The water pressure measured during slow, constant rate drainage in a two-dimensional porous model exhibits sudden jumps as bursts of air quickly displace water from a region. The measured size distribution of the pressure jumps is exponential. Invasion percolation (IP) simulations give a power-law size distribution of the connected regions invaded in bursts. In the experiments the meniscii of the fluid-fluid front adjust during a burst, causing the capillary pressure to decrease. Including this effect in a modified invasion percolation algorithm causes potentially large bursts to split up into smaller bursts that are exponentially distributed. From the experimental pressure curve it is possible to identify groups of bursts that would become a single, ‘‘composite’’ burst in a larger system. These composite bursts are power-law distributed, consistent with simulations and percolation theory. Different versions of the IP model result in different structures and power-law exponents. The best choice of model for the present experiment is discussed. © 1996 The American Physical Society.

• Received 16 March 1995

DOI:https://doi.org/10.1103/PhysRevE.53.966