successful Industry 4.0 strategy is to have the Operations Technology (OT) team, the Information Technology (IT) and the business units work together.
Figure 1 above shows the typical Industry 4.0 reference architecture. One component of the architecture must collect raw data.
This component typically falls under the responsibility of the OT team. They own the Program Logic Controllers (PLCs) that drive the manufacturing line. They also run and operate the control center that monitors the PLCs and line equipment. They rarely have a need to interact with the IT team besides getting network access for their PLCs.
To implement a successful Industry 4.0 strategy, the OT team will now have to expose and make its PLC data available to the larger organization and IT team. Once the data is collected, it needs to be shared with the component that will process the data. Processing of the data, at this level, will include some amount of cleansing, aggregating and filtering the telemetry data that comes from the equipment. There is no need to congest the network with bad or invalid data.
The final part, and perhaps the most valuable component, is where that raw data is turned into insights that will drive some process control point or report the insight to a person for action.
Driving a process control point is a very advanced part of Industry 4.0 and members of the OT team are not willing to go this far yet. They are unsure that they could trust a computer model to have the intelligence to know how to control a line when something goes wrong. However, they are willing to provide enough PLC data to implement a predictive maintenance use case. For now, this only requires OT and IT and not the business units.
Building a software and ecosystem ecosystem
Where do you start such a project?
When you drill into the next level of detail for a reference architecture, you begin to realize that an Industry 4.0 project is complex. It’s not just organizational boundaries that are crossed, but it also requires an ecosystem of vendors and suppliers whose software and equipment must work together.
As shown above in Figure 2, PLCs will need to communicate with small factory servers, sometimes called gateways. The gateway devices have to be ruggedized so they can withstand the harsh environment that exists on a manufacturing floor.
They also must have a secure stable operating system along with a software stack that can collect, filter and aggregate this telemetry data. Once collected and filtered, the data will need to be stored in a Data Lake where machine learning models and other systems can access the data for different use cases.
One of the recommendations for starting such a project was to evaluate what tools and capabilities existed already. Luckily, the OT team had recently updated all the PLCs on every line from using the regular OPC (Object, Linking and Embedding for Process Control [OLE]) Windows-based protocol to using OPC Unified Architecture (UA), which now allows the PLCs to communicate over HTTP.
Container orchestration with Red Hat OpenShift
Over HTTP, this data can now be easily shared with others throughout the company. Also, the IT team had implemented its first DevSecOps project that introduced Red Hat OpenShift and container orchestration.
Meeting datacenter needs with HPE Edgeline Converged System
The IT team also has a very strong relationship with Hewlett Packard Enterprise (HPE) for its datacenter needs. For the final piece of the puzzle, the company has relied on SAS for almost all of its statistical and analytical needs for a very long time.
There is a strong push from the industry and internally to take advantage of public cloud based services for machine learning. This is certainly on the roadmap but the team doesn’t feel this is the right project for it. There are so many moving parts to getting Industry 4.0 off the ground that they feel the need to keep things as simple as possible. So for now, that means keeping everything in their regional datacenters and taking on the public cloud later.Figure 3
They remember getting a presentation from their Red Hat representative about Red Hat Edge Reference Architecture for manufacturing, shown above in Figure 3. It had a similar three-tier architecture where Red Hat OpenShift was able to run all three tiers. From an IT perspective, this allows them to have a common platform to support from the edge gateways all the way to the datacenter.
This will make it much easier to manage and secure software applications and networks. Red Hat OpenShift at the edge also allows developers who are writing datacenter applications to write edge applications using the same DevSecOps methodology and tooling.
Red Hat already has a strong collaboration with HPE, and both Red Hat OpenShift and Red Hat Enterprise Linux (RHEL) support the HPE Edgeline Converged System series. This powerful series provides datacenter-like power and management, but in a smaller and ruggedized form that can support the tough conditions in a DSE manufacturing plant.
Deploying new models with SAS Viya
DSE also heard about SAS Viya running on OpenShift from this announcement. It already has a fairly large deployment of SAS Viya on RHEL in the datacenter. SAS Viya provides large scale parallel processing which can deliver results in seconds, not hours. Plus, it contains built-in governance that can help make decisions repeatable, explainable, transparent and trustworthy.
DSE has heard from others who have embarked on the Industry 4.0 journey that it could be challenging to deploy new models and make them operational. Because the IT development team is already well versed on SAS, building and deploying scalable and operational models in SAS Viya on OpenShift makes this task very straightforward and easy.
Based on the research, DSE is feeling confident that HPE, Red Hat and SAS together can get it started on the path to Industry 4.0. As a result, DSE builds its own reference architecture as a starting point for solving the predictive maintenance use case.
Figure 4 below shows the reference architecture it created.
Addressing the predictive maintenance use case
With reference architecture in hand and right vendors identified, DSE is ready to take its first step toward Industry 4.0 and all of its promises. There are still plenty of challenges but DSE is confident this is the right approach.
DSE teams will address the predictive maintenance use case in a pilot phase on one line. They will determine what telemetry data is required such as component temperatures, vibrations, activity time, etc. They will then need to build all the collection and filtering capability using Red Hat AMQ on OpenShift. When the data has been filtered, they will use the capabilities provided by SAS Viya to store the data and use it to train a predictive model.
This pilot will prove out the use case and technology. When the pilot is complete, they will then create a rollout plan that describes how they will roll out from pilot to all the lines in the plants.
What’s next for DSE?
Once this is rolled out successfully, they can start the planning for smart widgets. Smart widgets may require a different approach and will certainly involve the business units. Smart widgets will generate a whole new set of challenges but the lessons learned in this first project will certainly come in handy for any IoT and edge use cases in the future.
About the author
John Senegal is an ecosystem solution architect who works with strategic global partners to build joint and ecosystem solutions. His technology focus is around the AI/Ml and edge/ IoT partner ecosystems.