A New Fick-Jacobs Derivation with Applications to Computational Branched Diffusion Networks

Abstract: The Fick-Jacobs equation is a classical model reduction of 3-dimensional diffusion in a tube of varying radius to a 1-dimensional problem with radially scaled derivatives. This model has been shown to be unstable when the radial gradient is too steep. In this work, we present a new derivation of the Fick-Jacobs equation that results in the addition of higher order spatial derivative terms that provide additional stability in a wide variety of cases and improved solution convergence. We also derive new numerical schemes for branched nodes within networks and provide stability conditions for these schemes. The computational accuracy, efficiency, and stability of our method is demonstrated through a variety of numerical examples.