The influence of stacks on flow patterns and stratification associated with natural ventilation

We investigate the steady state natural ventilation of an enclosed space in which vent A, located at height hA above the floor, is connected to a vertical stack with a termination at height H, while the second vent, B, at height hB above the floor, connects directly to the exterior. We first examine the flow regimes which develop with a distributed source of heating at the base of the space. If hBhB>hA, then two different flow regimes may develop. Either (i) there is inflow through vent B and outflow through vent A, or (ii) the flow reverses, with inflow down the stack into vent A and outflow through vent B. With inflow through vent A, the internal temperature and ventilation rate depend on the relative height of the two vents, A and B, while with inflow through vent B, they depend on the height of vent B relative to the height of the termination of the stack H. With a point source of heating, a similar transition occurs, with a unique flow regime when vent B is lower than vent A, and two possible regimes with vent B higher than vent A. In general, with a point source of buoyancy, each steady state is characterised by a two-layer density stratification. Depending on the relative heights of the two vents, in the case of outflow through vent A connected to the stack, the interface between these layers may lie above, at the same level as or below vent A, leading to discharge of either pure upper layer, a mixture of upper and lower layer, or pure lower layer fluid. In the case of inflow through vent A connected to the stack, the interface always lies below the outflow vent B. Also, in this case, if the inflow vent A lies above the interface, then the lower layer becomes of intermediate density between the upper layer and the external fluid, whereas if the interface lies above the inflow vent A, then the lower layer is composed purely of external fluid. We develop expressions to predict the transitions between these flow regimes, in terms of the heights and areas of the two vents and the stack, and we successfully test these with new laboratory experiments. We conclude with a discussion of the implications of our results for real buildings.

On the transition from displacement to mixing ventilation with a localized heat source

We investigate the steady state natural ventilation of a room heated at the base and consisting of two vents at different levels. We explore how the air flow rate and internal temperature relative to the exterior vary as a function of the vent areas, position of the vents and heat load in order to establish appropriate ventilation strategies for a room. When the room is heated by a distributed source, the room becomes well mixed and the steady state ventilation rate depends on the heating rate, the area of the vents and the distance between the lower and upper level vents. However, when the room is heated by a localised source the room becomes stratified. If the effective ventilation area is sufficiently large, then the interface separating the two layers lies above the inlet vent and the lower layer is comprised of ambient fluid. In this case the upper layer is warmer than in the well mixed case and the ventilation rate is smaller. However, if the effective area for ventilation is sufficiently small, then the interface separating the two layers lies below the inlet vent and the lower layer is comprised of warm fluid which originates as the cold incoming fluid mixes during descent from the vent through the upper layer. In this case both the ventilation rate and the upper layer temperature are the same as in the case of a distributed heat load. As the vertical separation between lower and upper level vents decreases, then the temperature difference between the layers falls to zero and the room becomes approximately well mixed. These findings suggest how the appropriate ventilation strategy for a room can be varied depending on the exterior temperature, with mixing ventilation more suitable for winter conditions and displacement ventilation for warmer external temperatures.

The influence of repeated loading, residual stresses and shakedown on the behaviour of tribological contacts

Most tribological pairs carry their service load not just once but for a very large number of repeated cycles. During the early stages of this life, protective residual stresses may be developed in the near surface layers which enable loads which are of sufficient magnitude to cause initial plastic deformation to be accommodated purely elastically in the longer term. This is an example of the phenomenon of 'shakedown' and when its effects are incorporated into the design and operation schedule of machine components this process can lead to significant increases in specific loading duties or improvements in material utilization. Although the underlying principles can be demonstrated by reference to relatively simple stress systems, when a moving Hertzian pressure distribution in considered, which is the form of loading applicable to many contact problems, the situation is more complex. In the absence of exact solutions, bounding theorems, adopted from the theory of plasticity, can be used to generate appropriate load or shakedown limits so that shakedown maps can be drawn which delineate the boundaries between potentially safe and unsafe operating conditions. When the operating point of the contact lies outside the shakedown limit there will be an increment of plastic strain with each application of the load - these can accumulate leading eventually to either component failure or the loss of material by wear. © 2005 Elsevier Ltd. All rights reserved.

The modified Kanerva model for automatic speech recognition

A parallel processing network derived from Kanerva's associative memory theory Kanerva 1984 is shown to be able to train rapidly on connected speech data and recognize further speech data with a label error rate of 0·68%. This modified Kanerva model can be trained substantially faster than other networks with comparable pattern discrimination properties. Kanerva presented his theory of a self-propagating search in 1984, and showed theoretically that large-scale versions of his model would have powerful pattern matching properties. This paper describes how the design for the modified Kanerva model is derived from Kanerva's original theory. Several designs are tested to discover which form may be implemented fastest while still maintaining versatile recognition performance. A method is developed to deal with the time varying nature of the speech signal by recognizing static patterns together with a fixed quantity of contextual information. In order to recognize speech features in different contexts it is necessary for a network to be able to model disjoint pattern classes. This type of modelling cannot be performed by a single layer of links. Network research was once held back by the inability of single-layer networks to solve this sort of problem, and the lack of a training algorithm for multi-layer networks. Rumelhart, Hinton & Williams 1985 provided one solution by demonstrating the "back propagation" training algorithm for multi-layer networks. A second alternative is used in the modified Kanerva model. A non-linear fixed transformation maps the pattern space into a space of higher dimensionality in which the speech features are linearly separable. A single-layer network may then be used to perform the recognition. The advantage of this solution over the other using multi-layer networks lies in the greater power and speed of the single-layer network training algorithm. © 1989.

Simulation studies on the magnetization of (RE)BCO bulk superconductors using various split-coil arrangements

Simulation studies were conducted on the magnetization of (RE)BCO (RE-Ba-Cu-O, where RE represents a rare earth element) bulk superconductors using various split-coil arrangements by solving the critical state equation using the commercial software FlexPDE. A pair of coaxial coils of identical size is identified as an optimum arrangement for practical magnetization at 77K by the zero-field cooling technique. In general, the magnetization process is likely to be most effective when the outer radius of the coils lies between 100% and 50% of the sample radius. A relatively large coil pair is necessary for samples with either a smaller aspect ratio or larger values of J c0. Two different regimes of flux penetration are found to be involved in the magnetization process. For a sufficiently small sample, the penetration field is determined by flux propagation from beneath the coil to the centre of the sample; for a sufficiently large sample, the definitive propagation route is from beneath the coil to the periphery of the sample. Effective split-coil magnetization occurs only in the former regime, and both penetration regimes are completely different from that involved in the solenoidal-coil magnetization process. © 2012 IOP Publishing Ltd.