3.1 Preparation of Samples
The sample was oven dried for duration of 24 hours at a standard temperature of 105°C. After drying in the oven, the sample was sieved according to the requirement of the test standards. Bagasse ash collected as waste material from power plant is heated at temperature of 6500C termed as calcinations process to impart changes to the microstructure of soil upon amendment. The lime content and bagasse ash were adopted based on the prior studies by various researchers (Dash and Hussian, 2012; Ejune et al., 2017; Bahadori et al. 2019). To study the influence of lime on the BC soil blended with BA content, different amounts of BA by dry weight of the soil (5%, 10%, 15% and 20%) and 5% of lime were blended with the prepared dry BC soil.
3.2 Shrinkage characteristics of expansive soils
Accuracy of digital measurements is high when compared to manual measurements as the chance of crack initiation and minor cracks can be easily identified and measured which is successful in realistic shrinkage measurements even on repeatability. Advantage of considering minor irregular cracks is to provide accuracy and thus helps in best characterization of expansive soils (Puppala et al, 2004)
3.2.1 Manual measurements: All the manual measurements are made on Mitutoyo digital Vernier calipers with 0.01mm accuracy to compare the results obtained from the digital measurements. Measuring of average diameter and average height for each specimen after desiccation for 24 hours is done with the help of Vernier callipers with sharp edges to compare the linear shrinkage attained on image processing tools (Image J) and programming language (Phython).
3.2.2 Crack and volume measurements using programming language Python:
22.214.171.124. Data Acquisition: All the data obtained for the image study are obtained by photographs of specimens which are taken at a constant distance to the image at same focal lengths and the same images are used for image processing by Image J to avoid dissimilarity in measurements. Experimental platform and analysis includes the hardware that includes Laptop with Intel(R) Core(TM) i5-8250U CPU @ 1.60GHz, 64-bit processor with 8GB RAM. The software environments are python 3.9.7, VGG tool, and Open Computer Vision running on Windows 10 OS 21H1. The experiments are performed on data set of 8 images which were photographed at same focal lengths.
126.96.36.199. Pre-processing: Essential part of working in Python involves in importing libraries of numerical python, cv2 and matplot for displaying and processing images for computations. Pre-processing step in our method involves in removal of black background from the raw images. The raw images are read using Open CV python library. It includes reading image, creation of white background, binarize masks, blur alpha mask and alpha-blending to merge the foreground and background images. Reading image involves in adjusting the image size so that it matches the r channel of RGB image which is done by DeepLab V3 and by calling decode_map function. This operation is followed by creating a NumPy array of ones with the same shape as the RGB output map obtained from DeepLab V3 to create a white background and scaled to pixel intensity of 255 which is white background.
Binarized mask is generated by converting both the foreground and background of image to float type from uint8 and cv2.threshold function used to compare the pixels against the threshold value. Binary thresholding is the simplest form of global thresholding where each pixel location (x, y), the pixel intensity at that location is compared to a threshold value. The input code applies an adaptive threshold to an array. This is the function that transforms a gray scale image to a binary image according to the formulae:
Post threshold, background pixels have a value of 0 and a foreground pixels with value 1 which is essentially cracked part in the input image. The output of the threshold function is stored in the alpha. In order to take care of the quality of picture at edges a Gaussian blur filter is applied before using the masks to avoid undulating edges. Alpha mask is normalized to have an intensity between 0 and 1 by alpha blending foreground and background for this multiply foreground with alpha and background with the 1 – alpha respectively and the actual merging happens between masked foreground and background before displaying normalized output .
segment(dlab, './images/bgremoval/5BA.png', show_orig=False)
188.8.131.52.1. Coordinates generated through sublime editor:
184.108.40.206. Area computations using NumPy
Numerical python is used in area determinable functions from python library. As the images are represented in form of pixel elements with intensities in matrix form, NumPy library is used as it supports adding multi-dimensional arrays and matrices and high-level mathematical functions to operate on these arrays. Area coordinates fetched from the above process are used for the generation of crack area. The area of shrinkage dish and the pat are determined for volumetric analysis of the soil.
3.3. Digital measurements using Image J:
Novelty of this method lies in applying the study to the samples of shrinkage limit where the time can be saved and the same sample can be analysed for the volumetric shrinkage parameters without need to cast analyse and study specimens at larger scale (Manuel et al., 2021). Moreover the study is compared with the imaging software python for analysing shrinkage parameters and the impact of bagasse ash in volumetric shrinkage strains is evaluated. Lime content is varied in the preliminary image analysis of the specimens, where parameters like volumetric shrinkage strains (VSS), change in volume and linear shrinkages are determined. Dosages of lime content are tested for 0%, 1%, 3% and 5% in weight percentage for volumetric study. Raw images as input images after pre-processing to remove noise are thresholded to find the required volume as listed in Table 6 with corresponding VSS in Table 7. Based on the preliminary studies, conclusions are drawn for the major study in fixing the dosage of lime to 5% in order to study the reactivity among clay-bagasse ash blended mixes.
Image processing technique is performed by tool named Image J from NIH to study the volumetric and fractographic images of soil specimens adjusted to suitable scale. This suggests a novel approach in determination of shrinkage parameters occurred in expansive clays due to desiccation. The initial proportions which decide volumetric shrinkage strains such as Surface Area Ratio (SAR), Perimeter Ratio (PR) and Circular Area Ratio (CAR) images which are determined from image processing technique. Accurate crack width measurements that even measure minor shrinkage cracks helped in exact evaluation of shrinkage strains thus adopting digital image processing helps in choosing best stabilisation strategies for characterisation of expansive soils (Puppala et al., 2004).
Images are pre-processed so that for analysing them for the assessment of required parameters becomes easy (Patrick and Chakravarthy, 2020). Functions like binary threshold are utilised for the enhancement of image clarity in identifying the crack width. Threshold function helps in highlighting the region of interest making the background dark by setting a threshold value to measure crack area. Region selection tools are used to find the geometrical parameters such as area, perimeter, volume and diameter which are essential in determination of Volumetric Shrinkage Strains (VSS). Comparable chart of operations on raw images for computing volumetric shrinkage entities are listed in Figure 3.
3.4 Morphological and micro-structural characterizations
3.4.1 Field Emission Scanning Electron Microscopy (FE-SEM)
Influence of the waste material in imparting changes to mechanical and physical properties of soil is difficult to explain without keen observation of microstructural changes that occur within the soil matrix. Field Emission Scanning electron microscopy (FE-SEM) characterization is used to evaluate pre-treated and post treated soil samples by lime and bagasse ash for comparative analytical studies of soil samples. Hence SEM analysis of high resolution are required to keenly observe the changes that takes place.
To understand the microstructure and nano-particle structural distribution of soil, FE-SEM analysis is performed in thermo Fisher FEI QUANTA 250FEG for morphological understanding of different mixes adopted. The instrument is equipped with Schottky field emission electron gun as source of electrons with operating range 5kV-30kV offering high resolution of 1.2 nm at 30kV at high vacuum conditions adjoined with Everhart thornley detector for secondary electrons at high vacuum for coated samples. The sample preparation is done by placing 10 milligram of dried sample on stubs. Oven dried soil is selected to eliminate the charging effect on the microstructure and to obtain better illuminated image with high resolution. To avoid this, all the samples are sputter coated in Quantum Q150T S plus either with chromium or gold as the soil is non-conductive. Surface morphology and fractographic information helped in assessing the microstructural modification with the addition of stabilizer and which discusses change in shape and size of the particle.
3.4.2 crystal structure and mineralogical modifications of clay with proportional lime and bagasse ash by X-ray Diffractionn (XRD) analysis
Peak alterations in intensities with respect to the diffraction angle are keenly observed through illustrations of X-ray powder diffraction (XRD) of all the 8 mixes along with bagasse ash and expansive clay. Black cotton soil is primarily analysed to match with the other mixes to assess the rate of pozzolanic reaction happened. The prominence of the angle between incident ray and diffracted ray diffraction between 15°to 35° is reported by researchers for bagasse ash in deciding the mineralogy and crystal structure.
3.4.3 Fourier Transform- Infrared Radiation (FT-IR)
The amount of light that is transmitted through the elemental matrix of soil sample is plotted against Wave numbers. Expansive soil elements are analysed to check the variation in structure through amendment which is sensed through bond formation and distinct elemental vibrational behaviour to infrared radiation pertaining range 400–4000 cm_1.
The tetrahedral and octahedral sheets in clays are bonded by weak inter layer bonds and the presence of water molecule is the main reason for the expansiveness of soil minerals. Clays can only form bonds at the edges due to large surface area of clay exchange of cations adhesion to the clay particle due to difference in ion concentration. Soil Samples with addition of industrial waste such as bagasse ash and lime as stabilising agent are subjected to infrared radiations to inspect the chemical bond formation and reactivity between them along with soil mineralogy. Curing time is fixed as 7 days for the physiochemical changes to happen between the admixtures.