Lattice Boltzmann (LB) method is a relatively new approach to simulation of complex fluids and has a mesoscopic character, as opposed to conventional fluid dynamic approach based onNavier-Stokes equation. In LB fluid flow is modelled by distribution of particles moving on a regular lattice. The method has proven to be successful in modelling of complex fluid dynamical problem for which conventional macroscopic approach run into difficulties. This includes multi-phaseflows, solid-fluid suspension (collides) and flow in complex geometries, e.g. the flow of water an oil in porous media. A great advantage of the method is the inherent spatial locality of the updating rules. This property makes LB simulations ideal for parallelisation, We present a novel lattice Boltzmann algorithm for simulation of multiphase fluids involving amphiphilic molecules (surfactants). We discuss the parallelisation of the algorithm using a domain decomposition strategy and present results of simulations of oil-water phase separation in the presence of surfactant and the flow of a multicomponent fluid through porous media.

Subsea structures for the oil and gas production industries have multiplied in the past 25 to 30 years, reaching depths of over 400 metres. Diverless inspection and repair will be very important for both cost reduction of IRM (Inspection, Repair and Maintenance) for existing old platforms and also operator safety. Inspection is necessary for detection, location and sizing of defects. Repair is necessary in order to extend the remaining life of any members found to be defective. Repair by metal removal is often possible and can be achieved by electro-chemical machining. The DIRECT project has developed a prototype Electro Chemical Machining (ECM) tool to be deployed by a manipulator. The use of graphical modelling was thought to be a good approach to the design and placement of its support frame, which has to be tailored-built for each particular crack. The location of the crack in a particular node provides the necessary information for the positioning of the tool and for the calculation of the required tool path (effectively, the tool cam profile).

Individual graphical models for the different parts of the system are created using local co-ordinates (i.e.: co-ordinates defined for the particular partís local frame of reference). These models (node, tool and crack) can then be integrated together in one picture by converting their local co-ordinates into world co-ordinates using transformation matrices.

To ensure stable contact on the cylinder surface under inspection, three points of contact are required. They are found and defined both in tool and global co-ordinates. The global co-ordinates help in the positioning of the frame, while the tool co-ordinates allow the calculation of the dimensions of the frame.

Using a graphical simulation allows us to visualise the repair and check its viability in terms of reach, while providing essential information to the manipulatorís control system and to the toolís manufacturer. A similar approach can be applied to other problems where a graphic model is achievable and appropriate transformation matrices can be found.

Bandwidth consumption in distributed real-time simulation, or networked real-time simulation, is a major problem as the number of participants and the sophistication of joint simulation exercises grow in size. The initial success of SIMNET (Miller and Thorpe 1995) has prompted further growth over the past decade in the use of distributed real time simulation for the purpose of training. One of the most well known examples of distributed real time simulation is Distributed Interactive Simulation (DIS) (IEEE 1994). DIS attempts to provide high performance training environments by connecting individual simulators that are located in geographically distributed sites (STRICOM 1992). The principal element in such systems are the simulators and the connecting network (combinations of LAN & WAN).

This paper briefly reviews distributed real-time simulation and looks at bandwidth reduction techniques as a method of reducing the burden that DIS imposes on the connecting network. Although data independent reduction techniques are valid (comprising of compression and data bundling on the general flow of data), they tend not to scale well, so our work focuses on data-aware techniques instead. Data aware reduction methods use intelligent marshalling techniques that guide information to interested parties. These reduction techniques, of which there are five (simple approach, dead reckoning, network subscription, group subscription and relevance filtering) are evaluated specifically to be used in a real time simulation infrastructure.

Having looked at methods of reducing the demands placed on the network, the Generic Runtime Infrastructure for Distributed Simulation (GRIDS) (Saville and Taylor 1997,1998) is briefly presented as a research architecture for studying bandwidth problems. GRIDS uses Java abstract classes to promote distributed services called thin agents, a novel approach to implementing distributed simulation services, such as user-defined bandwidth reduction mechanisms, and to distributing the executable code across the simulation. Thin agents offer the advantages of traditional agents (Jennings et al. 1998) without the overhead imposed by mobility or continuous state, which are unnecessary in this context. We present our implementation and some predicted results from message-reduction studies using thin agents.

The aim of the simulation study described in this paper is to develop a tool to support the performance analysis of integrated electrical propulsion systems that are to be fitted in future warships. The concept of integrated full electric ship propulsion has already been adopted in several commercial ships successfully where it has been found seen to be a fuel efficient alternative to conventional mechanical propulsion systems. Other advantages which are of interest with respect to warships are the flexibility in the layout of the machinery, increased survivability through improved system redundancy, improved stealth operation and substantially reduced emissions. The 'all electric warship' programme has recently received substantial government money for its development is currently the subject of many worldwide research programmes.

The purpose of this simulation system is the study of steady state and transient conditions, stability analysis and fault conditions, parameter and component choice optimisation, and the development of control strategies and operating philosophies. The models described in this simulation system aim at an easy to use, flexible, integrated model of an electric propulsion system from the prime mover to the ship’s propeller. Such a system should allow the end-user to analyse the system as a whole, and should be capable of representing all effects of interest. Although the operation of the individual components is generally well understood the performance within an integrated electrical system has not been researched thoroughly yet.

Appropriately detailed mathematical models to represent the electrical, mechanical and thermodynamic processes of the propulsion equipment are developed and embedded within a computer simulation system. A graphical programming environment, Simulink, allows rapid development of complex dynamic simulations with a user-friendly interface. A coarse structure of the entire system is defined as a block diagram connecting the main components, which are the prime mover, generator, electrical power converter, propeller and ship dynamics. Individual component models are developed and tested separately, and compiled in a model library which is used in the integrated simulation of the entire system. New alternative component models e.g. of different complexity can be easily added to the library. The main blocks in the entire simulation system mask subsystems which use the models of the component library to hide the complexity from the end-user. System components and parameters can be easily changed to allow analysis of the effects of parameter or component changes on the whole system performance.

In the offshore industries underwater remotely operated vehicles (ROVs) are commonly used for inspection, maintenance and repair tasks of offshore structures to increase the efficiency and safety of such operations. ROV navigation is a difficult task and the operators need to be well trained and experienced. The navigation of the ROVs is performed by looking at video images and additional sensor readings, such as depth measures and acoustic sensors.

The concepts of virtual environments and graphical simulations are potentially valuable tools for future ROV operators. The aim of the virtual environment which is described in this paper is to aid the ROV operator in navigating the ROV in a known underwater environment, and to prove the usefulness of 3D acoustic cameras, which just recently have become commercially available.

The virtual environment shows the underwater environment, which consists of the a priori known rig structure and the ROV, alongside the real images from an optical and 3D acoustic camera. The 3D acoustic camera image is visualised in the virtual environment as 3D points of varying brightness according to the measured acoustic intensity. Additionally the viewing volume and the coordinate system of the 3D acoustic camera are visualised. The initial position of the ROV in the virtual envirinment does not correspond to the real position of the ROV with respect to the rig structure. Therefore the ROV has to be positioned manually to the actual position by matching the real images to the simulated optical and acoustic camera views. Once a match has been found the ROV position and pose can be extracted and the operator can choose any viewpoint by rotating, translating and zooming into the scene to get a clear virtual picture of the ROV position and pose with respect to the known environment. This allows him to easily evaluate the situation and plan his actions accordingly.

For an automatic recognition and positioning system where no or little manual intervention of the operator is required, simulated ideal optical and acoustic camera images derived from the virtual environment might be useful feedback information. The provision of an online navigation aid is the main aim discussed here, but other applications which could make use of the virtual environment include offline training of prospective ROV operators.

The capabilities of the World-Wide Web include the possibility of running simulations that are accessible by remote clients from anywhere in the world. A three-tier client-server simulation architecture (Browser:Web Server:Simulation Server) opens up the possibility of software rental and makes massive processing available to resource-constrained clients.

A major concern for a simulation service provider is to provide an environment that will allow registered users to perform authorised tasks while maintaining security and protection from unauthorised use.

This environment has been created through the use of server-side (servlet) Java programming to provide a secure, scalable gateway interface to an application. This paper discusses the use of a token-passing protocol to enable the secure provision of simulations and teaching material via the Web.

Increasing emphasis and competition for new pharmaceutical chemicals requires consistent, high quality products with shorter lead times to market. External pressures and the continual need for trial chemicals mean that effective process development is critical, with a key element for success being the ability to make the right decisions early on despite uncertain and incomplete information. There is often considerable uncertainty in the inherent physico-chemical mechanisms of pharmaceutical processes and therefore in the ensuing models and parameters. Since model predictions often form the basis for design decisions, it would be useful to consider uncertainty in the parameters characterising the model and evaluate the robustness of a process to this uncertainty using a framework of fundamental models within a suitable simulation environment.

A general methodology is proposed, incorporating a multiscenario stochastic flexibility analysis in which process feasibility with respect to uncertainty in the quality of the available data is assessed. The uncertainty of the model system, is quantified via a stochastic flexibility index, which represents the probability of feasible operation over the weighted scenarios of possible parameter realisations, subject to the System and product quality constraints. The uncertainty in the model parameters are described by normal probability density functions. The potential of this flexibility analysis approach is investigated using a simple equation system with low order uncertainty, modelled within the MATLAB environment. It is discussed how this potential can be extended to compliment the proposed methodology and form the basis for a computer aided decision support tool for the development of robust pharmaceutical processes.

As data storage costs come down and data retrieval times improve, the prospect of synthesising simulations from pre-computed and stored data is becoming more attractive. An already accepted case of such a strategy in electronic music replay is the production of complex sounds from wavetables. The question addressed in this paper is whether a similar strategy would bring advantages in engineering analysis and design, in situations where simulation of time-dependent phenomena plays a part. In particular, we consider the numerical construction of propagating stress pulses in elastic rods. In the time domain, the appropriate tools are either the use of a time-stepping algorithm on the matrix equations of motion, or the synthesis of the response via the Impulse Response Function and the convolution integral. Using the latter, the Impulse Responses of an extensive structure to inputs at various locations can be pre-computed and stored. The structural properties may be varied, and the corresponding sets of data may be organised as a data bank. The issue to examine is whether the simulation of a system to arbitrary inputs can be more efficiently generated by this technique than by the ab-initio calculation of the response by time-stepping for each different choice of structural parameters.

The same issue arises in frequency-domain calculations. In this case, the final result is obtained by inverse-Fourier transformation of a function which has to be built-up frequency by frequency by numerically evaluating some existing algebraic expressions. The strategy proposed here is to pre-compute and store these intermediate results for all frequencies, and for a range of structural parameters (including the distance between input disturbance and wave sampling point). From the point of view of the end-user of the simulation the tradeoff is between the amount of computation saved and the time spent in retrieving the stored intermediate results.

Adopting the frequency-domain method, an example is presented, involving the bouncing of a stress pulse from a boundary in a straight elastic rod. Conclusions are drawn from this preliminary study.

During the last decade we have witnessed an enormous growth in Internet Traffic. Different technologies have been developed since in order to cope with the capacity and QoS demands this imposes. Moreover as the foreseen convergence of the telecommunications and data communications industries is becoming real, the need for a suitable technology capable of providing quality of service assurances at a lower cost, high bandwidth and goodput even under congestion situations is becoming more evident. Developments in optical networking and the evolution of optic transmission technology from lower rate multimode fibre links to singlemode amplified fibre links, currently allow several 10 Gbits/s channels to be carried over a single fibre. Several approaches have been taken in order to take advantage of the different technologies available nowadays, this however has brought along several issues to be resolved.

When connecting two remote routers through a high speed backbone optical network, IP packets directed to the same remote IP network are gathered in optical packets carried by the network on an appropriate virtual connection to the destination IP router. Likewise all TCP connections between the same routers exclusively follow the same virtual path with link utilisation related to the aggregate activity of the TCP connections themselves. Thus a virtual path would exist between each pair of routers in the optical network, with every virtual path carrying flows of aggregate IP traffic reaching several gigabits per second.

A network failure, such as fibre cuts, may occur at any moment and the possibility should be always considered. The failure of a single fibre loaded with, i.e. 32 channels each one carrying an STM-16 signal will result in a loss of service capacity equivalent to more than one million telephone connections lost. Optical recovery allows the (re)construction of arbitrary virtual topologies resilient to fibre failures. Nevertheless the longer the time elapsed between the point the failure occurs and the

point the service is fully restablished, the greater the degradation in quality of service and service availability. This is mainly due to duplication of services including restoration at different levels, i.e. TCP congestion avoidance mechanism, that could lead to topology oscilations and race conditions degrading further the network performance.

In this article we present a performance study of a high speed backbone optical network based on a simulation model of an optical ring with up to 32 channels each one carrying an STM-4 signal, with a 60% load and four different types of traffic -- voice over IP, HTTP, FTP and email. The goal of this performance study is to measure the service degradation based on a metric defined within the study in order to find any correlation between the time elapsed during the outage period and the quality of service and service availability before, during and after the failure point.

Intro: Medical diagnosis may be considered as a study of the behaviour of n independent probabilities. Understanding changes in these probabilities is critical in improving diagnosis. However, the concepts in current use are not intuitively understood and are prone to mis-application and understanding. The use of a computer generated graphical representation of these probabilities could improve understanding, but to be of value in non-trivial cases requires a firm mathematical basis.

Concept: The incorporation of more advanced mathematical techniques for understanding the changes in these probabilities, such as discriminant analyses and elementary group theory, can be applied to this model, thus enhancing its power while retaining a user-friendly "front-end". The current application of the system is to present data on risk reduction in stroke and heart disease.

Discussion: The similarity between this system and other geometries remains unclear. Describing the system as a sub-space of an existing geometry could allow more use of existing mathematical knowledge and concepts. Conversely, the treatment of multiple probabilites as a geometry raises some interesting questions about concepts in probability. The application of this model in presenting data can lead to better presentation and understanding of the data.

In a previous paper [Al-Dabass et. al, 1999] the authors reported on the development of 3 parameter estimation algorithms using one or more higher time derivatives. In this paper the sensitivity of these algorithms to noise is investigated. The algorithms combine successive 1st order filters of specified cut-off frequencies, to provide smoothing and higher order derivative estimation, with non-linear static parameter estimators. Three hybrid methods for parameter estimation were proposed and tested for 3 categories of parameter sets: constants, variables with 1st order dynamics and variables with 2nd order dynamics. These three methods were developed using the following approaches: 1) By using three sets of estimated 1st and 2nd time derivatives of the measured output of the system; 2) By using two sets of estimated 1st, 2nd and 3rd time derivatives of the measured output of the system; and 3) By using a single set of estimated 1st, 2nd, 3rd and 4th time derivatives of the measured output of the system.

The effect of measurement noise on the estimation accuracy is investigated when the incoming trajectories are corrupted with random noise. Noise is derived from a random number generator with zero mean and added to the simulated system output. Analysis of the simulation results show varying abilities of the algorithms to cope with the noise perturbations. In some instances high prediction robustness were achieved, other simulations showed high sensitivity to noise.

D. Al-Dabass, A. Zreiba, D. Evans and K Sivayoganathan, "Simulation of Three Parameter Estimation Algorithms for Pattern Recognition Architecture", UKSIM'99, Conference Proceedings of the UK Simulation Society, St Catharine's College, Cambridge, 7-9 April 1999, pp170-176, ISBN 0-905488-38-5, available online, see http://ducati.doc.ntu.ac.uk/uksim/papers/moller/dad.doc

Machine vision is a useful robotic sensor since it mimics the human sense of vision and allows for non-contact measurement of the environment. A 3-D object gives rise to an infinite variety of 2D images or views, because of the infinite number of possible poses relative to the viewer. In order for a vision system to be effective in assisting a robot to approach an object autonomously, two things must be knownówhat the object is and where it is located, i.e. the object has to be recognised and its coordinates must be known.

The system developed utilises two cameras for obtaining 3-D information about the object. A hierarchical system has been developed in software for object recognition. Training of each object is done by presenting characteristic views of each polyhedral object. Ideally, the vision system on the robot arm should be moved around the object to obtain the characteristic views. In the simulations however, the objects are rotated and displaced to mimic robot movement. Each image of an object is processed and features such as corners and edges are extracted. The relationships between the features are determined to identify the types of surfaces. The relationships between these surfaces is then encoded and input into the artificial neural network. Incremental learning is done using Fuzzy ARTMAP for all objects. When a single novel image of an object is presented, the correct object can be recognised. Since stereo vision is used, the location of the object with respect to the cameras is also determined.

The vision system can now be implemented on a robot arm in the ëeye-in-handí configuration. The robot arm can be moved precisely to obtain the two images or a miniature stereo setup can be mounted close to the gripper. This system can thus be used to identify, locate and approach mechanical objects autonomously.

Robot programming can be difficult for complex operations. An alternative to conventional programming is the creation of self-adaptive robots based on intelligent techniques such as Artificial Neural Networks (ANN). ANN based control algorithms can benefit from early simulation by preventing unexpected robot behaviour during the testing stage.

The research presented in this paper shows how robots can operate in poorly structured environments with a minimum of instructions. The approach followed was based on the clustering of contact force patterns by a Robotic Assembly Controller (RAC) using the Adaptive Resonance Theory (ART). Initial results showed the effectiveness of the technique which led to successfully testing it during real robot operations.

First, a brief description of early simulations during mating operations of circular and rectangular geometries is given followed by the description of the testbed system. Information representing temporal force patterns during manual operations of assembly is encoded and then decomposed into spatial patterns and input into the RAC simulator. The simulator was trained with a different number of patterns and the learning time recorded. The simulation showed the effectiveness of the approach by quickly classifying all the contact states occurring during assembly operations. This demonstrated the feasibility of the technique.

From the experience gained during simulations, the RAC was enhanced by using the FuzzyARTMAP algorithm and implemented into the robotic system. Contact force patterns are now feed into the network as they occur during assemblies, which results in actual arm motions for the robot towards the assembly goal.

Information Gathering is a fundamental operation to monitoring and controlling distributed systems and it is therefore important to implement it efficiently. In particular, large systems rely upon shared large-scale communications networks, such as the Internet, which typically offer limited bandwidth and response time. Unfortunately most of the current information gathering systems follow a model based on centralised polling which, despite its popularity, does not allow for efficient use of network capacity and cannot provide timely responses to very dynamic systems.

The question addressed herein is whether alternative techniques developed in the Software Agent community can be employed effectively for information retrieval and filtering. Agents are one of the most important and exiting areas of research and development in computer science today. They are currently being applied in domains as diverse as computer games, e-commerce, autonomous vehicles, and industrial process control. However, despite their potential benefits, agents are not presently widely used in areas such as Network and Systems Management.

Some of the most common arguments generally brought against the application of agents to those areas include the lack of well-known design methodologies and standardisation, and the reduced level of security and safety of agents. Another important reason is the lack of quantitative data on the evaluation of agent systems. For instance, Mobile Agents absorb processing power and memory from their hosting entities and, during migration, they incur traffic in the network. Thus, the possibility of significant consumption of resources which could otherwise be used more efficiently needs to be quantified.

In this paper we present a methodology based on computer simulation for the assessment of a network monitoring system devised with Mobile Agent capabilities. In this system a monitoring task is delegated to different autonomous agents. Each agent is capable of computing its initial location within the network and of establishing which portion of the network to monitor. Agents can then gather, aggregate and filter raw data from the network and report them back to a central manager.

The assessment of the performance of the agent-based monitoring system and its comparison with a traditional system based on centralised polling is carried out by simulations. We have adopted the NS-2 simulator from U.C. Berkeley/LBNL to simulate IP network and protocol behaviour, and we have extended it with Mobile Agent capabilities, including two different agent deployment algorithms, agent cloning, and agent migration. Realistic network topologies, composed of routers, links, and hosts have been generated using the GT-ITM topology generator. Finally, we have adopted both traffic and response time metrics including the following: 1) traffic incurred around the management station; 2) overall traffic incurred in the network; 3) response time; 4) time to bring the system to steady state. Our initial results show that Mobile Agent solutions typically offer improved performance, often leading to a 90% reduction in steady-state traffic and response time. However, our analysis also shows that dramatic improvements in performance usually correspond to relatively long agent deployment periods, during which the system cannot operate. Therefore, the use of agents seem to represent a competitive alternative to traditional centralised monitoring systems whenever the agent deployment time is negligible compared to the duration of the monitoring task.

The insertion of a peg into a hole represents a typical and frequent assembly task in manufacturing industry. Automation has represented the continuous goal of many research activities during past decades, which have aimed to develop new techniques capable of coping with the uncertainties intrinsic to such processes. This has resulted in the introduction of new strategies based on the use of complex sensors, acting towards the creation of a self-adapting assembly system. Amongst these techniques, force control has proved its versatility in many peg in hole applications.

This paper illustrates how information from a force and torque wrist sensor during the one point contact stage of a peg in hole insertion, can represent the foundation of novel insertion strategies based on the localisation of the contact. A six axis F/T sensor was used to collect information during a simulated peg-hole contact. The custom based test rig was assembled on the support of a vertical milling machine, chosen for its capability of simulating contact states between peg and hole, providing accurate linear and angular offsets between the mating parts. These were manufactured from aluminium and consisted of a female component featuring chamfered (at 45ƒ) and chamferless hole, and a set of corresponding pegs. The cross sections considered in the experiments were circular and square. A three dimensional representation of the data acquired was obtained and used to show the consistency of such source of information. The same values, after being pre-processed, were used to form the input of a FuzzyART algorithm, a Neural Network based on Adaptive Resonance Theory (ART). This was used to show how the classification of the patterns may reflect the actual location of the contact on the surface of the hole.

The important conclusion drawn from the experience was that the contact information during the initial stages of the insertion can be used effectively to provide Neural Network based robot controllers with the required knowledge of the insertionís environment. This may contribute in reducing the amount of the low level information that the programmer needs to provide, hence increasing the degree of flexibility of robotic assembly systems.

Innovative housing association, Black Country Housing & Community Services Group, chose to adopt dynamic thermal simulation techniques as a tool to improve house design. Simulation results inform the consultation process, especially with prospective tenants. Design decisions now reflect tenants' reasonable expectations of affordable and comfortable housing. Three software programmes have been evaluated: IES 4D; Apache and ESDLís TAS Lite. Simulation results are compared with UK Building Regulations Standard Assessment Procedure for energy ratings that is based on the BREDEM model. Further validation is obtained through energy consumption records (meter readings), subjective assessment by tenant interview and a programme of energy use, temperature and weather monitoring.

Having experimented with the use of dynamic thermal simulations tools, innovative housing association, Back Country Housing & Community Services Group has identified the potential for a total immersion virtual simulation of buildings as a consultation tool. The Group sees a confluence of user consultation programmes with the ultimate, Disney World-like theme-park ride. Such a prototyping tool is envisaged to enable the prospective tenant to experience design proposals with all the senses. Using treadmill and virtual-reality ìsuitsî and headsets, the layout of a house can be literally walked through. The virtual visitor can bump into the furniture, smells the cooking and feels warmth. Damp or cold designs will be apparent and can be avoided by design rather than remedial action after failed design.

Aquarius is a computer program that assists engineers in the hydraulic design of linear pipelines. It replaces time consuming manual engineering calculations with a low cost, quick and easy to use, computerised solution. The pipeline project is described by its input data: its pipes, reservoirs, pumps and long section details. Aquarius validates this information for engineering consistency before executing hydraulic calculations. Upon request it produces system curves, pump curves, flow and head loss calculations and a hydraulic grade line. Results are output on screen in a graphical format. The user may revise the design of his input data in order to tailor the results (pressures and flows) to his precise requirements.

Aquarius is written primarily for use in the water industry but could be adapted for use in the design of pipelines conveying other fluids. It is written in JAVA 2.1 and uses object-oriented design.

The following features are highlighted as being of interest to designers of other cross-disciplinary simulation programs.

(i) The use of object-oriented design allows the reuse of functions and classes. An example is Aquarius's graph-drawing class. This has been written specifically to be of use to any program requiring two-dimensional graphs.

(ii) Classes that provide the graphical user interface (GUI) are separated from those that provide the engineering calculations. Elements of the GUI or calculation engine may therefore be extracted, updated or replaced independently of each other to suit a specific application.

(iii) Input data is validated in two stages: firstly for format, and then by a stand-alone validation class for engineering sense. This allows the calculation engine to be occupied only in performing calculations. Any inconsistencies in data input are identified before the calculation stage is reached.

(iv) The program uses a clear and simple graphical user interface for input and output.

(v) Extensive help is provided to the user by the use of tooltips and warning messages.

(vi) Stand-alone data libraries assist and advise the user in the selection of input data. Aquarius's Pumps' Library and Losses' Library demonstrate this. Such guidance would otherwise be sought from textbooks or experts, but may now be provided by computer programs themselves.

(vii) A choice of design or calculation methods is offered. Aquarius is able to perform pipe friction loss calculations using any of three popular formulae: Colebrook-White, Hazen-Williams and Manning's. The user may therefore select the method which is most relevant or with which he feels most comfortable.

(viii) The use of JAVA allows the compact uncompiled code of the program to be downloaded from a web page for compilation and use on any common platform. The program may therefore be distributed quickly and easily to a global market. Professional engineers, teachers, students and voluntary workers overseas are targeted as likely users of Aquarius

Field Programmable Gate Arrays (FPGA’s) provide an exciting new resource that can be used to implement simulations in hardware. Up until recently the only mechanisms that have been available to program these devices are the various schematic capture tools and VHDL compilers provided by the device manufacturers. Both of these methods involve the creation of a design by the user without reference to the internal structure of the device, the mapping being determined automatically by the software tools. There are many applications in simulation where the structure of the problem is a good fit to the internal device structure. Whilst one would hope that the place and route tools would be able to "recognise" this fit and translate the design accordingly, there is no guarantee that this will happen and no direct way for the designer to influence the outcome. One could of course write to the device "bitstream" (the file that determines the chip configuration) directly but the chip manufacturers have been unwilling to release the necessary specifications. One can only speculate as to why this should be but one possible reason lies in the significant differences in bitstream structure between chips with similar functionality which would make the maintenance of bitstream level interface applications

Recently Stephen Guccione and Delon Levi of Xilinx have developed JBits, a suite of Java classes that allows configuration of Xilinx 4000 series devices using the datasheet model. JBits is device independent within the 4000 family.

As part of a long term program of development of FPGA software technology centred around the "Graphite" language defined by R.J.Cant and C.S.Langensiepen, the present authors have developed "JBits UI" a graphically based user interface to the JBits classes (and thence to the Xilinx devices) that forms the subject of the present paper.

Computer-based interactive learning environments can help students to differentiate between their intuitive views on natural phenomena and the formalisms of Newtonian physics. The paper describes empirical investigations of a specific type of interactive learning environments, computer-based simulations. In many cases computer simulations deal with a simplified and idealised version of the natural phenomenon. Presenting the user with a simplification of reality is seen as one of the advantages of simulations, since too complex and too realistic simulations may sometimes be overwhelming for learners and may not permit the identification of the underlying model. Yet implications arise about the degree to which students either expect or perceive simulations to be real and how these expectations and perceptions affect their interaction with the simulation.

In this research, preliminary empirical studies of users interacting with a number of simulations highlighted some general issues. Two issues emerged as most interesting for describing aspects of simulations in physics: the underlying model, of which the students can catch a glimpse, either indirectly by observing the objects of the simulation in motion, or directly by taking into account the expression of variables into spreadsheets, digital counters and graphs, and how this underlying model is represented in the interface. The interface, as a visual representation of the physical reality, constitutes the medium between this physical reality and the user. In collaborative interactions of this type, the learners have to negotiate their understanding between them and with the computer. Evaluation of a number of simulations, two case studies and interviews with simulation designers and educators suggested these components.

Next, theories which described how a simulation imitated reality were examined. Reality for the purposes of this research is considered to be a construct comprising the visual fidelity (fidelity) and the complexity of the underlying physical model (complexity) of the simulation. It seemed that an appropriate classification of learning environments (for the purposes of this research) would be in terms of fidelity and complexity.

On the basis of this rationale an overall statement of a research hypothesis was proposed:

Altering the fidelity level of a simulation and the relation between this fidelity and the complexity of the underlying model affects students' conceptual learning and contributes to the students' perception of reality.

The full consequences and implications of this general statement were outside the limits of this research, during the period of study for a PhD. Consequently this overall statement was broken down into three specific statements. The particular statements referred to high fidelity simulations, to low complexity ones and to the use of multiple representations in the interface’s design.

The next stage involved learners in two empirical studies testing hypotheses to explore this relationship between fidelity and complexity using a number of simulations. The simulations were different representations of physical reality, dealing with the same phenomenon with different degrees of fidelity and complexity. Newtonian mechanics was chosen as the content area to explore because it is an important area of physics in which many students have difficulty visualising aspects of phenomena and predicting outcomes. Also the use of mechanics (e.g. visualising motion) takes advantage of the computer’s graphics and interactive capabilities. The learning environments involved simulations which displayed collision of masses on a screen.

Analyses were carried out on videotapes and questionnaires of students interacting collaboratively with the simulations (40 hours of computer based activity). The empirical approaches to these studies, reports on work done, including the emerging data in multiple forms (questionnaires, video and audio tapes of the students interaction) and its analysis are presented in this thesis. The work reported looks at students’ interaction with the simulations (pre to post test learning gain and issues concerning pre and post testing), their comments on the interface and the model underlying the simulation.

The empirical studies indicated that collaborative schemes or an instructional cycle were the route to follow for optimisation of the interaction. Also, the use of simulations facilitated learning and the design of computer based environments can influence the users' perception of the physical phenomena they represent. The central hypothesis also impinges upon the design process of interactive computer learning environments. The production of successful simulations is intrinsically connected with the use of carefully designed software. The research was planned to investigate what design issues are essential to the creation of successful simulations, thus exploring the implications for design and discussing the development of design criteria for how simulations might be built.

The outcomes of this research support the view that well designed computer-based simulations can promote learning and that design issues are essential to the creation of successful simulations. The findings claim that:

a) enhanced fidelity of an instructional simulation has positive effects on the learner outcome,

b) interfaces which use multiple representations offer valuable information which facilitates problem solving strategies and

c) low complexity simulations are better suited to novice learners.

These outcomes are presented as implications for simulation design and the use and development of a syntax in simulation design is also discussed (design criteria for how systems might be built).

This paper is based on the outcomes of doctoral research carried out in the Institute of Educational Technology at the Open University.

Publications related to this research

Hatzipanagos, S. (1999). Simulations for physics learners: reality vs. abstraction. Journal of Interactive Media in Education (in press).

Hatzipanagos, S., Scanlon, E., & Li, Y. (1997). Typologies of interface design which provide effective support for learning physics. ALT-C 97, Wolverhampton.

Whitelegg, E., Scanlon, E., & Hatzipanagos, S. (1997). Multimedia Motion - Motivating Learners. ALT-J, 5(1), 65-69.

Education (ICCE), Singapore, Conference Proceedings (nominated as "Best paper" by members of the programme committee).