Enhancing the resolution of the angular orientations of the flow controlling blades on a sustainable house by training an artificial neural network


METADATA ONLY
Loading...

Date

2021-02-15

Publication Type

Journal Article

ETH Bibliography

yes

Citations

Altmetric
METADATA ONLY

Data

Rights / License

Abstract

Among all its effects, the development of the boundary layer, its separation, and formation of the wake region could lead to higher convective heat transfer over the body, if the flow conditions cause high gradient velocity profiles in the surface vicinities of the field. And also, a low-pressure region in the downstream of the geometry is formed, which increases the pressure drag exerted on it. The influence of the aforementioned issue on the zero energy house design has been tackled by introducing a new flow control mechanism. The so-called flow controlling blades (FCBs) were recently designed and investigated on a smart sustainable house, in order to control the flow field around the house, prevent the separation, and decrease the wake intensity, targeting a lower level of convective heat loss and drag force exerted on the body. The angular orientation of these FCBs was formerly determined for 12 different free wind directions (30° increments), as a look-up table for the main control system of the house. To increase the resolution of the orientations, we make use of a recently successful tool in machine learning called neural networks to estimate the desired orientation of the blades for the wind directions that do not exist in the said look-up table. Consequently, all the sample investigated sub-intervals not originally covered by the CFD data, showing great coincidence with the data driven from the neural network utilized in this study.

Publication status

published

Editor

Book title

Volume

14 (2)

Pages / Article No.

25

Publisher

Springer

Event

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Building aerodynamics; Flow control; Artificial neural networks; Sustainable buildings; Universal function approximator; Independence assumption; Model capacity

Organisational unit

Notes

Funding

Related publications and datasets