The sound speed of sand has been shown to vary with frequency, yet in many instances in geoacoustic inversions, sand is modeled as a frequency-independent effective fluid. This paper investigates the effect to which assuming a frequency-independent fluid model that neglects poroelasticity can skew parameter estimation in a laboratory layered waveguide consisting of 1-mm diameter water-saturated glass beads (WSGBs), suspended in a water-filled glass tube. The phase speed in the waveguide was measured from 1 to 7 kHz and compared with phase speeds predicted in a finite element simulation of the experiment, where the WSGBs were treated as either a fluid with constant bulk density and frequency-independent or frequency-dependent sound speed, or by an effective density fluid model (EDFM) that includes poroelasticity. Measurement-simulation agreement occurred when using the EDFM to model the WSGB, although neglecting poroelasticity in the simulation only led to a maximum phase speed discrepancy of 8 m/s. However, this effect was significant when an inference process was used to determine the effective fluid properties of the WSGBs. Finally, high-frequency (150 to 450 kHz) direct sound speed measurements of the WSGB were obtained, and best matched the mid-frequency inference results obtained using the EDFM.
