While many of our posts cover topics in computational fluid dynamics, you may have noticed an increasing focus on process digitalization. Here, Founder and Managing Director Martin Weng describes how this aligns with the aixprocess strategy
The Challenge of Inconsistent Data
In 2017, we first explored the potential of big data. When collecting process information from a brownfield plant for computational engineering, we were often challenged with inconsistent data sets. Heat and species balances were not to be closed, sometimes even the mass balance was open. This was obviously not due to missing expertise by plant engineers, but rather a lack of holistic information.
Consider a simple example: the calorific value of a fuel. With regular coal-based fuels, there might be only one sample available from a shipload, but storage and transport can alter the moisture content significantly. With alternative fuels, the situation is even more complex.
Our Approach to Comprehensive Data Collection
That was when we asked clients for full process data of weeks, months and years from PLC systems, quality control and additional sensor data available from kiln cameras, emission measurements and others - the plant Big Data. Utilizing readily available data analytics tools, we found great value in applying:
-
Statistics
-
Various correlation methods
-
Multivariate analysis
These techniques helped us identifying closed balances and more: trends and drifts in plant operations that were assumed to be stationary.
Customer Feedback and Insights
Despite initial successes and setbacks, the main feedback from customers was clear: the combination of process engineering expertise and data analytics provides real added value. This contrasts with the typical promise of "give us your data and our black-box analytics will increase your yield by XY%."
The "aixProM" Data Platform
In 2019, we started developing a data platform with the working title "aixProM." The name "aix" originated from the ancient name of the hot springs in our hometown Aachen, "Pro" stands for process, and "M" for modeling. Interestingly, we later noticed the embedded "AI". As often the case with working titles, they tend to stick…
The Synergy of Process Engineering and Data Analytics
Through this journey, we've learned that integration of process engineering and data analytics is more than a trend - it's a necessity. But a bumpy road, too. In one of our follow-up posts we will share more experience about obstacles and setbacks along that digital journey.