Determination of hydraulic conductivity and shear strength properties of unsaturated residual soil from flysch rock mass

Slopes in flysch deposits areas wide within Europe are highly prone to landslide occurrence. Depending on the material properties and climate conditions, instabilities in a form of earthflows, shallow and deep-seated landslides were observed in these formations. Typically, slope instabilities occurred after prolonged periods of rainfall. The Rječina River Valley, Croatia, built in flysch formation, is well known by several landslides occurred in the past. The weathering process of flysch rock mass and local climate conditions resulted in a specific engineering geological profile of the valley, with the unsaturated residual soil covering the bedrock. Although the behaviour of residual soil is important for a landslide triggering both through the rainfall infiltration process and (unsaturated) shear strength, hydro-mechanical properties of this material in unsaturated conditions were not investigated in the past. This paper summarizes the results of different laboratory tests performed on intact samples for hydro-mechanical characterization of the residual soil from flysch rock mass. It was found that the unique shear strength envelope could be used to determine failure conditions both for saturated and unsaturated conditions. The results obtained from strain-controlled and wetting tests performed in conventional and modified direct shear apparatuses indicated high values of the apparent cohesion that the near-surface soil can experience due to the increase of matric suction. The hysteresis effects and hydraulic paths to which soil was exposed to in the past were found to affect the soil behaviour, while the soil formation process results with a complex soil structure that imposes the necessity of using intact soil samples for proper hydraulic characterization of the soil.


Introduction
This paper presents some of the results of laboratory tests performed within a novel study that focuses on the role of the unsaturated zone in rainfall-induced landslides in flysch deposits [1]. While the effect of rainfall on the landslide triggering was investigated through the build-up of (positive) pore-water pressure along the sliding surface (e.g., [2][3][4]), very few studies have investigated the physical process of rainfall infiltration in a flysch slope (e.g., [5][6][7]) and how it affects the stability of flysch slope with time (e.g., [7]). Soil-water retention curve (SWRC) and the hydraulic conductivity function (HCF) are the unsaturated soil property functions required to analyse the transient process of rainfall infiltration. In order to determine how the redistribution of water content and change of matric suction affect the available shearing resistance along the (part of) sliding surface above the phreatic line, the functional relationship between the matric suction and shear strength has to be known. The latter can be related to the SWRC through effective stress (e.g., [8][9][10]).
Determination of unsaturated soil property functions involves the usage of expensive laboratory devices that enable control or measurement of (matric) suction during a test. Change of volume of a tested specimen due to change of matric suction, maintenance of the hydraulic contact, air diffusion and cavitation of water are some of the commonly encountered problems when performing laboratory or field measurements in partially-saturated soils (e.g., [11,12]). Test duration is another challenge when determining unsaturated soil properties experimentally. This especially refers to fine-grained soils, where low hydraulic conductivity further reduces under increasing soil suction conditions. Thus, aside from the need for covering a wide range of matric suction that fine-grained soil can typically exhibit, the above-mentioned problems are even more pronounced due to prolonged testing time.
To overcome these shortcomings, various estimation methods for the prediction of unsaturated soil properties were developed. For example, grain-size distribution or basic index soil properties can be used to predict SWRC (e.g., [13,14]). Regression-based approaches can be used to determine HCF based on observed flow quantities (e.g., [15]). Another common approach is to estimate the HCF from the SWRC and some other soil properties (e.g., [16,17]). SWRC can be used to predict the unsaturated shear strength as well (e.g., [18,19]).
In this study, the hydraulic conductivity and shear strength properties were determined experimentally on intact samples of residual soil from flysch rock mass. Firstly, the saturated hydraulic condu determined on intact and remoulded s outflow tests performed in conventiona triaxial apparatuses. Measurements co samples tested in the HYPROP device evaporation method [20] were used to o a considered range of matric suction. was estimated from the SWRC and k s . controlled direct shear tests were perfo conditions, both on intact and rem Suction-controlled and wetting tests w intact samples by using the modified based direct shear apparatuses (DSA). T were used to define a unique shear stre terms of the Bishop's effective stress fo general conclusions on hydro-mechani the investigated soil are outlined in the e 2

Investigation area characterization
In recent years, persistent rainfall has landslides in the Rjeþina River Valley, larger landslide occurred in February slope of the valley above the Valiüi landslide was chosen as a pilot ar comprehensive study that investigates t unsaturated zone for rainfall-induced la slopes. The superficial samples of the re from different parts of the Valiüi landsl that the material covering the slope ha in terms of the grain-size distribution limits, and could be classified as lean Intact samples were collected from located in the central part of the Valiüi drilled samples, collected from the soi few meters of depth, were used to deter natural water content, and basic soil ( Table 1). Several details of specime outlined in Fig. 1.
The formation process of the resid slopes of the valley results with t structure, characterized by the occasi macro voids and siltstone grains that ca weathering degree. Measurements for the SWRC showed that different SW when remoulded and consolidated sa instead of intact ones [22]. Best-fit SW intact soil are summarized in Table 2.

Experimental procedures
Measurements for the determinatio conductivity and shear strength prope conditions were performed on both inta samples. As it was found that k s valu remoulded samples differ significantly, for determination of unsaturated soil performed on intact samples. Thu unsaturated soil properties of the re flysch rock mass are representative for i uctivity (k s ) was samples using the al oedometer and ollected on intact and the extended obtain the HCF in Finally, the HCF . Series of strainormed in saturated moulded samples. ere performed on d, axis-translation The obtained data ength envelope in ormulation. Some ical properties of end.

and soil
triggered several Croatia. The last 2016, at the SW Lake [21]. This rea for the new the importance of andslides in flysch esidual soil, taken ide body, showed as similar features n and Atterberg's n clay (CL) [22]. the sampling pit i landslide. Handil surface up to a rmine soil density, index properties en collection are dual soil covering the specific soil onal presence of n vary in size and determination of WRC is obtained amples are used RC parameters of on of hydraulic erties in saturated act and remoulded ues for intact and all measurements l properties were us, the obtained esidual soil from in situ soil.    Fig. 4 shows intact and remoulded for the determination of k s with the fallin

Measurements of k w
The intact cylindrical specimens of dimensions 80/50 mm were installed device (HYdraulic PROPerty analyze HCF within a limited range of matri intact specimens were tested, starting water content (HYP wn ); (ii) saturated p immersion into the water with (HYP 1 ); application of a small vacuum (HYP 2 ). evaporation recorded by the precise suction readings from two mini-tensiom to calculate the coefficient of permeabi water phase k w according to the Darcy-B where ȥ mean is the mean tension avera interval of ǻt i , ǻm sample mass differen by the scale (g), ĳ w density of wate depending on the soil type, A cross-se specimen (cm 2 ), and ǻi m average hydrau Table 3 presents the water con saturation and the void ratio for each and after saturation. The HYPROP (UM used in the study is shown in Fig. 5.  f diameter/height in the HYPROP er) to define the ic suction. Three from (i) natural prior to testing by ; and, without the Water loss due to scale and matric meters were used ility respective to Buckingham law: aged over a time nce in ǻt i recorded er, a flux factor ection area of the ulic gradient [23]. ntent, degree of specimen before MS GmbH) device ampling pit.

Shear strength tests
Conventional DSA was used t properties of intact and remou conditions, while two axis-tra used to investigate shear str samples in unsaturated conditi

Shear tests in saturate
Strain-controlled tests were specimens using the conven (Controls S.p.A.). After the box at preserved w n , specimen aired water and consolidated u of vertical stress. Following specimens were sheared at th When the maximal shear di material from each testing separately, thoroughly mixed w to rest overnight. The same am ed conditions e performed on intact ntional DSA 27-WF2160 installation into the shear ns were submerged into deunder 50, 100, or 200 kPa g the consolidation stage, he rate of 5E-03 mm/min: isplacement was reached, g device was collected with distilled water and left mount of material was mbH) at the UniLj (see [22]). installed in the same testing device, testing procedure was repeated on specimens as it was in a case for the adopted testing program was aimed to differences in shear strength propert specimen disturbance effects. Fig. 6 details about testing with conventional D

Shear tests in unsaturated con
Unsaturated shear strength properties on intact specimens using two axis-tran apparatuses.  samples was approx. two orde for intact ones. As expected HYPROP device indicated a increase of matric suction, wh obtained for specimens satur vacuum-saturated specimen values, k w and k s match close parameters determined on inta to estimate the HCF represen soil, according to the equ Genuchten [25] and Fredlund e ݇ ௪ ሺ߰ሻ ൌ ݇ ௦ ൛ଵିሺఈటሻ షభ ሾଵା ሾଵାሺఈటሻ apparatuses at the UniSa (GDS Inst. Ltd.) (bottom).
w ) values determined on intact nd HCF estimated from the eq. rs of magnitude lower than d, measurements with the a decrease of k w with an hile higher k w values were rated prior to testing. For and low matric suction ely. Finally, k s and SWRC act samples [22] were used ntative for the investigated ations proposed by van et al. [26]: where Į, n, and m are constant, e y is raised to the dummy variable power derivative of the Fredlund and Xing's S Unlike for the hydraulic conductivi in saturated shear strength test results o and remoulded samples was not that sig Fig. 10 represents the behaviour of sheared in saturated and unsaturated obtained results indicate that, for the pressure, increasing matric suction resu of shear strength, while compressive shearing in saturated or nearly satu changes to dilative with an increase o The threshold value of matric suction patterns of behaviour was found to i increasing of the ʍ net . Conditions at fai saturated and suction-controlled tests ar with the stress paths from two wetting te The effective stress was calculated formulation proposed by Bishop with an effective stress parameter assum the current degree of saturation (Ȥ=S) [ (Fig. 9). intact specimens conditions. The e same confining ults in an increase behaviour during urated conditions, of matric suction. n separating two increase with the ilure obtained for re presented along ests in Fig. 11. according to the (4) med to be equal to 8,27]. e, all saturated viour. The volumetric behaviour of un affected by both the net ve applied during the shearing. parameters were estimated fo intact (c'=5 kPa, ‫)°7.13=‬ (c'=6 kPa, ‫.)°6.82=‬ The re samples indicate that the ev stress as proposed by Bishop shear strength parameters (c'= saturated and unsaturated cond is more suitable to be co conditions of intact samples process. Stress paths of the tw Fig. 11 reveal that the failure only when the stress path failure envelope.  (Fig. 11). The latter ompared with the stress s undergoing the wetting wo wetting tests shown in e upon wetting is reached approached the obtained opes for intact and remoulded ditions.
Pa (right) of net vertical stress.

Conclusions
Some of the test results for the h characterization of residual soil form are presented in this manuscript. The suggest that soil formation process and to which the soil was exposed in its hist need for using intact samples to cor hydro-mechanical properties. For a con matric suction, measured k w was foun with the estimated HCF for intact samp of shear strength parameters was obtai effective stress formulation proposed by for saturated and unsaturated conditions strength envelope was found to be able conditions at failure in case of wetting all cases, an increase of matric suction c of shear strength, and change from cont soil behaviour.
The obtained results will be impo researches of rainfall-induced lands deposits. The obtained soil properties useful for investigation of the rainfall in and landslide triggering, especially in c scale physical model of flysch slope, strength component associated with th may represent a dominant part of the shear strength along the sliding surface. frame of Project "Research Infrastructure Laboratories at the University of Rijeka", c the Ministry of Science, Education and Spo of Croatia and the European Fund for Reg (ERDF).