Oceans of Data – An Online Hack

Back in October 2021, CACI took part in the Northumbrian Waters Innovation Festival. One of the events we participated in was the Oceans of Data hack sponsored by Sia Partners – Chris, Ian, Matt and Myself (Paul) all attended remotely representing CACI. The data hack wasn’t scheduled to start until the Tuesday, but a kick-off event was being held on the Monday morning, the highlight of which was an impromptu workout session run by none other than Mr Motivator himself; it was high energy fun and exhausting! If you want to see just how exhausting… watch this.

Taking the rest of the day to recover from that, we got hold of the data sets being provided for the data hack under an NDA and set about our challenge. For the Oceans of Data challenge, we were asked how can we leverage insights from our ocean of operational data to make our services even more reliable and resilient?

“Data introduction: In 2019, NWG invested in Cloud technology to capture and store the vast volumes of data generated by our 30+ SCADA systems – now creating almost 45 million data records per day. While great progress has been made to extract value from this data, we are confident that much more insight can be generated about the performance of our assets, their rate of deterioration and the way they respond to changing operational conditions”.

The data provided was broken down into two broadly distinct sets of data:

1. Work Order information out of their Asset Management System (Maximo).
2. Sensor data from their extant SCADA networks across several sewage treatment plants. There was also some supplementary information provided, including an asset classification hierarchy that Northumbrian Water have crafted and a series of SCADA Mimics (high level process flows).

The stated objective of the data hack was to provide a path to holistic analysis of the data, particularly, correlating sensor data to work order data providing a history of activity leading up to corrective work orders. This historic information could then be used in predictive modelling of future failure; thereby turning all the reactive maintenance activity (which is relatively time consuming and costly) into scheduled maintenance activity (which is planned, less disruptive and much cheaper).

Day 1

The hack consisted of presentations by Northumbrian Water around the operating of sewage treatment plants, to provide us with a basic understanding of the business, forming the teams for the hack and a first pass at data analysis; working out what could and could not be done in the rest of the time available.

The event had an amazing mix of people attending, ranging from data scientists and machine learning (ML) experts to software engineering expertise. My team consisted of software engineering specialists, which worked well for our chosen idea as our teams abundant Java and React experience proved invaluable. Along with the experts attending, we could also call on our own ML and data scientists within CACI if we needed any advice.

Having dived into the data, Team CACI decided to play to our strengths and develop something, focusing on the data sets in isolation and hopefully providing Northumbrian Water with a new way of interpreting their data.

Day 2

We spent the day preparing the data into machine readable formats and finalising the approach for the application and then actually building it. We chose to provide a heatmap analysis of the two datasets, with Work Orders being visualised across the asset classification hierarchy and Sensor data being visualised across the SCADA Mimics (high level process flow diagrams) that had been made available.

Day 3

Day 3 was focused on refining the application – bug fixing, visual styling, cutting bits we had no time for (such is the way of the hackathon!) – and creating the presentation, which included the demo that we would show the judging panel.

Unfortunately, we ran out of time on the sensor data visualisation, so that hit the cutting room floor. This allowed us to focus on bug fixing the application and making it look more like the asset classification hierarchy we had been shown – making it relatable to Northumbrian Water.

The afternoon of Day 3 was spent presenting the solutions from all the teams and being judged. As the presentations progressed (Team CACI had opted to go last) a common theme arose: correlating the two data sets was very difficult and to do it with any degree of accuracy/success would require more data.

We were the only team who had made a working proof of concept, while others had focused on presentations. Our interactive website that the judges could play with was well received. The fact that we had used Northumbrian Water’s own asset classification hierarchy was positively acknowledged and provoked quite a bit of interest. We were then subjected to some questioning by the judging panel before they retired to select the category winners.

I am delighted to announce that we won the Best Data Visualisation category:

The CACI data hackathon team really impressed the judges with their solution, a prototype data visualisation to help direct our maintenance teams toward our most problematic assets, structured to represent our asset hierarchy in clear format to enable rapid insights. We enjoyed working with the team who remained enthusiastic despite the challenging task we’d set for them! Thank you and well done.

Northumbrian Waters

Takeaways from the Event

There were a few takeaways from the data hack and the event as a whole:

• With preparation, virtual events of this kind can really work, congratulations to Northumbrian Water for putting on such a successful event.
• As ever, time is your enemy, the teams only had about 16 working hours available to us to produce something and everybody made the absolute most out of the time allowed.
• Finally, this was a team effort, but massive thanks to Chris, Ian and Matt who did all the hard work, I just presented it at the end. Not forgetting Ryan who nominated us, and the Critical National Infrastructure team, for selecting the event.

For more information on the event and our proof of concept expertise, or our wider CACI capabilities, contact us at cni@caci.co.uk.