Published in Glass Worldwide, March/April 2015
How good would it be to have all that historical and realtime process data available in one system, extremely easy to access and retrieve using accompanying sophisticated analysis tools? All data in a single place; from raw materials to warehouse and beyond.
In particular, melting, refining and conditioning processes are hard to manage because of the extremely high temperatures and relatively poor sensor technology available.
Managing these processes boils down mainly to the availability of experienced personnel and since experience does not come overnight, these individuals are normally not from the younger generation. Fortunately and historically, glass manufacturing employees tend to stay forever with ‘their’ company but that behaviour is changing over time. Today, the fact must be faced that a huge group of people are close to retirement and experienced replacements are extremely difficult to find.
Considering all this, it can be concluded that the glass industry will suffer from process control competence flaws, to which there is no solution from a human resource perspective; the industry’s senior process technologists are almost impossible to replace!
It could be worth considering replacing the longterm, detailed experience of senior technologists with automation, although it seems that the development of intelligent systems, centred mainly on model-based predictive strategies, is only slowly moving forward.
Implementations of these advanced systems are still relatively rare and focus primarily on the melting and conditioning part of the process.
The author’s personal history and experience with fuzzy logic and MPC (Model Predictive Control) goes back at least 15 years and still, these systems are failing to be recognised as accepted technology. Compared to oil refining process industries, in which model-based predictive control has been installed almost universally, perhaps only 5% of the glass industry has installed advanced process control successfully. This shows that the glass melting process is relatively complex compared to the refining processes, that the glass industry is more risk averse and that available budgets for this technology are rare.
It is assumed and hoped that young process technologists entering the glass industry will change this soon and it is essential they are provided with the tools to do so.
Data management and analysis tools are unquestionably of great importance but the industry needs to find solutions and control strategies to solve a complexity of problems, more specifically:
Installing model-based predictive control systems might be helpful but they cannot replace and should not be considered as a replacement for human process intelligence, at least not yet. The glass industry still needs people to study and improve the process, which will of course eventually result in the improved automation of processes but the industry is probably at least 10 years away from full automation of melting, fining and conditioning processes.
Since the industry will soon lose, (if it has not already lost) most of its senior technologists due to retirement, it needs to ensure that their replacements will find an attractive working environment, enabling them to make ‘a flying start’. Moving fast and data-driven technology is attractive for young people and both they and their employers will profit from a sophisticated working environment of this kind.
Next to the capability of such a system to interface, collect and store all process data, that environment also needs to be able to act as an interface between the different ways older and younger generations tend to approach process problems. It should be kept in mind that the availability and processing of process data in the early days was difficult. Today, data storage and data processing is no longer an issue. Consequently, seniors and juniors approach problems in a different way. Seniors rely on their long-term experience and juniors will try to find evidence in data. The solution is to capture the seniors’ knowledge together with the historical and real-time process data in a single system, accompanied by analysis tools to enable both old and young technologists, as well as operators to study, understand and control their process. In other words, match and confirm the long-term experience of close to retirement glass technologists and operators by using analytical strategies based on historical and real-time data. The advantages of having experienced plant personnel closely involved in these developments are obvious:
The availability of process data has always been essential to understand and improve glass industry processes. In earlier days, however, it was difficult and expensive to manage huge amounts of data. Today, these issues have been overcome and capturing, converting and storing huge amounts of data is no longer a problem.
A single historian industrial database is perfectly capable of storing all process data at a sufficient resolution, from batch to warehouse and a whole furnace campaign, without problem. The key of course is to design such a system using a method that stores data in a way that can be retrieved easily. Decent TAG-name conventions need to be in place, along with the availability of easy-to-use data analysis tools.
Once this kind of system becomes available and has been well-designed by knowledgeable glass manufacturing experts, plant personnel will automatically start using it, learn from it and eventually it will speed up their learning curves and process control improvements. When that has been done, it will be a perfect data source for all kinds of sophisticated process control enhancements. Humans have already entered the data-driven era, like it or not. Now it is time to do the same in a professional context because ‘it is a capital mistake to theorise before one has data’!
* Source: Sir Arthur Conan Doyle.
About the Author
René Meuleman is Global Glass Business Development Manager at Eurotherm by Schneider Electric