In the old days, condition monitoring was conducted standing in front of equipment using some combination of the five senses. Fast-forward to today, when assorted sophisticated and increasingly affordable condition monitoring technologies are available to automate data collection, even from afar.
The latest development – the industrial internet of things (IIoT) – allows all of that information to be streamed and centralized in a repository shared by other Big Data, where advanced analytics, algorithms, and artificial intelligence combine to improve and expedite asset management decisions.
Some companies have already begun taking their condition monitoring (CM) and predictive maintenance (PdM) practices to this next level. The examples that follow, representing six different industries, illustrate the promise of IIoT-enabled CM and how real-world companies are moving in that direction.
San Diego County-based Stone Brewing Company, the 10th largest craft brewer in the United States, uses Inductive Automation’s Ignition as its supervisory control and data acquisition (SCADA) system and industrial IoT platform. Ignition is used to monitor pumps and valves, milling systems, conveyor systems, and package lines. This is done via run hours and amp draws monitored on pumps, valve cycle counts, tension switches, and general fault alarms for conveyors.
The brewer first implemented the SCADA system four years ago. “Once we saw what we could see and do with the real-time information, it became our go-to idea for further expansion,” says Garrick Reichert, senior engineer at Stone Brewing Co. “We tied it into our ServiceNow computerized maintenance management system (CMMS), which receives alerts from Ignition to generate and assign work orders, and then into our enterprise resource planning (ERP) system to see finished goods and inventory tracking information in real time.” He adds, “It allows us to really stay ahead of issues that affect reliability.”
His team looks for conditions such as overheating pumps and worn valves and addresses trips and jams as soon as they appear. Work-order generation is automated, and escalation plans and call trees have been developed to expedite issue resolution. “We repair or replace whatever we can before it fails, because downtime is a killer here,” explains Reichert.
Current and future uses include control of pumps and valves, statistical process control (SPC), historical analysis, overall equipment effectiveness (OEE), management of recipes and work orders, tracking of critical downtime, tracking of key performance indicators (KPIs), and transaction of finished goods out to the warehouse. Stone Brewing also uses Ignition to control its water reclamation facility.
Based on the results of the solution at the Escondido, CA, brewery, the company now plans to roll it out to its Richmond, VA, location. “Having this ability to see what we can now see has saved our behinds a few times,” says Reichert. “Once you have the data, you’ll know how your assets are really performing, and you won’t always have to be scrambling to find parts or to buy parts or to react to downtime that could have been prevented,” he adds.
Offshore drilling contractor
Songa Offshore, an international midwater drilling contractor headquartered in Cyprus, is rolling out a digitized IIoT approach to maintenance. It plans to move from calendar-based to condition-based maintenance for all of its semi-submersible rigs to enable more-reliable and cost-effective operation.
Signals from sensors attached to engines, pipes, and other critical equipment will be collected and transmitted via the IFS IoT Business Connector to the company’s IFS Applications software, where work orders will be automatically generated as needed.
The first phase of the project involves 600 assets on each rig, where maintenance will be triggered based on an asset’s hours in use. “Imagine collecting 600 asset readings and then registering these in a maintenance system every hour, manually, on every rig – it says something about the potential improvement” from switching to the simplified and more automated IoT approach, says Cato Sola Dirdal, IT director at Songa Offshore.
“When you operate advanced, complex, and fully integrated digital assets, it would be extremely workload-demanding to maintain these assets the conventional way,” explains Dirdal. “Instead, we want our intelligent equipment to tell us when it needs to be repaired.”
Electric power company
Duke Energy is using the IIoT to reduce unplanned downtime and improve reliability. Bernie Cook, formerly the director of maintenance and diagnostics for central engineering at Duke Energy, described the company’s SmartGen program in a 2016 ARC Industry Forum presentation titled “Application of IIoT for Predictive Maintenance.”
The company, which supplies and delivers electricity to more than 7 million U.S. customers, conceived its SmartGen/IIoT initiative as way to avoid major catastrophic failures following one such failure in 2010. That year, a transformer explosion in Florida resulted in more than $10 million in damages and a significant loss of power generation capacity and associated revenue.
Manual condition monitoring data collection and analysis needed to be replaced with smart automated monitoring and diagnostics (M&D) technology. “We wanted to shape our future with technology innovation and workforce optimization,” says Cook. “In IIoT-speak, what we wanted were smart-plant connected assets.”
SmartGen goals included the use of critical equipment sensors; smart diagnostics and prognostics; data integration and visualization; enhanced reliability; zero safety and environmental events; and workforce optimization.
Duke Energy has so far looked at more than 10,000 assets across 50 critical plants; installed more than 30,000 sensors; put in a data acquisition and M&D network with 2,000-plus nodes enabling remote analysis; built an M&D center based on the existing Schneider Electric Avantis PRiSM APM software; and with PRiSM’s machine learning and the added sensors, increased the number of prediction models to 10,000. It has also added integration and visualization to identify the most-critical issues.
“All of this infrastructure is to help us understand, of the thousands of assets across the fleet, these are the top 10 or 20 things you need to look at right now,” explains Cook.
In its first three years, the SmartGen Program saved Duke Energy $31.5 million in avoided repair costs. One example “find” avoided an estimated $4.1 million when the M&D center picked up a small change in vibration in one of the turbines. It was sub-synchronous information that would not have been detected without the advanced sensors and diagnostics on the turbine. Plant engineers were notified; they determined the root cause and applied corrective maintenance, saving the turbine from failure.
The rate of cost avoidance with SmartGen is expected to increase with further machine learning and newer sensor technologies.
Ciner Resources, a trona mining and soda ash production business in Ciner, WY, enabled specific IIoT-enhanced condition monitoring solutions plantwide on certain critical equipment over the past two years. The goal was to predict and reduce process downtime. Conventional analytics had been unable to warn of preventable equipment failure, but the new solution has shown potential to significantly reduce unplanned downtime.
The company chose PI from OSIsoft to perform analytics and generate notifications via email or back via the distributed control system (DCS). Falkonry is a pattern-recognition artificial intelligence system used on top of PI to help with more-complicated analytics for condition monitoring. Most of the systems monitoring vibration, infrared radiation, and oil come through the Foxboro I/A 700 series DCS into PI.
“We monitor many different asset types” for predictive maintenance, says Jolene Baker, IT applications specialist/project engineer at Ciner Wyoming. “For example, we were able to stop a rebuild of a Schwing pump by capturing an ‘in-progress’ failure condition. The repair was roughly one-third of what it would have cost without proactive action.”
Vertimills are another example. When a Ciner Resources Vertimill vertical grinding mill is down, 60% of total production is lost. To avoid this, predictive analytics were needed with insights into ore grade quality, mechanical issues, and process anomalies. The new solution found patterns to create prediction models for conditions such as bad ore and inadequate media charge, allowing plenty of time for a corrective response instead of responding reactively to failure.
Baker now believes they are rounding the corner on the pilot mode and are ready to start sitewide solutions. “We will expand our CM to more critical equipment and processes and start building a condition-based maintenance (CBM) model that will work with a Microsoft solution to automate emergency/nonemergency work orders,” she says.
“The entire company, from our CEO on down, is supportive of the SMART Plant initiative,” adds Baker. Ciner Resources CIO Scott Schemmel implemented the companywide solution strategy, and Baker works as the lead for the initiative with help from continuous improvement manager Jeremy Coffi. “Proving value can be difficult and is sometimes only seen in hindsight, so it takes a positive attitude and willingness to succeed to keep the ball rolling on the initiative,” she observes.
Gerdau, a leading producer of long steel in the Americas, is implementing IoT-enabled software and services from GE Digital to help it identify and correct impending equipment failures and process problems. The company tested a small-scale deployment before deciding to roll out the solution to connect 600 assets in 11 plants across Brazil.
It is a digital transformation leveraging the cloud, Big Data, condition monitoring technologies, online diagnostics, artificial intelligence, and predictive analytics as a means to generate early, actionable warning of potential asset issues. Adding lead time allows the steel manufacturer to be proactive in its maintenance processes, rather than searching for and reacting to problems.
For Gerdau, the GE asset performance management (APM) solution consists of GE’s SmartSignal and Historian software, services, and remote monitoring and analytics out of GE’s Industrial Performance & Reliability Center (IPRC) in Illinois.
The pilot test quickly persuaded the steel company to roll out the solution. During proof-of-concept monitoring of 50 of its assets, two “catches” produced immediate value. The emerging issues were addressed with planned corrective maintenance, avoiding impact to the business and producing savings equal to the cost of the pilot.
“Gerdau is incorporating greater agility and autonomy in operational decisions via digitalization,” says Andre B. Gerdau Johannpeter, CEO of Gerdau. “We are focused on creating value and enhancing the competitiveness of our operations, and our partnership with GE will definitely provide important support for this challenge.”
Power transmission substation
American Electric Power (AEP), one of the largest electric utilities in the United States, is using the IIoT to improve the maintenance of its fleet of substation equipment without significantly increasing maintenance costs. Jeff Fleeman, director of advanced transmission studies and technologies at AEP, discussed the process of transitioning to PdM with the AEP Asset Health Center Program in a presentation at the 2016 ARC Industry Forum.
AEP works hard to ensure high reliability of its transmission network assets, Fleeman said. Though the power lines are designed for redundancy, the substations and transformers may not be, and the highly distributed nature of transmission assets makes service call efficiency an imperative. Other concerns include aging personnel and infrastructure, stringent regulatory oversight, and limited maintenance budgets.
To replace time-based maintenance practices, AEP partnered with ABB to develop the ABB Asset Health Center (AHC). AHC automates data collection and analysis, calculates remaining life, and proposes actions based on an asset health index. Circuit breakers, transformers, and batteries are among the asset types monitored using the new AHC platform.
AHC’s performance models contain algorithms to assess asset health based on online asset health sensor data, real-time SCADA operating parameter data, and offline data such as the maintenance history and supplier information.
The solution was put in place in phases over the past few years and is providing new insights for the AEP team. For instance, dissolved gas and oil had been seen as the primary cause of failure for power transformers, but after a brand-new transformer violently failed due to undetected shipping damage in 2014, AEP learned the premise is true only for slow-evolving failures.
“We had hypotheses that partial discharge (PD) detection for transformers of a certain type was critically important, and that has been proven now through AHC,” Fleeman says. “We can predict fast-evolving failures with PD before we’ll ever see a gas alarm go off or a gas trend indicating a problem.”
“We’re proud that we’ve committed ourselves to it and got a system that’s working, but we would like to see many transmission owners moving the same way,” says Fleeman. He believes that great adoption would “help to develop the market where people can make the more low-cost, low- or no-maintenance, simple sensing needed to detect issues that will lead to the failure or degradation of equipment, and impact the customers and public.”
Will it work for you?
These six companies, all realizing the value of IoT-enabled CM, are enthusiastic about its potential. “If you don’t already have it, it’s where you should be, considering where we are heading with automation,” suggests Stone Brewing’s Reichert. “As our director of engineering puts it: ‘What gets measured, gets managed, gets improved.’ ”
For those considering implementing IoT-enabled CM, Ciner Wyoming’s Baker recommends a “positive attitude, patience, and belief in what you are doing. And interns—interns are excellent analyzers.”