Advanced Statistical Refinement of Surface Layer’s Discretization in the Case of Electro-Spark Deposited Carbide-Ceramic Coatings Modified by a Laser Beam

Abstract

Carbide coatings have numerous industrial applications due to their high abrasion, sliding and erosion resistance. The paper presents results of an advanced statistical analysis – involving auxiliary simulation methods like smooth bootstrapping and imputing of missing data – executed on surface layer profile of modified carbide-ceramic coatings. Source data were gathered in the previous research by Radek and Bartkowiak focused on microstructure analysis (SEM Joel JSM-5400), microhardness (Vickers method), roughness (FORM TALYSURF-120L) and adhesion (CSEM REVETEST)) tests. Anti-wear coating were first deposited on carbon steel C45 from WC-Co-Al2O3 electrodes in the process of the electro-spark alloying (ESA) by the EIL-8a apparatus. In the next step the coating were laser melted using impulse mode of Nd:YAG laser (BLS 720 model). Due to significant irregularity of collected data, the special methods of smoothing and imputing were involved based on Monte-Carlo methods. The collected data set was several times randomly divided into analytical and verification sub-sets and mentioned methods were applied. The results were used to calculate descriptive statistics like average values, variances, confidence intervals and smoothed histograms of probability distributions. The validity of the proposed approach was positively verified and it significantly improved quality of the results. The smoothing and imputing of data allow to avoid numerical artifacts that may arise during the classical statistical calculations on irregular data.