Edge Computing For Manufacturers

Introduction

New technologies are revolutionizing how manufacturers connect processes and collect data on the shop floor.

Among the most promising is Edge Computing.

Edge computing is a method of improving data aggregation and processing by placing computing resources close to where data is collected.

This guide will introduce you to edge computing, and explain how manufacturers of all sizes can benefit from the use of edge devices to power their IIoT and machine monitoring initiatives.

Edge computing refers to the allocation of computing workloads to the “edges” of a network–to devices and resources closer to network endpoints than a centralized data center or cloud.

In manufacturing, edge computing resources can include machines, gateways, protocol converters, or other types of industrial controllers. Most often in manufacturing, edge computing refers to the use of a dedicated edge device.

By distributing computing resources, manufacturers can:

1. improve data processing speed

2. filter signal from noise early

3. improve data accessibility

The philosophy behind edge computing is that it’s faster and more efficient to process information at its source.

Consider how information travels from an industrial asset to a data center. The machine generates data—temperature, vibration, rpm, resource consumption, etc—as it runs. That data likely needs to undergo some sort of format or protocol conversion to make it useful to engineers. Then the data is shuttled to either an on-prem server or to a cloud database. If it’s sent to the cloud, then it may need to travel considerable distance to one or several geographically dispersed, physical servers.

While cloud computing and storage is essential for the Industrial Internet of Things, this relay between data generation, processing, and return can create problems with latency, bandwidth, and data management.

For sensitive manufacturing processes, even a small degree of latency can have significant effects on efficiency and quality.

Edge computing prevents this by ensuring that data processing happens in close geographic and network proximity to data creation.

Understanding why edge computing is important in manufacturing requires understanding some challenges that arise with cloud architectures.

Manufacturing has long used centralized information architectures. In a centralized system, information generated throughout an operation is processed and stored on a backbone server. This has been important for handling high volumes of data, as well as for managing sensitive controls.

With the advent of the cloud, this “backbone” has moved from on premise resources to geographically removed locations. The journey data takes to the cloud, however, is essentially the same.

Advances in edge computing mean that manufacturing is returning to a decentralized model.

Rather than concentrate data processing in a single location, decentralized systems like edge computing take advantage of the compute and storage capacity extant at nodes throughout the network. This type of structure is “decentralized” because computing is done where there are available and sufficient resources. With decentralized systems, proximity is a leading determinant of where processing occurs.

Even if the growth of edge means a growth in decentralized computing, it’s important to note that this does not mean that edge will replace the cloud.

Rather, the edge and the cloud exist in a symbiotic relationship.

For many manufacturers, cloud infrastructure is necessary for making sure that data is stored in an accessible, scalable location. It’s necessary for supporting operations with sufficient computing power without expensive outlays for on-prem equipment.

Without the cloud, the logistics of storing, pulling, and analyzing data would be significantly more complicated. Yet the realities of modern machine monitoring make edge computing desirable.

Thus, manufacturers can reap the most benefit from their digital initiatives by marrying a cloud infrastructure with edge computing as operations require.

This kind of hybrid architecture--where a centralized infrastructure supports decentralized computing--is often referred to as fog computing. Put slightly differently, fog is how you bring the cloud close to the ground.

Indeed, the best estimates of each technology’s future suggest that market size for cloud and edge will both grow significantly over the next decade.

Edge devices can be sensors, IoT-enabled machines, gateways, or single-board computers. For small and large manufacturers alike, these offer a quick, lightweight method for streamlining digital initiatives. These devices run on minimal power, consume low quantities of energy, and can be extremely cost-effective.

Edge devices are an essential component of any machine monitoring system. They are important for several reasons.

Protocol Translation - Not every machine communicates over the same protocols. While newer machines can communicate using protocols like MTConnect and OPC UA, either natively or with translation products like Kepware, others require devices to make sensor data legible to an end-user. Legacy machines, in particular, need an intermediary device to convert sensor data into useful information. Edge devices can capture information from IoT sensors and translate them into whichever protocols you need to turn data into insights.

Data from Multiple Sensors - In many cases, understanding machine performance and health requires taking data from several sensors simultaneously. For example, if you want to understand the root cause of a part failure in a CNC mill, it’s important to have data on all of the parameters that could contribute to a specific part’s degradation.

Edge devices help to compile data from multiple sensors to put machine health in a more complete perspective. This enables you to get past speculation to root causes.

Data from Multiple Machines - Understanding each machine as individually important. However, the most impactful insights come when data from a single machine is measured against the data taken from a full department. Edge devices help aggregate and centralize data from a number of different machine sources, making it easier to understand performance in real-time.

Filtering Signal From Noise - Not every piece of information generated by a machine is useful. Edge devices help separate important information–like when a machine exceeds an established parameter or when a specific event is detected–from the deluge of data created during scheduled uptime. Increasingly, machine learning algorithms are running on edge devices to facilitate this filtering process.

Real-time Insights - For many manufacturers, machine monitoring is a way of understanding production in real-time. While historical data is important for understanding machine performance and health over time, it’s equally useful to have visibility into machine status during operating hours. Edge devices make it possible to route data from machines into visual analytics dashboards. By passing machine data through an edge device, manufacturers can access the analytics necessary for real-time insights.

Increasingly, success in manufacturing requires marshaling operational data for continuous improvement. In short, it requires a machine monitoring program.

But machine monitoring isn’t always enough. Manufacturing systems generate a tremendous amount of data. And data isn’t useful if it can’t be stored and accessed in an actionable form.

Of those manufacturers tracking their machines’ performance, few have a data-management strategy. Even fewer (14%, by some estimates) report having no problems wrangling the abundance of data generated on their shop floors.

This is where edge devices can help.

Edge devices help guarantee that your machine data is 1.) legible, 2.) accessible, 3.) secure, 4.) relevant to the KPIs you want to track.

To demonstrate how edge devices can be used in real manufacturing contexts, let’s look at how a Tulip customer uses our IoT gateway to collect data from their legacy machines.

This manufacturer faced a common problem. Though their legacy machines had no operational flaws, they couldn’t connect to the internet natively. This meant that measuring machine performance was a manual operation. In many ways, their calculations of OEE were subject to error, and their analyses of bottlenecks struggled to identify root causes.

In order to better understand their production, this customer used IoT sensors and Tulip edge devices to bring their analog machines online.

Using edge devices and sensors, this manufacturer was able to measure RPM and other key machine parameters. With little upfront investment, they were able to improve their understanding of production processes enough to identify bottlenecks and better balance their lines. The IoT gateway lets this manufacturer collect data from several machines simultaneously. The manufacturer aggregates this information in a visual production dashboard through Tulip analytics. Now, everyone on the shop floor understands production in real-time.

By connecting this machine data to human performance data, they were able to view their operations from a holistic perspective, identifying areas for improvement, and understanding exactly how WiP flowed through their lines.

In the end, this manufacturer increased the number of units produced by 15%. They’ve met their ambitious goals for expanded production while reducing the cost of goods sold. By bringing a brownfield facility online, they were able to improve uptime and visibility in a way that made a real impact on the business.