Industrial processes are complex. Modelling them can be even more
so. Yet this is what the MAPP project partners managed to do. They
developed integrated software to help reduce the environmental
impacts of industrial processing while lowering production costs
and improving product quality. Their software includes all steps
from data mining, data capture, database and environmental
modelling, to real-time implementation.
The multivariate real-time system was evaluated using three case
studies - wastewater treatment at a municipal sewage treatment
plant, metal fuel tank production and the phosphating process
during truck cabin fabrication. MAPP (multivariate approach for
statistical process control and cleaner production) project
coordinator, Jonas Rottorp, from the IVL Swedish Environmental
Research Institute Ltd explains that MAPP software bridges the gap
between process modelling and monitoring.
"There has always been a gap between researchers and the end-users
in online process control, particularly using multivariate,
real-time systems," he says. "We showed that it is possible to use
advanced modelling techniques for process monitoring. Typically
this has failed because implementation in the industrial processing
environment is too difficult."
By using models describing the relationship between process input
and product performance - including economic performance, quality
and environmental impact - better process control can be achieved.
This results in both cost savings and reduced environmental impacts.
The three case studies showed that by using MAPP software, online
predictions replaced time-consuming and expensive laboratory
analyses. At the same time, operators gained a deeper knowledge of
the process and were able to exert tighter process control.
All round better wastewater treatment
Municipal wastewater treatment is essential to control the
environmental impacts of human behaviour. It reduces the emissions
of organic matter and nutrients. The difficulty with wastewater
treatment is to continuously produce clean effluent because the
composition and the flow of the influent constantly changes. In
addition, the consumption of precipitation chemicals and energy
must be considered as their use also has a negative impact on the
environment.
The new multivariate modelling techniques, used in combination with
dynamic database modelling, solved the difficulties inherent in
this very dynamic process, which requires high quality monitoring
and control systems. MAPP real-time predictions of the effluent
quality and process status was accomplished by using a software
sensor for predicting phosphorous in the composition of incoming
water.
Rottorp says this is a "real scientific breakthrough" that resulted
in savings in precipitation chemicals and energy supply as well as
improving the quality of the effluent. MAPP resulted in several
significant achievements at the Borlaenge Wastewater Treatment
Plant, including the development of an object-oriented database to
improve the ability to model the process and the creation of a
sophisticated lifecycle inventory of precipitation chemicals that
enables online monitoring of environmental impacts.
More efficient fuel tank production
Metal fuel tanks are more resistant during car crashes and safer
because they reduce the likelihood of fire. However, their
production is more difficult because metal is less malleable and
more expensive than plastic.
The MAPP model for the welding of fuel tanks demonstrates the risk
of leakage by using the concept of a complex form of analysis. If
the analysis yields a value greater than -1, the tank will not leak.Welding experts at Fiat's ILAS factory near Turin, where most
tanks are made, pointed out that the junction along the perimeter
of the tank is the most critical. This information was factored
into the process model.
Application of the MAPP model resulted in fewer actual tests, as
there is no need to test tanks in which the model gives a value
greater than the threshold. The MAPP model is statistically
significant and the results quite easy to interpret. As a result,
fewer leak tests are needed due to the initial screening model of
the welding result.
Tighter control of painting processes
The phosphating process is an essential metal surface treatment.
The phosphating layer serves both as a corrosion protection and as
a pre-treatment process for painting. The quality of the layer is
critical to the final finishing of the metal product. Because it is
an energy and chemical demanding process, optimisation leads to
economical and environmental savings.
Knowledge about which process changes are needed to achieve a
better product with less impact on the environment can be obtained
using MAPP's multivariate model based monitoring and control system.
Operators could follow cost and environmental impact per produced
unit in real-time and monitor all process movements. MAPP's
monitoring system also revealed the predicted contribution of
greenhouse gases, eutrophication potential and cost per produced
truck cabin. Quality parameters were measured in real-time using
multivariate models.
Using MAPP software at Volvo Truck Corporation resulted in
significant cost savings and reduced environmental impacts. Says
Rottorp: "Volvo was evaluating the wear and tear caused by weather
to the paint used on their trucks every second week. MAPP models
enabled the company to do this evaluation three times a day. This
resulted in cost savings, improved quality and a reduced
environmental impact. Using process control, trucks can be produced
using less paint and chemicals. It's a win-win situation."
MAPPing the future
The prototypes developed and the successful demonstration projects
at full scale industrial plants clearly shows that problems in
using multivariate process models to control various processes have
been solved. Future plans including developing the prototype into a
new generation of software for process monitoring and control.
"We were monitoring the process but we plan to put more research
into how we can better control the process," explains Rottorp. "We
will be merging scientific disciplines to enable real-time process
control and decision making. This should take about two to three
years."
The MAPP consortium included IVL, the University of Warwick, Centro
Ricerche Fiat, the Volvo Truck Corporation, Borlaenge Energy,
Kenwater (Kemira Chemicals) , BioBalance, DELTA, Sigma iSoft and
Umetrics.
Subject Descriptors : Communications networks, Information transmission
Subject Index Codes : Telecommunications, Information Processing, Information
Systems
Subject Class : IT, telecommunications
Remarks : Results entry form
Collaboration Sought : Further research or development support, Licence
agreement,
Marketing agreement, Financial support, Venture capital/spin-off funding
Collaboration Detail : support in developing the protype into new generation
software for
monitoring and control
Sources of Support : CEC
Programmes : IST
Projects : IST-1999-11990
Project Title : Multivariate Approach for statistical Process control and
cleaner Production
Contact Details
Contact Name : ROTTORP, Jonas
Contact Organisation : IVL Swedish Environmental Research Institute Ltd.
Address : Haelsingegatan 43
PO Box 210 60
City : Stockholm
Region : STOCKHOLM
Postcode : 100 31
Country : SWEDEN
Telephone Number : +46-8-59856302
Fax Number : +46-8-59856390
Electronic Mailbox : jonas.rottorp@ivl.se
