Shokri, AliPittari, Adriande Graaf , KimNikghalb Ashouri, Saeed2025-05-022025-05-022024https://hdl.handle.net/10289/17351The most critical hydraulic property of soil is the saturated hydraulic conductivity (Ks) of its matrix. Accurate determination of Ks is vital for understanding and simulating flow and transport phenomena. Hydrologists require accurate estimations of soil moisture content and retention to better discern rainfall distribution between runoff and infiltration; agronomists similarly rely on this data to inform crop yield models and formulate appropriate irrigation timetables. A diverse range of methodologies currently exists to determine the Ks, with the constant head test in the laboratory being a common approach. Traditionally, it is presumed that the Ks values obtained from both field and laboratory assessments remain constant. In addition, Ks is considered as a constant in various applications, such as solute transport modelling and surface water as well as groundwater simulations. Given that soils can undergo saturation and desaturation they can be subjected to gradual and/or rapid changes in the hydraulic head, and thus the Ks of the soil will be affected. Consequently, any changes in Ks will affect the hydraulic properties of the soil. Therefore, to fully understand the hydraulic behaviour of a soil requires investigating the changes in Ks during short and long timeframes. A primary objective of this study was to systematically investigate the changes that could occur in Ks during short-term and long-term constant head tests. Two soil columns were prepared and subjected to continuous constant head tests for a duration exceeding 50 days The hydraulic heads were increased gradually in a similar way as for a constant head test, with the only difference being that the hydraulic heads were increased every 5 to 7 days. The outcomes reveal that Ks exhibits negligible variations over short intervals, showing minor daily fluctuations. However, the long-term variations highlight significant shifts in Ks, with a maximum decrease in value of approximately 90% from the onset of the experiments, followed by intermittent brief periods of slight increase. These findings challenge the notion of Ks as a constant value over time. Moreover, a comparative analysis of soil properties before and after the tests indicates notable modifications in the sample characteristics. Specifically, the particle size distribution and particle density differ pre and post-experiment, with an observable migration of fine, low-density particles towards the upper section of the sample closer to the outlet, and aligned with the flow direction. This phenomenon results in an increased particle density at the lower part of the sample adjacent to the water inlet. The migration of particles is evidence of the constant head test impact on the distribution of particles. The change in the distribution of the particles pre and post experiment, caused by the standard procedure of the tests, shows that the Ks value determined in the lab through constant head tests is influenced by the tests, and might not reflect the actual Ks of the sample. In addition, a series of short-term experiments were carried out on 10 samples that were prepared with different particle size mixtures. These mixtures featured varying proportions of fine and coarse materials to study the range of changes in the Ks in different mixtures. The hydraulic heads of the tests on these samples were systematically adjusted in ascending increments, to investigate the range of changes in Ks under low and high hydraulic heads during short-term tests. This test series revealed a distinctive 3-phase pattern exhibited by the samples: a low and relatively consistent Ks was initially observed prior to a transition through a phase change as the hydraulic head escalates, leading to a stable constant Ks under high hydraulic gradients. These observations underscore the significance of the changes of Ks during constant head tests, and show a need to consider that Ks has a range of values over time rather than being a constant value. Determining a suitable value for Ks in the lab and field is time consuming, it can be costly, and test results can be impacted by the test conditions. For instance, in the lab, ensuring the full saturation of the samples before the tests needs determination of solid particle density, which requires a pycnometer. Moreover, due to the spatial variations in Ks, the number of samples needed to represent the typical Ks in a study area can be costly to retrieve. Similarly, to investigate the spatial variation of Ks, the field tests (slug tests, pumping tests, or tracer tests) add extra cost and time to projects. Additionally, taking soil samples or performing Ks field tests in remote areas or harsh weather conditions is an arduous task. Therefore many researchers have endeavoured to establish predictive relationships to estimate Ks accurately without relying on resource-intensive laboratory or field tests. Indirect estimation methods are generally based on easily obtainable data such as soil texture, particle size distribution, specific surface area, or porosity. These relationships are developed into pedotransfer functions and aim to provide estimations of Ks. However, the reliability of these functions remains uncertain due to discrepancies observed when comparing the results obtained using different pedotransfer functions with actual field or laboratory measurements. To address this issue, the Ks derived from Darcy's equation for the short-term tests were compared with those computed using 7 of the most commonly used pedotransfer functions. The objective of the comparison was to evaluate the effectiveness of these equations in predicting Ks. The analysis revealed that the majority of pedotransfer functions tend to underestimate the Ks values of the soil samples. Subsequently, a novel method was employed to adjust the parameters of one of the equations, which is suitable for non-plastic sandy soils as used in the earlier experiments, by linking the equation parameters to the coefficient of uniformity and void ratio of the samples. This modification led to a more precise estimations of Ks for the range of sandy samples used in the research. Given the inaccuracies related to the constant head tests and the impact on Ks determination (such as the disturbance in connection of drainable pores due to varying hydraulic heads, inaccuracies in reading the flow rate and sample head gradient, improper sample saturation, effect of air entrapment in the sample), it was necessary to look for improvements in determining the Ks in the soil. Tracer tests on soil samples in the laboratory and field have been used by several researchers to estimate the soil hydraulic properties. These tests offer essential insights into site hydrogeology, aiding in the delineation of crucial factors such as natural groundwater velocity and hydraulic connectivity between neighbouring aquifers and aquitards. When coupled with groundwater flow and transport models, tracer tests can enhance the precision of defining Ks and its spatial distribution accurately. In order to acquire soil hydraulic properties from tracer tests, a tracing element is injected into the samples, and the changes of concentration in the outflow is plotted over time. The resulting graph is called a break through curve (BTC). In order to acquire soil hydraulic properties from BTCs, advection-dispersion equations (ADE) and the models that are built upon ADE equations are widely used. The ADE links soil hydraulic properties (Ks and dispersivity coefficient) to the BTC. Trial and error is required to achieve the best fit between predicted and measured BTC which can be challenging and may require several trials to process multiple sets of BTCs. In this study, the use of a statistical distribution, specifically the Inverse Gaussian distribution (IG), was examined for fitting BTCs obtained from 94 tracer tests utilizing sodium chloride as the tracer compound. The results of fitting the BTCs using the IG distribution revealed that solute transport in porous media could be effectively articulated using this distribution, particularly given its suitability for positively skewed data. The majority of the tracer tests fitted with the IG distribution exhibited model efficiencies exceeding 0.9 (based on Nash–Sutcliffe model efficiency coefficient), indicating a strong fit of the BTCs. A comparison of the IG method with alternative methodologies from existing research further underscored the superior fit achieved by the IG distribution in modelling BTCs. The test series outlined in this thesis represent a significant advancement in the comprehension of saturated hydraulic conductivity within porous media. The noticeable long-term changes of Ks in the laboratory show this parameter cannot be considered as a constant, and the assumption of linearity between the hydraulic head and flow in the samples is only valid in the short-term tests. Additionally, Ks can have a range of low and high values, during rapid changes of the hydraulic head. The implications of these findings offer potential revisions to the existing approaches for integrating saturated hydraulic conductivity within equations of surface and groundwater models. Furthermore, the study highlights the viability of employing statistical methods to analyse the BTCs from tracer tests. The use of statistical methods with the tracer tests can save time and cost in the post-processing of the tracer test results.enAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.Thesis