Andrew Ogden-Swift reviews the pressing challenges in Process Automation
Over the last 40 years, there has been continual growth in the use of digital technologies in process manufacturing. This has ranged from increasingly reliable and low-cost sensors, digital field networks, control and shutdown systems, process historians, model-based optimising controllers, and manufacturing information systems, to enterprise resource planning systems. These technologies have enabled higher process performance, reduced operating risk and improved staff productivity.
Improvements in these technologies continue:
In the broader world, there are trends that will impact the process industries and enable further performance improvements and help mitigate loss of expertise. They will also contribute to improved sustainability of these industries. This article focuses on digital technologies. Many of these will be addressed in Advances in in Process Automation and Control 2019 in November in Manchester, UK. The following is a high-level view of these.
The trends in digital computing and mobile technologies now mean there are huge amounts of information about people and their behaviours, appliances and equipment, and the environment. Increasingly, companies are analysing these data to help make better decisions, faster. We are all familiar with internet-
based systems recommending other purchases based on the items we have selected, or adverts presented to us based on entries made.
Similar opportunities exist in manufacturing with some challenges that are specific:
1. To be successful, subject matter experts need to use modern analytical tools. Subject matter experts such as engineers, managers and operators seldom have a strong background in statistics, machine learning and AI. Similarly, data scientists seldom have good understanding of process manufacturing. Tools are needed that enable subject matter experts to perform analyses and yet collaborate with data scientists on the more complex problems.
2. Data is stored in disparate systems. There are usually data quality issues including missing data, different sampling rates and temporal misalignment, signal and process noise, measurement uncertainty and the relationship with business risk. Tools need to make it easy to access, align and cleanse data.
3. Process analysis needs to make it easy to answer common questions but in ways that make sense by, for example, allowing for grades, operating modes, and feedstocks.
4. Analysis needs to allow fast results to be achieved, effective collaboration, issues to be communicated quickly and improvements to be driven rapidly.
Increasingly, tools are becoming available that meet these needs and enable engineers and managers to analyse performance without the grind of trying to use spreadsheets or become data science experts.
Catch up on the latest news, views and jobs from The Chemical Engineer. Below are the four latest issues. View a wider selection of the archive from within the Magazine section of this site.