Rolling Mill - Periodic Disturbances and Observer Design
Summary
This laboratory is concerned with the study of the impact of eccentricity (of the rolls) on the performance of the BISRA Gauge for estimation exit thickness in a Rolling Mill. Remedial strategies will be studied based on observer design which will take us to another level of performance.
Of particular relevance to this virtual laboratory is the problem of soft sensor design and periodic disturbance rejection in rolling mills.
Figure 8.1: Screenshot of Program
The Physical Apparatus
You have found in the previous laboratory that the BISRA Gauge performed very well but in practice the performance gains are certainly not perfect. Indeed when the BISRA Gauge is deployed in practice it is found that there are oscillations in the thickness of the exit strip. Thus while the BISRA Gauge results in some improvements these have a ceiling. This situation is common in control systems design where successive improvements ultimately reach a limit. This is only remedied by further investigation of the process and formulation of further remedial strategies which again lead to improvements in the system.
In the case of the BISRA Gauge the performance ceiling is due to the fact that the rolls of the mill are not perfectly round. Indeed in practice the source of the difficulty was not immediately obvious and pinpointing the difficulty as eccentricity of the rolls was a significant breakthrough. In this laboratory we will study the impact of eccentricity of the rolls on the performance of the BISRA Gauge. Having understood the problem we will then implement remedial strategies based on observer design which will take us to another level of performance improvement.
This virtual laboratory is a companion to an earlier two laboratories on the rolling mill. In the first laboratory, Rolling Mill (A), we looked at system modelling and classical control. In that laboratory we found that the delay between the mill and the exit thickness measurement was a major source of performance degradation. In Rolling Mill (B) soft sensors were investigated. In that laboratory we found that the use of a soft sensor gave significant improvements.
Please refer to the other two rolling mill laboratories for further background on the physical apparatus. Of particular relevance to this virtual laboratory is the problem of soft sensor design and periodic disturbance rejection in rolling mills; a problem that has been the subject of much investigation over the last five decades. The source of the disturbances is the rotation of the rolls. Thus the period is rolling speed dependent.
In keeping with the evolutionary improvement in technology this virtual laboratory will successively investigate a number of control architectures aimed at dealing with periodic disturbance issues raised in this application.
Disturbances
The most severe disturbances in terms of off specification product, are those originating in the strip (setup errors, hydrodynamic backup roll bearing speed effects and input strip surface roughness are also significant but are beyond the scope of this lab). Input strip thickness and hardness variations can have a particularly severe waveforms that approximate step functions. These basic disturbances have been discussed in Laboratory 6 (Rolling Mill (A)). Another disturbance, eccentricity, discussed in detail here, originates from deviations in the radius of the rolls due primarily to manufacturing imperfections, heating and wear. We will see in this laboratory that these disturbances have a significant effect on the performances of the soft sensor. We will also investigate the ways of reducing this effect based on Observers.
Prerequisites
This virtual laboratory uses more advanced ideas including observer design and periodic disturbance compensation. It would thus probably be more suitable for a second course in control system design. It may also be helpful, although not strictly necessary, to have done the first two Rolling Mill Laboratories in Chapters 6 and 7 (at least in rough form).
Learning Objectives
The objectives of this laboratory are:
- To introduce the student to periodic disturbances.
- To gain appreciation of valuable control system design tools including Soft sensors.
- Periodic disturbance compensation via an observer.
- To gain insight into fundamental performance limitations associated with various control system architectures applied to this system.
