ACCURACY CONTOURS IN (NT,LAMBDA) SPACE IN ELECTROCHEMICAL DIGITAL SIMULATIONS

In finite difference simulations of electrochemical transport problems, it is usually tacitly assumed that lambda, the stability factor D-delta-t/delta-x2, should be set as high as possible. Here, accuracy contours are shown in (n(T), lambda) space, where n(T) is the number of finite difference steps per unit (dimensionless) time. Examples are the Cottrell experiment, simple chronopotentiometry and linear sweep voltammetry (LSV) on a reversible system. The simulation techniques examined include the standard explicit (point- and box-) methods as well as Runge-Kutta, Crank-Nicolson, hopscotch and Saul'yev. For the box method, the two-point current approximation appears to be the most appropriate. A rational algorithm for boundary concentrations with explicit LSV simulations is discussed. In general, the practice of choosing as high a lambda value when using the explicit techniques, is confirmed; there are practical limits in all cases.