5 Steps Towards a Digital Water Transformation
“Collecting lots of data, but not leveraging it effectively.”
In a Black & Veatch survey of more than 200 U.S. water industry stakeholders, 50% said they were not leveraging data effectively, despite collecting lots of it. Only 5 percent described their data management practices as a “robust, fully integrated approach.”
It’s time for change in the water industry. Unfortunately, the intersection of customer expectations, climate pressures, an aging workforce, and an increasing amount of data has put utility leaders in a reactive state.
What is Machine Learning for Water?
A recent report by Bluefield Research forecasts that water operations are turning to AI and machine learning to address network and customer challenges, including water loss, aging infrastructure, and unforeseen climate events. Their report also found that spending on connected hardware, software solutions, and digitally-enabled professional services will reach a combined $92.6 billion over 12 years –– with (AI) technologies representing $6.3 billion of the share.
In the increasingly lean water industry, artificial intelligence allows treatment centers to gain a competitive edge and develop fully intelligent systems.
What Can You Do With Machine Learning and Water Resources?
Machine learning is the creation of systems and processes that improve automatically using data. When applied to water treatment, machine learning enhances the ability and knowledge of your existing team by using advancing technologies to help with menial tasks. For example, you can use machine learning to:
- Predict water asset failures
- Prevent future downtime
- Automatically analyze problems
- Automate manual work
- Detect Underperformance
- Crease User-Editable Analytics
What Are the 5 Steps To Start a Digital Water Transformation in Utilities?
By working with renewable operators across tens of thousands of assets, we have identified five key steps to kickstarting an operational digital transformation.
Step One: Start With a Specific Use Case
Start by identifying the problem that you want to solve. Starting too broad (like “we want to derive value from our data”) can add to the feeling of data-overwhelm and does little help direct us to where to start. Instead, pick problems like, “we want to have more visibility into why we are losing pressure, we want to know if a pump is failing, if there is blockage in the line, an electrical problem?”
Step Two: Review the Data
Next, review what data you have. It's okay if there is minimal data – it's better to start small and work towards the goal by adding new data loggers.
Once you review where your data is being generated from, you should review where the data is stored, how often it is collected, how much historical data you have, and how frequently it is measured. Chronicling this is important because it can determine the types of analytics you create, additional data you collect, or sensors that need to be upgraded.
Step Three: Focus on Culture and Expertise
Now that you know what problems you are trying to solve and what data you are starting with, you can determine who is spearheading the digital transformation. Whoever is in charge of the digital transformation needs to have three characteristics:
- Be willing to adopt new technologies
- Be open to new ideas
- Be strategically adaptable
Additionally, an integral aspect of a successful digital water transformation is having company-wide buy-in. Picking the right leaders is a significant first step to accomplishing this.
Once you choose who is leading the charge, educate and align with additional critical members throughout the organization. Educate everyone about the importance of making this shift and what is needed from each member of the operation. By aligning with the organization as a whole, everyone can understand that this is a culture change that they should be excited about.
Step Four: Choose the Right Technology
As important as picking the right leader is the choice of software that you decide to operate. Does it compliment your existing expertise? Is it making life easier and your operation more efficient? Just as the person you pick to lead the charge needs to be nimble, the technology you choose to complement needs to be easy to implement and not static.
Choosing software that is easy to edit and adapt on an ongoing basis increases the likelihood of adoption and ease of transition.
Step Five: Build Analytics, Adapt and Iterate
Now that you know what you want to solve and with who, you can assess the types of analytics you need. While this is something that your strategic software should be able to manage for you, it is interesting to outline the possibilities:
- Anomaly detection - doesn’t need much or any historical data - models are built to run going forward.
- Machine Learning models - require extensive historical data and need to be assessed as to their need and practicality.
The most important thing to remember with this process is that it is iterative. Continue to assess, reassess, change and adapt. Being agile will allow you to derive the most substantial results.