Interpretation of measured profile losses in unsteady wake-turbine blade interaction studies

The interaction of wakes shed by a moving bladerow with a downstream bladerow causes unsteady flow. The meaning of the freestream stagnation pressure and stagnation enthalpy in these circumstances has been examined using simple analyses, measurements and CFD. The unsteady flow in question arises from the behaviour of the wakes as so-called negative-jets. The interactions of the negative-jets with the downstream blades lead to fluctuations in static pressure which in turn generate fluctuations in the stagnation pressure and stagnation enthalpy. It is shown that the fluctuations of the stagnation quantities created by unsteady effects within the bladerow are far greater than those within the incoming wake. The time-mean exit profiles of the stagnation pressure and stagnation enthalpy are affected by these large fluctuations. This phenomenon of energy separation is much more significant than the distortion of the time-mean exit profiles that is caused directly by the cross-passage transport associated with the negative-jet, as described by Kerrebrock and Mikolajczak. Finally, it is shown that if only time-averaged values of loss are required across a bladerow, it is nevertheless sufficient to determine the time-mean exit stagnation pressure.

A Physical Interpretation of Stagnation Pressure and Enthalpy Changes in Unsteady Flow

This paper provides a physical interpretation of the mechanism of stagnation enthalpy and stagnation pressure changes in turbomachines due to unsteady flow, the agency for all work transfer between a turbomachine and an inviscid fluid. Examples are first given to illustrate the direct link between the time variation of static pressure seen by a given fluid particle and the rate of change of stagnation enthalpy for that particle. These include absolute stagnation temperature rises in turbine rotor tip leakage flow, wake transport through downstream blade rows, and effects of wake phasing on compressor work input. Fluid dynamic situations are then constructed to explain the effect of unsteadiness, including a physical interpretation of how stagnation pressure variations are created by temporal variations in static pressure; in this it is shown that the unsteady static pressure plays the role of a time-dependent body force potential. It is further shown that when the unsteadiness is due to a spatial nonuniformity translating at constant speed, as in a turbomachine, the unsteady pressure variation can be viewed as a local power input per unit mass from this body force to the fluid particle instantaneously at that point. © 2012 American Society of Mechanical Engineers.

Use of contexts in language model interpolation and adaptation

Language models (LMs) are often constructed by building multiple individual component models that are combined using context independent interpolation weights. By tuning these weights, using either perplexity or discriminative approaches, it is possible to adapt LMs to a particular task. This paper investigates the use of context dependent weighting in both interpolation and test-time adaptation of language models. Depending on the previous word contexts, a discrete history weighting function is used to adjust the contribution from each component model. As this dramatically increases the number of parameters to estimate, robust weight estimation schemes are required. Several approaches are described in this paper. The first approach is based on MAP estimation where interpolation weights of lower order contexts are used as smoothing priors. The second approach uses training data to ensure robust estimation of LM interpolation weights. This can also serve as a smoothing prior for MAP adaptation. A normalized perplexity metric is proposed to handle the bias of the standard perplexity criterion to corpus size. A range of schemes to combine weight information obtained from training data and test data hypotheses are also proposed to improve robustness during context dependent LM adaptation. In addition, a minimum Bayes' risk (MBR) based discriminative training scheme is also proposed. An efficient weighted finite state transducer (WFST) decoding algorithm for context dependent interpolation is also presented. The proposed technique was evaluated using a state-of-the-art Mandarin Chinese broadcast speech transcription task. Character error rate (CER) reductions up to 7.3 relative were obtained as well as consistent perplexity improvements. © 2012 Elsevier Ltd. All rights reserved.