Columbia Pipeline Group (CPG)’s odyssey from being a low-cost supplier that might not deliver to a valued part of TransCanada began January 2, 2010 with the failure of a $50 part. “It cost us millions because we didn’t fulfill our contracted purchase orders,” said Brian Sloan, manager of automation and electrical engineering, TransCanada. “Let’s just say we were challenged to get to where we could predict that sort of failure before it occurred.”
Sloan presented “Columbia Pipeline Group Reaches 99.5% Reliability with a Connected Control System” in the Oil & Gas Forum at Automation Fair this week in Atlanta.
TransCanada owns 31,000 miles of pipe with 6 million installed horsepower and 550 billion cubic feet (BCF) of storage. It delivers one-fourth of all the natural gas moved in the United States. Over the past few years, gas production and consumption has been rising. Industrial facilities rely on gas deliveries to make production, and commercial and residential deliveries are also critical, especially during the cold winter months. “So we need to be reliable to deliver on our contractual obligations,” Sloan said.
In 2010, the group’s 100 pipeline compressor stations used multiple control systems, with controllers and HMIs from a broad array of suppliers. These disparate control systems were difficult to support due to the wide range of required expertise. They were also expensive to maintain, with multiple support contracts, many components – and difficult to upgrade to modern security levels.
[sidebar id =1]“Lack of reliability was a major risk factor for ongoing operations,” Sloan said. “Our control system was outdated—we were buying components off of eBay. We needed a platform everyone could troubleshoot—to be able to train them on one platform, not multiple platforms with multiple support contracts.”
The company also wanted to rationalize data. “We had multiple control systems with multiple tag systems and databases, and multiple hands touching them,” Sloan said. “We needed data integrity.”
Solution: The Connected Enterprise
The search for a solution led the CPG engineering team to standardize on a virtualized PlantPAx distributed control system (DCS) from Rockwell Automation. The high-availability, scalable, modern DCS would share more information across facilities and up to the executive level, enable remote access and improve security.
“We configured the systems with visualization, historian, domain controller and asset management,” Sloan said. At first, they used separate Dell servers, but the cabinets were filling up and they became concerned about long-term IT support. So they decided to implement virtualized industrial servers.
High-availability, virtualized servers from Stratus Technologies reduced the physical space of the hardware, removed the need to procure hardware for each new application, and reduced the likelihood of downtime due to hardware errors.
Each virtualized server runs multiple Rockwell Software FactoryTalk applications including Historian, View Machine Edition, View Site Edition, AssetCentre and ViewPoint. ViewPoint allows operators to access the HMI applications from any location through a Web browser, so they can make real-time decisions.
Combined with historian software, the visualization capabilities of the system enable monitoring, reporting and data recording for immediate and future access. The historian platform integrates with the corporate-level OSIsoft PI data warehouse system.
“Virtualization allows for hardware independence, improves longevity, increases flexibility and scalability, and increases uptime,” Sloan said.
The information-enabled system runs on EtherNet/IP, allowing data to flow easily from engines to the main control panel, the motor control centers (MCC) to the programmable automation controllers (PACs), one compressor station to another, and from stations up to executive offices.
Allen-Bradley Stratix industrial Ethernet switches manage the secure, real-time information sharing. “It gives us one solution for data from the PLCs to the historian,” Sloan said. “We worked with the instrument group to name the tags and lock them down, and now the tag in the process is the tag in PI.”
Instead of multiple networks, “Everything’s EtherNet/IP,” Sloan said. “At first, our network was flat, and as we added capabilities, speeds dropped. Rockwell Automation evaluated it and advised us to split it out in to separate segments for the HMIs and OIs, for reliability, and for emergency shutdown.”
Rollout and results
With the new system installed, “We upgraded 82 stations in 16 weeks using virtual machines (VMs) that we built offsite and installed very quickly. We can flip the old one off and the new one on, and if it works, great. If it doesn’t, we flip it back while we figure it out,” Sloan said. “Now the IT guy’s job is simple, because all the stations are the same. He doesn’t have to pay attention to which one he’s working on or updating.”
[sidebar id =2]The group has just 17 people supporting 100 stations in 14 states. “With multiple stations apiece, it’s important that they can remote in to the stations,” Sloan said. All personnel are able to monitor stations remotely via FactoryTalk ViewPoint, and service personnel can use Remote Desktop Protocol to work on stations remotely. “We like FactoryTalk ViewPoint because they can walk around with iPads and see from inside the plant whether or not a fix is working.
“Data is now valid, equipment is supported, and we can get replacement parts. And we can now make good operational decisions. Before, we were able to project our needs for maintenance at the local level, but without a system-wide view, operating requirements would vary from what we expected. Now we can see into the future, and when we see a problem coming up, we can take equipment down and maintain it before it’s needed.”
For example, six weeks ago, an alarm on temperature and vibration trends predicted a failure on January 3. “We got our compressor manufacturer to replace it now, while it’s still warm outside. That prevented a potential $13 million fine if we couldn’t deliver our contracts in January.
Compressor availability went from 86% in 2010 to 98% this year. “From 2011 through June of this year, preventing unplanned outage events has saved us $9.8 million, and that’s a very conservative estimate,” Sloan said. “In fact, it’s now probably closer to $14 million.
“Without data, none of this would happen. We’re making more educated decisions based on real-time and historical operations data that we can access from anywhere, including from our headquarters. As a result, we’ve been able to significantly improve our company’s profitability.”
[sidebar id =3]
