Special Report: Prognostics software identifies potential issues

version introduces an improved interface that includes an "at-a-glance" feature showing a particular component or machine's estimated useful life. It presents a clear message to the user that a component, such as an electric motor, is at risk of failing within a certain number of days. It also displays the risk of failure as a percentage within that time.

Senseye says its product is unique because it offers automated condition monitoring combined with remaining useful life analysis without requiring any condition monitoring experience from the user.

Knowing the remaining useful life of a piece of equipment helps companies adopt cost-effective maintenance practices, typically allowing for a 10 percent to 40 percent reduction in maintenance costs and a downtime reduction of 30 percent to 50 percent, according to the company.

The company says the system collects operational data and analyzes it using machine learning, advanced statistical algorithms and the Internet of Things (IOT) to identify abnormalities that require further inspection or repair.

"We're currently working on injection molding machines, looking at things like the motors, screw piston, melting times and temperatures, although the technology could also be used with extruders," company co-founder Alexander Hill said.

"We're trying to make predictive maintenance really accessible to people who don't have a background trained as a condition monitoring analyst or data analyst," he said. "As far as possible, we want to remove humans from this loop of understanding maintenance data or condition monitoring data and relating that back to the health of the machine."

The software collects and analyzes data from machines and third-party sensors, and examines historic performance when making its predictions. Senseye's software can monitor an unlimited number of machines, something that would be expensive or impossible with humans analyzing the data, Hill said.

Oden Technologies

Oden Technologies, which develops process-monitoring software and hardware, is helping processors monitor their equipment, diagnose the root causes of equipment failures and increase efficiency.

For more than two years, Oden Technologies has been helping extrusion companies collect equipment data. Its core missions are eliminating waste, increasing production rates, reducing machine failures, optimizing settings and helping manufacturers run their operations more efficiently, said Willem Sundblad, company co-founder. After getting its start in extrusion, the company is now expanding into all types of plastics processing.

The company is investing heavily in artificial intelligence (AI) and machine learning to predict machine and prevent product quality failures.

In addition, Oden recently began leveraging Google Cloud services to help the company develop algorithms to accurately predict machine failures. Oden already helps its customers diagnose problems when they occur through root-cause analysis, so they can get the equipment back up and running quickly.

But, with its customers' cooperation and assistance from Google, the company is collecting machine data that could help diagnose conditions that lead to failures. Collected data is hosted on Google's servers, Sundblad said.

"We work very closely with Google's infrastructure, and we're hosted in their data center, so we collaborate on Internet-of-Things applications and specifically machine learning and artificial intelligence," Sundblad said. "Their service enables our engineers to work with our customers, our partners and machine manufacturers on getting to zero unplanned downtime and getting to zero failures. That's our utopian vision, but we're not there yet."

Accurate predictive maintenance depends on AI and machine learning. It's not a simple plug-and-play technology, Sundblad said. It's important for customers interested in predictive maintenance technology to know how long a supplier has been working in the industry, details about its dataset and the accuracy of its predictions.

"I often feel it sounds more like a snake-oil salesman because I know how difficult machine learning and artificial intelligence is," Sundblad said.Oden's machine monitoring collects millions of pieces of data every day from every production line. All that information is analyzed through the company's proprietary platform.

"Without those real-time data points and access to historical production data, you can't even begin to develop machine learning technology, and thus, predictive maintenance tools, in a cost-efficient way," he said.

Oden's software tracks a variety of data, including melt temperature, melt pressure, motor load, revolutions per minute, screw speed, amps going into machines, and speed of production. Information is gathered directly from a machine, PLCs, additional equipment sensors or any combination of the three, and is fed into a small data-collection device known as the Hugin, which sends information wirelessly to Oden's platform. Customers can then access the data on their own computers or smart devices.

"Our platform allows [customers] to solve quality issues and machine failures in a matter of minutes rather than weeks or months just through having access to the data in our platform, where they can analyze it and they can do very detailed root-cause analysis," Sundblad said.Oden will offer additional predictive maintenance services in the future through a higher-tiered product offering.

Semeq

Semeq, which deploys a team of field technicians to monitor customers' equipment at more than 250 plants around the world, has expanded its services. In recent months, the company started offering online vibration monitoring equipment that can identify potential problems without the need for on-site inspectors.

"We've added a new wrinkle this year, which is an online, wireless vibration accelerometer that you attach to your equipment or we can attach it, and we take a reading every 30 minutes," said Andrew Rodes, director of business development. "It goes up to the cloud, and our centralized laboratory analyzes the data and if there is anything going on, you find out right away."

This vibration-monitoring equipment can more quickly identify developing equipment problems. The company began offering the service because the costs of providing the technology have been falling. Vibration monitoring works with a variety of equipment, including larger injection molding machines, extruders and all types of gearboxes.

"It's wireless, and it's online," Rodes said. "Those are the big things. There are no wires attached."

It uses Bluetooth technology to transmit information wirelessly that is then uploaded to the cloud for analysis by Semeq.

"Essentially, we will set some alert levels," Rodes said. "If some of those alert levels are reached, a human being will get a notification and analyze the data. If they find the alert is credible and serious, they will contact the plant right away."

If a more slowly developing fault is detected, Semeq can generate a report online and send an email to the company.

IQMS

IQMS provides predictive maintenance tools to customers through its IQMS manufacturing execution system (MES) that includes Real-Time Production and Process Monitoring software.

The system collects data from existing PLCs on production equipment, as well as additional sensors that can be installed by customers or the equipment manufacturer.

The IQMS MES system also can assist manufacturers with planning and scheduling, measure overall equipment effectiveness and key performance indicators, track when equipment begins operating outside normal parameters, and assist with inventory optimization. However, increasingly, the company is seeing users take advantage of the predictive-maintenance features, said Ed Potoczak, director of industry relations.

The company is working with customers to identify sensor readings that might suggest a coming equipment failure. In addition, customers are installing more sensors on their equipment. It is becoming more common for customers to not only monitor the injection molding machine where the plastic flows, but to also put sensors on pumps and motors that can predict failures by detecting unusual noises, vibrations and abnormally elevated temperatures.

"We're seeing people creating more granularity or more detail where they are monitoring more than just the ram or the screw or the nozzle only," Potoczak said. "Those are certainly very important aspects of injection or melt control, but they're looking at things like hoppers, feeders, gates, cooling systems, and monitoring the robots themselves to make sure they are picking, placing, trimming, doing all of those things well. It's more of a holistic approach to monitoring."

Not only does the extensive monitoring allow manufacturers to respond more efficiently to an equipment failure, but it also is beginning to predict failures.Also, customers are increasingly interested in viewing more than one cell at a time on a screen that depicts most or all of the production machines on the floor, Potoczak said.

In addition, IQMS has been building the data sets necessary for plastics producers to use their monitored data for predictive maintenance. The knowledge of equipment and what parameters are linked to equipment failures is key.

C-Labs

In September 2016, C-Labs introduced Machine Monitor, a new product that provides operational awareness to machine operators and plant managers. The information obtained by C-Labs software can then be used with third-party machine learning or AI software to detect trends that could indicate the need for maintenance.

"Once one understands the basic state of a machine, action can be taken to change an undesirable condition instead of letting the undesirable condition persist," said John Traynor, the company's COO. "At C-Labs, our experience is that many firms in North America are just starting with automated collection and analysis of machine data."

C-Labs Machine Monitor software provides basis information for connected machines and can be connected to sensors to retrofit older equipment. This allows plant managers and supervisors to gather information about how their machines are operating.

Once baseline machine information is available, companies can use machine learning or AI systems to recognize when a machine's operating state is drifting away from normal, Traynor said.

"Our primary focus is to provide live access to industrial equipment and IoT data without compromising or side-stepping corporate IT security policies," he said. "C-Labs software allows some basic processing rules that provide for an "if-then" construct. For example, if the temperature of a piece of equipment gets too high, the software should send an alert.

"This capability can be extended with third-party [machine learning/artificial intelligence] applications for more complex situations. There are many [machine leaning] and AI tools available."

Bruce Geiselman, senior staff reporter

[email protected]

For more information

C-Labs Corp.Bellevue, Wash., 425-999-3295, www.c-labs.com

IQMSPaso Robles, Calif., 805-227-1122, www.iqms.com

Oden Technologies Inc.New York, 800-230-9063, https://oden.io/

Semeq Inc.Arlington, Texas, 214-269-5287, www.semeq.com

SenseyeSouthampton, England, +44 845-838-8615, www.senseye.io