Analytics That Engineers in Water Can Really Use

Water utilities have no shortage of data – real time and historical, and the promise of analytics to harness this data is well documented and widely promoted. However, how does the hype compare with reality when it comes to non-data scientists using these tools and driving meaningful outcomes? The benefits are clear for analytics, mine your existing data to reap the rewards:

• Drive down cost
• Improve efficiency
• Improve accuracy for water demand planning
• Decrease unplanned downtime
• Improve chemical management

Wrangling data, building models and algorithms are often considered the realm of data scientists, resulting in analytics projects that limit the number of use cases and therefore the wider benefit across a water utility. As an expected outcome, until now most engineers and operators have lacked confidence in being able to take advantage of analytics software.

Industrial analytics are becoming more accessible for water utilities using out-of-the-box training models tuned for specific use cases, with analytics solutions like GE Digital’s Proficy CSense.  Using a self-service user interface, engineers can combine data across industrial data sources and rapidly identify problems, discover root causes, predict future performance, and automate actions to continuously improve quality, utilization, productivity, and delivery of operations.

In a recent example, GE Digital helped a mid-sized water utility predict pump failure up to 16 days in advance, using Proficy CSense. Of note, this was achieved without writing a single line of code and resulted in a trend chart that helps engineers identify failure.

The component identified as causing the pump failure was a critical bolt prone to corrosion - however it was challenging to visually inspect based on its location.  When the bolt eroded, its threads would loosen and lose contact, allowing the impeller to wobble.  The extra vibrations created by this movement would result in more damage to the motor and its coupling. Eventually, the bolt head would separate, resulting in the impeller dropping out of the housing and causing a catastrophic failure. The outcome is this cheap bolt takes this very expensive pump out of commission for weeks.

This was an expensive point of failure that needed to be eliminated.

Using analytics and a trained data model, patterns and changes to vibration signals are now being monitored, and future failure can be detected. The outcome is the water utility has two weeks to schedule preventative maintenance versus wasting resources on unplanned downtime, resulting in only a one-day disruption versus weeks.

All of this was possible without writing a single line of code and using historian data that was readily available.  The magic happened when the GE Digital water algorithms were trained.

In this case, text-based maintenance records were ingested. Analysis of those records found two instances of pump failure and correlating vibration changes that had previously been overlooked. Using this insight, the data was cleansed and a statistical method known as “principle component analysis” was used to find the minimum number of tags needed to accurately predict failure reducing extraneous data noise. We didn’t need hundreds of data points or dozens of data sources. We just needed the right set of data that was readily available.

This new information was then combined with off-the-shelf learning models to train the algorithm for optimum, repeatable performance. The resulting data visualization easily identifies anomalies that can then be fed into scheduling preventative maintenance.

This example of how GE Digital and this water utility were able to predict prove it’s possible to pump the precise amount of water that is demanded, where it’s demanded, at specification - while maintaining the lowest operational costs possible.

If you’re collecting and storing data from your operations, the opportunities are endless. To get your imagination started, below are some use cases where analytics can make a huge impact:

• Optimizing chemical usage (e.g. ammonia)
• Reducing energy costs based on utilization of assets in processes
• Improving the accuracy of water demand and optimizing flow
• Asset failure prediction

Now that you’ve seen that analytics are within your reach, where do you want to take it?

For more information on this topic watch our video, Predicting Asset Failure and Process Changes with Self-Serve Analytics.