|
|
CONFIDENTIAL
REPORT
| Project
Title |
: |
The
Development of an Intelligent Control System for the Drayton Glost
Kiln at Portmeirion Potteries Ltd. |
| Client
|
: |
Portmeirion Potteries Ltd. |
| Contact
|
: |
Mr Philip White |
| Report
Prepared by |
: |
Andy Lowe |
| Project
Manager |
: |
Andy Lowe |
| Issue
Authorised by |
: |
Seija Aalto
Operations Manager
CERAM Process & Materials |
| Date
of Issue |
: |
29 August, 2001 |
| Circulation
|
: |
Seija Aalto
Chris Woolgar (Neusciences) |
The Development of an Intelligent Control System for the Drayton Glost
Kiln at Portmeirion Potteries Ltd.
CONTENTS
1.0 Introduction
2.0 System Specification
3.0 Data Acquisition
4.0 Performance of the Data Collection System
4.1 Recommendations
5.0 Results to Date
6.0 Project Status
7.0 Contacts
1.0 Introduction
The use of intelligent technology, specifically neural networks, to control
kilns was first addressed at CERAM in the recent DTI funded "Intelligent
Kiln Control" project. Portmeirion Potteries Ltd. identified the concept
as a key technology for optimising the performance of their kilns and included
it as a work package in their "PROMOTE Life" project. This project,
which is being part funded by the 1999 European "Life" initiative, aims
to demonstrate that tackling environmental issues actually saves, not increases
costs.
To enable the successful development of an intelligent control system, Portmeirion
has formed a partnership with CERAM and Neusciences. Neusciences have been
working in the field of intelligence technologies since 1989 and are providing
their expertise in the application of advanced mathematical and statistical
techniques to solve this complex process situation.
The Drayton glost tunnel kiln was identified as the best proving ground
for this idea. This kiln operates in a lean manufacturing environment and
therefore has to maintain an accurate time-temperature profile independent
of the type of ware being processed.
Temperature control is achieved by using conventional feedback controllers,
one for each of the six burner groups. The control loop consists of a thermocouple
which measures the actual temperature within the kiln, a feedback control
unit which compares this value with a pre-set value and a mechanism to control
the flow of energy into the kiln. This system is very effective and maintains
quite tight temperature control in the main firing zones.
Many of the factors influencing temperature variation are related to product
loading. This glost kiln produces combination of flatware and hollow-ware
fired on refractory batts. The height and weight of the ware on each batt
varies unpredictably, depending on the type of product available. In some
cases empty batts are processed adjacent to fully loaded batts.
The feedback control system is able to respond to changes in the loading
pattern by adding more or less energy as required but this response can
cause large fluctuations in temperature in the preheat zone. Consequently
the actual temperature profile will shows little variation in the main firing
zones but large variations in the heating and cooling zones.
The intelligent kiln control project at Portmeirion aims to develop a system
that can predict and advise on the control actions necessary to maintain
a set temperature profile independent of the car loading.
2.0 System Specification
The core of the system is the "SpecView" SCADA software installed on the
PC (Pentium 200MHz) supplied by CERAM. This software was the data acquisition
tool used during the DTI sponsored intelligent kiln project. The operating
manual and relevant software now belongs to Portmeirion Potteries Ltd and
is presented with the original copy of this report. The original software
has been upgraded on a regular basis and is Y2K compliant. The version currently
installed at Portmeirion is 582. Future upgrades can be obtained by visiting
the SpecView web-site at www.specview.com.
Kiln data from the monitoring thermocouples is routed through an Integrated
Control Unit (ICU) supplied by FGH. The ICU unit has three data acquisition
cards installed; two analogue input cards and one digital input card. Only
the analogue cards are being used at this time, there are 4 analogue inputs
still available for use. The configuration software for the ICU is stored
with the SpecView manual and is installed on the PC.
Communications cards in each control unit are linked to the PC via the ICU,
enabling the actual temperature in each control zone and the control parameters
set to be monitored and recorded. It is possible to set the control parameters
from the PC, therefore the software is password protected for use by myself
and Dave Joines. No changes have been made to any kiln setting since installation.
My password is alpxr5.
All of the information is carried on screened twisted pair cable type RS
368-738 (4 wires).
The kiln loading data was collected manually (procedure described in the
next section) and combined with the temperature data for analysis.
3.0 Data Acquisition
Temperature and controller information is contained in the logfiles generates
by the SCADA system.
Various systems for collecting the load data were investigated:
- The most basic used the output from the video cameras already installed.
However a video recorder was not available and the camera view was not
adequate for a detailed analysis.
- For this initial data collection exercise a manual approach was adopted.
Whilst this was time consuming it allowed a better insight into the
problem. The following categories were used to describe the kiln loading:
Table 1. Kiln Car Loading Classification
| Code
|
Description
|
Weight
Code |
| HLT
|
Large
plate stack in refractory setting with a single layer of product on
top |
1.03 |
| HLO
|
Large
plate stack in refractory setting |
1.00 |
| HMT
|
Medium
plate stack in refractory setting with a single layer of product on
top |
0.93 |
| HMO
|
Medium
plate stack in refractory setting |
0.90 |
| HST
|
Small
plate stack in refractory setting with a single layer of product on
top |
0.48 |
| HSO
|
Small
plate stack in refractory setting |
0.45 |
| L |
Single
layer of ware (cups, mugs, bowls etc…) |
0.48 |
| HD |
Dishes
in refractory stack |
0.90 |
| RL |
Large
plate refractory |
0.50 |
| RM |
Medium
plate refractory |
0.45 |
| RS |
Small
plate refractory |
0.25 |
| E |
Empty
refractory batt |
0.00 |
The information was recorded on paper then transcribed into an Excel spreadsheet,
where a rough estimate of the product load and height was produced. This
data was supplied to Neusciences in conjunction with the temperature data.
Early results indicate that there is a correlation between the temperature
change in the preheat zone and the kiln load. If this proves to hold true
it will become unnecessary to measure the kiln load, relying only on the
kiln response.
- The advent of inexpensive CCTV technology and image analysis software
will provide the solution for the future data collection system should
this be necessary.
4.0 Performance of the Data Collection System
Consistent and unbroken communication between the instruments and the PC
has been the major problem. The first issue was damage sustained to the
communication hardware during the Christmas (2000) shutdown when part of
the system was short circuited blowing the COMMS boards at the ICU and the
RS422 connector. Some of the existing intermittent problems are probably
due to this also.
Interference from the adjacent power circuitry is an issue however the use
of twisted screened cable has proved to be beneficial.
The manual collection of the load data proved adequate for the initial evaluation
by Neusciences.
4.1 Recommendations
- Replace the COMMS cards within each control unit to eliminate the
intermittent faults.
- Portmeirion should consider replacing the SpecView SCADA software
with a more user-friendly system eg Datamanager from FGH**.
- Communications between kiln hardware and PC is fraught with difficulty
and the above recommendations should be carried out by one service provider
eg FGH.
- Verify the relationship between the preheat temperature change and
the kiln loading.
**Note, FGH was stated because the existing controllers were supplied by
FGH and continuity is important for COMMS.
5.0 Results to Date
A preliminary report based on kiln data collected in April 2000 compiled
by Chris Woolgar (Neusciences) is attached. Chris is now preparing a more
detailed report.
6.0 Project Status
Temperature and load data has been sent to Chris Woolgar at Neusciences
for analysis. He will produce a report that will indicate the relative importance
of input variables, model performance, likely performance of the finished
system, and budgets and timescale for any further modelling work required.
In addition, budgets and timescale for implementation of the control system
based on that model will be included.
7.0 Contacts
| Neusciences |
Chris
Woolgar |
| Phone
|
023
80 664017 |
| Mobile
|
07967
740383 |
| e-mail |
chris.woolgar@neusciences.com |
| FGH |
Chris
Bootherstone |
| Phone |
01782
208448 |
| SpecView |
Various
contacts |
| Phone |
01825
766566 |
| Web-site |
www.specview.com |
|
