Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/67245
Title: Sound propagation in performance halls with balconies
Authors: Cheung, Liu Yee
Advisors: Tang, Shiu-keung (BSE)
Keywords: Architectural acoustics.
Theaters -- Design and construction.
Theater architecture.
Issue Date: 2017
Publisher: The Hong Kong Polytechnic University
Abstract: The development of the West Kowloon Cultural District and the growing demand of performing facilities for the locals, more world-class performance halls with balconies are expected to be built in the future in Hong Kong. This urges the need to understand more about halls with balconies. Nowadays, to understand the variation of the acoustics in large performance hall in a detailed manner, seats to seats measurement is unavoidable. However, full scale measurements are very time consuming. This study started with real hall measurements as a hall survey aiming at building a hall database locally for this study and future research. Four different performing halls with different sizes and designs were measured using impulse method using Room Acoustic software. In three of the halls, both its concert and proscenium setting were measured. The measurement results was then reviewed and commented. With the hall geometric data and the first reflection path difference calculated from the point of reflection, the measured hall data were then used for establishing a simple framework using neural network analysis. Testing the four training schemes with a simple feed-forward network, an artificial neural network for the evaluation of performance hall acoustics was successfully established. This network predicted the parameters measured in Hall A successfully. Hall B's data was used to validate this prediction approach. With the validation results, this framework of using a small number of training/measured inputs to predict other hall parameters were founded reliable for halls with similar level of reverberance. Furthermore, the real hall measurements results of Hall A were used to test and build various regression model. For simplicity, the regression models generated are formed by linear combinations of polynomials of these parameters without any inclusion of cross-products of different parameters. Once the source-to-receiver distance, azimuthal and elevation angle are included, the regression model predicts more accurately than the neural network approach. However, the symmetry of the hall affected the formation of the best performing model. A model consisting of quadratic polynomials in source-toreceiver distance and elevation angle and a linear function of the azimuthal angle magnitude performed best in symmetrical halls while a quadratic function of source-toreceiver distance and a linear function of elevation angle, a polynomial in azimuthal angle is the best for asymmetrical hall. Since asymmetrical hall design yet common in Hong Kong and lack of measurement data, further validation is required. To study the design of the balcony to various parameters in a hall, a 1/10 scale model of Hall A was done to evaluate the balcony effect. The model architecture was based on the geometry of Hall A. Plywood panels were used to construct the model on top of a raised timbre framework that allow access from below. Both the concert and proscenium setting of the hall was tested with and without the balcony. The results show that the balcony affect the energy received at different location of hall, especially the seats underneath the balcony.
Description: PolyU Library Call No.: [THS] LG51 .H577P BSE 2017 Cheung
xxviii, 202 pages :color illustrations
URI: http://hdl.handle.net/10397/67245
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
b29616724_link.htmFor PolyU Users208 BHTMLView/Open
b29616724_ira.pdfFor All Users (Non-printable)5.02 MBAdobe PDFView/Open
Show full item record

Page view(s)

40
Last Week
8
Last month
Checked on Oct 16, 2017

Download(s)

8
Checked on Oct 16, 2017

Google ScholarTM

Check



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.