With 1000s upon 1000s of customer datapoints to map to Honeywell tag names, chemical engineers would start with a blank Excel template. In the Excel screenshot below, the green columns are dropdowns whereas the red columns require manual entry of the customer's alphanumeric codes, or tag names.

Honeywell’s chemical engineers would have to enter customer tag names found in the P&ID one-by-one, mapping each to its Honeywell counterpart. This would eventually create a digital twin of the entire plant, including its miles of pipes, compressors, heaters, pumps, and other equipment, within Honeywell’s database.

To be frank, this process was awful. It was extremely time-consuming, tedious, error-prone, and time-consuming. Because chemical engineers spent months on this setup process, they had little to no time to lend their optimization expertise to customers. What they were actually hired for, none had the time to do.

As soon as I was tasked with this initiative, I and my design team's research manager interviewed 17 colleagues and users in chemical engineering, product, software development, and data science roles. We received countless demos and current process feedback along with walkthroughs of an ideal workflow. Next, we designers conducted workshops with data scientists and developers all over Honeywell worldwide to make this mayhem an automated operation using custom algorithms.

Raw tag names and instrument data from the customer would be uploaded, then smartly categorized based on the instrument type. Using this method, chemical engineers would just have to confirm smart suggestions from the system. Thanks to the algorithm, tags could be mapped 100s at a time instead of one-by-one. No manual data entry required.