To say that engineers have a love/hate relationship with Excel spreadsheets might be an understatement. As helpful as they are complicated, spreadsheets are home to a gold mine of data and subsequent insights just waiting to be unearthed. The problem is the unearthing can take hours, days, or even weeks. For a major manufacturing company, that was simply too long and the process too cumbersome. With 38 PI Systems collecting operational data from over two million tags, they knew the pain of working in Excel. With the help of the PI System and BI tools, they moved beyond spreadsheets and developed a new way to quickly get the right operational insights to optimize paper production.
From manual processes to version control to data duplication across spreadsheets, using Excel for reporting was cumbersome for the team. Even the morning report spreadsheet consisted of 12 visible spreadsheets, 39 hidden worksheets, and eight VBA modules-not to mention numerous VLOOKUP functions and over 110 named ranges. After an engineer asked if Excel was really the right way to run the morning report, the team wondered if that engineer was right.
“What if we were to try to get Excel out of the picture? Or get Excel out of the picture as much as we could?” a Manufacturing Process Information Manager pondered during the 2020 PI World Online conference. Amidst converting four more mills from a different historian to the PI System, the company began to explore options. Ultimately, the answer would not only enable engineers to get the insights they needed; it would give the data scientists the ability to use BI tools to run analytics.
The goal was to leverage a data lake to store the massive amounts of operational data so the company's BI tools could take over several Excel functions. Using a mixture of PI System tools including Asset Framework (AF), the contextualization layer of the PI System, the team migrated spreadsheet calculations and metadata into AF to standardize those calculations and categorize assets. After linking AF to external data sources and exporting data via PI Integrator for Business Analytics, they linked business analytics tools to PI System data and had a solution. Now it was time to put it to the test.
One of the firm's engineers was working in an Excel spreadsheet to calculate CTO process flow efficiency. She had already been working on her analysis for three days. She was looking at a variety of flows and trends, trying to identify when batches started and stopped. However, a flow valve right after the discharge pump could be moving up towards the brine tank or down towards to CTO tanks depending on whether the settling or filling decanter was running. The flow valve's ability to move in either direction made it challenging for the engineer to parse only the data she needed.
The team was easily able to separate out each flow based on which decanter was running using analytics. The single flow meter now functioned as two flow meters for analysis purposes, enabling them to view trends separately for each decanter. After identifying the production phases, they calculated events and associated each event with the specific production process. From there it was easy to filter out any production flows associated with the filling decanter to visualize just the settling decanter information in PI Vision.
While this process took a day to setup, it provided engineers with a tool that they could run for any time period, significantly reducing future analysis time. “The proof was in the pudding that when I averaged my data and compared it to her data, we were different by .03 percent,” the engineer said. “Which is essentially we got the same answer.”
It's clear that the new process is working, but it's not without challenges. From protecting the operational data store, including PI Data Archive and the MES databases, to gaining the trust of engineers and data scientists so they would adopt, the team has worked tirelessly to address any hurdles along the way. To learn more about how they overcame these challenges to put their company on the path to moving beyond Excel, watch the full 2020 PI World Online presentation here.