Water content (%) Dry density (kN/m3)
Chapter 5 Hydraulic characteristics of soil-WH composite
5.2 Soil-water retention and desiccation characteristics of soil-WH fiber composite
5.2.1 Test Setup and Procedures
Schematic test setup used in the current study is depicted in Fig. 5.1. All soil types (bare and soil-fiber composite) were compacted in a Poly Vinyl Chloride (PVC) mold, which has an inner diameter of 290 mm and height of 250 mm. The diameter was chosen to meet the requirement of preparing a representative elementary volume (137.5 mm in diameter) sample for studying crack development pattern (Li and Zhang 2010). The mold was placed on a perforated base plate, where a filter paper was placed to prevent loss of soil particle. All soil samples were statically compacted to 0.9 MDD by applying a vertical compaction force of 60 kN in the mold (Reddy and Jagadish 1993). The maximum dry density (MDD) of bare soil, coir fiber reinforced soil, jute fiber reinforced soil and WH fiber reinforced soil are 16.5kN/m3, 15.8 kN/m3, 15.7kN/m3 and 15.6 kN/m3, respectively. The OMC varied by less than 1.2% for all fiber reinforced soil. It was noted that the difference in standard Proctor compaction states of fiber-reinforced soil and bare soil was negligible. Hence, the compaction parameters of bare was used for fiber reinforced soils. The compaction level of 0.9 MDD is commonly used in embankment soil (Li et al. 2016). Before compaction, a thin layer of lubricant was applied on the inner surface of the mold to reduce soil- PVC interface friction. To achieve uniform soil density, the compaction procedure was divided into three layers with equal thickness. Initially, the soil was oven dried and fibers of 0.75% (by mass; typically used in soil-fiber composite applications) were dry-mixed to create a soil-fiber mixture. Initial water content of 17% (w/w) was sprayed on the soil-fiber mixture and the mix was placed in the mold for compaction. Despite of thorough mixing between the soil and the fibers, it is acknowledged that compacting the specimen vertically in the confined mold would make the fibers in the composite orientated predominantly sub-horizontally (Ibraim et al. 2012). The final
height of each sample was 170 mm. For each soil type (unreinforced; reinforced by coir, jute and WH), three replicates (in three separate molds) were prepared (i.e., 12 molds in total).
After soil compaction, the specimens were kept outdoor (exposed to natural environment), inside an open steel frame with a transparent roof for the entire monitoring period. This arrangement allows free air circulation over the specimen and the environment. The purpose of transparent roof was to eliminate direct precipitation while allowing sunlight to pass through it. A sprinkler system was attached on top of the soil column for applying controlled wetting. The system was connected to a Mariotte’s bottle so that a constant water head can be maintained during testing for producing controlled intensity and duration of wetting. Wetting events are not intended to relate with natural precipitation events but rather to stimulate crack closure and surface wetting of the soil types. A digital camera (model: Canon EOS 700d) with an adjustable frame was also mounted on top of the soil column to obtain periodic time-lapse images (method described in next section) of the soil surface for studying the crack development patterns during drying cycles. All suction and moisture sensors were installed diametrically at opposite ends at 40 mm soil depth, where cracks were expected to be developed. Although the exact crack propagation path and cracking depth were unknown prior to testing, the choice of sensors’ depth was based on laboratory observation from previous desiccation research on compacted silt (Cui et al. 2014, Song et al.
2016). It is recognized that it is not easy to use discrete sensors, like TEROS21 sensors in this study or tensiometers, for measuring suction at the air-soil surface due to stringent boundary conditions required. However, for better understanding of the behavior of crack evolution in relation to soil moisture regime, it is not uncommon to assume the soil suction and water content in the crack zone to be uniform during data interpretation (Trabelsi et al. 2012; Cui et al. 2014).
To prevent water leakage during each wetting event, a layer of anabond adhesive was applied to
gaps around the sensor installation holes. Both TEROS 21 (formerly known as MPS6) and EC-5 sensors were connected to an EM-50 data logger system (Decagon Devices 2016a) for data logging. The SWRC of each soil specimen was obtained by relating the suction and moisture content measured during the monitoring period.
After compaction, all the soil columns were wetted immediately. A total water volume of 1500 ml was applied, and it is equivalent to a rate of 6 mm/min and a return period of 100 years of rainfall in India (Guhathakurta et al. 2011) and Hong Kong (Lam and Leung, 1995). After wetting, suction ranges between 9 and 12 kPa in all specimens. During this wetting process, negligible soil volume change (less than 2% reduction of plane specimen area) was observed.
Subsequently, the columns were subjected to a continuous drying period of 7 days, when suction reached the range of wilting point (i.e. around 1500-3000 kPa; Feddes 1982, Cassel and Nielsen 1986, Lazarovitch et al. 2018) beyond which the natural fibers would have lost their mechanical strength against crack formation (Mesbah et al. 2004). After the drying event, all columns were wetted again, following the identical procedures described above. This wetting/drying cycle was repeated by 15 times for all soil columns. During the entire monitoring period, minimal soil volume change (both radially and vertically in the mold) was observed. The meteorological data, including solar radiation, air temperature and air relative humidity were monitored by a microclimate monitoring station (Decagon Devices 2016a). The climate monitoring station was placed next to the specimens at a distance of only 1.5 m. Since the steel frame where the specimens were tested was open to the natural environment, the climate data measured is thus representative of the conditions experienced by the specimen. Fig. 5.2 shows the variation of weather parameter during the 105-day monitoring period. The average relative humidity and temperature along with the error in mean were 82.6 ± 11% and 27.8 ± 6.80C, respectively, while the solar radiation varied from 10
to 20 MJ/m2/day. Camera images of the soil surface were captured daily to analyze the desiccation crack formed during all drying cycles.
Fig. 5.1. Schematic representation of soil column setup
Fig. 5.2. Weather conditions under which the soil columns are exposed