Practical Manufacturing 4.0 and the future
Manufacturing industry is one of the fuels that drives the economy vehicle for a country. Over the past few decades, some of manufacturing has shifted to nations with lower production costs, labor costs, and less stringent laws on environment control. In the united states, industries are bringing back Manufacturing for multiple reasons. Given this current state, there are huge opportunities to increase the efficiency, quality, and throughput of running a plant. Automation and the use of Cloud and Machine Learning technologies contribute to realize a truly digital manufacturing business, often referred to as the Industry 4.0 revolution. The ability to apply these techniques on top of existing platforms enables tangible business outcomes with minimal investments. Traditionally, industry equipment have been using sensors that follow the standards such as OPC UA (Open platform communications Unified Architecture) and SCADA (Supervisory control and data acquisition). These machines are able to integrate with the modern Internet of Things (IoT) stack to gather real time data for monitoring and efficiently managing environments digitally.
The image from McKinsey describes the 4 different key areas that are ripe candidates for disruption in this space and can be used as a maturity model to see how a business is progressing in it’s digital transformation journey.
Let us look at the first two areas in the above picture to understand building a digital enterprise using Cloud and AI capability. If we evaluate an end to end manufacturing process, there are three key steps and each step has its own desired outcome.
- Design — In this ideation phase, we are interested in identifying the right fit and validation of the product request
- Manufacturing — In this phase, we are interested in improving the quality of the product, safety of the facility, optimize occupancy etc.
- Operation — In this phase, we are interested to reduce downtime, reduce warranty recalls, optimize performance etc.
The next obvious question is how does Cloud and AI enable improvements in each of these phases? Most of the leaders in the manufacturing space are reinventing themselves as a business that builds specialized software products for manufacturing domain. This mindset enables us to look at a digital view of the overall process that drives the underlying equipment and machinery. This thought process has led to coining a term referred to as the Digital Twin. It can be thought of a virtual representation of a physical asset. The asset can be a car, a plane, a wind turbine, a factory, a building, a ship, an energy plant etc. operating in a remote environment. An engineer can analyze, monitor, and make a change in the behavior of this asset without being next to it physically. This is a huge advantage to monitor and maintain an equipment from a control room. This enables us to manage the asset when it is cost prohibitive or dangerous to be in its proximity to control. This also gives us an option to see the response behavior under unknown conditions. This typically would have traditionally required use of deterministic mathematical and Physics models. With the emergence of capabilities in AI and ML, heuristic models have been able to accomplish the same. If we map this idea and look at the three steps with a new lens, we can think of three different kind of digital twins.
1. Product Design requires a Product Twin
2. Manufacturing requires a Production Twin, Process Twin, Facility Twin
3. Operations requires a Service Twin
With this background and understand, the next question is how would we go about implementing a digital manufacturing solution?
The above diagram is a reference architecture built with core components from AWS. IoT Sitewise and IoT Analytics are both managed services that are built for scale and can ingest, manage, monitor, and provide intelligent insights for devices in real time. The AWS reference architecture helps us to map to the first two (Connectivity, compute, data, and analytics using ML and AI) areas in our 4 areas of the Manufacturing Maturity Model.
The Microsoft Azure platform also presents a similar capability for implementing Digital Twins in manufacturing industries. Irrespective of which platform you pick, it is easy to plug and play the services that you choose to consume. It is important to start with a clear objective and figure out the appropriate services in the architecture from the chosen hyperscaler. With the increase in the use of microservices to design a decoupled architecture, there is flexibility to build and evolve an architecture as the business needs change. This also enables us to have a multi cloud solution and pick the right combination of services based on the business need.
The Microsoft IoT platform referenced in my earlier articles (here and here) provides a robust platform to generate end to end turn key solutions. Here are a couple of examples showing a Smart Meter solution and a fully remotely managed Solar Panel management solution built using the Azure IoT framework.
I like this architectural stack from Azure which does a good job of showing how the physical world of machines and sensors meet the digital world of a manufacturing facility managed through a software solution. Microsoft Azure refers to this as the Metaverse stack.
If we focus on the remaining two (interaction with humans and advanced engineering) areas of our Manufacturing Maturity Model, we notice that these capabilities can be naturally built on top of the existing stack.
We have seen the emergence of Virtual Reality and Augmented Reality over the past few years. The idea of a Mixed Reality that we can achieve using something like a “HoloLens” not only enables us to have digital annotations for our physical world but also be able to interact and modify the subject and environment. The power to remotely monitor machinery in a remote oil drilling rig or a solar power plant located at a place with abundant sunlight is pretty powerful. This ability to interact seamlessly gives us the flexibility to not only control the production and labor costs but also provide safe working environments when operating complex machinery in a hazardous environment.
It already is a giant leap to manage an environment fully remotely and be able to detect failures before they occur to manage a resilient plant operation. But, the icing on the cake would be to use a self healing mechanism in the industry as we practice in software development (using Highly Available and Disaster Recovery strategies). Using a digital equivalent of a physical asset, we can create a replica of it and replenish a faulty asset using a combination of software modeling and 3D printers. This is an example of how additive manufacturing using 3D capability (as shown in the fourth area of our maturity model) plays a role in the manufacturing pipeline for resilient operations.
Having said all that, it is exciting to know that we are barely scratching the surface of how Cloud, AI, Mixed Reality, and 3D printing capabilities can enhance the future of manufacturing. Hope you are able to leverage some of these capabilities when thinking of disrupting your current business.
