E-News January 2012 Edition

Reducing Costs with Optimization by Professor Mike J Grimble

The use of offline and online optimization methods has increased over recent years with new and more effective algorithms and software tools. Static optimization may be used for economic performance improvement and dynamic optimization may be used in control systems to provide best performance (in terms of the criterion selected).

The upper level optimisation in predictive control algorithms is often cited as the most effective component in achieving cost reductions or efficiency improvements. However, optimisation is now used in all aspects of system model development, control design and real time control. For example, roll force models in metal processing can utilise static models that are computed using optimization in a function space setting.

There is increasing interest in so-called intelligent or optimized adaptive control algorithms. The aim of such tools is to reduce the effort involved in tuning the system and to improve the quality of control as system parameters and conditions change. The numerical optimization algorithms need to be both fast and reliable for such systems. The additional features they can provide such as use of future set-point information and constraint handling provides a competitive advantage over other control solutions. The ADEX Adaptive Predictive Controller is an example being utilized successfully on power plant applications amongst many others.

Optimization is also used very effectively in the analysis of systems to reduce energy costs or select the most energy efficient solution given a range of possibilities. For example, the decision as to whether to use variable speed pumps or fix speed solutions in water distribution systems can have a large impact on the total cost of operation (ISC Pumpsim modelling tool). The optimization methods can be used to enable specifications to be precisely tuned to system needs helping to guide management decisions.

ACTC members have had optimization based studies in areas such as wind turbines, automotive engine control, power and chemical plant control.

In addition, the ACTC offers one, two or three days training in both static and dynamic optimization methods. This is accompanied by hands-on simulation experience to enable algorithms to be compared and benefits observed. These training courses are normally held at company premises and they are tailored to the application area of most interest to the company involved. Further details may be obtained from Meghan McGookin.

Mike Grimble


Video: Understanding the User Case for CIF

A new study has been completed by ISC that examines how Foundation Fieldbus CIF operates and where it is likely to offer the greatest advantages over traditional host-based process control. In the video. Dr. Andy Clegg explains the parameters of the study, and the basic findings.

His report discusses how ISC carried out the independent evaluation, and the circumstances under which the high determinism of CIF can outperform conventional loops driven by a PLC or DCS.


Planned Future Training Programs

Many companies, in particular large companies, are planning their future training activities more carefully. To provide better service, the ACTC started to change the way it undertook training after discussions with Boeing in Seattle. It was determined that there was a need to structure courses better and to have the levels of attendee more clearly identified. With this in mind the ACTC introduced basic, intermediate and advanced level training courses . This has been very valuable in the aerospace and automotive industries to enable training to be organised to coincide with career progression.

The ACTC started to change the way it undertook training after discussions with Boeing in Seattle. It was determined that there was a need to structure courses better and to have the levels of attendee more clearly identified. With this in mind the ACTC introduced basic level, intermediate level and advanced level training course activities and attempted to remove too much overlap between courses. This has been very valuable in the aerospace and automotive industries so that training could be organised to coincide with career progression.

To discuss how your company can benefit from this approach to training, please email Mike Grimble , Meghan McGookin. or call 0141 847 0515.

This does not of course preclude companies purchasing one-off individual courses either outside or within the ACTC program. Bespoke courses are still being assembled in very specialist areas to address particular company needs. For example, intensive courses have been developed for the wind energy, oil industry, as well as metal processing and automotive industries. These courses are held all over the world for companies and are tailored to the specific industry needs.

For further information please contact Meghan McGookin.


New Training Courses in Wind Energy

Two new training courses are being developed for presentation at company premises on wind energy systems. The first concerns the design of controllers for individual wind turbines and the use of condition monitoring and fault detection systems. The second focuses on offshore and onshore wind farm controls. Both of these courses include tutorial style presentations and involve hands-on simulation experience and experiments. The courses can be tailored either for design engineers/researches or for those more concerned with the operation and commissioning of wind turbines. These courses are in addition to the two international workshops the ACTC plans on these topics.

Further details may be obtained from Meghan McGookin.


Training Courses for Boeing, March 2011

Prof Mike Grimble and Dr David Anderson from the University of Glasgow presented two training courses for Boeing Commercial Airplane Group (BCAG) in Seattle. The first was a 3-day training course at basic level, namely ‘Control Fundamental’. This course was designed for delegates who required a refresher in control theory and/or newcomers who needed to become familiar with basic control engineering. The second was a 2-day intermediate course called ‘Optimisation Methods’. This course was aimed at engineers who needed to apply optimisation methods especially for optimal control and parameter identification. This course included examples that were specific to aerospace application. Both courses were very well-received and were attended by new and experienced aerospace design engineers.


Training Courses for Chrysler, March 2011

The ACTC successfully ran two training courses for Chrysler, Detroit in March 2011 to two different departments within Chrysler’s: an intensive 4-day training course consisted of a 3-day course on ‘Control Fundamental’ and a 1-day course on ‘Kalman Filtering and System Identification.’

The first training course was delivered to about 40 engineers from the Powertrain group and the second to the Electric Vehicles group and hybrid systems. Both training courses were well-received and there was a lot of interaction between Prof Mike Grimble, Petros Savvidis and the delegates to discuss real-life automotive issues and possible solutions.


ACTC Training Course for Garrad Hassan, Bristol, June 2011

During the summer, a 3-day advanced training course on Multivariable Control and Application was delivered to a group of Garrad Hassan control engineers from the area of wind turbine control applications. The delegates found the hands-on session valuable as it helped in understanding the theory. Very positive comments were received from the delegates such as ‘Yes, highly recommended.’


Training Courses for Boeing, June 2011

In June, two advanced control training courses were delivered by Prof Mike Grimble and Santo Inzerillo to a large group of aerospace engineers who had good background knowledge and experience in control theory. The first 2 days of the course covered ‘Robust and Reliable Control System Design’ and the final 2 days involved ‘Multivariable Control and Applications’. Built around aerospace applications, this course was very well received and good group discussions occurred.


Wind Farm Control Workshop at RWE Npower, Swindon, September 2011

Since our popular 2-day wind farm control workshop in May 2011, RWE Npower requested a related Wind Turbine Control course to be delivered in Swindon in September 2011. The windfarm workshop was delivered twice to two different groups of control engineers. Both groups found the course useful and ‘offered lots of insights and new ideas’.


Nonlinear Predictive Control – Is it Practical? by Tecspert

The predictive control of linear system has been applied very successfully in the process and petrochemical industries. Commercial manufacturers have produced predictive controls for nonlinear systems but often using empirical methods or approximations. Over the last decade, however, significant advances have been made in nonlinear predictive control. For example, the old idea of scheduling has been brought up to date.

University groups have developed a range of nonlinear predictive algorithms that were initially too complex both theoretically and for implementation. However, in recent years they have become more pragmatic and practical. For example, considerable effort has been put into obtaining relatively fast algorithms. In some methods most of the calculations are performed offline so that the real time implementation is much simpler.

Unfortunately, there is not as yet a main stream nonlinear predictive control algorithm that is accepted by all as the standard for the industry. In the case of linear systems most of the algorithms relate back to so-called Generalised Predictive Control or Dynamic Matrix Control class of algorithms, whereas in nonlinear systems no such common trend has yet emerged. This makes it difficult for equipment manufacturers to decide where to invest in new algorithms and facilities.

The main advantage of true nonlinear predictive algorithms is that performance can be maintained across a much wider range of operating conditions without having to detune the controllers. The fact that the control algorithm fits the system model better also provides an additional level of robustness in a rather indirect way.

The concerns which still exist are to do with ensuring performance and stability in non-ideal situations. However, for systems which are well understood and where the range of operation is well defined it should be possible to produce tailored predictive nonlinear algorithms to provide significant advantages. This may be particularly useful in high speed servo systems or other electrical mechanical systems where high performance is the major requirement.


Brett Ninness Plenary

Professor Brett Ninness, School of Electrical Engineering and Computer Science, The University of Newcastle Australia, presented his plenary at the IFAC World Congress in Milan on Computational System Identification.


Symposium in Honor of Edoardo Mosca and Roberto Genesio

The Symposium in Honor of Edoardo Mosca and Roberto Genesio was held in Firenze, September 5, 2011.


Advanced Control Conference in Florida

The major annual IEEE conference on control is the Conference on Decision and Control. This year the Conference reached its fiftieth birthday and was held in cooperation with the European Control Conference in Orlando, Florida. The very prestigious Bode keynote lecture was given by Professor John Baillieul of Boston University. In fact there was an excellent set of plenary and semi-plenary talks and the one that stood out was by Professor Thomas Parisini of Imperial College/University of Trieste.

It is not surprising that the attendance was much greater than usual and the range of topics considered was also very interesting. The conference included some very good application papers in areas such as wind energy and in automotive systems but there was not a large number of special sessions which one might find at an American Control Conference on applications.

For the first time the conference proceedings was provided on a memory stick rather than a CD. In years gone by we did of course struggle with multiple volumes of proceedings but I guess the cost of carrying on airlines would now make this prohibitive even if we were able to handle the large number of volumes and weight.

For those who enjoy water, sand and pina coladas, I would draw your attention to the 51st IEEE Conference on Decision and Control which will be held Monday through Thursday, December 10-13, 2012 at the Grand Wailea, Maui, Hawaii. The conference will be preceded by technical workshops on Sunday, December 9, 2012

Full details can be found here


New Nonlinear Industrial Control Book Website

Some of you may be aware that we have written a new book on nonlinear industrial control which should be published next year. You may like to know that we are establishing a website for this new book. The website will include the NGMV software which is an output of EPSRC supported projects and industrial studies kindly supported by ISC.

The software is numerous industrial application examples and control design methods ranging from the simple NGMV to robust H-infinity, predictive control and even controls which, when the system is linear, become LQG. It also includes the new nonlinear estimators.

The main reason for sending you this note is to make the suggestion that you put a link in from our website to the book website since I think it reflects well on us all.

In the meantime, anyone who would like access to the software can download the design package here


Invitation to Perspective Authors

Professors Mike Grimble and Mike Johnson of the University of Strathclyde are joint editors of an industrial monograph series on applied industrial control published by Springer. Any ideas for future monographs or offers to write such a text can be submitted to Mike Grimble.



We have had recent enquiries from companies interested in using polynomial or frequency domain based control solutions and wondered how they could obtain Matlab based software for solving the polynomial matrix equations. In fact a toolbox referred to as PolyX has been developed for polynomial systems by Professor Michael Sebek and his team in Prague. Further details of the toolbox can be found at http://www.polyx.cz/.