A novel method to inverse the water retention curves with consideration of volume change during centrifuge testing

Measuring the soil water retention curve (SWRC) using the centrifuge method often results in compressibility and an increase in bulk density. The compressibility and bulk density of the specimen are always assumed to be constant while deducing the SWRC from centrifuge testing, which may cause inaccurate SWRCs. This study proposed an improved genetic algorithm (GA) to inverse the hydraulic parameters and further constructed the SWRC by taking bulk density into account for soils with varying grain size compositions. For this study, sandy loam soils (i.e., seven treatments T1-T7) were prepared based on different volume ratios of coarse (0.075-2 mm) to fine (<0.075 mm) particles. SWRCs including compressibility and bulk density changes were measured using the centrifuge method. For verification, SWRCs were also independently deduced from one-dimensional infiltration experiments by inverse analysis using the improved GA that combined multi-objective optimization and elitist strategy. Upon the final shrinkage ratio of 18.32 to (similar to)26.38%, the bulk density during centrifugation increased by 22.33 to (similar to)35.72% correspondingly, thereby causing a difference in soil moisture of 9.58 to (similar to)16.53% between the corrected and uncorrected SWRCs. The comparison between simulated SWRC and two types of measured SWRCs (i.e., corrected and uncorrected SWRC) demonstrated that the corrected SWRC was more reasonable regardless of soil particle compositions. The findings suggest the corrected SWRC considering bulk density changes provides more accurate hydraulic parameters, which can be inversed from the infiltration data using the improved GA.