“To improve data quality, you must make it specific, relevant and create a goal for it,” said Cecelia Adair, senior operations research analyst at the Las Vegas Valley Water District. With this straightforward formula, Adair highlights the core of what lies at the center of any digital transformation—the data.
This explains why Adair is so focused on data quality. After all, without quality data no application of advanced automation technologies can improve your operations. Her experience shows that, as much as technology has advanced over the past several decades, the old mantra of GIGO (garbage in, garbage out) still holds true.
And for the Las Vegas Water District, its maintenance data is critically important to manage its assets that deliver 200 million gallons of water per day to more than 1.5 million people through the use of 54 pumping stations, 79 reservoir basins, 6500 miles of water transmission and distribution pipelines, 76 production wells and 26 recharge wells.
Describing the district’s evolving focus on data quality at the AVEVA World Conference in Dallas, Adair said it all began with the district’s implementation of AVEVA’s Avantis Maintenance Management software in 2003. Prior to the implementation, Adair said the Las Vegas Water District had been using work management processes to maintain its assets. “But with AVEVA, we could do better. So, we started looking at our reactive maintenance data,” she said.
The devil is in the details
When they dug into the data they discovered discrepancies in reactive maintenance data between work types and work priorities. “When we plotted the percent of our reactive maintenance work, it was high—at around 25 percent—as recorded in our work types. But our priority types were showing it to be only around 10-13 percent,” Adair noted. “The data wasn’t telling us the same story, which indicated a problem.”
Digging further, Adair said they discovered that their Level 1 and 2 maintenance activities were largely comprised of reactive work maintenance, as expected. But they also discovered Level 3 maintenance activities here too. These activities are supposed to be routine work, she said, which meant that finding this work coded under reactive maintenance work meant that it wasn’t being coded correctly.
With a clear coding problem on their hands, the water district needed to fix this issue. To do this they needed a trend chart. To create an effective trend chart for maintenance activities, Adair said you need “to choose a time period appropriate to your data, and then graph the data regularly over time.”
In the process of choosing this data and finding out where it could be wrong, Adair said it’s important to cross check the data by talking to operators and maintenance personnel, and by using Excel pivot tables to explore the data. “Odd things will jump out” when you scan the detailed data in this way, she said.
The next step is to review your codes by defining a purpose for each code field and choosing the codes you’ll allow. “Consider using entity classifications to filter your code list,” Adair said, “and then train people on what the codes are for and which ones to use. Then measure code use compliance. Remember, the codes are all about how to get information out of your system in a way that’s meaningful for everybody. They’re not created so that everybody can have every word they want in the system. That’s what leads to duplicate codes.”
She also stressed the importance of entity hierarchy to data management. “Without this you might miss things,” Adair said. The Las Vegas Water District uses a three-level process covering location, equipment and system. In a multi-level process like this, “check for the use of a consistent structure across entity type and classification,” she advised.
A better understanding of processes
Even after specific data goals and procedures were in place and communicated to the water district’s staff, Adair admits that code compliance was not great in the early days. However, she said it did improve over time as the staff saw how regularly the data was reviewed, and compliance was required.
Feeding their AVEVA software with better quality data delivered significant results. For example, the cycle time to close priority 1 maintenance issues could take nearly 300 days in some instances, before the data coding improvements. By the third year of mapping the data to activities and constantly improving the data inputs, priority 1 issues were being completed in less than 40 days. Six years into the program, the district was resolving these issues in less than 4 days.
With these data improvement basics in place, you can go further by adding even more info into your CMMS so that you know everything about your equipment, Adair said. For example, by properly coding their vibration program in maintenance, it has “helped us know more about our processes—not just for maintenance, but for engineers too,” she said.
Based on the success of this data quality program, the water district is extending it to other maintenance operations, such as its wells, which are used in summer to maintain water levels as the Colorado River recedes. “We use aggregate data to determine the highest cost pumping stations and see if they are the the ones we thought would cost us the most,” Adair said. They also use this process to track failures across the district’s 400,000 service lines.
“By measuring your business, your employees understand more, which leads them to put in more and better data,” Adair said, “because they get to see the value of the data. As a result, their work is better documented, asset management is achieved, and you’ve created a synergy to continue to improve.”