The benefits of two-leg versus three leg power control, in three phase heating applications
Mathematical modelling, advanced process control and model-based predictive control strategies are in constant use across the downstream oil and gas industries. These technologies have been embraced and installed in their process control systems for many years. The use of Model-Based Predictive Control (MPC) and fuzzy logic have also been studied for glass melting and conditioning. However, the glass industry has been hesitant in its implementation of model-based control strategies.
Recently, some leading glass container manufacturers have started to implement advanced control and it is becoming more widespread in glass processing. The processes of melting, fining and conditioning are extremely complicated due to the combination of chemical reactions and complex thermo-dynamics. When the relatively poor inertia of the process is also considered, it is clear that running a furnace by hand can be considered a ‘mission impossible’.
Glass processes are very slow, so often an operator will implement a control action and not see the result before handing over to the next shift. Consequently, it is very difficult to have a standardised process for controlling a furnace when relying solely on operators. Furthermore, operators sometimes introduce more process instability than they are aware.
Glass melting tank control can seem simple but it is not! In these control applications, the number of parameters available for manipulation are low, the precision of actuators is sometimes poor, the number of usable process values are limited and sometimes, even the process values are inaccurate and subject to drift. Twenty years ago, studies were carried out to find consistent ways of controlling and operating a furnace. It began with fuzzy logic, capturing and transferring operator technical expertise into a standardised and automated control strategy. This takes out the fluctuations experienced between different operators and shifts, providing better performance results.
Single operating environment goal
There have been many developments in automation and control strategies over the years, an area of expertise for Eurotherm by Schneider Electric. Eurotherm is well known for its high precision programmable automation control (PAC) systems and has a good ‘price-to-performance’ fit for the glass industry.
The company’s PID control algorithms and tuning strategies are highly respected. Furthermore, third party advanced control has been integrated into its PAC systems for many years. Eurotherm understands that its customers want a deeper integration of the MPC process with their overall control system, so the company is working to create a single operating environment that integrates the results from MPC directly into the control process, removing the necessity for multiple standalone environments.
Traditional methods of ‘training’ an MPC model are also unfavourable, due to the real-time actions required on furnaces and forehearths. CelSian, a company closely linked to Dutch research organisation TNO-Eindhoven Institute, has been working on a process called rigorous MPC. This process uses CelSian furnace modelling technology for step response data acquisition, instead of obtaining step response data through experimental observation.
The build of these operational models does not usually require interruption of on-site processes. It is obvious from these strengths that a CelSian-Eurotherm partnership brings many benefits and advantages to customers.
In working together, this arrangement will draw on expertise across all areas. The MPC modelling is run on a furnace model created by CelSian, utilising its vast furnace and forehearth expertise.
The Finite Impulse Response model will be built and run using a leading software product from AVEVA. Eurotherm will supply the hardware and software of the DCS process control environment, as well as the engineering expertise to make this an integrated advanced process control solution.
This CelSian and Eurotherm partnership will make institutional mathematical modelling knowledge accessible at the daily operational level of glass manufacturing, providing improved understanding, reliability and operational efficiency of glass processes for many years.
This article appeared in the May/June 2019, issue 83 of Glass Worldwide.
About the Authors:
René Meuleman is Business Leader for Global Glass
West Sussex, UK
tel: +44 1903 268500