Quantifying the errors of the particle-source-in-cell Euler-Lagrange method

TL;DR

This paper develops an expression to estimate errors in the PSIC-EL method for simulating particle-laden flows, relating errors to particle size, mesh spacing, and flow regime, thus guiding accuracy expectations.

Contribution

It provides a systematic error estimation formula for the PSIC-EL method based on particle size, mesh ratio, and Reynolds number, which was previously lacking.

Findings

Error bound approximated by (6/5) * (d_p / h)

Expression accurately estimates errors for d_p / h ≤ 1/2

Valid across particle Reynolds numbers up to 500

Abstract

The particle-source-in-cell Euler-Lagrange (PSIC-EL) method is widely used to simulate flows laden with particles. Its accuracy, however, is known to deteriorate as the ratio between the particle diameter~($\smash{d_\text{p}}$) and the mesh spacing~($h$) increases, due to the impact of the momentum that is fed back to the flow by the Lagrangian particles. Although the community typically recommends particle diameters to be at least an order of magnitude smaller than the mesh spacing, the errors corresponding to a given $\smash{d_\text{p} / h}$ ratio and/or flow regime have not been systematically studied. In~this paper, we provide an expression to estimate the magnitude of the flow velocity disturbance resulting from the transport of a particle in the PSIC-EL framework, based on the $\smash{d_\text{p} / h}$ ratio and the particle Reynolds number, $\smash{\text{Re}_\text{p}}$. This, in turn, directly relates to the error in the estimation of the undisturbed velocity, and therefore to the error in the prediction of the particle motion. We show that the upper bound of the relative error in the estimation of the undisturbed velocity, for all particle Reynolds numbers, is approximated by $\smash{(6/5)\,d_\text{p} / h}$. Moreover, for all cases where $\smash{d_\text{p} / h \lesssim 1/2}$, the expression we provide accurately estimates the value of the errors across a range of particle Reynolds numbers that are relevant to most gas-solid flow applications ($\smash{\text{Re}_\text{p} < 500}$).