On the Load Dependence of Micro-Hardness Measurements: Analysis of Data by Different Models and Evaluation of Measurement Errors

Abstract

The influence of applied loads between 0.09807 N and 0.9807 N on measured values of micro-hardness was evaluated by Meyer's index n, proportional specimen resistance model (PSR) and Hays -Kendall methods, Total Dispersion Zone and Analysis of Variance (ANOVA). The measurement was repeated 6 times using the same hardness reference block with standard hardness Hc = 327 HV0.05 as a sample. The influence of the load on the measured value of micro-hardness is statistically significant, and the relationship between applied load and micro-hardness manifests reverse indentation size effect (ISE) for most of "measurements". The high value of the uncertainty of results can affect the existence and nature of ISE, especially at low loads.

The load dependence of the micro-hardness of the blast furnace slag

“…Meyer's power law, proportional specimen resistance (PSR), and Hays -Kendall approach were used for the quantitative description of the ISE. The procedure for calculating ISE characteristics using references of Sangwal et al [18], Li & Bradt [25] and Michels & Frischat [26], and in particular, the procedure of the measuring the ceramic micro-hardness by Kim & Kim [27] presented in more detail Petrík [20].…”

“…There is a "normal" ISE, whose causes were summarized, i.e. Petrík [20] under the authority of, i.e. Sangwal et al [18], Gong et al [21], Ren et al [22].…”

The fatigue stress on laser welded joints of DC06EK steel

“…3. Ferritic grains are elongated in the direction of rapid heat dissipation into the weld metal [12]. The structure of the weld metal (WM) in Fig.…”

Effect of Non-Fulfillment of Assumptions on Gage Repeatability and Reproducibility Study Evaluation

“…For most studies, three operators were chosen to measure each sample three times. To increase the explanatory power of all studies, the number of measurements of each sample was increased to ten by each operator (Petrík, 2016). As a result, the new data set no longer came from the normal distribution.…”