E-News August 2016 Edition

The Importance of Simulation by Professor Mike J Grimble

If on some future occasion I am to be sent to a desert island and can take only one control technique with me then it will have to be the very trustworthy simulation capability. PID control would of course be helpful on the sandy beaches but simulation is probably the most important tool for the control engineer. In years gone by this was certainly not the case and in fact only 30 years ago simulation facilities were what might be described as primitive. However, in recent years simulation has become a valuable tool, certainly for those employed in advanced control systems design.

In many industries simulation is needed to reduce the time is takes implementing advanced model based controls. The tuning of such systems can be performed in a safe and fast manner and what if questions can be answered. Because of the rapid model change in automotive applications there is a real need to reduce time consuming activities like engine calibration. The generation of engine maps and the tuning of controllers can be reduced in time if good simulation tools are available. Since the trend is to use model based control, the models needed for simulation will of course exist and tools like the LabVIEW based simulation facilities and the ubiquitous Matlab/Simulink are very common.

The need for simulation in for example energy systems is often to improve the total system control. The requirements for modelling and control in individual wind turbine control are surprisingly similar to those in automotive engine control. However other problems like total wind farm power control, whilst minimising fatigue and wear and tear requires a rather different approach. For individual wind turbines high fidelity models are required for each turbine here as for wind farm control the complexity of the models needs to be reduced and in any case sophisticated models are nor needed for the much slower dynamics regulating total wind farm power outputs. In this case problems arise due to the scale of the modelling and simulation problem.

Getting back to the sandy beach on the desert island you may have wondered why simulation was so important. The main reason is often associated with the comfort a good simulation provides. By knowing how the system is very likely to perform and by having a range of possible controller designs all if which give adequate performance the visit to the application is a lot less worrying and fraught. In my case since I can choose the desert island I think I will make it one in Hawaii and I am sure there are many applications to explore.

Mike Grimble


Control Fundamentals, 11th - 13th October 2016, Glasgow

ISC is pleased to announce that registration is now open for the Control Fundamentals Training Course being held on 11th - 13th October 2016 in Glasgow City Centre.

This three-day course introduces the basic control engineering concepts in terms of Linear Systems Models, Classical Control Theory and Practical Aspects in Control. The course presents analysis of a linear control system in terms of stability and performance. In addition, Classical Control Method such as PID controller (including tuning methods), lead-lag and root locus are also covered. Significant hands-on examples are used to reinforce the lectures.


Can Engineers Be Replaced by Software?

There is a thought that in a utopian world control design can be accomplished with super adaptive design software which does not require detailed knowledge of the system but is based upon data collected. Such a concept has been considered for many years but has been decades away from ever being realised. However, there are now developments in nonlinear system identification and control which makes such a possibility more likely.

One of the most costly stages of any control system design is the additional modelling and understanding required. If it were possible to replace much of this work by model identification and reliable control strategies, this would provide a considerable breakthrough. It would enable advanced control techniques to be applied where currently the effort required is daunting. Watch this space for developments in these ideas.


Model Based Control for Automotive

There are not many industries where a significant change in control philosophy has arisen leading to major changes in control design philosophy. However, the automotive industry seems almost universally to have accepted the need for model based advanced control and the use of the corresponding model based control algorithms. This is not to suggest that actual applications at the present time reflect this shift but it does appear that in future years it will be the dominant design philosophy.

It will not be a surprise that amongst the control design philosophes adopted, predicted control has a major role, and indeed nonlinear predictive control and simpler variants of the philosophy, has been dominant in recent research.

There are many reasons for the adoption of model based control that have persuaded automotive company management to support such developments. One of the driving forces is the need to produce new products more frequently which require new or modified engines, using innovative actuator and measurement systems.

Using a conventional approach significant effort is required by engine calibration teams to produce engine maps for the feedforward and feedback controls. By the use of dynamic models some of this work can be avoided and such models can be modified more easily when changes arise.

Another factor stems from the regulatory authorities that require engines to be more fuel efficient with less harmful emissions. To optimise the system models are required and suitable performance criteria specified. Classical empirical tuning methods do not provide such a capability.

There is also the rather specious argument that we should do it because we can, meaning that the computing power available in engine management systems will in the future allow much more sophisticated solutions to be employed. This is not of course a real reason for using advanced controls but it does answer the question why there is such enthusiasm and excitement at present.

The ideal in future will be to have good engine models, simulated using modern tools such as Simulink or LabVIEW, where the structure enables changes to be made quickly and easily by relatively inexperienced control design engineers. Of course, engineering insights and intuition will still be required but the initial development of simulation and design facilities could be accomplished by very experienced research and development teams. The software tools will be provided in a form where much less experienced colleagues can use the algorithms effectively. This will enable new staff to be absorbed into a company more quickly and enable expertise to be transferred more effectively. It will also enable new actuator or measurement devices to be modelled and added without requiring major design or control structure changes.

To explain the latter point a technique like model predictive control can easily have additional outputs added that must be controlled and additional actuators represented without changing the basic philosophy or tuning approach.

Further information may be obtained from Professor Mike J Grimble.


Quantitative Feedback Theory

The subject of Quantitative Feedback Theory (QFT) sounds highly theoretical but in fact it provides one of the most practical robust control design methods available. It is true that it has been used by a rather limited community of researchers but it is now becoming a much more accessible and valuable method, mainly because the availability of good design software.

Professor Mario Garcia-Sanz who is now at Case Western Reserve University, has developed a very easy to use design package particularly designed for multivariable systems. In the early days of QFT it was only really suitable for single-input single-output design but over the year's various software tools and design methodologies have been developed, which provide practical multivariable control solutions. One of the most successful packages and a text book were produced by Professor Constantine Hopes and Dr Steve Rasmussen. These both worked at the Air Force Institute of Technology in Ohio, USA in fact Professor Garcia-Sanz also worked with these researchers at various times (see Quantitative Feedback Theory: Fundamentals and Applications, Second Edition by Constantine H. Houpis, Steven J. Rasmussen, Mario Garcia-Sanz).

The main advantage of QFT is that it provides a robust control design tool and the robustness is meaningful. That is, there are a number of control design methods which provide some form of robustness, sometimes of only a theoretical nature. However, QFT really does allow the engineer to understand the robustness margins that are being achieved. In the early days a Nichols chart was used and a rather laborious process was needed to establish the boundaries of uncertainty which then allowed the robust control design to be achieved. In recent years the computer aided design packages make the whole process much faster and more efficient. Moreover, the type of experience need by the engineer is to have a good instinct for frequency domain control design. Classically trained control engineers often have very good instincts in this respect and hence the design method is very accessible.

Further details of software tools can be obtained from Professor Mike Grimble.


Use of Simulation for Training Simulators

One area of applications that ISC engineers specialise is creating dynamic training aids backed by physics-based simulations of real-life systems or processes. One recently completed operator training tool, in partnership with CPI Biologics (Darlington, UK), is a prototype biosimulator with an instructor and a trainee module. The instructor module supports customisation of the training needs and specific normal and fault conditions. The trainee module is equipped with a dynamic model of the targeted bio-process and enables the process operator to run training scenarios in real-time or higher speed as needed, experience the process quantitatively as far as possible during normal and fault scenarios, and learn about the diagnosis and correction of specific process abnormalities. Developed in LabVIEW, both the instructor and trainee modules can be deployed as standalone executables to run using the freely available LabVIEW Run-Time Engine, removing the need for LabVIEW development licenses thus lowering cost.

Another training tool currently being developed in partnership with PaleBlue Ltd (Stavanger, Norway) and Norsk Yrkesdykkerskole NYD (Oslo, Norway), and awarded Eurostars program (funding by EU Eureka) in 2015, is a Dive Control Simulator which combines dynamic modelling of the underlying physical processes, photo realistic control panels and video of diver operations for the training of supervisors in the field of saturation diving. The Dive Control Simulator presents training scenarios including normal operations, what-ifs and emergency cases. For more info, visit project page at: http://www.paleblue.no/#!dive-control-simulator/wwg8h.

Please contact Xiaohong Guan if your company has any needs for training simulators.


ISC and ACTC Training Courses

EDF Bespoke Control Engineering Practice for Nuclear Power Stations

ISC Ltd has successfully developed a bespoke 4-days training course for EDF in Barnwood. The bespoke training course took over a year to develop and it has been tailored to meet EDF’s objectives which included engaged thinking, open & collaborative relationships and knowledge transfer. This training course focuses on ‘control engineering practice’ for EDF Nuclear Power stations and provides good understanding of theoretical concepts reinforced with EDF specific application examples.

The course also designed to support operational safety & excellence, and reinforce the key principles of nuclear professionalism. A pilot course has been delivered and it has been well-received by the relevant audience. This training course is officially scheduled to run in September 2016.

Phillips 66: Introduction to Process Control, Humber Refinery

ISC Ltd has successfully delivered a one-day training course in ‘Introduction to Process Control’ to Phillips 66, Humber Refinery. This is one of our most popular training courses as it aims at technicians and/or operators that require a basic understanding to process control and the importance of good control engineering. This training course provides the fundamental knowledge of applied control engineering without all the heavily involved mathematics. This training course is usually well received and the technicians find the course beneficial in terms of relating the lecture material to their day-to-day job.

Thales Optronics, Glasgow: 4-days Nonlinear, Multivariable System, Predictive Control and Estimation

Thales, Glasgow has been an invaluable client to ISC Ltd and recently they have requested a tailored 4-days training course in advanced control engineering. The topics were carefully selected by Thales to suit their engineers’ current needs in terms of research & development and knowledge transfers. This is an intense training course and a prerequisite to control engineering is required.

Perceptive Engineering, Daresbury: 2-day Introduction to Pharmaceutical Process Control

ISC ltd has been working closely with Perceptive Engineering to deliver a 2-days training course in ‘Introduction to Pharmaceutical Process Control’ as part of the ADDoPT program (https://www.addopt.org/). The training course is tailored to allow engineer from Pharmaceutical / Crystallisation industry to have better understanding of control engineering and how this knowledge can be applied to yield better (and consistent) production. The training course is very well-received and this course will be re-run in November 2016.


What is the Best Multivariable Controller?

30 or 40 years ago the multivariable control problem still did not have a very good solution. In recent years many multivariable control methods have proven to be successful in a range of applications. The best known method is model based predictive control. Such a controller is easy to program and can handle both soft and hard constraints. It does of course have a few difficulties in its common form, which is dynamic matrix control or generalised predictive control. That is, stability is not guaranteed for all cost choices, even if the plant is linear and known exactly.

There is also the point that even in the unconstrained version the controller does not have a traditional structure. It is more a black box numerical calculation rather than something that has an intuitive control structure. Having said that predictive control is still by far the most popular multivariable control method and it is one of the few where the calculations are sufficiently simple that very large systems can be controlled without massive computational requirements.

Linear Quadratic Gaussian (LQG) control has also been successful in applications, particularly in the aerospace and marine industries. For example, dynamic ship positioning systems almost universally employ this approach (except for low priced basic systems). The main benefit of LQG control is that it is truly a stochastic control method that is can handle disturbances, measurement noise, and bias terms in a very convenient manner. It is well known that Kalman filter is part of the LQG solution. The multivariable control action is provided naturally since there are no assumptions which restrict the system to be single-input single-output. This does of course also apply to the predictive control case. Its main weakness is that it does not handle hard constraints directly and robustness problems can be an issue, although they are not common.

The Hinfinity so called robust control theory provides another good multivariable controller which also does not include constraint handling directly. The H? design methods are perhaps better in theory than in practice, since the robustness they provide is to uncertainties that are not very representative of that usually encountered. However, they also provide a gift which was not asked for. That is, H? controllers provide remarkably good multivariable non-interacting control action in the case where the plant model is well known. The step-responses are sharper and the interaction less than with most other methods. This is because the cost-function involves sensitivity functions which not only effect robustness but also determine time responses.

Classical methods of multivariable control are of course still used and there are attempts to use Bristol's relative gain array and decoupling matrices in various ways. However, good multivariable control requires dynamic controllers rather than static compensators and for the majority of cases, that require high performance, it is normally better to use one of the model based optimal control design methods.

The big question is therefore which of the above approaches or one of the other techniques is likely to provide the best multivariable control solution. Since most multivariable systems are also affected by significant nonlinearities questions arise whichever of the above approaches are used. If stochastic properties dominate then LQG control is a good place to start but if the system is very nonlinear, large scale, or if constraint handling is required, then predictive control is probably a good option.

Further advice on which multivariable design method may be best for you may be obtained by contacting ISC


Advisory Service

ISC Ltd., the parent company of the ACTC, was established by the University of Strathclyde almost 30 years ago to provide a technology transfer service for industry. The company has always been willing to provide advice on control problems to engineers in industry.

If you have a control problem and would like feedback please contact us at ISC Ltd.


Ideas for Industrial Control Text Books and Monographs

The ACTC news often has book reviews on texts that may be of interest to engineers in industry. Unfortunately, control engineering tends to be dominated by control theory texts rather than applications topics. Professor Mike Grimble has therefore invited ideas from potential authors, or anyone that has a suggestion for a text book or monograph. Professor Grimble of the University of Strathclyde's Industrial Control Centre, is of course a Director of ISC, the parent company of the ACTC. He and Professor Michael Johnson (who is also a Professor from Strathclyde) jointly manage both the Springer Advances in Industrial and Control Monograph series and the Springer Control and Signal Processing text book series. Any ideas to fill gaps in industrial control applications knowledge would be welcome.

This seems a good opportunity to thank the various publishing houses for providing texts to the ACTC to enable book reviews to be conducted.


Book Review: Digital Control Applications Illustrated with MATLAB®, by: Professor Hemchandra Shertukde, Published By: CRC Press, Cost: £84

Books on control that use MATLAB® and Simulink are very popular with engineers in industry. This new book on Digital Control Applications, Illustrated with MATLAB® is by Professor Hemchandra Shertukde from Hartford University, Connecticut. This is a book clearly designed for use by students on University courses, but it should be valuable for engineers in industry, because it deals with the very important topic of digital control systems analyses, synthesis and control design. The numerous examples are very helpful and the text also includes problems which are helpful to lecturers and instructors.

Some books on Digital Control include material like the rather dated sampled data control systems analysis books popular two decades ago. These were valuable at the time but have an emphasis that is somewhat different to what is now required for Digital Control systems courses. This book on the other hand provides a very nice tutorial introduction covering the range of topics which are really needed in both courses and applications.

The spirit of the book is more concerned with the basics of digital control systems and the reference to MATLAB® in the title is a little misleading since most of the text is on the theory of such systems. However, this is a very valuable text for anyone involved with digital control systems design. The problems that arise in digital systems are discussed in a clear and precise manner and both the layout and content of the book ensure it is easy to read and understand. This is highly recommended for the book shelves of engineers and for graduate course students.

The book is published by CRC Press and includes 9 chapters together with various appendices concerned with MATLAB®. The cost of the hardcover text is reasonable at £84.

Book Nerd.


Book Review: Stability Regions of Nonlinear Dynamical Systems - Theory, Estimation, and Applications, by: Hsiao-Dong Chiang and Luís F. C. Alberto, Published By: Cambridge University Press , Cost: ££89.99

The text is spilt into 4 parts covering theory, estimation, advanced topics and applications.

The subject of dynamical system is of course of great importance in industrial applications. If systems are considered to be almost liner the stability conditions are reasonably straight forward but if systems are nonlinear or time varying the stability conditions are more complex to determine. It might be thought that stability is more of an academic consideration since if the system is well defined it should not enter regions of instability. However, Evermore complicated models are used for larger industrial control systems and high performance is sought leading got grater difficulties with total system stability.

The text is well illustrated with examples including power system transient stability analysis. The book is mainly of value to researchers in advanced control systems but it is a well written additional text for the industrial library.

Book Nerd.


Book Review: Game Theory with Engineering Applications, by: Dario Bauso , Published By: SIAM, Cost: $82.50

This new text by Dario Bauso is concerned with game theory with engineering applications and is published by the Society for Industrial and Applied Mathematics (SIAM). The subject of game theory does of course arise in several areas of engineering and particularly in control systems engineering. For example, it is one method by which H? robust controllers can be obtained. This text begins with chapters that provide an introduction to game theory and to the various techniques which are needed in this optimisation problem. From a control systems point of view the disturbance input aims to maximise cost whilst control action aims to minimise cost. The game problem can therefore be set up to represent a robust control problem whereby control action attempts to minimise the worst effect of disturbances.

This text is quite comprehensive in providing a reasonable overview of the subject together with a number of sections dealing with different applications areas.

This book is a useful addition to any engineer's bookshelf who regularly deals with system optimisation and who wishes to understand the background theory of some of the robust control papers that involve such a solution technique.

Book Nerd.