A dimensionless frequency number, N-f, (fd(p)(2)/alpha(e)), is proposed to understand better and predict the bed-to-surface heat transfer coefficients in dense gas-solid fluidized beds. A total of 756 sets of pressure signals were acquired simultaneously with the measurement of local, time-averaged bed-to-surface heat transfer coefficients. The experimental matrix included 3 different locations for the pressure probes, 13 different powders, (with d(sv) in the range 20 mu m-1789 mu m), 3 powders at least, in each of Geldart A, B and D categories, two different distributor plates, and a wide range of fluidization velocities (up to 2.4 m/s). Fast Fourier Transforms were utilized to analyze and interpret the pressure signals. First principle models that retain the physical picture of Mickley and Fairbanks's [H.S. Mickley, L.R. Fairbanks, AIChE J. 3 ( 1955) 374] packet theory were scaled into mesoscopic correlations. The experimental data were compared with the mesoscopic correlations derived from first principles models.