How HPE is Putting IoT on the Shop Floor




Hewlett Packard Enterprise (HPE) is an enterprise information technology company that emerged from the split of Hewlett-Packard in 2015. HPE has been defining itself in the IoT market through its specific approach to industrial IoT. This approach is focused on the Intelligent Edge and relies heavily on what they call converged IoT systems, most specifically their line of Edgeline servers, which incorporate data acquisition hardware from National Instruments.

In edge applications, more of the processing power and analytics are housed where work is actually getting done: on pumps, machinery, and vehicles that are on the shop floor or out in the field. This reduces communication and processing delays (usually called latency) and bandwidth use, but does require a lot of processing under more difficult physical conditions. Edge processors need to deal with shock, vibration, extreme temperatures, solvents, and dust.

Predictive maintenance and remote guidance

HPE’s workhorse Edgeline 1000 plays a number of roles in its solutions for this client. For predictive maintenance, for example, one EL 1000 runs four virtual machines: KEPServerEX, Kafka (a distributed streaming platform from Apache), an agent server, and ThingWorx. A ThingWorx mashup outputs information into graphs, dashboards, and alerts.

Alerts can be sent to specific people with defined responsibilities, as well as providing remote access for technicians and links to HPE’s MyRoom Visual Remote Guidance, an interface for wearable tech devices.

Building a baseline capability

HPE has become a kind of stealth IoT competitor, with a much lower profile than some other players. It has focused on solving specific pain points, building expertise and partnerships over the long term.

In this case, they established a partnership with a global manufacturer of consumer goods, mostly foodstuffs. They identified some of the most common IoT project requests and formed a multi-disciplinary team across the two organizations. They picked a typical manufacturing site for their pilot project.

Operational engineers are desperate to gain efficiency. A quarter of a percent gain in efficiency, over the scale of a manufacturing operation, is a big deal. This is where IoT can really show value.

At LiveWorx 17, Steve Fearn, HPE’s Chief Technologist IoT, presented HPE’s experiences with this client.

Starting point

The chosen manufacturing site was a modern production facility. Machinery was controlled by programmable logic controllers (PLC), which connected via routers and data concentrators to a Manufacturing Execution System (MES), a real-time control system. It was a typical complex manufacturing environment, with a mix of equipment of various ages from a variety of vendors.

The plant was large, with widely separated production lines. The building was steel-framed, making a Faraday cage that blocks EM, and is thus a perfect place for Kepware, since you don’t have to worry about signal interference.

Integrating legacy systems

Engineers have been trying to get situational awareness by connecting up machines with sensors and controllers for decades. An IoT implementation has to work with these existing control systems while adding new sensors and overlaying deeper analytics—a process called “legacy interworking.”

An example is a production line powered by compressed air which needs to wait until the compressor comes up to pressure. It then sends a signal saying it is ready, and the production line starts. This system works; there is no reason to replace or modify it, and its legacy communication needs to be worked into the new IoT implementation.

KEPServerEX connects the various sensors and controllers without changing their various communications protocols. The data can be forked to both the existing destination, and to new analytic and control software, overlaying the IoT onto what is already on the shop floor.

RFID asset tracking

It’s surprisingly easy to lose track of assets, even critical assets, on a large and complex factory floor or in associated storage locations. The client particularly wanted to keep track of assets related to reconfiguration or repair. They were losing work cycles trying to track things down when they were needed.

Radio frequency identification device (RFID) prices have plummeted, making them economical for tracking large numbers of items. RFID tags were placed on the assets to be tracked. A MySQL database stored each tag and the piece of equipment associated with it.

Factories have a lot of metal equipment and structures, which block RFID signals, so RFID readers were placed in doorways and narrow hallways, where equipment being moved would pass close by. Scripts and a C# program kept track of which tags had moved where.

A visualization was created through a ThingWorx mashup, showing all factory locations and where each piece of equipment was.

Interestingly, one great benefit was changing employee behavior. Knowing the locations were being tracked made employees more likely to return them to their proper location.

IoT for the real world

HPE has deliberately positioned itself in providing rugged, workable solutions for factories seeking incremental process improvement and efficiency gains. Each project provides information on what IoT solutions will show results.