With data science developing at a rapid rate, water utilities are poised to transform the way they analyse SCADA records, according a utility systems analyst.
At the upcoming Ozwater’16, Coliban Water Systems Monitoring Manager Peter Prevos will be presenting on the use of virtual tags to analyse SCADA data.
“With the whole data revolution outside the water industry, a lot of new thinking has developed that we, within the industry, can use,” Prevos said.
“The more we dig into our data, the more things we find out about our water and wastewater treatment plants that we normally wouldn’t consider looking at.”
Prevos said there is a view in the water sector that using SCADA data for analysis, rather than operational direction, is too difficult.
“I’ve come across the opinion before that it’s one for the too-hard basket, because SCADA data is so messy,” he said.
While this use of virtual tags is an industry first for the water, Prevos said this approach has been available in other industries for some time.
“Before I came to the water industry, I worked on dredging ships around the world. This approach was very common 20 years ago in the dredging industry,” Prevos said.
“I wanted to replicate that for the water industry. What we have used is a standard data science approach. You can apply this approach to any data set. What we want to do is create more virtual tags for other SCADA related problems.”
While the use of virtual tags extracts, analyses and reports on data available through SCADA, the approach also allows utilities to pinpoint, sort and clean specific data, and offers a clearer contextual backdrop for analysis.
“My position at Coliban Water is to create value from data. When I started looking at the SCADA data, I heard all of these horror stories about how bad it was,” Prevos said.
“SCADA was designed to give operators information about today, maybe yesterday, to know whether processes are under control and how to optimise the running of their plants, and it is the perfect system for that. But from a data science perspective, that information is not really suitable for post-hoc analysis.”
Prevos said one of the most important aspects of virtual tag use in analysis is its ability to combine different data sets, allowing utilities to customise their analytic approach for each different plant.
“What we have been able to do with our software is identify where the messy parts are and, when it comes to actual plant performance, each of the treatment barriers that contributes to local performance is able to be assessed at a level that we previously didn’t have.
“The biggest value-add is knowing more about how we collect data through SCADA. Meaningful data collection is at the crux of this software.”
The software was designed specifically to help in analysing SCADA data in accordance with the Microbial Health-Based Targets (HBT) manual published by the Water Services Association of Australia in 2015.
“The industry is definitely on the right path. There is a lot of work being done, but different utilities are at different stages,” Prevos said.
“Specifically with respect to SCADA, there is a lot of science now being developed in regards to analysis. That is the way to solve problems.
“But now, this field of data science is emerging, which is really adding systematic approaches to what has come before.”
Prevos will discuss Coliban Water’s new data analysis software at the upcoming Ozwater’16, including an in-depth look at virtual tag use and analysis.
See the full program here.