PPCL's Newsletter Archive

Visualising and Reducing Operational Variability in Process Plants

Variability in process operations costs a plant in lost production, increased energy usage, poorer product quality, increased waste and/or recycle, increased emissions and in many cases also reduces catalyst and/or equipment life.

The first step in reducing variability is to understand the size of the problem and identify special and common causes. Many plants have realised that to do this they needed data and have already implemented process historian and lab information systems. But then they discovered that they had so much data available that they couldn’t see the woods for the trees; And the methods available to extract information from all this data jumped from a spreadsheet and simple univariate statistics, which by definition ignore interaction between variables, directly to highly mathematical multi-variate mathematics with nothing in-between so all that costly data is hardly used and plant operations continue to be plagued with variability.

The solution was to start again from a completely new mathematical basis that did not require an advanced mathematical education to understand and use. The answer was n-dimensional geometry. This has allowed us to harness the amazing visual pattern recognition capability of the human brain in a way that allows anyone on a site to visualise variability and see cause and effect relationships.

Many large process plants are now using our methods, some for everyday ‘problem solving’ and others within formal process improvement programs such as 6-Sigma. Some have come with us all the way to the frontier of realtime Operating Envelope models for closed-loop control, batch process control and realtime optimisation entirely without maths. These models are as much as an order of magnitude cheaper to develop and maintain than conventional Model-Based Control (MBC) models hence are economic to apply to a much wider range of plant and equipment. They give equally good if not better results too.

To find out more about this topic please email us.

• < Prev
• Next >