National renewable energy provider Infigen Energy is looking at boosting the granularity of the data it obtains from its generation facilities as the next step in an analytics revamp that began two years ago.
Infigen has six wind farms as well as a small solar farm. All up its facilities, split between New South Wale, South Australia and Western Australia, have a combined capacity of 557 megawatts. (The company also has a 1200MW development pipeline.)
As a participant in the highly regulated east coast power market, Infigen receives instructions (via a flat file FTP transfer) every five minutes by the Australian Energy Market Operator (AEMO). AEMO can direct Infigen to increase, decrease or maintain the power the utility it is feeding into the grid.
“If we don’t obey those signals then there could be implications on a business level you can be fined and potentially lose your license,” said Victor Sanchez. Sanchez is an application architect at Infigen and oversaw the rollout of the Splunk data visualisation platform at the utility, which he credits with delivering significantly more effective analysis of generation operations.
Sanchez said that although interaction between AEMO, Infigen’s systems and the SCADA systems that control and monitor the company’s windfarms is fully automated, the power company has a 24/7 Operations Control Centre that monitors both the market signals and the generating facilities.
AEMO sends a token with a market signal that is fed into Infigen’s data centre. A bespoke automated SCADA control system retrieves the signal and then issues instructions to the windfarms.
Data from the windfarms is fed back to the SCADA control system as well as to AEMO, with the end-point SCADA systems aggregating and sending data every 10 minutes. In addition, there are one-minute heartbeat signals between the SCADA systems and the control system.
If one of the automated systems fails, then potentially a member of the OCC staff would have step in to ensure Infigen obeys the AEMO signals.
The initial proof-of-concept Splunk implementation delivered more timely analysis of the data generated by the SCADA systems and the control system, Sanchez said. Previously the process would involve an engineer retrieving the semi-structured SCADA data and massaging the data before it could be matched to data on the control system side.
“That could take days or weeks depending on availability of resources on-site, and resources at our offices, and the time actually processing the data,” Sanchez said.
“It was taking too much time to figure out what was happening. We operate on a five-minute basis, and within five minutes something can happen,” he said.
Working with Katana1, Infigen has built dashboards that can reveal significant incidents, such as a loss of heartbeat, and operational and market events, at a glance. The utility can also review historical data or drill down to get a more detailed view.
The single-machine proof-of-concept project monitored three of the company’s windfarms, and the implementation took a handful days. After the pilot, the company decided to go “all-in” with the platform, expanding it to all generation facilities, as well as its IT systems (including delivering security monitoring and integration with the single sign-on product used by the utility).
Infigen went with a distributed mode production implementation of Splunk to give it flexibility to grow. Although the end users were originally intended to be the OCC and IT team, it has since spread throughout the business.
“Word got out about the implementation — people saw some of the dashboards we were able to produce — and we decided to open access to people across the company,” Sanchez said.
“Right now it’s not only the OCC operator or the IT personnel that will see something, but, for example, the risk and compliance manager will get an alert if something goes wrong, or any of the executives, such as the CFO, will be able to log in and see what is happening with the windfarms.”
Sanchez said that business continuity planning and disaster recovery capability have been boosted because even in the event of the OCC being evacuated, staff can continue to monitor the interaction between generation facilities and AEMO signals from any device.
The company has a licence that allows it to ingest up to 10GB of data a day, and it is currently averaging 6-8GB across all the monitored systems. In the medium term the plan is to increase the amount of data ingested as the company experiments with more analytical tools such as machine learning.
Sanchez said that although the control system receives aggregated windfarm data every 10 minutes, the SCADA systems themselves receive sub-second data.
“We have some potential projects where we will be able to ingest more granular data – to the minute level,” Sanchez said.