A discussion on the performance of seven existing models proposed to describe induced polarization

Abstract

Many types of classical models of the induced polarization (IP) effects of rock ore have been introduced in previous studies. Our focus is on determining the most effective model for describing the rock ore IP effects based on a numerical simulation calculation and a rock mineral spectrum experiment. We have constructed a Delphi 7 program for seven models to calculate the complex resistivity spectrum of phase. We have also evaluated the influences of the parameters on the spectrum and the spectral scope. More than 50 pieces of various natural rock ore samples from 16 regions were selected, and the frequency sweep measurements were completed using an SI-1260 Solartron spectrometer. Another program of seven models was developed using a differential evolution algorithm and least-squares for fitting the experimental spectral data. The purpose was to evaluate the ability of these models for fitting the IP effects. The results found that the influence laws on the spectrum of the different parameters, as well as the spectral scope of the models, were different. The natural samples measured revealed unimodal and bimodal phase spectra. Dias and multi-Cole-Cole models were optimal for characterizing the unimodal and bimodal phase spectra of the selected metal minerals, with the highest fitting precision root-mean-square (rms) error of 2.5% and 1.16%, respectively. Warburg, Cole-Cole, and generalized Cole-Cole models were suitable for the unimodal phase spectra, with the highest fitting precision of 3.67%, 3.01%, and 3.27%, respectively. Madden and Cantwell model was suitable for characterizing the silver-ore- and gold-bearing pyrite, but the rms was [Formula: see text] for other samples. The Debye model had an rms [Formula: see text] for characterizing all our experimental samples.