05 Apr Edge Computing and MES
The importance of harnessing large volumes data and the value its analysis creates has been the rhetoric behind Industry 4.0 and seen as one of the major drivers towards achieving the benefits associated with it. What is perhaps not discussed to the extent it should be, is how the large volumes of data that IoT-enabled devices, sensors and systems create every second should be handled in the most productive way, reducing associated costs, network and decision latency.
Today, we will explore the concept of edge computing as it relates to manufacturing and how an ideal MES, with the ability of processing data at the edge, may help manufacturers in truly creating value from copious amounts of data generated from their shop floors, which would otherwise be lost.
‘Edge computing’ as a term has gained prominence over the last couple of years and it simply means the ability of a system or broader application to process data generated near its source, so that vital insights from collected data aren’t lost and conveyed in near real-time to stakeholders who might benefit from the information and act on it. It is estimated by Gartner that by the end of 2021 there will be over 25 Billion IoT enabled devices. This means there will be a LOT of data generated, and while lots of data might be considered a good thing from a big data perspective, handling it becomes increasingly a challenge, especially if all the data collected must be relayed to a central data repository (cloud or local), then analyzed and reported back. This could cause unavoidable quality issues, network lag and require higher costs to increase network bandwidth.
In order to overcome the challenges posed by millions of devices connected over the internet always communicating and generating data, edge computing came across as a solution. Data generated by any IoT device may now be collected and analyzed locally, either on the device itself or in an external system, allowing for only relevant and important data to be sent to the central database and at the same time allowing for any immediate action, if warranted to be taken then and there.
How does edge computing figure in manufacturing though? Well, it is most relevant in manufacturing from an Industry 4.0 perspective. On any shop-floor which has IoT enabled devices, data is being generated constantly. As materials or parts move on a production floor, every transaction which happens creates data; every piece of equipment while operational is creating data; every operator while performing an action is creating data. Imagine a production line with hundreds of machines, constantly processing material and creating products and imagine the amounts of data being created every second- it is huge.
For a real-world example of manufacturing big data analytics in action, let’s look to the skies. Sensors incorporated into Rolls-Royce aircraft engines gather 70 million data points a year for real-time analysis by AI, ML, and sophisticated analytic tools.
When it comes to manufacturing every transaction has potential value, especially in case where something happens which is outside the set standards. The faster an issue is detected, the better the chances are of containing its impact and taking actions which help prevent future issues. This is where edge computing comes into the picture. But before we get into more specifics, let’s zoom out a bit and understand how an Industry 4.0 enabled shop floor operates.
Industry 4.0 in Manufacturing
Industry 4.0 in manufacturing is all about hyper-connectivity; it is about process equipment and automation-level applications connecting with applications like MES, relaying information continuously for both real-time and detailed analysis based actions. At the same time, traditionally the MES is also connected and communicating with enterprise-level applications, such as PLM, ERP or CRM, utilizing data received from such applications to create context to enable operational level decision making, and sharing operational information with higher level applications (such as BI) to enable business level decision making.
A critical aspect in Industry 4.0 manufacturing is the ability of the operation to self-govern. The hyper-connectivity of IT applications along with hundreds of IoT enabled devices creates data, which when analyzed through AI enabled Machine Learning to provide insights. These insights in turn help the operation predict issues before they happen and take actions in order to avoid disruptions. Similarly, the intelligence gained and the ability to change course of action in real-time makes the entire business more agile and resilient, thereby allowing demand or supply related events reported from beyond the shop floor to result in savings and profits instead of losses and costs.
When we consider autonomous actions of the shop floor in a granular manner, we can quickly conclude that the data being generated continuously and in large amounts on the shop floor needs to be analyzed as it is generated. Why? First, to generate actions which prevent and contain any issues detected and second to provide insights which will lead to process improvement and even innovation. With all of the data being harnessed, the goal is to reach a level of autonomy where the operation functions more independently and the MES orchestrating the whole process is able to make or prompt decisions which prevent losses and improves efficiency.
MES and Edge Computing Capabilities
An ideal MES application thereby must possess the capability of capturing data at manufacturing’s edge, analyze it and ‘react’ to the data: prompting corrective action; raising an alarm condition; recommending a route change; notifying maintenance; or simply sending the data to a repository, as-is or in enriched form, depending on the architecture and need for redundancy.
In its simplest form, edge capabilities would reflect the ability of the MES to capture data and report an OOS (out of spec) event, which would allow process owners to arrest and contain the issue, without the lag associated with a traditional process where an OOS event might be discovered only after a batch or product is finished and being inspected. Such capabilities have existed for a while and MES applications, through their evolution, have become more adept data handling, manipulation and storage. This includes capturing and reporting shop floor data to the top floor in real or near real time. However, what changes with Industry 4.0 and IIoT is the sheer volume of data, as the MES is not just capturing data from the automation level applications; there are mobile devices, sensors on materials and pallets. Even support equipment, such as trolleys and bins, are creating data– and any of this data might prove valuable to the process for improvements, optimization and efficiencies.
Adding to the high volume of data is process complexity. In complex manufacturing environments, edge computing becomes more and more significant, as hundreds of product variants may get manufactured on the same shop floor, with certain aspects of each product customized. This means that every single action performed is unique, and every transaction needs to be monitored. Managing the flow of manufacturing and ensuring each product is being manufactured as per the set recipe/route/regulation/standard becomes a challenge, unless every bit of data is captured and made sense of right at the source.
Errors which go undetected might prove extremely costly–more than the rework itself. The disruption to production schedules might become extremely difficult to manage, given the customizations and shorter lead times which dictate the manufacturing landscape in modern times.
To ascertain complex manufacturing flows, a well-managed MES application needs to have a robust and well architecture ‘data platform’ which is capable of collecting data at the edge.
Edge collection ensures nothing worth noting is missed, and that AI-enabled processing results not only in preventive action but even predictive ones. At the same time, data processed at the edge should also be stored for subsequent detailed analysis, which would allow for further gains to happen. For manufacturers having aspirations towards an Industry 4.0 enabled set-up, the importance having a MES equipped with manufacturing data platform, capable of both edge computing and more detailed analysis, can’t be stressed enough.
Specifically where edge computing is concerned, data generated at the edge is always high volume, raw and lacking in value, as it doesn’t have the context required to extract the intrinsic value to the organization. The need to capture and process edge data was initially done to lower the network lag and costs associated with storage and bandwidth. As edge computing solutions became more robust and MES applications leading the charge in Industry 4.0 become more capable of using data at the edge and creating more value through inherent context and AI, edge computing became more mainstream and relevant.
Data captured and analyzed at the edge leads to both immediate and long term benefits. The real impact is how the data is used by the MES. As more and more sensors are added to the shop floor, as more IoT enabled devices become active and connected to the process each day, the ability or the lack thereof to successfully and securely capture all the hidden value in this data, at high speed, will be the differentiating factor between a world-class MES and an application which just isn’t ready for Industry 4.0 level performance.
You should be wise while making the choice on a potential MES, an application which only provides low level functionality when it comes to edge computing might not be an ideal application for you, or lead you to the levels of improvement you need to reach true digital transformation.
Capturing data before it loses its value is key, and not all MES applications can preserve, leverage and scale this value!