Development And Application Of The Quantitative Analysis System For Soil Crack Network And Microstructure

Soil desiccation cracking is a common phenomenon in nature.The developments of desiccation cracks in soil will destroy the complete structure of the soil,weaken the physical and mechanical properties of the soil and cause damage to the structure of rock and soil which will lead to a variety of engineering geological problems directly or indirectly,consequently causing serious economic loss.Therefore,it has great instructional significance to quantify the cracking network of soil and master the laws of cracking development and the influence of desiccation cracking on engineering properties.In this paper,image processing technologies are applied on the MATLAB platform to develop the software:Crack Image Analysis System of Nanjing University(CIAS-NJU)based,which can be used for crack image processing,automatic identification and quantitative analysis of crack networks and soil blocks to obtain various geometric information of the crack networks and blocks in crack images.At the same time,wetting and drying(W-D)cycle tests were carried out to have a quantitative study on the laws of cracking development during W-D cycles.Unconfined compression tests were carried out to measure the influence of desiccation cracking networks on the engineering properties.Computed tomography(CT)technology is an effective means for observing the internal structures of geological materials.Thus,some CT tests were carried out in this paper to measure the geometrical characteristics of internal crack patterns in the specimen quantitatively during a drying procedure.CIAS-NJU was applied in these researches and some achievements have been obtained.The soil microstructure always reflects the formation process and depositional environment of soil,so that it is an intrinsic factor that determines the engineering properties such as soil deformation and strength.Therefore,it is of great significance to study the macroscopic physical and mechanical properties of the soil by conducting a quantitative study of the microstructure of the soil.In this paper the results of predecessors were summarized and image processing technologies were applied with MATLAB software to develop the software SEM Image Processing and Quantitative Analysis System of Nanjing University(NJU-SEM)for the processing of SEM image and the automatic identification and quantification of particles and pores in SEM images.Therefore,various geometric information of soil particles can be extracted from the SEM image with NJU-SEM.The NJU-SEM has been applied successfully in the microstructure of the loose sandy soil from the actual project and microstructural parameters were extracted to analyze the spatial differences of the sandy soils in different strata.The dissertation was completed under the financial support of Professor Tang Chaosheng's National Excellent Young Scientists Fund(41322019)and the National Natural Science Foundation of China(41572246,41772280).A large number of results were obtained in the paper,some main achievements are shown as follows:(1)The CIAS-NJU system has been developed independently includeing two image processing modules:single image processing and massive images batch processing.The processing of a single image includes three steps:image preprocessing,clod identification and quantification,and crack networks identification and quantification.The cracks and clods can be segmented by a series operations such as graying,binarization,and denoising,which are functions of the image preprocessing before geometrical information extracted.The identification of cracks requires the some operation to be done first,such as crack boundary smoothing,skeletonization,and branch cropping operations.The extraction of the soil and crack information is achieved by using a labeled connected region method.A quantitative indicator system for assessing the characteristics of clods was proposed,such as soil block area,soil block perimeter,number of soil blocks,ratio of area and perimeter,shape coefficient,form factor,equivalent side length and equivalent diameter.A quantitative index system was proposed for the evaluation of the geometrical characteristics cracking networks,such as crack length,crack width,crack area,surface crack ratio,angles of bifurcation,number of crack segments,number of intersections,number of endpoints and fractal dimension.At the same time,statistical description-probability density function is introduced to describe and analyze the distribution characteristics of crack network parameters.The procedure of image batch process is the same as the single image,but it's requires images at the same format under a folder to be read and the image processing threshold to set before the system automatically completes the processing of all images and saves the data and images in the procedure.The digital image processing method effectively overcomes the disadvantages of the traditional measurement methods and improves the efficiency and accuracy of the crack pattern morphology parameters.CIAS-NJU provides a variety of binarization methods and redundant processing methods,meanwhile two batch processing methods are integrated to meet the actual needs of users in various aspects.(2)W-D cycle tests were carried out on compacted expansive soil samples and CIAS-NJU was used to quantitatively analyze the cracking evolution characteristics of soil during the W-D cycles.The results showed that:the development of desiccation cracking in expansive soils tends to increase as the water content decreases in the drying process.The surface crack ratio,the number of crack segments,the total crack length and the shrinkage of the soil increase with decreasing moisture content.There is no consistent relationship between crack width and water content,which may due to the shrinkage of the soil that leads to the closure of the cracks in a certain degree in the later drying period.At the same time,W-D cycles have a significant effect on the development of cracks.With the increase in the number of W-D cycles,the surface crack ratio,the number of crack segments,the total length of cracks and the shrinkage of soil volume increase,while the average crack width decreases,which mainly is a result of a large amount of micro-cracks developed during the thirddrying process in the sample.(3)The unconfined compression tests on soils were carried out under W-D cycling conditions and CIAS-NJU was used to quantitatively analyze the crack development in the soil.The results showed that:W-D cycles,moisture content,and crack development all affected the unconfined compressive strength.Unconfined compressive strength,initial stiffness and brittleness of the specimen gradually decreased with the evaporation of moisture during drying.The increase in the number of W-D cycles will lead to the attenuation of the unconfined compressive strength of the soil,and the intensity of the attenuation after the first cycle is the most obvious.This phenomenon is more significant at the low moisture content.At the same time,the stiffness and plasticity of the sample subjected to the W-D cycles increase,which is related to the soil structure adjustment and internal crack development in the W-D cycles.There is a clear linear relationship between the increase in the number of crack segments and the attenuation of the strength of the specimen,but there is no obvious relationship between other crack parameters and strength.In the follow-up study,the quantitative relationship between cracks and strength should be further studied under the control of crack development..(4)Based on the CT test,the development of internal cracks in the sample during the drying process was studied and CIAS-NJU was used to quantitatively analyze the CT images obtained from the tests.The results showed that:there is a certain spatiotemporal feature in the development of desiccation cracks in the sample during drying process.The parameters such as crack ratio,crack area,average crack width,total crack length,and number of crack segments increase with the moisture content decrease during the drying process.The development of cracks shows differences at different depths that there are several depth intervals where intense desiccation cracking developed which indicates the uniformity and stability of the soil structure are destroyed.Macroscopic physical properties of soil such as the soil volume,porosity ratio,saturation,density,and dry density soil were calculated by the CT scanning results and manual measurements.It's found that the results of CT image quantitative analysis based on image processing techniques are more accurate than the results of manual measurements.Meanwhile,comparing the physical indices calculated from the CT images with cracks and without cracks,it is found that the development of cracks has a significant influence on the physical properties of soils.The void ratio of the sample calculated without cracks are smaller than that with cracks,while the saturation,density and dry density calculated without cracks are larger than those with cracks.(5)The SEM image processing and quantitative analysis NJU-SEM system was developed independently,mainly including three modules:image preprocessing,soil particle identification and quantification and statistical analysis.The process and principle of the image preprocessing and soil particle identification and quantification are similar to the preprocessing and clod identification and quantification in CIAS-NJU.The soil particles and pores are segmented to extract the information of soil particles by the processes such as graying,removing redundant information,binarization,opening operation and removing noise.To facilitate the processing of statistical information,the NJU-SEM system provides a statistical analysis module including several operations like data selection,sorting,screening,and calculation of probability density functions in various ways.By summarizing the quantitative research results of soil microstructure,an index system for quantitative evaluation of soil particle structure was proposed:particle area,particle perimeter,number of particles,equivalent length,equivalent diameter,shape factor,morphology ratio,apparent porosity and fractal dimensions.(6)In terms of loose sandy soils,the technique of sample preparation using the vacuum freeze-drying method and the plastic injection method was proposed.It can effectively ensure the original microstructure not to be disturbed and the ideal sample for scanning electron microscope observation can be obtained.The NJU-SEM system was used to quantitatively analyze the SEM images of undisturbed sands taken from different strata and the microstructure parameters of the particles were extracted.It is found that there are certain relationships between different parameters.There are significant spatial differences in the microstructure parameters between different soil layers,which are mainly affected by the formation history and the depositional environment of the soil.These differences in microstructures determine the macroscopic physical and mechanical properties of different soil layers to some extent.