Hardware capabilities impact edge computing success

Industrial PCs offer a range of processing capabilities, form factors, and certifications that engineers must consider when implementing edge computing strategies in new applications or legacy systems.

By Eric Reiner March 6, 2019

Edge computing in cloud-connected plants has many advantages. By gathering and analyzing process data on local controllers before sending it to the cloud, engineers can visualize production data on human-machine interfaces (HMIs), monitor machine health, and schedule predictive maintenance as well as minimize data upload costs. Edge computing also heavily relies on networks, software, algorithms and communication protocols, such as message queuing telemetry transport (MQTT), advanced message queuing protocol (AMQP) or OPC Unified Architecture (UA).

However the controller hardware is an often-overlooked aspect. While many factors affect the success of edge computing applications, the hardware determines if the applications are even possible. It is important to install a high-quality PC-based controller that can handle the advanced data acquisition, processing, and cloud communication tasks that are crucial when designing and commissioning new Internet of Things (IoT)-ready machines. Open PC-based hardware is even more crucial when retrofitting legacy equipment with edge computing capabilities, which requires additional steps to obtain plant data.

As a result, there are many key considerations when selecting an appropriate edge computing device including:

  • A device’s ability to fit into cramped control cabinets
  • The quality of housing materials
  • Processing power
  • The ability to divide tasks among CPU cores.

The right hardware can vary by application as well. Understanding the available options is the first step to ensure a successful edge computing project.

Processing power, customization

Controllers used for edge computing must handle large amounts of data while completing other automation tasks. Today’s industrial PCs (IPCs) provide many levels of processing power and memory, ensuring appropriately sized hardware for each application and price point. The IPCs also should offer enough flexibility for customization in how they use available processor cores.

Edge computing devices can feature a variety of processors from a single-core to a quad-core processor, in speeds from 400 MHz to 1.6 GHz. Many-core IPCs and embedded PCs feature more advanced options, with four to 40 cores and speeds of 2.2 GHz.

When examining these specifications, it is also important to look for automation software that allows users to dedicate tasks to individual cores and make the CPU run at peak effectiveness. This will also ensure the IPC can multitask and accommodate more functions on one device. For example, with a quad-core CPU, isolating the programmable logic controller (PLC) on core 0, motion control on core 1, and HMI on core 3 allows an engineer to dedicate the final core to edge computing activities.

Local data storage and RAM capabilities also fall on a wide spectrum. Some IPCs offer from 512 MB MicroSD cards to 960 GB solid-state drives with the option to add a second storage device for additional space if needed and from 1 GB up to 64 GB of DDR4 RAM. On the other end are even more capable IPCs that provide up to 1 TB DDR4-RAM EEC and 4 TB or more on 3 1/2-in. hard disk drives. As with processing power, the specific application determines the amount of storage required. For example, cutting-edge installations with advanced vision systems will require more memory and processing power, while less complex projects will not.

In greenfield edge-computing applications, IPCs will likely control entire machines or lines in addition to analyzing and making corrections based on real-time process data. Implementing this level of control in a brownfield setting would require a retrofit, which may not be worthwhile to engineers hoping to gather and leverage more process data. Without resorting to a rip-and-replace strategy, an IPC can easily gather data from the legacy fieldbus and/or PLCs, filter it, analyze it using advanced algorithms, and send the required information to the cloud.

Form factor, material choices

The edge device must be well suited to the factory and enclosure. Production environments, depending on what is being manufactured, run the gamut from very hot to very cold, and existing control cabinets often cannot accommodate much additional heat. IPCs and embedded PCs in compact form factors can withstand temperature extremes and support many application types.

IPCs with rugged metal housings can, for example, integrate into a variety of spaces with the option for cabinet installation or DIN-rail mounting. The durable metal construction ensures they are ready for use in many plant environments, and when outfitted with a heat sink or fan, it also minimizes the risk of overheating the enclosure. In addition, selecting a certain processor series can minimize the risks involved with high temperatures. A broad selection of connector ports, such as gigabit Ethernet, USB 2.0 and 3.0, DisplayPort (a Video Electronics Standards Association connector standard), along with scalable memory and RAM make these IPCs excellent edge devices for new applications that demand the highest performance levels and for existing systems that require edge capabilities.

When implementing a new control system, embedded PCs offer additional benefits. These DIN rail-mountable controllers connect directly to input/output (I/O) modules in the control cabinet, further minimizing hardware footprint and cabling requirements. Embedded PCs equipped with either industrial-grade plastic or metal housings can operate with minimal heat dissipation and in a range of temperatures, often spanning as high as 50°C and as low as -25°C.

For engineers looking to upgrade or implement new PC-based control systems, it will be worthwhile to invest in the minimal effort required to retrofit the cabinet. However, for those who want increased data acquisition and analysis on the edge, with no fundamental changes to the system architecture, IPCs offer a solution with a wide variety of options to choose from to meet the application’s needs that is almost plug and play.

Certifications, software assistance

Examining these hardware, software, and networking factors will lead engineers in the right direction, but they may encounter similar options. In these instances, examining product certifications and software capabilities can break a tie.

Depending on the cloud service in use, Microsoft Azure Certified IPCs or controllers approved by Amazon Web Services (AWS) can provide a better option and peace of mind. Controllers with multiple certifications are designed to reassure engineers hardware changes will not be necessary if there’s a change in the cloud service. Regardless, PC-based controllers are well suited to accommodate public and private cloud systems.

In addition, PC-based controllers offer the best operating system and processing power for edge computing applications. Engineers should ensure any edge device supports Microsoft Windows 10 IoT and has the capability to handle future upgrades.

The goal of edge computing is to make continuous improvements far into the future. With the proper software and hardware selection, a successful edge computing implementation will boost machine performance, reduce downtime, increase throughput, and help ensure factories remain as efficient and competitive as possible. Carefully considering the many PC-based controller options will give controls engineers an advantage.

Eric Reiner, IPC Product Specialist, Beckhoff Automation. Edited by Emily Guenther, associate content manager, Control Engineering, CFE Media, eguenther@cfemedia.com.

MORE ANSWERS

KEYWORDS: Edge computing, controllers, industrial PC

Factors that impact edge computing success

What to consider when selecting an edge computing device

Consider this:

What capabilities does your IPC need for successful operations?


Author Bio: Eric Reiner, IPC and MX-System Product Manager, Beckhoff Automation LLC.