Three duplex stainless steels (Avesta 2205, Sandvik SAF2507 and Zeron 100) were successfully welded automatically at a range of pressures from 1 to 32bar. The gas tungsten arc (GTA) welding process was chosen as it allows a high degree of control to be exercised during welding. Initial autogenous bead on plate welds established the effects of pressure on the welding process and allowed the process parameters to be determined for subsequent experiments. Analysis of the effects of pressure on the weld thermal cycle showed that at higher pressures the precipitation of phases deleterious to the weld quality was less likely than at ambient pressure. It was also found that the arc melting efficiency increased as the pressure increased, which was taken into account when the process parameters for the joints were selected. A V-butt design with a 'land' on each side was chosen for the joints to counteract any tendency for the welding arc to wander at higher welding pressures. The root welds were performed using pulsed current welding techniques to overcome the difficulties in achieving consistent penetration that were encountered when welding at lower pressures. It was found that by employing standard welding consumables commonly used for welding duplex steels at ambient pressure satisfactory austenite-ferrite phase balances could be achieved in the weld metal at all pressures. Metallographic examination of the welds showed that the joints did not have any microstructural complications that were related to pressure and mechanical testing revealed that, in terms of impact toughness, the weld metal and heat affected zone (HAZ) performed as well as, if not better than, the parent plate material. This work shows that welding of duplex stainless steels using the hyperbaric welding method is a viable option for subsea operations up to a depth of at least 320m, automated hyperbaric welding being advantageous at depths greater than 40m.

Machining processes, such as milling, drilling, turning and grinding, concern the removal of material from a workpiece using a cutting tool. These processes are sensitive to parameters such as cutting tool properties, workpiece materials, coolant application, machine selection, fixturing and cutting parameters. The focus of the work in this thesis is to devise a method to monitor the changing conditions of a machining process over time in order to detect faulty machining conditions and diagnose fault types and causes. A key aim of this thesis is to develop a monitoring regime that has minimal cost of implementation and upkeep in a production environment, therefore an unsupervised monitoring system which applies non-intrusive sensing hardware is proposed.

In this thesis a grasp metric based on stability and kinematic feasibility is introduced. The preshaping paradigm is extended to include consideration of the trajectories that the digits take during closure from preshape to final grasp. The resulting <I>grasp strategy </I>is dependent upon task requirements and is designed for a set of "ideal" object-hand configurations. The grasp strategy couples the degrees of freedom of the dextrous hand in an anthropomorphic manner; the resulting reduction in degrees of freedom makes the grasp planning less complex. Grasp strategies are fitted to real objects by optimisation of the grasp metric; this corresponds to fitting the real object-hand configurations as close to the ideal as possible. First, the <I>preshape aperture, </I>which defines the positions of the fingertips in the preshape, is found by optimisation of an approximation to the grasp metric (which makes simplifying assumptions about the digit trajectories and hand kinematics). Second, the full preshape kinematics and digit closure trajectories are calculated to optimise the full grasp metric. Grasps are planned on object descriptions built from laser striper range data from two viewpoints. A surface description of the object is used to prune the space of possible contact sites and to allow the accurate estimation of surface normals, which is required by the grasp metric to estimate the amount of friction required. A voxel description, built by ray-casting, is used to check for collisions between the object and the robot hand using an approximation to the Euclidean distance transform. Results are shown in simulation for a 3-digit hand model, designed to be like a simplified human hand in terms of its size and functionality. There are clear extensions of the method to any dextrous hand with a single thumb opposing multiple fingers and several different hand models that could be used are described. Grasps are planned on a wide variety of curved and polyhedral objects.

Automated lay-up of prepreg is slowly replacing manual layup during the production of composite parts for aerospace, automotive and renewable energy applications. This is driven by a need for higher manufacturing rates from aerospace manufacturers and a need for high quality manufacture for wind and tidal power blades. The main methods for automated layup are Automated Tape Laying (ATL) and Automated Fibre Placement (AFP). However, despite its industrial importance the amount of research into both processes is limited, in particular with respect to productivity and process reliability. The study outlined in this presentation aims to investigate feasible processing conditions for automated layup that may improve laminate quality and productivity both analytically and experimentally whilst curing out-of-autoclave. Prepreg properties relevant to automated processing are identified. Lastly, requirements for an optimised layup system are proposed. A detailed study of the resin content of automation grade prepreg shows a significant variability of resin content that may affect tack and lead to unsuccessful layup. High product quality is governed by the amount of voidage entrapped in the uncured plies. A study of the surface roughness of uncured prepreg shows the potential for significant entrapment of air during layup. Since debulking is commonly omitted for continuous processing a large amount of voidage is entrapped in the uncured laminate and may have detrimental impact on mechanical performance. A model was developed that takes into account the prepreg interface and forming behaviour during automated processing to predict the amount of entrapped air and allow optimisation of processing conditions. A two-stage layup model was implemented in Abaqus/Implicit to study the process in detail. First, the interaction between ply and layup system was studied in detail, the results were then used to predict the forming behaviour of uncured prepreg on a microscopic scale. To understand the effect of processing conditions in more detail a laboratory Iayup system was built that enables layup at conditions currently not achievable with industrial equipment. The interaction between layup temperature and pressure was studied experimentally to validate previously developed models

This thesis describes an approach to robotic assembly planning based on a logic of <i>plan specification</i> formulae. These formulae consist of a term representing the plan, a description of the states before and after the plan is performed (using a simple but structured world model), and a set of literals which can be evaluated as Prolog goals to test preconditions and generate terms to appear in the postconditions. The specifications for the atomic actions are given as axioms and planning takes place by attempting to prove the specification for the desired plan, decomposing it by the reverse application of inference rules that state how the specification for a plan can be deduced from those of its subplans. These are derived from the definitions of the temporal operators that are used to construct plans. The initial plan may be partially instantiated due to prior constraints on the form of the plan, and the remaining plan variables and atomic action parameters will be instantiated during the course of planning. The planning strategy is expressed using goal-decomposing 'tactics'. The representation of world states is defined using equational logic and (equational) unification is used to 'match' these <i>state specifications</i>. The form of this logic is designed to meet the requirements of <i>behaviour-based</i> systems such as Edinburgh's SOMASS system, which is described in this thesis, and to allow various temporal languages to be used to represent plans, in order to investigate their uses and to develop appropriate planning strategies.

This thesis reports the results of fifteen months of participant observation study of a major technological research and development project. The project, in the area of advanced factory automation, was part of the British government's Alvey Programme. The findings bear upon two main bodies of theoretical literature. The first is Marxist literature on technology, the state and the labour process. Participant observation study of this project reveals technological change to be a much more chaotic process than this literature assumes. The process, for example, is not guided by clear capitalist interests. The other body of literature is the 'actor-network' approach of Callon, Latour and Law. In common with them, it is found that technological change is not merely a technical process - it is 'heterogeneous engineering' of both 'technical' and the 'social' simultaneously. However, the actor network theorists overstate the possibilities for this 'heterogeneous engineering'. It is neither as thoroughgoing or as successful as these writers might be read as asserting. A further conclusion is that the significance of gender for participant observation studies of science and technology has been underestimated. In particular, the gender of the researcher appears to have an important bearing on the research process.

A novel application of entropy to support Systems Engineering is presented with the specific objective of providing unambiguous and objective decision support to the engineer from the earliest stages of system development. Three facets of entropy are identified: How information passes around the system, How elements are oriented within the system, How the structured of the system is realized. From the outset, the Unified Modelling Language (UML) has been used as the language of choice for capturing system description and the work described in this thesis has made a significant contribution to its use within Thales for Systems Engineering. At the beginning of the project this was a novel application but the use of UML is now commonplace across the SE community. In order to capture the measures of entropy within the UML, a new generalised framework for interface definition is proposed covering six aspects: Information Protocol Management Security Carrier Physical, The work explores a range of techniques that are in common use, and through discussion of two specific, real-world examples conclude that minimising the entropy is entirely consistent with improvements suggested by other means; A wireless house alarm will reduce entropy but only for the larger house Clustering at the lower levels of networks reduces their entropy This new, unambiguous and objective measure of entropy provides engineers with a decision aid that can be applied early in the development lifecycle. The UML is used to describe systems and lends itself to incorporating this entropy measure. However, three simple rules result from the thesis that can be applied without formal calculation: 1 Minimise the number interfaces between entities and their bandwidth. 2 Either ensure all interfaces of an entity the same or ensure each is unique. 3 Ensure an entity has no more than one dependency on other system entities.

No description available.

No description available.

In production planning for strong seasonal demand products, it is uneconomical to configure the supply chain for throughputs equivalent to the demand peaks. Instead, a holistic approach to supply chain optimisation is adopted where forward demand forecasts drive the production planning process. In this thesis, the medium-term supply chain planning components of forecasting, production planning and evaluation are addressed through studies on a paints production facility. With a large number of specialised products, family-level forecasting is adopted for its simplicity and practicality in applying forecast techniques, coupled with its benefits on the inception of new products into markets. A time-series component was incorporated into traditional clustering techniques for segmenting products into families. The dominant cluster profiles identified are attributed as the seasonal component for the subsequent generation of demand profiles. In multi-purpose batch plants, production planning involves the twin decisions of batch sizing and lot sizing, often performed in series. This campaign is optimised through augmenting the batch sizing operation within a lot-sizing model. In the Mixed Integer Linear Programming model developed here, the degrees of freedom are the monthly batch sizes of each product, integer number of batches of each product produced each month, amount of monthly overtime working and outsourcing required as well as the time-varying inventory positions across the chain. Values for these are selected to balance the trade-offs in batch costs and inventory costs as well as the overtime and outsourcing costs. The final section sees the development of stochastic, dynamic supply chain models to predict the effect of different inventory policies, taking into account forecast accuracy, as derived from clustering. Using Monte Carlo based simulations, the various supply and production decisions are assessed against process manufacturing performance indicators. These planning components are then reconfigured to derive an optimal paints supply chain.