From: "Saved by Windows Internet Explorer 7" Subject: UTC for Advanced Instrumentation Date: Fri, 23 Jan 2009 00:47:44 +0800 MIME-Version: 1.0 Content-Type: multipart/related; type="text/html"; boundary="----=_NextPart_000_0000_01C97CF4.34208260" X-MimeOLE: Produced By Microsoft MimeOLE V6.0.6001.18000 This is a multi-part message in MIME format. ------=_NextPart_000_0000_01C97CF4.34208260 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Location: http://seva.eng.ox.ac.uk/publications_journals_academic_abstracts.html UTC for Advanced Instrumentation
3D"University Invensys UTC for Advanced Instrumentation =

Academic papers with abstracts

=20
M. Zamora, H. Wu and M. P. Henry, ``An FPGA = Implementation=20 of Frequency Output,'' IEEE Transactions on Industrial = Electronics,=20 Accepted for future publication.
Digital frequency input and output (typically in the range = 1 Hz=20 to 100 kHz) for data transmission is employed in many industrial=20 applications. This paper provides the following elaborations of the = ISIE 07=20 conference paper [1]: a thorough literature review suggests previous = techniques can be classified into three basic approaches. = Theoretical=20 expressions for the errors of each are derived and compared with the = new=20 approach developed by the authors. Each method has been implemented = in a=20 more recent FPGA architecture (Spartan 3), and the results are = consistent=20 with the theoretical values. The new method provides a precision of = 6 x 10=20 -6 % or better for all frequencies, based on a 40 MHz clock. =
=20
M. Zamora and M. P. Henry, ``An FPGA implementation of a = digital=20 Coriolis mass flow metering drive system,'' IEEE Transactions on=20 Industrial Electronics, vol. 55, pp. 2820-2831, July = 2008.
Coriolis mass flow metering provides a direct measurement = of=20 mass flow and is generally regarded as the most accurate and precise = flow=20 technology in common use in industry. This paper describes the role = of the=20 field-programmable gate array (FPGA) hardware, which is programmed = using the=20 Handel-C language, in the implementation of a =93digital=94 Coriolis = meter,=20 which replaces the conventional analog positive feedback system used = to=20 maintain flowtube oscillation. The FPGA is coupled to a = microprocessor,=20 which carries out conventional measurement tasks and selects the = drive=20 parameters to be used by the FPGA. The resulting meter is able to = maintain=20 the operation in more difficult process conditions, including = two-phase=20 flow, which has previously caused Coriolis meters to cease = oscillation. The=20 system described in this paper is used in a commercialmeter that has = been=20 successfully applied to two-phase industrial applications. =
=20
M. D. Duta and M. P. Henry, ``The fusion of redundant = SEVA=20 measurements,'' IEEE Transactions On Control Systems = Technology,=20 vol. 13, pp. 173-184, March 2005.
The self-validating SEVA sensor carries out an internal = quality=20 assessment, and generates, for each measurement, standard metrics = for its=20 quality, including online uncertainty. This paper discusses = consistency=20 checking and data fusion between several SEVA sensors observing the = same=20 measurand. Consistency checking is shown to be equivalent to the = maximum=20 clique problem, which is NP-hard, but a linear approximation is = described. A=20 technique called uncertainty extension is proposed which causes a = smooth=20 reduction in the influence of outliers as they become increasingly=20 inconsistent with the majority.
=20
M. P. Henry and R. Mercado, ``Advances in Coriolis mass = flow=20 metering technology,'' ATP International, March 2005.
=20
I. Gyongy and D. W. Clarke, ``On the automatic tuning = and=20 adaptation of pid controllers,'' Control Engineering Practice, in=20 press, 2005.
A simple approach to the automatic tuning of PID process=20 controllers is proposed. Like the relay-based autotuner, its = objective is to=20 attain a design-point on the Nyquist diagram. By injecting sinewaves = and=20 employing a phase/frequency estimator, closed-loop adaptive tuning = is=20 possible and there is exact convergence to the design-point without = the=20 approximations of describing-function theory. The variant discussed = achieves=20 a required phase margin and imposes a carefully chosen constraint on = the=20 controller parameters, leading to consistent behaviour for a wide = variety of=20 generic test-cases. A real-life demonstration on a non-linear flow = rig is=20 provided.
=20
C. Clark, R. Cheesewright, S. Wang, M. P. = Henry,=20 M. E. Zamora, and M. S. Tombs, ``A radically new dynamic = response=20 capability for Coriolis flow meters,'' Sensors and Actuators A:=20 Physical, 2005. accepted 10 March 2005. Available online 6 May = 2005.
The dynamic response of flow meters is significant in many = applications, including fast control operations, e.g. short duration = (less=20 than 1 s) batch filling and for tracking the periodic flow = fluctuations=20 produced by positive displacement devices. The factors which = determine=20 Coriolis dynamic response have been elucidated. It has been shown = that the=20 meter flow tube response time cannot be less than the duration of = one drive=20 cycle of the tube vibration (i.e. reciprocal of drive frequency). = This gives=20 the potential of a response time of order 1 ms for the fastest = currently=20 available meters. However, the delay-time and update rates from the = user=20 output depend upon flow transmitter technology and design. Flow tube = dynamic=20 response has been investigated theoretically (simple straight tube), = by=20 finite element simulation (complex flow tube shapes) and = experimentally.=20 Commercially available meters were tested to determine the flow tube = dynamic=20 response to step changes in flow rate and the response to low = frequency=20 (compared with meter drive frequency) flow pulsations. Generally, = dynamic=20 flow events have been found to introduce contaminating signal = components at=20 one or more frequencies, other than that of the meter drive. The = paper also=20 presents details of the signal processing used to extract the = required=20 phase-difference and a method for reducing the contaminating signal = noise. A=20 new fast-response meter is currently being developed and some of the = significant advances in the technology of a novel digital = transmitter are=20 described.
=20
M. P. Henry, ``Digital Coriolis improves polyethylene = productivity,''=20 IEE Computing & Control Engineering Journal, = vol. 15,=20 pp. 4-5, December 2004.
=20
M. S. Tombs, M. P. Henry, and C. Peter, ``From = research to=20 product using a common development platform,'' Control = Engingeering=20 Practice, vol. 12, pp. 503-510, April 2004.
This paper describes a hardware and software platform used = to=20 develop prototypes of advanced instrumentation in a university = research=20 environment. It consists of a commercial processor board, running a=20 real-time operating system, linked to analogue electronics via a = highly=20 flexible field-programmable gate-array (FPGA). The use of Handel-C = to=20 configure the FPGA means that the prototypes are essentially = =91software=20 products=92 and that the intellectual property created is readily = ported to=20 commercial applications using components with appropriate price, = performance=20 and power consumption. Two examples, from the process and transport=20 industries, demonstrate how the platform can be used to move from = research=20 prototype to near-commercial product.
=20
M. P. Henry, ``Coriolis transmitter technology and dynamic = response=20 performance,'' Measurement and Control, vol. 36,=20 pp. 278-281, November 2003.
The paper deals about the fundamental limits on the = dynamic=20 response of Coriolis meter flowtubes. Commercial meters exhibit much = poorer=20 dynamic response times due to time delays. The primary benefit of = the=20 Coriolis metering is the direct measurement of mass flow where = commodity=20 value is related to mass rather than volume. In practice, the = response time=20 of commercial flowmeters has been slower and found to be delays in = excess of=20 1s for one mass flowmeter. The limitations led to the development of = a new,=20 all-digital, design which eliminates classes of fault and offers a = much=20 improved dynamic response. The rapid dynamic response of both the = control=20 and measurement function of the digital transmitter has led to = improved=20 performance in difficult industrial conditions.
=20
M. P. Henry, ``Coriolis meter digital transmitter = technology,''=20 IEE Computing & Control Engineering Journal, = vol. 14,=20 pp. 34-35, August 2003.
Recent years have seen significant advances in Coriolis = flowtube=20 technology Arguably, however, it is advances in transmitter = technology that=20 will ultimately prove more significant in providing users with the = desired=20 level of measurement performance and robustness. The key to = transmitter=20 enhancement is new digital technology, enabling improvements in = flowtube=20 control, signal processing, and measurement quality. The Sensor = Validation=20 Research Group at the University of Oxford, in partnership with = Invensys=20 Foxboro, has been developing Coriolis transmitter technology for = many years.=20 The latest Oxford prototype is the basis for the new Foxboro CFT50 = Coriolis=20 transmitter product. Key features include digital synthesis of the = drive=20 waveforms, and maintaining operation during two-phase or aerated = flow.=20
=20
G. Tortora, B. Kouvaritakis, and D. W. Clarke, = ``Fault=20 accommodation with intelligent sensors,'' Automatica, = vol. 39,=20 pp. 1227-1233, July 2003.
Optimal sensor fault-accommodation is considered for = faults=20 modelled by an increase in measurement noise. This noise is taken to = be=20 bounded and its probabilistic properties unknown. It is assumed that = intelligent (e.g. self-validating) instrumentation is in use and = estimates=20 of the noise bounds are available. The fault-tolerant controller is = designed=20 to optimize a noise rejection and a nominal reference tracking index = and=20 leads to a mixed norm minimization problem (l/sub 1//l/sub 2/). We = exploit=20 known results and a particular feedback configuration to show that = it is=20 possible to optimize simultaneously without a trade-off the two = performance=20 indices. The results are applied to systems where the presence of = auxiliary=20 measurements allows for an optimal fault-accommodation strategy. = Using=20 properties of the optimal solution, we define a factorization for = the=20 optimal controller alternative to the Youla parametrization, leading = to an=20 algorithm which is optimal, transparent and efficient.
=20
M. P. Henry, C. Clark, M. D. Duta, = R. Cheesewright,=20 and M. S. Tombs, ``Response of a Coriolis mass flow meter to step = changes=20 in flow rate,'' Flow Measurement and Instrumentation, = vol. 14,=20 pp. 109-118, June 2003.
Trials have taken place to determine the response of a = prototype=20 Coriolis mass flowmeter to step changes in flow rate. The meter = typically=20 exhibits a delay of 16 ms, and tracks step changes well. Comparison = with=20 previously published results suggests that this performance = represents an=20 improvement over several types of current commercial flowmeters. The = effect=20 of a correction technique to reduce noise is demonstrated. This is=20 considered valuable, even though it is responsible for 6 ms of the = overall=20 delay. A good dynamic response is important in process industry = applications=20 with short batch times.
=20
M. P. Henry, C. Clark, and R. Cheesewright, = ``Pushing=20 Coriolis mass flowmeters to the limit,'' IEE Computing & = Control=20 Engineering Journal, vol. 14, pp. 24-28, June 2003.
Coriolis meters are well known for their steady state = accuracy=20 and repeatability but they suffer from poor dynamic performance. New = developments are about to improve on this shortcoming.
=20
D. W. Clarke and T. Ghaoud, ``A dual phase locked loop = for=20 vortex flow metering,'' Flow Measurement and Instrumentation, = vol. 14, pp. 1-11, March 2003.
A vortex-shedding flow meter produces an approximately=20 sinusoidal signal whose average frequency is proportional to = volumetric flow=20 rate. The frequency of vortex shedding varies over a large range (up = to=20 1:100), depending on the size of the flow meter. The conventional = frequency=20 estimator uses a zero-crossing algorithm (ZC), but this approach = fails at=20 low flows, where the signal-to-noise ratio is poor. An alternative = is to use=20 a phase-locked loop (PLL), but typical single PLL designs cannot = cope with a=20 wide frequency range at high noise levels. This paper describes a = dual PLL=20 structure that can be used to track the vortex shedding frequency = over the=20 flow meter=92s full range of operation. This dual PLL provides = improved=20 accuracy and tracking ability compared with the conventional = zero-crossing=20 algorithm.
=20
D. W. Clarke and J. Park, ``Phase-locked loops for plant = tuning=20 and monitoring,'' Proc.IEE, Part D, vol. 150, = no. 2,=20 pp. 150-169, 2003.
=20
D. W. Clarke, ``Pretuning and adaptation of pi controllers,'' = Proc.IEE, Part D, vol. 150, no. 6, = pp. 585-598,=20 2003.
=20
D. W. Clarke and T. Ghaoud, ``Pretuning and adaptation = of pi=20 controllers,'' Flow Measurement and Instrumentation, = vol. 14,=20 no. 1, pp. 1-11, 2003.
=20
G. Tortora, K. Kouvaritakis, and D. W. Clarke,=20 ``Fault-accommodation with intelligent sensors,'' Automatica, = vol. 39, 2003.
=20
M. S. Tombs, ``Advanced Coriolis mass flow metering,''=20 Introduction to Special Issue published in IEE Computing and = Control=20 Engineering Journal, August/September 2003.
=20
M. E. Zamora, M. P. Henry, and C. Peter, = ``Generation of=20 frequency output for instrumentation applications using digital = hardware,''=20 Sensor Review, vol. 23, no. 2, pp. 143-149, = 2003.
The use of frequency output for measurement transmission = remains=20 common in the design of smart transmitters. Conventional methods of=20 frequency generation, based on counting clock cycles, have a = precision which=20 is inversely proportional to the frequency to be generated. = Consequently,=20 frequency output precision could be much lower than the measurement=20 precision. This paper describes a simple frequency generation = technique=20 which, when implemented in low-cost hardware, provides a precision = of 10_-6=20 per cent for all frequencies. The method represents an intermediate=20 non-available frequency by dithering between two exact frequencies.=20 Averaging over some reasonably short timescale provides the desired=20 frequency to high precision.
=20
D. W. Clarke and T. Ghaoud, ``Validation of vortex = flowmeters,''=20 IEE Computing & Control Engineering Journal, = vol. 13,=20 pp. 237-241, October 2002.
A vortex-shedding flow meter produces an approximately = sinewave=20 signal whose average frequency is proportional to volumetric flow = rate. The=20 frequency of vortex shedding varies over a large range (up to = 1:100),=20 depending on the size of the flow meter. The conventional=20 frequency-estimation method-a zero-crossing algorithm-fails at low = flows,=20 where the signal:noise ratio is poor. This paper describes a dual=20 phase-locked loop structure that tracks the frequency over the = flowmeter=92s=20 full range with improved accuracy. The lock indicators associated = with each=20 PLL provide the SEVA metrics.
=20
F. Zhou, N. Archer, J. Bowles, M. Duta, = M. P.=20 Henry, M. S. Tombs, M. E. Zamora, S. Baker, and = C. Burton,=20 ``Remote condition monitoring and validation of railway points,'' = IEE=20 Computing & Control Engineering Journal, pp. 221-230, = October=20 2002.
After recent railway tragedies in the UK, the condition of = the=20 railway infrastructure has become a matter of rising concern, = focusing on=20 railway safety and the optimal management of railway maintenance. = Recent=20 developments in microelectronics and IT make possible low-cost, = high-quality=20 condition monitoring for trackside equipment, to reduce the = likelihood of=20 breakdown and to enhance railway service quality.
=20
D. W. Clarke, ``Designing phase-locked loops for = instrumentation=20 applications,'' Measurement, vol. 32, no. 3,=20 pp. 205-227, 2002.
Many transducers provide a pulsed output whose frequency = depends=20 on the specific measurand=92s property. Conversion of the signal = from a=20 frequency-analogue to provide a magnitude-analogue output requires = an=20 accurate estimate of frequency, which is difficult when there is = significant=20 noise. Two common approaches to this problem-zero-crossing (ZC) = detection=20 and phase-locked loops (PLL)-are analysed and compared. The = classical PLL=20 however has difficulty handling baseband signals, where the expected = input=20 frequency ratio is large. A carefully chosen PLL structure, using=20 heterodyning and a Hilbert transform phase-sensitive detector, is = shown to=20 cope with input data having high noise and a wide spread of possible = frequency. The corresponding PLL design equations are derived: these = indicate the improvement in performance over that of a classical = PLL, an=20 all-digital PLL and a ZC algorithm.
=20
T. Ghaoud and D. W. Clarke, ``Modelling and tracking a = vortex=20 flow-meter signal,'' Flow measurement and instrumentation,=20 vol. 13, no. 3, pp. 103-117, 2002.
The piezoelectric sensor in a vortex-shedding flow meter=20 produces a raw signal whose average frequency indicates volumetric = flow=20 rate. Imposed on this signal are amplitude and frequency variations, = together with noise caused by plant- specific pressure pulsations. = The=20 objective of this paper is to provide and verify a simulation model = of the=20 overall signal, so that different frequency-tracking methods can be=20 compared. In particular the performance of a zero-crossing algorithm = is=20 evaluated. A prefilter is designed to extend the tracking ability to = lower=20 flow rates, where the signal:noise ratio is poor. Experimental tests = in real=20 time on a flow rig demonstrate the improvement in turn-down by a = factor of=20 two.
=20
G. Tortora, B. Kouvaritakis, and D. W. Clarke,=20 ``Simultaneous optimization of tracking performance and accommodation = of=20 sensor faults,'' Int.J Control, vol. 75, no. 3,=20 pp. 163-176, 2002.
Sensor performance can be monitored using Self-Validating = SEVA=20 devices. These supply the user with an estimate of the measurement=20 reliability as well as the measurement value. This paper presents a = sensor=20 fault-accommodation strategy exploiting information regarding the=20 measurements quality, such that indices of dynamic performance and = noise=20 rejection are continuously optimized as the sensor reliabilities = vary. An=20 important result proved in this paper is that the minimization of = these two=20 indices is completely decoupled and hence noise rejection can be = maximized=20 without trading-off dynamic performance. Two further = fault-accommodation=20 strategies based on interpolation are then derived with the intent = of=20 trading-off optimality for computational simplicity and = transparency.=20 Numerical examples are provided showing that one of these = (stabilizing=20 interpolation) is particularly attractive, being only negligibly = suboptimal=20 for the examples considered and much better than a =91do-nothing=92 = strategy.=20
=20
M. P. Wen, Y. Henry, ``Time frequency characteristics of = the=20 vibroacoustic signal of hydrodynamic cavitation,'' IEEE Journal of = Vibration and Acoustics, pp. 469-475, 2002. 124.
=20
M. P. Henry, ``On line compensation in a digital Coriolis = mass flow=20 meter,'' Flow Measurement and Instrumentation, vol. 12,=20 pp. 147-161, April 2001.
For many years it has been demonstrated that the use of = digital=20 technology, particularly embedded microprocessors, can improve = flowmeter=20 performance in various ways (e.g. temperature compensation, = elimination of=20 drift, generation of engineering units, diagnostics). A = self-validating or=20 SEVA sensor uses in- built processing power to generate generic = metrics of=20 measurement quality, based on on-line uncertainty. This uncertainty = includes=20 all factors effecting the on-line measurement, including diagnostics = but=20 also manufacturing aspects such as the components used and the = calibration=20 procedure. This paper describes a prototype SEVA Coriolis mass flow = meter=20 transmitter built primarily from digital components, which provides=20 compensation for both faults and manufacturing limitations. A = technique to=20 compensate for drift and imbalance in the transmitter front-end = circuitry is=20 described in detail. The prototype has been developed using a = hardware/=20 software co-design approach in which virtually all aspects of = instrument=20 design are described in software, but which can be implemented = flexibly in=20 either hardware or software according to economic requirements. This = approach offers manufacturers the opportunity of incorporating the = latest=20 components into their products rapidly in order to remain = competitive.=20
=20
R. P. Liu, M. J. Fuent, M. P. Henry, and M. D. = Duta,=20 ``A neural network to correct mass flow errors caused by two phase = flow in a=20 digital Coriolis mass flowmeter,'' Flow Measurement and=20 Instrumentation, vol. 12, pp. 53-63, March 2001.
Coriolis mass flow meters provide accurate measurement of=20 single-phase flows, typically to 0.2%. However gas-liquid two-phase = flow=20 regimes may cause severe operating difficulties as well as = measurement=20 errors in these flow meters. As part of the Sensor Validation SEVA = research=20 at Oxford University a new fully digital Coriolis transmitter has = been=20 developed which can operate with highly aerated fluids. This paper = describes=20 how a neural network has been used to correct the mass flow = measurement for=20 two-phase flow effects, based entirely on internally observed = parameters,=20 keeping errors to within 2%. The correction strategy has been = successfully=20 implemented on-line in the Coriolis transmitter. As required by the = SEVA=20 philosophy, the quality of the corrected measurement is indicated by = the=20 on-line uncertainty provided with each measurement value. =
=20
M. P. Henry, ``Self validating sensors: torwards standards = and=20 products,'' Automazione e Strumentazione, vol. 49,=20 pp. 107-115, February 2001.
Oxford University and the Foxboro Company have been = developing=20 the concept of the self-validating SEVA sensor. Accordingly over the = last=20 year there have been a number of developments leading to a British = Standard=20 for online measurement quality assessment based upon Seva. The paper = covers=20 each of these topics in turn. Firstly, the fundamentals of SEVA are=20 described. The Coriolis standardization activities are then = discussed,=20 followed by the moves towards commercialization. Finally, some = research=20 results from Oxford are presented.
=20
D. W. Clarke, ``Adaptive control of servohydraulic = materials-testing=20 machines: a comparison between black- and grey-box models,'' = Annual=20 reviews of control, vol. 25, pp. 77-88, 2001.
=20
D. W. Clarke, ``On the design of adaptive notch filters,'' = Int.J.=20 Adaptive Control and Signal Processing, vol. 15, = pp. 715-744,=20 2001.
In instrumentation and other applications, the online = estimation=20 of the frequency and amplitude of a noise-corrupted sinewave is of = great=20 practical interest. An adaptive notch filter (ANF) with global = convergence=20 properties has been developed, and is a candidate approach to our = problem.=20 This paper analyses the transient and noise properties of this ANF = and=20 equips the method with design equations. Using frequency ranges = greater than=20 (up to 2 decades) and signal/noise ratios less than (down to -16 dB) = those=20 commonly found in the ANF literature, it is verified by extensive=20 simulations that the new frequency estimator has excellent tracking = and=20 noise-rejection properties, provided that the signal/noise ratio is = not too=20 small. A comparison is made of its behaviour with that of a = phase-locked=20 loop, a method commonly used in practice.
=20
D. W. Clarke, ``The road to plant validation,'' IEE = Computing and=20 Control J, pp. 169-178, 2001.
Effective production requires serviceable unit processes, = under=20 stable control with credible sensed data and precise actuation. = Validation=20 of these components is not just an aid to safe operation, but can = also=20 enhance the availability of the plant. Cost-effective validation = uses=20 standard metrics for the intercommunication of results via a network = and=20 starts at the sensor and actuator level. In particular, greater = availability=20 comes from limp-home features that provide a degree of operability = even in=20 the presence of faults. This can be achieved at the device level, by = adjusting a loop=92s PID regulator, by using auxiliary variables and = data-fusion, or by the reconfiguration of a model-predictive = controller=20 (MPC). Examples are given of self-validating SEVA field instruments = and the=20 effect of adjusting a control loop using SEVA data. Plant validation = is a=20 combination of SEVA devices and of change detection based on the = residuals=20 derived from an MPC model.
=20
M. P. Henry, ``Recent developments in self validating SEVA = sensors,''=20 Sensor Review, vol. 21, no. 1, pp. 16-22, = 2001.
Provides a brief overview of developments in the field of=20 self-validating SEVA sensors. The fundamentals of SEVA are = described, using=20 the example of a dissolved oxygen sensor. UK Coriolis = standardization=20 activities are then discussed along with moves towards Coriolis=20 commercialization. Finally, some research results based on a = Coriolis mass=20 flow meter are presented.
=20
M. P. Henry, ``Plant asset management via intelligent sensors = digital, distributed and for free,'' IEE Computing & Control=20 Engineering Journal, vol. 11, pp. 219-227, October = 2000.
Instrument redesign based on a validation analysis of the=20 current product can deliver better performance and the elimination = of fault=20 modes. A new Coriolis mass flow meter transmitter, developed using = audio=20 technology, is significantly out-performing the commercial product = from=20 which it was derived.
=20
D. W. Clarke, ``Intelligent instrumentation,'' = Trans.Inst.=20 Measurement and Control, vol. 22, no. 1, pp. 3-27, = 2000.
With the falling cost of microelectronics, intelligent=20 instruments are increasingly used in process control. They are = designed to=20 integrate into data networks such as Fieldbus, so that remote users = can=20 access internal values and reconfigure their function. It is argued = that=20 such instruments should exploit models of the plant=92s data = generation to=20 achieve optimal filtering, of their own internal behaviour as = verified by=20 self-testing, and of the user=92s needs. A figure of merit for = sensor signal=20 processing is defined, and some fundamental results indicating the = benefit=20 of appropriate signal models are deduced. The value of validation, = in which=20 diagnosed faults affect the measurement status and its uncertainty, = is=20 emphasised. Examples of a self-validating thermocouple, a = dissolved-oxygen=20 sensor and a flow-control valve are described together with some = guidelines=20 about practical implementation.
=20
M. P. Henry, D. W. Clarke, N. Archer, = J. Bowles,=20 M. Leahy, R. Liu, J. Vignos, and F. Zhou, ``A=20 self-validating digital Coriolis flowmeter: an overview,'' Control = Eng.=20 Practice, vol. 8, pp. 487-506, 2000.
A new implementation of a Coriolis mass-flow meter = transmitter=20 is described. It is based on digital components, and has improved=20 performance compared with the commercial, mostly analogue, = transmitter using=20 the same flowtube (transducer). Improvements are found in flowtube = control,=20 measurement precision, and performance with two-phase and = partially-empty=20 conditions, including batching from empty. The new transmitter is = viewed as=20 a second-generation sensor validation SEVA demonstrator, in which = experience=20 from validating the commercial analogue transmitter has led to a = redesign=20 using digital technology. The resulting SEVA transmitter provides = improved=20 measurement performance and reduced vulnerability to fault = conditions, as=20 well as online estimates of measurement quality and fault = compensation.=20
=20
A. Kuznetsov, R. Bowyer, and D. W. Clarke, = ``Estimation of=20 multiple order models in the delta domain,'' Int.J.Control,=20 vol. 72, no. 7, pp. 629-642, 1999.
=20
J. Yang and D. W. Clarke, ``The self-validating = actuator,''=20 Control Eng. Practice, vol. 7, no. 3, = pp. 249-260,=20 1999.
Correct functioning of actuators is of great importance to = industrial processes. Conventional actuators are hampered by their = inherent=20 characteristics (including non-linearities) and fault conditions, = which=20 degrade loop performance. A new generation of =91smart=92 actuators = have=20 self-diagnostic capabilities and can compensate for non-linearities = and=20 fault conditions. A generic model and interface for self-validating=20 actuators is proposed.
=20
R. Bowyer and D. W. Clarke, ``Multiple model = least-squares using=20 a lagrange multiplier approach.,'' Electronics Letters, = vol. 34,=20 no. 3, pp. 311-312, 1998.
=20
C. Chow, A. Kuznetsov, and D. W. Clarke, = ``Successive=20 one-step-ahead predictions in multiple model predictive control,''=20 Int.J.Systems Science, vol. 29, no. 9, = pp. 971-979,=20 1998.
=20
D. W. Clarke, ``Nonlinear control of the oscillation = amplitude of a=20 Coriolis mass-flow meter,'' European J.Control, vol. 4,=20 no. 3, pp. 196-207, 1998.
In a Coriolis mass-flow meter (CMFM) the process fluid (of = potentially variable density) flows through a pipe undergoing yawing = vibration to generate the required transverse Coriolis acceleration. = Reliable control of the amplitude of vibration is necessary for this = and=20 other applications employing a deliberately oscillating = mass-spring-damper=20 where the mass and damping are subject to variation. The method = proposed=20 uses positive feedback of the output velocity to cancel the internal = damping, with a gain determined by an outer sampled-data amplitude- = control=20 loop. The insertion of two non-linearities into the outer loop is = shown to=20 induce overall global linearity. Hence a straightforward=20 proportional-plus-integral control design procedure gives results = which are=20 uniformly satisfactory over desired amplitudes spanning several = decades,=20 unlike na=EFve methods which are effective for only a limited range = of=20 amplitudes. A simple desaturation strategy overcomes potential = problems=20 caused by power limits in the actuator. A practical example of = successful=20 use of the overall controller on a CMFM indicates the improvement = over an=20 existing analogue design.
=20
P. Fraher and D. W. Clarke, ``Fouling detection and = compensation=20 in clark-type dox sensors,'' IEEE Trans.Inst and Meas, = vol. 47,=20 no. 3, pp. 686-691, 1998.
Clark-type dissolved oxygen (DO) sensors are used in = process=20 control and environmental monitoring applications. A common fault = affecting=20 the accuracy of DO sensor measurements is fouling: the build-up of = surface=20 agents on the membrane of the sensor. This paper outlines a = technique to=20 detect and compensate for DO sensor fouling, and hence make these=20 instruments more robust and reliable. The technique uses an accurate = model=20 of the sensor=92s transient response to provide a value of the = oxygen transfer=20 time tau in the membrane. This new value of transfer time is = compared with=20 the initial unfouled value and the measurement is compensated by a = factor=20 related to the fouling-induced changes in tau. The measurement = uncertainty=20 associated with both normal and corrected measurements is examined.=20
=20
D. W. Clarke and C. J. Hinton, ``Adaptive control of = materials=20 testing machines,'' Automatica, vol. 33, = pp. 1119-1131,=20 June 1997.
A general-purpose servohydraulic machine used for testing=20 materials or manufactured components is a high-performance = mechatronic=20 system: wide bandwidths and fast sampling are equired for typical = tests.=20 Accurate results need good consistent control performance, but = changes of=20 stiffness of the test specimen markedly affect closed-loop natural = frequency=20 and damping. Moreover, a new specimen type often necessitates = controller=20 returning-a procedure that is unfamiliar to most materials = engineers. An=20 adaptive controller is required to maintain performance irrespective = of the=20 specimen type or of changes of its stiffness during the cycle. This = paper=20 describes a successful adaptive =91grey-box=92 controller that has = been used on=20 over 300 different testing machines. A simplified machine model, = based on=20 physical principles, has parameters dependent on the specimen = stiffness, and=20 a test procedure using an arbitrary specimen provides initial model = and=20 controller settings. Two single-parameter recursive estimators track = changing stiffnesses, and the derived values are used to retune PI=20 parameters according to a simple algorithm. This adaptive controller = maintains bandwidth even if the specimen stiffness varies during a = single=20 test cycle, unlike black-box adaptive controllers, for which lack of = persistency of excitation precludes the required parameter tracking. =
=20
M. Leahy, M. P. Henry, and D. W. Clarke, ``Sensor=20 validation in biomedical applications,'' Control Engineering=20 Practice, vol. 5, no. 2, pp. 1753-1758, 1997.
The sensor validation or SEVA project (Henry and Clarke = 1991;=20 Henry and Clarke 1993) promotes the use of intelligence in = =91smart=92 sensors=20 and the use of standard metrics to efficiently communicate = self-diagnostics=20 to the outside world. The standard metrics describe the status of = the sensor=20 including on-line uncertainty and a status flag to describe how the = current=20 validated measurement value has been derived. The end result is to = provide a=20 compact generic description of the quality of a measurement to the=20 controller, with which decisions as to how to use the measurement = can be=20 made. This paper proposes the use of SEVA principles in the = interpretation=20 of data from biomedical instrumentation, in order to aid the = decision-making=20 process, particularly in critical care. For these purposes the pulse = oximeter and polarographic oxygen tension meter will be used as = working=20 examples of typical =91intelligent sensors=92 because they make use = of a=20 microprocessor to perform self-diagnostics, as well as implementing=20 measurement algorithms.
=20
J. Yang and D. W. Clarke, ``A self-validating = thermocouple,''=20 IEEE Trans. Control Systems Technology, vol. 5, = no. 2,=20 pp. 239-253, 1997.
Traditionally the sensor has been the system component = neglected=20 by control engineers, yet it has long been recognized that accurate = and=20 reliable sensor readings are vital for good controller performance,=20 Self-validating sensors can improve reliability through = self-diagnosis and=20 by the provision of on-line metrics (measurement accuracy and its=20 trustworthiness). One benefit of validation is the potential to = sustain=20 satisfactory loop performance even in the presence of sensor faults, = This=20 paper describes a self-validating thermocouple that can discover = several=20 types of internal fault, including the practically important aspect = of=20 loss-of-contact with the process environment, and shows how the = associated=20 uncertainties can be calculated, The operation of the device is = demonstrated=20 through a series of experiments.
=20
D. W. Clarke and P. M. A. Fraher, ``Model-based = validation=20 of a dox sensor,'' Control Engineering Practice, vol. 4, = no. 9, pp. 1313-1320, 1996.
An optimal sensor validation SEVA procedure should = generate a=20 measurement uncertainty associated with any detected fault and where = possible correct for the effect of the fault on the derived = measurement data=20 so that the transmitter always provides the best estimate of the = underlying=20 value of the physical variable. A dissolved oxygen sensor, often = used for=20 environmental sensing, uses a permeable membrane which might become = fouled=20 under adverse conditions. A test procedure, based on a physical = model of the=20 diffusion processes in the sensor, provides fault detection and = correction=20 capability. The theory behind the test and its application to a = commercial=20 Dox sensor made by Foxboro =96 the 871DO =96 are described. = Practical results=20 comparing data from a fouled but compensated sensor with data from a = separate clean sensor show that the procedure is extremely = effective.=20
=20
M. P. Henry, N. Archer, M. R. A. Atia, = J. Bowles,=20 D. W. Clarke, P. M. A. Fraher, I. Page, = G. Randall,=20 and J. C.-Y. Yang, ``Programmable hardware architectures for = sensor=20 validation,'' Control Eng. Practice, vol. 4, = no. 10,=20 pp. 1339-1354, 1996.
A previous paper by M. P. Henry (1995) introduced the = technique=20 of hardware compilation as the basis for developing highly flexible=20 programmable hardware platforms for control applications such as = sensor=20 validation. This paper describes two PC-hosted architectures for = sensor=20 validation research. The first holds up to two FPGAs and supports a = daughter=20 board with application-specific circuitry. The second is based on = the=20 transputer TRAM standard, and consists of programmable hardware = modules=20 providing interfacing and low-level signal processing between the = transputer=20 and arbitrary I/O components. Three applications are described, = based upon a=20 thermocouple, a dissolved oxygen probe and a Coriolis mass flow = meter.=20
=20
A. Kuznetsov and D. W. Clarke, ``The performance of = generalised=20 predictive control with interval constraints,'' European = J.Control,=20 vol. 2, no. 4, pp. 260-277, 1996.
=20
J. Yang and D. W. Clarke, ``Control using = self-validating=20 sensors,'' Trans.Inst.MC, vol. 18, no. 1, = pp. 15-23,=20 1996.
The correct functioning of sensors is of paramount = importance to=20 the operation of industrial processes. The concept of a = self-validating SEVA=20 sensor, capable of detecting and correcting for its own faults and=20 accompanying each measurement with an estimate of its uncertainty = and=20 standard quality indicators, has been proposed. This paper aims to=20 investigate how the extra information provided by SEVA sensors can = be used=20 to enhance the performance of simple three-term (PID) controllers in = the=20 single-loop configuration. An implementation of the SEVA interface = on a=20 dissolved oxygen sensor and the device response to a common fault is = described. SEVA sensor measurement qualify indicators can be used to = decide=20 whether to continue closed-loop operation and, if so, how the = controller=20 should be retuned to counteract the effects of sensor faults: = control=20 strategies in the presence of a range of sensor malfunctions are = illustrated=20 by simulation examples.
=20
T.-W. Yoon and D. W. Clarke, ``Improved recursive estimator = with a=20 relative dead zone,'' Electronics Letters, vol. 32, = no. 9,=20 pp. 854-855, 1996.
=20
M. P. Henry, ``Sensor validation and fieldbus,'' IEE = Computing=20 & Control Engineering Journal, vol. 6, pp. 263-269, = December 1995.
Intelligent sensors are now able to generate diagnostic=20 messages, and Fieldbus can carry them, but how are they to be = integrated=20 into a control system? It is argued that a standard format must be = adopted,=20 and online uncertainty is proposed as a metric for data quality. =
=20
M. P. Henry, ``Keynote paper: Hardware compilation a new = technique=20 for rapid prototyping of digital systems applied to sensor = validation,''=20 Control Engineering Practice, vol. 3, pp. 907-924, = July=20 1995.
This paper provides tutorial introductions to = field-programmable=20 gate arrays (FPGAs) and the concept of hardware compilation-the = translation=20 of a high-level programming language directly into a hardware = design. As an=20 illustration, a simple stepper motor control program is presented. = The=20 research aims of the sensor validation programme are described, and = the=20 benefits of using hardware compilation techniques are presented. = This leads=20 on in the conclusion of the paper to a more general discussion of = the=20 interaction between research and technology, and in particular the = influence=20 of information technology upon control engineering.
=20
C. Chow, A. Kuznetsov, and D. W. Clarke, = ``Application of=20 generalised predictive control to the paper machine benchmark,'' = Control=20 Eng. Practice, vol. 3, no. 10, pp. 1483-1486, = 1995.
=20
D. W. Clarke, ``Sensor, actuator and loop validation,'' = IEEE=20 Control Systems, vol. 15, no. 4, pp. 39-45, = 1995.
Economic pressures are dispersing machine intelligence = away from=20 centralized computers toward distributed Fieldbus devices. = Simultaneously,=20 the control concept is being extended to include other operational = factors=20 such as quality assurance, process data management, just-in-time=20 maintenance, and plant safety and availability, As credible sensing = and=20 actuation are essential prerequisites for advanced control, = validation of=20 the sensor and actuator interface is necessary. As such subsystems = are=20 linked to a range of different overall plant systems, we must = consider=20 generic validation and its reporting to the =91=92next level = up.=92=92 This article=20 discusses validation features which are best embedded in local = devices.=20
=20
T.-W. Yoon and D. W. Clarke, ``Observer design in = receding-horizon=20 predictive control,'' Int.J.Control, vol. 61, = no. 1,=20 pp. 171-191, 1995.
=20
T.-W. Yoon and D. W. Clarke, ``A reformulation of = receding-horizon=20 predictive control,'' Int.J.Systems Sci, vol. 26, = no. 7,=20 pp. 1383-1400, 1995.
=20
D. W. Clarke, ``Self-tuning control,'' in The Control=20 Handbook (W.S.Levine, ed.), CRC Press, Boca Raton, FL, USA, = 1995.
=20
D. W. Clarke, E. Mosca, and R. Scattolini, = ``Robustness of=20 an adaptive predictive controller,'' IEEE = Trans.Autom.Control,=20 vol. 39, no. 5, pp. 1052-1056, 1994.
=20
P. Scokaert and D. W. Clarke, ``Stabilising properties = of=20 constrained predictive control,'' Proc.IEE, vol. 141,=20 no. 5, pp. 295-304, 1994.
=20
T.-W. Yoon and D. W. Clarke, ``Adaptive predictive control of = the=20 benchmark plant,'' Automatica, vol. 30, no. 4,=20 pp. 621-628, 1994.
=20
H. Demircioglu and D. W. Clarke, ``Generalised = predictive=20 control with end-point state weighting,'' Proc.IEE, = vol. 140,=20 no. 4, pp. 275-282, 1993.
=20
M. P. Henry and D. W. Clarke, ``The self-validating = sensor:=20 rationale, definitions and examples,'' Control Eng. Practice, = vol. 1, no. 4, pp. 585-610, 1993.
Traditionally, the industrial sensor has been viewed as a = simple=20 signal generator. The application of microprocessor technology, = digital=20 communications and fault detection techniques, coupled with = increasing=20 demands for measurement quality assurance, have rendered inadequate = such a=20 simplistic view. In this paper a new sensor model is proposed which=20 encompasses new demands and capabilities. This self-validating = sensor=20 performs self-diagnostics and generates a variety of data types, = including=20 the online uncertainty of each measurement. A demonstration system = is=20 described, based upon a Coriolis mass flow meter.
=20
A. Ordys and D. W. Clarke, ``A state-space description = for gpc=20 controllers,'' Int.J.Systems Sci, vol. 24, no. 9,=20 pp. 1727-1744, 1993.
=20
T.-W. Yoon and D. W. Clarke, ``Receding-horizon predictive = control=20 with exponential weighting,'' Int.J.Systems Sci, = vol. 24,=20 no. 9, pp. 1745-1757, 1993.
=20
H. Demircioglu and D. W. Clarke, ``Cgpc with guaranteed=20 stability properties,'' Proc.IEE, vol. 139, no. 4,=20 pp. 371-380, 1992.
=20
D. W. Clarke and R. Scattolini, ``Constrained = receding-horizon=20 predictive control,'' Proc.IEE, vol. 138, no. 4,=20 pp. 347-354, 1991.
=20
B. Robinson and D. W. Clarke, ``Robustness effects of a=20 prefilter in generalised predictive control,'' Proc.IEE,=20 vol. 138, no. 1, pp. 2-8, 1991.
=20
D. W. McMichael, ``Robust recursive lp estimation,''=20 Proc.IEE, vol. 137, no. 2, 1990.
=20
C. Mohtadi, ``Bode's integral theorem for discrete-time = systems,''=20 Proc.IEE, vol. 137, no. 2, pp. 57-66, = 1990.
=20
C. Mohtadi, ``On the role of prefiltering in parameter = estimation and=20 control,'' in Adaptive Control Strategies for Industrial Use=20 (S.L.Shah and G.Dumont, eds.), Springer-Verlag, 1990.
=20
D. W. Clarke and C. Mohtadi, ``Properties of generalized = predictive control,'' Automatica, vol. 25, no. 6,=20 pp. 859-875, 1989.
=20
K. Kim and D. W. Clarke, ``Robust adaptive = pole-placement=20 control using the pseudo-plant method,'' Int.J.Systems Sci,=20 vol. 20, no. 11, 1989.
=20
S. Yung and D. W. Clarke, ``Local sensor validation,''=20 Measurement and Control, vol. 22, no. 5, = pp. 132-141,=20 1989.
=20
D. W. Clarke, ``Self-tuning multistep optimization = controllers,'' in=20 Adaptive Control Strategies for Industrial Use (S.L.Shah and=20 G.Dumont, eds.), Springer-Verlag, 1989.
=20
D. W. Clarke, ``Application of generalized predictive control = to=20 industrial processes,'' IEEE Control Systems Magazine, = vol. 8,=20 no. 2, pp. 49-55, 1988.
=20
F. Liann and D. W. Clarke, ``Application of explicit = criterion=20 minimization in adaptive control,'' Int.J.Systems Sci, = vol. 19,=20 no. 4, pp. 551-571, 1988.
=20
T. Tsang and D. W. Clarke, ``Generalised predictive = control with=20 input constraints,'' Proc.IEE, vol. 135, no. 6,=20 pp. 451-460, 1988.

This file has been generated by bibtex2html = 1.74

------=_NextPart_000_0000_01C97CF4.34208260 Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Location: http://www.ox.ac.uk/i/reverse-small.gif R0lGODlhLQA1ANUuAIqIsgAAZv///5SSuggIa1RRjaSjxHRxo1xZk317qZuZvUVCg4WDrmxpnRcX dJybwKyryWVimUlFhVFNi2dno05KiWBdlY+NtbSzznBtoCUlfFhVkKCfwnh1pVJSl4B+q2lmm5iW vJKQt6imxnJyq0hIkWNgl1xcnX19sUFBjaqoyHp4p7Cuy0tHhwAAZgAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACH5BAEAAC4ALAAAAAAtADUAAAb/QJdw SCwSUijOyKAYKJQcVIpQrFqvQg2Dk6gEvmDwZMUBBLDoYuABWHwRk014HlgAOOf0lQC4fL0BHyJh EWAHDQ1gAAMOemoQXhMhhV8DbnR+HwVgCxApjkIeeF8LCAlgFnRfESYfXxMXbgMUjid+FolfFxJ1 CIIKwAMMEZdfIR1ygSRpHn51IptfBSIGAA0FEgsSBSB3InIWB8UBCbRXa3WkA3UDHCaqYQjA42AX JVYEBgEtGRuXJixAwFOFQEUGMZcMNCqCYtOCDwwEHhg1EJ6wLy0uLQhRRMM6MAgCZBhBKQALCCAD FLjUAcOlFhwY0Elwb0isBS3AbFARAIOA/wgKBAgQYUEAhgAMBDQAIRRDAQMCNh4k9YXDkACDPgCI MCEABF4bjLIQCsFnMqFQjaoQ4OyryAuJOnxykaFrAggKAq0AYwCDXwwsMIwAwwBDXwEs+vIKYCHv QQQhRQjJSyrRYDARhGoWkIqUT6NCnX15UKACgzYBQjjQIPpLh1MgA18Q4TdZnREYONDG4ArVoDAZ PHggtkGCFwU5072aUGECvQUTCsShF4DiFwkAUEiIMMDaAsotBADocEGctgXaEFTAiX5BBAB7wwDY UADBig4BDFjCGcFCAWcLYHBaAhE00EEDCcA3wQEifPABCP04OE4G+DlRiH5zNNCbBGwx0P9BBB3A 18cFBXQQAgAHfHBBBhkkMA4CDCxwUE4K+AFIAAfgV4d4DBxA3Ab0BeCFBf55sUGK40wzxwB+bGAC fhR6dtoBJixgQIIAVMBLAgcoAIAFFSAZhpKLYbVOBBecEsGGHVK5wErQLeBFBS1sEFIBYoLUhggi mBGCCG9e8p+UDGRAJAMIgNABCOipIEwGIIT5wTgIriIQByREQ8oDpAggzAEWLBhBBnDK6CCLJU4a xjCc5FfCAQEgkMEgA3S1gAAMFHqAdAXYWcEGv24wQQstNHCAi2FwgN5iGzmQlQW8NCBTTycmcMEA ISiQ7QghGPAAMCH8mQCy0oSA44oBmED/y0dhXObXAfAekIG8FN4nbwf4rrACbJWAmYpAF2jgwgmd LXDKATIJG0YxAYZU0QSXqRjSA0IQwCkIDLoxgqZ/CMBTgEJ58YEAU7Vr2yonDLFCVwcAkIE2LNDx gFGf0WwUJq7EyBcVQjjAaQedMYZSGBUYVZhRAEBV8hcfODYCB6lEYM4QEQiESgARuAXGbqD5NbQi lC0AgggCjcAzEQ+4sUAHExxUAAvifFFUU5rFJ+QIMr0jTQADzFWEA/pABsvWGbyDgF8jXKDAuwE0 IMIIISEwQNAZLHMOHhtR1UIB10yASAQFfAA0AoFZ/RDfckieRgoU2RHCBzcigF62bng538cKHxiu wNlYBHBZABtYMAAlC5AYgT4BSBAzHW4cgAHvaAD+zgIW+PPFBk1DcIHsCcKTgAA1gUIACRzUF4ZZ LazwwAiJ+0ikCRmsYABPoHTUX3LQNVBSKS07oa0gHqifFVIQgBTIAV+c0MiE8iDAK2ggAh5IQdIS YIGMLCwCIfBbAx24EAP4QgQPwAC2BiAF6G0QCynQUSVOeMJ1RGcTDzAhC6MHgQckaAUXOMoM60cA AnikABmghQx3iA/DmIABBXhAzIgIiu50pQIZeB4THUGA1EzgASlYyBSpyKkZBgEAOw== ------=_NextPart_000_0000_01C97CF4.34208260 Content-Type: text/css; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Location: http://seva.eng.ox.ac.uk/basic.css HTML { MIN-WIDTH: 600px } BODY { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } DIV { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } TD { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } TH { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } H2 { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } H3 { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } H4 { FONT-SIZE: small; COLOR: #333; FONT-FAMILY: verdana,sans-serif; = voice-family: inherit } BODY { FONT-WEIGHT: 700; FONT-SIZE: 10pt; MARGIN-LEFT: 10pt; COLOR: #000000; = MARGIN-RIGHT: 10pt; FONT-STYLE: normal; FONT-FAMILY: Verdana, Myriad = Web, Syntax, arial, helvetica, sans-serif; BACKGROUND-COLOR: #ffffff } #banner { PADDING-RIGHT: 0px; PADDING-LEFT: 0px; FONT-WEIGHT: bold; FONT-SIZE: = 12pt; BACKGROUND: url(images/banner_03.jpg) #ffffff no-repeat left top; = PADDING-BOTTOM: 0px; VERTICAL-ALIGN: top; WIDTH: 950px; COLOR: #ffffff; = PADDING-TOP: 0px; FONT-STYLE: normal; FONT-FAMILY: Syntax; POSITION: = relative; HEIGHT: 60px; TEXT-ALIGN: left } #banner IMG { PADDING-RIGHT: 2px; PADDING-LEFT: 2px; PADDING-BOTTOM: 2px; = BORDER-TOP-STYLE: none; PADDING-TOP: 2px; BORDER-RIGHT-STYLE: none; = BORDER-LEFT-STYLE: none; BORDER-BOTTOM-STYLE: none; align: middle } #banner SPAN { PADDING-RIGHT: 20px; PADDING-LEFT: 20px; PADDING-BOTTOM: 20px; = PADDING-TOP: 20px; TEXT-ALIGN: center } .clearhack { DISPLAY: inline } A { COLOR: #06c } A:hover { COLOR: #333 } A:active { COLOR: #000 } P { LINE-HEIGHT: 140% } H1 { FONT-WEIGHT: bold; COLOR: #333; FONT-FAMILY: trebuchet ms } H2 { FONT-WEIGHT: bold; COLOR: #333; FONT-FAMILY: trebuchet ms } H3 { FONT-WEIGHT: bold; COLOR: #333; FONT-FAMILY: trebuchet ms } H1 { FONT-SIZE: 180%; MARGIN: 0px } H1 A { COLOR: #333; TEXT-DECORATION: none } H1 A:hover { COLOR: #000; BORDER-BOTTOM: #666 1px dotted } H2 { FONT-SIZE: 140%; PADDING-BOTTOM: 2px; BORDER-BOTTOM: #ccc 1px solid } H3 { FONT-SIZE: 140%; PADDING-BOTTOM: 2px; BORDER-BOTTOM: #ccc 1px solid } P.note { PADDING-RIGHT: 4px; MARGIN-TOP: 0px; PADDING-LEFT: 4px; FONT-SIZE: 85%; = BACKGROUND: #eee; PADDING-BOTTOM: 4px; LINE-HEIGHT: 130%; PADDING-TOP: = 4px; FONT-FAMILY: tahoma } H3 { FONT-WEIGHT: bold; FONT-SIZE: 14pt; BACKGROUND: none transparent scroll = repeat 0% 0%; FONT-STYLE: normal } H5 { FONT-WEIGHT: bold; FONT-SIZE: 12pt; BACKGROUND: none transparent scroll = repeat 0% 0%; COLOR: #054da5; FONT-STYLE: normal } LI.horiz { PADDING-RIGHT: 20px; DISPLAY: inline; LIST-STYLE-TYPE: none } A.invensys_link { TEXT-ALIGN: left; TEXT-DECORATION: none; vertical_align: top } A.invensys_link:link { COLOR: #ffffff } A.invensys_link:visited { COLOR: #ffffff } A.invensys_link:active { COLOR: #ffffff } A.webmaster { FONT-SIZE: 7pt; FONT-STYLE: normal; TEXT-DECORATION: none } A.webmaster:link { COLOR: #0000a0 } A.webmaster:visited { COLOR: #0000a0 } A.webmaster:active { COLOR: #0000a0 } A.facilities_list { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: normal; FONT-FAMILY: = Syntax; FONT-VARIANT: small-caps; TEXT-DECORATION: none } A.facilities_list:link { COLOR: #0000a0 } A.facilities_list:visited { COLOR: #0000a0 } A.facilities_list:active { COLOR: #0000a0 } A.prototypes_list { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: normal; FONT-FAMILY: = Syntax; FONT-VARIANT: small-caps; TEXT-DECORATION: none } A.prototypes_list:link { COLOR: #0000a0 } A.prototypes_list:visited { COLOR: #0000a0 } A.prototypes_list:active { COLOR: #0000a0 } A.research_list { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: normal; FONT-FAMILY: = Syntax; TEXT-DECORATION: none } A.research_list:link { COLOR: #0000a0 } A.research_list:visited { COLOR: #0000a0 } A.research_list:active { COLOR: #0000a0 } A.pub_list { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: normal; FONT-FAMILY: = Syntax; FONT-VARIANT: small-caps; TEXT-DECORATION: none } A.pub_list:link { COLOR: #0000a0 } A.pub_list:visited { COLOR: #0000a0 } A.pub_list:active { COLOR: #0000a0 } A.external_link { FONT-WEIGHT: bold; FONT-SIZE: 10pt; TEXT-DECORATION: none } A.external_link:link { COLOR: #000077 } A.external_link:visited { COLOR: #000077 } A.external_link:active { COLOR: #000077 } A.people_list { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: normal; FONT-VARIANT: = small-caps; TEXT-DECORATION: none } A.people_list:link { COLOR: #0000a0 } A.people_list:visited { COLOR: #0000a0 } A.people_list:active { COLOR: #0000a0 } LI.people_list { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: normal; FONT-VARIANT: = small-caps; TEXT-DECORATION: none } SPAN.bottom_menu { FONT-SIZE: 7pt; FONT-STYLE: normal; TEXT-DECORATION: none } SPAN.keywords { FONT-WEIGHT: bold; FONT-SIZE: 10pt; FONT-STYLE: italic; = TEXT-DECORATION: none } ------=_NextPart_000_0000_01C97CF4.34208260 Content-Type: text/css; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Location: http://seva.eng.ox.ac.uk/tabs.css #main { CLEAR: both; BORDER-RIGHT: #cccccc 1px solid; BORDER-TOP: #cccccc 1px = solid; BACKGROUND: #cccccc; PADDING-BOTTOM: 0px; BORDER-LEFT: #cccccc = 1px solid; WIDTH: 950px; PADDING-TOP: 1.7em; BORDER-BOTTOM: #cccccc 1px = solid } #contents { PADDING-RIGHT: 10px; PADDING-LEFT: 10px; FONT-WEIGHT: 200; FONT-SIZE: = 10pt; MIN-HEIGHT: 100px; BACKGROUND: #fffdf3; PADDING-BOTTOM: 0px; = WIDTH: 930px; PADDING-TOP: 10px; FONT-FAMILY: verdana,sans-serif } #header { BACKGROUND: #ffffff; WIDTH: 950px; POSITION: relative; HEIGHT: 2em } #header UL#primary { PADDING-RIGHT: 0px; PADDING-LEFT: 0px; BACKGROUND: #ffffff; = PADDING-BOTTOM: 0px; MARGIN: 0px; WIDTH: 950px; COLOR: #ffffff; BOTTOM: = -1px; PADDING-TOP: 0px; POSITION: absolute } #header UL#primary LI { DISPLAY: inline; COLOR: #ffffff; LIST-STYLE-TYPE: none } #primary A.current { BORDER-RIGHT: #aaa 1px solid; PADDING-RIGHT: 0px; BORDER-TOP: #aaa 1px = solid; DISPLAY: block; PADDING-LEFT: 0px; FONT-SIZE: x-small; FLOAT: = left; PADDING-BOTTOM: 4px; MARGIN: 1px 2px 0px 0px; BORDER-LEFT: #aaa = 1px solid; WIDTH: 8em; COLOR: #000077; PADDING-TOP: 4px; BORDER-BOTTOM: = #aaa 1px solid; FONT-FAMILY: tahoma, verdana, sans-serif; TEXT-ALIGN: = center; TEXT-DECORATION: none } #primary SPAN { BORDER-RIGHT: #aaa 1px solid; PADDING-RIGHT: 0px; BORDER-TOP: #aaa 1px = solid; DISPLAY: block; PADDING-LEFT: 0px; FONT-SIZE: x-small; FLOAT: = left; PADDING-BOTTOM: 4px; MARGIN: 1px 2px 0px 0px; BORDER-LEFT: #aaa = 1px solid; WIDTH: 8em; COLOR: #000077; PADDING-TOP: 4px; BORDER-BOTTOM: = #aaa 1px solid; FONT-FAMILY: tahoma, verdana, sans-serif; TEXT-ALIGN: = center; TEXT-DECORATION: none } #header UL#primary A { PADDING-RIGHT: 0px; DISPLAY: block; PADDING-LEFT: 0px; FONT-SIZE: = x-small; FLOAT: left; PADDING-BOTTOM: 4px; MARGIN: 1px 2px 0px 0px; = WIDTH: 8em; COLOR: #000077; PADDING-TOP: 4px; FONT-FAMILY: tahoma, = verdana, sans-serif; TEXT-ALIGN: center; TEXT-DECORATION: none } #header UL#primary SPAN { PADDING-RIGHT: 0px; DISPLAY: block; PADDING-LEFT: 0px; FONT-SIZE: = x-small; FLOAT: left; PADDING-BOTTOM: 4px; MARGIN: 1px 2px 0px 0px; = WIDTH: 8em; COLOR: #000077; PADDING-TOP: 4px; FONT-FAMILY: tahoma, = verdana, sans-serif; TEXT-ALIGN: center; TEXT-DECORATION: none } #header UL#primary A.current { PADDING-RIGHT: 0px; DISPLAY: block; PADDING-LEFT: 0px; FONT-SIZE: = x-small; FLOAT: left; PADDING-BOTTOM: 4px; MARGIN: 1px 2px 0px 0px; = WIDTH: 8em; COLOR: #000077; PADDING-TOP: 4px; FONT-FAMILY: tahoma, = verdana, sans-serif; TEXT-ALIGN: center; TEXT-DECORATION: none } #header UL#primary SPAN { MARGIN-TOP: 0px; BACKGROUND: #cccccc; PADDING-BOTTOM: 6px; = BORDER-BOTTOM-STYLE: none } #header UL#primary A.current { MARGIN-TOP: 0px; BACKGROUND: #cccccc; PADDING-BOTTOM: 6px; = BORDER-BOTTOM-STYLE: none } #header UL#primary A.current:hover { MARGIN-TOP: 0px; BACKGROUND: #cccccc; PADDING-BOTTOM: 6px; = BORDER-BOTTOM-STYLE: none } #header UL#primary A { BORDER-RIGHT: #aaa 1px solid; BORDER-TOP: #aaa 1px solid; BACKGROUND: = #ffffff; BORDER-LEFT: #aaa 1px solid; BORDER-BOTTOM: #aaa 1px } #header UL#primary A:hover { MARGIN-TOP: 0px; BORDER-LEFT-COLOR: #666; BACKGROUND: #dddddd; = BORDER-BOTTOM-COLOR: #666; PADDING-BOTTOM: 5px; BORDER-TOP-COLOR: #666; = BORDER-RIGHT-COLOR: #666 } #header UL#secondary { PADDING-RIGHT: 0px; PADDING-LEFT: 0px; LEFT: 1px; PADDING-BOTTOM: 0px; = MARGIN: 0px; WIDTH: 950px; BOTTOM: -1.4em; PADDING-TOP: 0px; POSITION: = absolute } #header UL#secondary LI A { PADDING-RIGHT: 10px; DISPLAY: block; PADDING-LEFT: 10px; BACKGROUND: = none transparent scroll repeat 0% 0%; FLOAT: left; PADDING-BOTTOM: 0px; = MARGIN: 0px; WIDTH: auto; BORDER-TOP-STYLE: none; PADDING-TOP: 0px; = BORDER-RIGHT-STYLE: none; BORDER-LEFT-STYLE: none; TEXT-ALIGN: center; = BORDER-BOTTOM-STYLE: none } secondary LI A:visited { PADDING-RIGHT: 10px; DISPLAY: block; PADDING-LEFT: 10px; BACKGROUND: = none transparent scroll repeat 0% 0%; FLOAT: left; PADDING-BOTTOM: 0px; = MARGIN: 0px; WIDTH: auto; BORDER-TOP-STYLE: none; PADDING-TOP: 0px; = BORDER-RIGHT-STYLE: none; BORDER-LEFT-STYLE: none; TEXT-ALIGN: center; = BORDER-BOTTOM-STYLE: none } #header UL#secondary LI SPAN { PADDING-RIGHT: 10px; DISPLAY: block; PADDING-LEFT: 10px; BACKGROUND: = none transparent scroll repeat 0% 0%; FLOAT: left; PADDING-BOTTOM: 0px; = MARGIN: 0px; WIDTH: auto; BORDER-TOP-STYLE: none; PADDING-TOP: 0px; = BORDER-RIGHT-STYLE: none; BORDER-LEFT-STYLE: none; TEXT-ALIGN: center; = BORDER-BOTTOM-STYLE: none } #header UL#secondary LI A { FONT-WEIGHT: bold; COLOR: #000077; TEXT-DECORATION: none } #header UL#secondary LI A:hover { PADDING-RIGHT: 10px; PADDING-LEFT: 10px; FONT-WEIGHT: bold; BACKGROUND: = #cccccc; PADDING-BOTTOM: 0px; COLOR: #777777; BORDER-TOP-STYLE: none; = PADDING-TOP: 0px; FONT-STYLE: normal; BORDER-RIGHT-STYLE: none; = BORDER-LEFT-STYLE: none; BORDER-BOTTOM-STYLE: none } #header UL#secondary LI A:active { BACKGROUND: none transparent scroll repeat 0% 0%; COLOR: #fff7cd } #header UL#secondary LI:unknown A { BORDER-TOP-STYLE: none; BORDER-RIGHT-STYLE: none; BORDER-LEFT-STYLE: = none; BORDER-BOTTOM-STYLE: none } ------=_NextPart_000_0000_01C97CF4.34208260--