Good Data for Better Process

The better the data, the better a utility can control and improve processes. Many utilities invest thousands of dollars into data storage, enrichment, integration, presentment and analytics, but it’s all for nothing if the results are not actionable.

4 Criteria of Actionable Data 

For data to be considered actionable, it must be relevant and specific enough to translate into practical steps or strategies that support decision-making. To do that, the data must have the appropriate level of accuracy, resolution, volume and latency.

1. Accuracy
   

   This is the first major hurdle for many water utilities. Often, operators have data that isn’t validated, which can lead to conclusions that are not only fruitless but potentially costly. To ensure data is accurate, instruments must be calibrated and tested regularly. As water utilities incorporate analytics into their workflow, these systems can also estimate and validate the data from various sources.

2. Resolution
   

   Data must be granular enough to match the purpose. For example, detecting a residential leak requires readings reported as often as every 15 minutes for 24 hours or more. Meanwhile, to detect a pressure transient, a reading resolution of 128 Hz or more is required. In other situations, simply knowing when readings change is necessary.

3. Volume
   

   The amount of data needs to be proportional to the size of the problem that operators are trying to solve. In other words, the amount of data needed to go from 50% efficiency to 60% efficiency is far less than the amount needed to go from 95% to 99% efficiency. Volume can be quantified in terms of the number of instruments in the network taking samples or in terms of data accumulated over time (months, years, etc.), depending on the goals in question.

4. Latency
   

   This describes the amount of time from measurement to communication, which is dictated by the problem at hand. When producing a monthly water bill, for example, there is a lot of latitude. However, when there is a possible contamination breakthrough, every second counts.

Once the data is in hand and accessible through a centralized interface, it must be parsed and analyzed. While this can be done manually, there are a range of solutions that can help operators make sense of their data to discover insights and determine the best course of action.

Types of Data 

The point of data is to define the action—a truck roll, a valve turn, a pipe replacement—that ultimately gets results. The types of data that water utilities regularly collect fall into four general categories: 

Operational, Maintenance and Engineering Data

This is information about the movement, treatment, and asset management in the utility. It can be further divided into the following groups:

• Device data: This is information produced by sensors or other instruments. In addition to the parameter being measured, these devices generate data in the form of timestamps, device status, settings and diagnostics results.
• Utility operational data: This is data collected during day-to-day operations, such as from the laboratory, customers and SCADA systems, as well as data stored in computerized maintenance management systems, etc. 
• Utility contextual data: Utilities also store information about assets they manage, account affiliation, land parcel, service territories, etc. Although this data is largely static, it can be important for a wide range of applications.

Business Data

This includes enterprise data, such as inventory, purchase agreements and asset management systems. 

Account Data

Customer-specific data, like parcels, billing history and purpose of use.

Master Data

This comprises information relevant to multiple assets and functions, including geographical information, elevation and service territories.

Where to Begin

Generating actionable data won’t happen in a vacuum; it requires a champion who will begin by defining the goals that the utility hopes to achieve. Then, that person must ensure the necessary personnel are aligned with those goals. If the utility isn’t sure what data it will need to achieve its objectives, then operators may need to do some research, including speaking with other utilities that have tackled similar projects.

Piloting a solution might be worthwhile to see if it meets the targeted need. If nothing else, the pilot could reveal what data or other factors may be missing before a full-scale solution can be implemented.

It's important to remember that data and analytics don’t solve problems; subsequent action does. The key is to ensure that the person undertaking the task has the right information to take the right action at the right time. Ultimately, data doesn't solve problems—but action does.