Bending wave loudspeaker technology!

Very interesting reading. In his Master’s Thesis in the Master’s programme in Sound and Vibration Kuonan Li critically explores of bending wave loudspeaker technology and implementation being used by highly acclaimed Göbel High End and similar approaches used by Manger, German Physiks, NXT etc…
“The interest in flat panel loudspeakers has increased in recent years that various forms of bending wave loudspeakers (BWL) and distributed mode loudspeaker (DML) are of great interest to loudspeaker producers and have been developed for many different applications. Both BWL and DML are based on the use of bending waves. The main difference is that the BWL uses an infinite plate approach while the DML uses a finite plate approach, or preferably, a modal approach. 
An introduction to these technologies is given in the beginning of this thesis. Then follow the bending wave theories, including famous Euler-Bernoulli beam equation, Kirchhoff plate theory and the corrections for these two simple cases. The elementary radiators that can be used for describing different approaches will be shown, as well as several particular concepts relevant to the measurements and simulations.
Afterward, some commercial products applying bending wave theories are discussed: the Bending Wave Loudspeaker from Göbel, the DDD driver from German Physiks, the sound transducer from Manger, and the commercially most important, the DML from NXT. The first three use an infinite plate approach while the last one is based on modal approach. 
In order to study the relative performance of a DML, a sample provided by NXT is used as together with a conventional electro-dynamic. The two speakers are subject to a series of measurements, such as impulse response - which also implies frequency response - , directivity, sound power, efficiency, sensitivity, and nonlinear distortion. In addition, sound pressure levels are measured at 30 positions in a listening room to study sound distribution. The mechanical properties of the DML, including density, Young’s modulus and mechanical impedance are vital for further simulation; therefore they are measured as well. A blind listening test is also done to investigate the listeners’ responses to DML and the preference between DML and electrodynamic loudspeaker tested earlier. 
At the end of the project, some simulations are carried out to optimize DML performance. The ratio of length to width is important for the distribution modes in frequency, as well as for choosing the optimum position of the exciter. Finally, finite element methods are employed using of Comsol’s Mutiphysics software to investigate the effect of panel’s size to sound pressure response at low frequencies. Also, the boundaries of the plate are varied as roller, free, or fixed in Comsol’s model, to see how boundary conditions influence the frequency response.
With the result of the experiments mentioned above, one can conclude that DML has lower sensitivity and efficiency than the conventional electrodynamic loudspeaker tested here. The DML lack of low frequency is due to low modal density and the lack of high frequency is due to the transition of bending waves to transverse waves. The directivity of the DML is not better than the electrodynamic loudspeaker’s in fact it measured less at mid-frequencies. The result of the simulations is that the best length/width ratio is between 0.95~0.99, depending how the optimum is defined.”
Complete article: here.