Partners test a data-sharing platform and learn a new style of project development.
By David Wagman, Managing Editor
A power plant may be an unlikely sandbox to play in, but Xcel Energy last year made its 715 MW Cherokee Station in Denver a place for vendors to experiment with new technology having potentially larger implications for the utility’s entire generating fleet.
The invitation to experiment at Cherokee came as a result of Xcel Energy’s Utility Innovations program, launched in 2004. The program’s executive director is appointed by CEO Dick Kelly and serves for just a year in that capacity. The program’s current director is Michael Lamb, who came from the distribution side of Xcel Energy. His predecessor, Corey Hessen, was a generation guy.
Utility Innovations is built for speed. Expectations are that the time from project conception to deployment should be 90 to 120 days. Partners are expected to be completely open when it comes to information sharing, including cost details. “We want absolutely no information boundaries,” Lamb says.
The program is also proving cost-effective for the utility. Besides providing a venue for testing innovative ideas and partnerships, Xcel Energy invests relatively little of its own capital. Last year it spent around $2 million to fund half a dozen Utility Innovations projects, Lamb says. Corporate partners spent close to $17 million. They stand to benefit if a project proves successful and can be taken to market with identifiable benefits.
Convincing power plant operators to take part in fast-track installations of new technologies can be a challenge. Lamb credits CEO Kelly for supporting risk-taking across the utility. And, in the case of Cherokee, the plant’s operators have a reputation for accepting innovation. What’s more, only around two dozen people at the plant were expected to be affected by the technology experiment.
Xcel Energy’s 715 MW Cherokee Station near Denver. Photo courtesy Xcel Energy.
The idea being tested at Cherokee was whether or not independent data management systems could be integrated to improve plant operations, including maintenance and equipment failure prediction.
Xcel Energy’s power generation business, Energy Supply, had been using data sources such as Maximo and plant distributed control system historians for several years to help predict equipment failures and improve plant performance. However, the process of combining this data proved complex and time-consuming. As a result, the full value of these data sets was not being fully realized.
The innovation tested at Cherokee was to create a software system to organize and analyze the Maximo and plant historian data independently. In addition, it combined this information to help personnel working in Energy Supply reduce equipment failures and improve plant performance.
Design components included Meridium software which is a scalable, multi-tiered, XML Web Services based application that uses Microsoft .NET technology; SmartSignal which is a .NET-based application that runs SQL Server database; and components for data exchange between SmartSignal and Meridium.
Several cases were tested using the Cherokee plant’s Units 3.
In one case, Cherokee’s 3A boiler feed pump was analyzed. The analysis showed its current probability of seal failure was 18.8 percent. Based on historic performance and assuming normal operating conditions, calculations suggested that in 200 days the probability of failure would be around 55 percent. While the seals were expected to fail every 318 days, when the ratio of unplanned to planned maintenance cost was considered, calculations showed that the optimal preventive maintenance (PM) date for seal work should be 274 days to minimize risk.
Equipment performance data from SmartSignal was used to determine if 3A was healthy enough at the 274-day point to extend maintenance even further. Here, new mean-time-between-failures statistics were developed. Additionally, information from SmartSignal was combined with Meridium’s reliability-centered maintenance analysis to reduce time to repair, determine equipment criticality and appropriate overhaul intervals.
Reducing System Operating Risk
In another test, the same Cherokee 3A Boiler Feed Pump was analyzed, this time along with pumps 3B and 3C. The system-model design assumed that two pumps run simultaneously, with a third pump as a spare to maintain the unit at full load. In calculating the expected results over 1,000 days, analysts determined that total failure related costs would be around $100,000. A more detailed analysis showed that both the 3B and 3C pumps were expected to fail during this period, but not at the same time preserving system functionality.
Power plant operators determined that $100,000 in failure costs for this system were more than justified based on revenues generated during a 1,000-day run. Alternatively, the analysis also pointed to potential opportunities to perform critical tasks during shorter outages during off-peak windows that minimize failure costs while maximizing generation. When this system-model is evaluated with a larger system model, the scheduled outage can possibly be postponed.
Reducing Unplanned Downtime
During the pilot, Generation Availability Data (GADS), Maximo data and SmartSignal data were analyzed. Integrating data in one system greatly facilitates its analysis as it allows users to make more timely and accurate decisions. Alerts from SmartSignal can be verified and steps taken to use the combined system. More intensive and rigorous reliability analyses such as root cause analyses and failure-and-effect analyses can also leverage data in a single system. It also allows users not only to measure confidence levels of expected results from these calculations, but also to have increased confidence in decisions to defer maintenance, which will ultimately result in a reduction of unplanned downtime.
Lamb says the Cherokee project brought together two vendors who previously had not worked together and added “significant value” to their previously separate systems. Based on results, Xcel Energy is now weighing the feasibility of installing the analytical system across its fleet of generating plants.
“The concept is one brick at a time,” Lamb says.
Not all bricks fit, however. A project last year would have delivered an “executive dashboard”, which was intended to power plant data to decision-makers in real time. The idea was to bypass a series of filters that sometimes can delay the transmission of information. The delay resulted in decisions being made in something less than real time.
The pilot proved it was possible to feed information from core data collectors directly to the dashboard. But it also showed the important role data filters play in providing useful information.
“It proved that interim steps are important,” Lamb says.
The Utility Innovations program is itself proving to be increasingly popular across Xcel Energy. A year ago, program managers had to approach hesitant power plant operators with project ideas. “They reluctantly agreed,” Lamb says. This year, operators are coming to innovations program managers with ideas of their own. The goal, Lamb says, is for business units eventually to “own” the process themselves.
In June, Xcel Energy’s Utilities Innovation program won an Edison Award from the Edison Electric Institute for its work to date.
“We want to be a change agent in Xcel,” Lamb says, “and across the industry.”