Introduction
The road subsidence on the underground pipe is caused by two reasons in most cases. One is the progressive erosion of soil on water inflow and drainage cycles from the defective part of the underground pipe and continuously the propagation of soil erosion is extended to the road surface. The other is that the underground pipe is aged and damaged suddenly with overburden pressure during the aging process to the concrete materials [
1-
3]. Movement of soil particles on erosion process due to water inflow and soil drainage is clogged by arching effect of due to rearrangement of the soil particles when the soils loses the effective stress at the process of the water inflow and soil drainage. However, the large-void space is progressively developed in the road base [
4-
11]. Eventually, the road with inner eroded soil should be collapsed by overburden pressure of road base/traffic road or lack of bending loads of asphalt road. In these two decades, the number of road subsidence has been increased [
12,
13] and they were mostly caused in the urban area. From other view point, the aging of concrete material should be chemically progressed and the increase of the defective underground pipe was accelerated by chemical effect except for the damage with earthquake. National institute for land and infrastructure management (NILIM) summarized the number of the accidents with road settlement in Japan. Reynolds and Barrett [
14] also summarized an age profile of sewers in England and Wales. Figures 1(a) and (b) show their summary obtained from NILIM 2006. The number of road subsidence increased from 1980 almost linearly till 2005. These data almost correspond to over 40 ages of underground pipes installed. Besides, the road subsidence was caused in rainy and summer season in Japan as shown in Fig. 1(b). During heavy rainy season, the rain water flew into the underground pipe as sewerage pipes and the pipe would be filled with rain water with high pressure. Then, it can be considered that water infiltrated into the ground or were drained through or into the defective part.
The significant factor in the project design of infrastructures should be material durability or serviceability. In fact, American Association of State Highway and Transportation Officials (AASHTO) recommend that a life cycle analysis should be performed for a material selection [
1,
15]. A material ability to perform its indented functions, structural, and hydraulic in the case of pipe, becomes irrelevant if the material cannot perform satisfactorily for an economically acceptable period. Hence, the issue of defective underground pipe should be more focused before critical damage is caused. The city official has agreed to begin making repairs but there is much concern as to the financial cost. Therefore, the key issues in this topic are 1) to find the defected part of the underground pipe and the cavity of the ground under the traffic road, 2) to avoid the fatal road subsidence, and 3) to propose the countermeasure method to the cavity generation of the ground under the traffic road.
Mukunoki et al. [
10] performed a series of model tests using Toyoura sand subjected to monotonic water inflow test, monotonic soil drainage test and the cyclic test of the water inflow-soil drainage to visualize the ground failure due to each hydraulic application using X-ray computed tomography (CT) scanner. The density distribution of the engineering materials such as soils and rocks can be visualized in three dimensions without any destruction by X-ray CT scanner. The interest conclusion was that the factors causing the cavity in the ground were water flow toward the defect part after the lost of capillary force. This conclusion indicates that the lost of fine grains due to water inflow-soil drainage cycles might cause the fatal failure of road base. To clarify the key factors on the failure mechanism to cause collapse of road base, authors focused on 1) relative density and, 2) grain property of the backfill soil and 3) the relation between failure degree and the dimension of crack width of defective part and soil properties in this paper as the continuous study of Mukunoki et al. [
10].
The objective of this paper is to clarify the effect of relative density and grain property of soil to failure mechanism due to water inflow and soil drainage from defected underground pipe using X-ray CT scanner. In particular, a radiation slot along the direction of diameter on the underground pipe was prepared in the model test as the defective pipe since the number of cases was the greatest as shown in Fig. 1. Besides, two key issues based on performing the model test with cyclic water inflow and soil drainage are focused as follows; 1) how relative density of model ground, and 2) how fine grains affect to the failure of soil. In this paper as a continuous work of the previous study in Mukunoki et al. [
10], the model tests using soil as a common soil with greater uniformity coefficient than Toyoura sand were conducted with same the approaches as Mukunoki et al. [
10].
Test method
X-ray CT method
X-ray CT scanner and X-ray CT image
Table 1 shows the specification of X-ray CT scanner used in this study. A voltage of X-ray tube is related to the intensity of X-ray beam into the specimen. Meanwhile, the current is related to the number of photon generated in the X-ray tube. To avoid causing X-ray beam hardening (e.g., Kethcham and Calson [
16]) in the X-ray CT image with respect to the intensity of X-ray beam, the combination of 300 kV and 2 mA was selected in this study. The thickness of the X-ray beam was collimated as 1 mm. All specifications of X-ray CT scanner used in this study were able to be referred in same as Mukunoki et al. [
10].
The diameter of the scan area was selected
ф = 150 mm and the number of voxel in the image was 512 × 512; hence, the dimension of one voxel was 0.293 × 0.293 × 1 mm
3 because what the diameter of 150mm in scan area divided by 512 was 0.293 mm. An X-ray CT image is composed of CT-value in each voxel with proportional to the material density and a digital image with a gray-scale image with 256 levels between black and white colors. Black area in the two-dimensional (2-D) image indicates the lowest density area, while white color means the greatest density in the CT image. 3-D CT image can be reconstructed by convoluting these 2-D images. All typical issues about use of the industrial X-ray CT scanner can be referred in Otani et al. [
17] and Kethcham and Calson [
16].
Test apparatus
Figure 2 shows a view of an apparatus for model test and Fig. 3 illustrates schematics of a test apparatus and a defected model tested, which was exactly same as what Mukunoki et al. [
10] used. In this study, a model underground pipe was installed at the bottom of the model ground and defective type was a type of a radiation slot as shown in Fig. 3(d). The radiation slot along the direction of diameter on a half pipe model with two different widths of 5.0 and 2.1 mm was installed at the bottom of the soil box to evaluate the influence of crack width and grain size to the failure degree. A slot was established in a half pipe model so it was concerned about the loss of soil particles through the defected part in the underground pipe during the process of soil preparation in the different soil box. To avoid the extra mass losses of the soil, sugar with natural moisture contents was well placed in the model pipe at first and then, the soil was compacted from the top surface of the pipe model [
10]. Once water was flown through the underground pipe in the first step from the bottom of soil box, sugar was melt gradually and therefore, water could be seep to the ground without any mass losses of soil particle in the initial condition. The overburden pressure of 10 kPa was applied through the plate placed at the ground surface by air pressure to take account the confining pressure in the soil because the underground pipe was usually installed at the depth of 1 m [
13]. Also there were holes in the plate to be able to drain water from the ground surface so the overburden pressure applied indicates effective stress.
Test cases and procedure
Sandy soil was used with regulating relative density and distribution of grain size for each case of the model test. Figure 4 shows the grain distribution curve of soils tested. The initial moisture content was regulated as the values of 11%. The basic concept of the model test conducted in this study is to apply the cyclic water-supply and drainage from the defected part. To understand the soil behavior due to the condition of water inflow and model ground, the two kinds of test condition (i.e., difference of relative density and grain size distribution) were prepared as shown in Table 2. The parameters to compare the test results are defined as follows:
1) Relative density: Case 1, Case 2 and Case 3;
2) Grain size distribution regulated: Case 4, Case 5 and Case 6;
3) Crack width of defective part: Case 7.
Relative density
The first item would give information how the initial condition of backfilling at the installation process affect to the failure of soils. To evaluate the initial density condition three different relative densities were considered in this study: 80% as Case 1, 40% as Case 2 and 60% as Case 3, and then the uniformity coefficient and the coefficient of curvature were 4.2 and 2.3 for each case.
Grain size distribution and crack width of a defective pipe
The second and third items should be needed for the case that the crack width of the defective pipe is similar to the mean diameter of the soil used for backfilling. It is difficult to imagine what the underground pipe has a huge hole as long as the underground pipe made of concrete. The aging of concrete should be progressed with time, so the defective part is also generated gradually. Probably, the dimension of the defective part would be related to drainage phenomena of soil subjected to grain size distribution. If the crack width of defective pipe is less than mean diameter of soil, the degree of ground collapse under the road should be small; however, if not, the soil drainage should be progressed and it causes the fatal failure of road embankment. In this case, water was supplied and soil was drained repeatedly through the defect part on the underground pipe. To be concerned about this potential issue, 3 kinds of maximum grain size (850 m as Case 4, 2.0 mm as Case 5 and 4.75 mm as Case 6) were regulated as shown in Fig. 4. Also, two different crack widths of defective part (5.0 mm as Case 1-6 and 2.1 mm as Case 7) was prepared for three different maximum particle sizes as shown in Fig. 4. The uniformity coefficients were 13.9, 18.0 and 21.3, and the coefficients of curvature were 1.7, 2.1 and 2.5 for the tests with changing the maximum grain size.
Test procedure
The model ground was scanned to confirm the initial condition and then water of 100 mL was supplied by the hydraulic pressure of 10 kPa. Once water of 100 mL was supplied into the model ground, the water valve as shown in Fig. 3 was closed and then the model ground was scanned to investigate the behavior due to the water inflow. After scanning, the water valve was opened and it was kept until no dropping water with soil was confirmed. Again, to observe the inner condition after water draining effect, the model ground was scanned. This process was defined as one cycle, and was repeated until the surface of the model ground was submerged.
Calibration line between CT-value and bulk density
Figure 5 shows the calibration line to convert CT-value to dry density of model ground in the soil box used. As shown in Fig. 5, CT-value is proportional to the material density and, the CT-value change of 100 indicates the density change of 0.1 t/m
3 [
17]. Because of the beam hardening issue with respect to X-ray physics [
16], when the calibration line is produced, the sample should use same dimension of a soil box as model test. As long as the calibration line is applied to the image analysis of X-ray CT, some fundamental parameters of soil can be obtained from CT value.
Results and discussions
Comparison on the relative density of model ground
As the test result for Case 1, Fig. 6 shows cross-sectional X-ray CT images of the model ground with relative density of 80% after 1st and 100th cycle of water inflow and drainage test. Mean CT-value of model ground at the initial condition was approximately 450-500. As shown in Fig. 6(a), the mean CT-value increased to 560-650, and it indicates that water went into the model ground so the wet density of model ground was increased. However, the mean CT value at 1st cycle was almost same as that at 100th cycle as shown in Figs. 6(a) and (b). This means that soil particles were not almost moved because of well interlocking enough since the relative density was 80%. Eventually, the model ground with relative density of 80% was not destroyed due to cyclic behavior of water inflow and soil drainage. Mukunoki et al. [
10] conducted the same test using Toyoura sand with uniformity of 1.3 and its model ground failure after 2 cycles of water infiltration and soil drainage. Soil particles were well interlocked because the uniformity coefficient was at least 2.4 times greater than one of Toyoura sand used in Mukunoki et al. [
10] so it can be considered that the seepage force could not make soil particles transferred. Therefore, the cavity was not generated in the model ground. These results concluded that the regulation of backfilling soil should be conducted severely with high-relative density at least
Dr = 80%.
As the test result for Cases 2 and 3, Figs. 7 and 8 show X-ray CT images of model ground with relative density of 40% and 60% at the different height in the last stage of the cyclic test. In particular, the maximum grain size of Toyoura sand was 425 m, and on the other hand, that of sandy soil regulated was 2.0 mm in this study. Hence, the factors of heterogeneity are variability of 1) shape of grain, 2) spatial distribution of grain size in the model ground and 3) spatial distribution of density locally more than the ground of Toyoura sand. Then, cavity area was grown continuously along the height of the model ground from the model pipe and the large hole was observed around the top of the model ground. In interest, the valve for the soil drainage was plugged on the model pipe with well enough interlocking phenomena because of the rearrangement of soil particles. X-ray CT images at the height of 30 and 50 mm as shown in Fig. 8 looks similar to those in Fig. 7.
Figures 9(a) and (b) show 3-D X-ray CT image of cavity area in the model ground at the last stage of model test for Cases 2 and 3, respectively. The cavity volume for Case 2 is greater than that for Case 3. As for Case 3 in Fig. 9(b), it can be observed that the large cavity was formed uncontinuously from the model pipe. In the last, the continuous cavity would be formed as a large hole of ground surface to the Case 3. However, the number at the last cycle of the test for Case 2 was different from Case 3; hence, the difference of cases between the relative density 40% and 60% would be affected to the degree of interlocking of each soil particle and the process of the cavity generation in the model ground.
Quantitative analysis of X-ray CT images
Figures 10 and 11 show the cross-sectional images of the model ground at the 27 mm height from the bottom with relative density of 40% and 60% respectively at the last stage of cyclic process. Authors defined that the inside of solid line was region
A and its outside was region
B to distinguish the region in the CT image. Focused the inside of solid line area as shown in Fig. 10(a) and Fig. 11(a), the area of region
A became a large hole at the drainage stage as shown in Fig. 10(b) and Fig. 11(b). The mean CT-value of region
A in Fig. 10(a) was 462 and its standard deviation was 214. Also, the mean CT-value at the same area at the initial condition was 169 and its standard deviation was 309. These CT-values indicate that the bulk density of model ground as shown in Fig. 10(b) was increased from the initial condition. If the pore structure in the cross section of model ground at the 27 mm height was not changed due to the water infiltration, the factor of increment of CT-value should be condideredby two reasons as follows; pores was filled with water and fine grains. The subtraction of CT-value is more than the value of 300 and this indicates that the bulk density is increased approximately 0.3 t/m
3. The both of CT-value in regions
A and
B increased almost same density according to the CT-value analysis. The model ground at the initial condition had the 11% moisture contents and dry density of 1.5 t/m
3; hence, the initial saturation degree was 38%. Assumed that the saturation degree of model ground at the second cycle becomes 100%, the wet density is 1.9t/m
3 in case without change of pore structure. If so, the CT-value should be 560 because of the 0.4 t/m
3 increase of bulk density. In fact, the mean CT-values in regions
A and
B were 462 and 477 in Fig. 10 for Case 2; however, the standard deviation of the region
A decreased more than the value of region
B. The reduction of the standard deviation in region
A means that the region
A was homogenized. The key issue in these results is that the bulk density in each region
A and
B was increased but the variation of CT-value was different in each other. Therefore, the mechanism of density change in region
A can be distinguished from the region
B. Otani et al. [
18] discussed that variation reduction of the standard deviation to mean CT-value in one region meant that its one region was homogenized as CT-value and the pore was filled fine grains. Same trend could be observed in Fig. 11 as Case 3 with relative density of 60%. Once water flew into the model ground, the seepage force carried fine particles to upward locally in the model ground; and then, the soil particles just above the model pipe would lose the effective stress due to decreasing the contact stress with losing fine particles at the process of water inflow. At the process of the soil drainage, the soil particles were carried into the defective part, and finally the large cavity was formed.
It is concluded that the soil particles in the model ground with relative density of 80% and the uniformity coefficient greater than 3 was not transported easily because of well interlocking between coarse and fine particles even though the water inflow and drainage were repeated 100 times in the model ground. The cyclic number of this test should be related to the water pressure in the infiltration process. The viewpoint of this result is to reduce the effective stress due to fine particle transportation progressively and eventually to lose the effective stress near the defective part and the cavity was formed in the ground on the defective part.
Relationship between defected width and maximum grain size
Figures 12-14 were a series of X-ray CT images at the initial and the 2nd cycle of water-inflow and soil drainage in Cases 4, 5 and 6. In Case 6, the maximum grain size i was 4.75 mm so that it is not meaningful to define the relative density because the relative density is defined to the soil with the maximum particle size is less than 2 mm in the criterion of Japanese Geotechnical Society (JGS). In this section, the density condition for cases 4, 5 and 6 is fixed as dry density of 1.41 t/m3. The key issue of this section is the ratio between maximum grain size (Dmax) and crack width (B) of underground pipe. As for Case 4 and Case 5, the ratio of Dmax/B is 5.9 and 2.5, respectively. On the other hand, it is 1.05 for Case 6. As for Cases 4 and 5, the model ground was collapsed at the 2nd cycle of water infiltration and soil drainage test. However, the cavity area for Case 6 was observed at the 13th cycle as shown in Fig. 14. These images result that the ratio between Dmax and B of underground pipe is a significant factor for the collapse level of the ground. Until the 12th cycle, there was almost no observation in X-ray CT image for each height; however, it was found that fine particles localized on the model of underground pipe. The cyclic behavior of water inflow and drainage affect to the seepage force to the fine particles and then, they were drained eventually; but the largest particles could not be drained easily because the ratio of Dmax/B is only 1.05. Hence, the fatal collapse of model ground for Case 6 which was shown in Cases 4 and 5 was not observed.
To discuss the movement of fine particles due to cyclic water infiltration and drainage, the crack width was changed from 5 to 2.1 mm as Case 7, which the ratio between Dmax and B of underground pipe is 1.05, and hence, the maximum grain size was regulated 2.0 mm. Figure 15 shows the X-ray CT images of model ground at initial, 19th and 23rd cycles in each height. It can be observed that fine particles were localized around the center of the model ground along the defect model of underground pipe. As shown in Figs. 15(b) and (c). However, the large cavity such as shown in Figs. 12-14 was not confirmed . Eventually, there was no critical collapse of model ground although the ratio between Dmax and B of underground pipe is the same as Case 6.
Figures 16 (a) and (b) as B/Dmax of 5.9 and (3),(4) as B/Dmax of 1.05 show X-ray CT images for Case 7 at 40 mm height of the model ground at each process of 2nd cycle. To emphasize the fine particle area, window level (as herein as WL) of all images were different from WL of X-ray CT images as shown till Fig. 15. In case of B/Dmax of 1.05 for Cases 6 and 7, there was no significant change in the process of water inflow and soil drainage as shown in Figs. 16(c) and (d) as B/Dmax of 1.05. This indicates that soil particles around the defective part could not move out from the defective part. Meanwhile, the different observation was seen in Figs. 16(a) and (b) as B/Dmax of 5.9. These different results indicate that the cyclic number of water inflow and soil drainage is significant factor, in particular, whether or not soil particles can be moved out from the defective part at the 1st cycle should fix the generation of cavity. Figures 17 and 18 analyze CT-values in X-ray CT images for Cases 4 and 5 as shown in Figs. 16(b) and (c). CT value of areas A at the cycle of water inflow as shown in Fig. 17 is approximately 100 less than area B but the standard deviation in are A was 60 less than area B. In obvious, the density condition in area A is less than area B and some part of area A became a cavity as area C as shown in Fig. 17(b). Area C is 19.6% of area A as shown in Fig. 17(a) and the mean CT-value of area B did not change. Moreover, the standard deviation of the region A in Fig. 17(b) was greater than the region A in Fig. 17(b). Namely, the region A in Fig. 17(b) should be changed progressively to the cavity because the increase of standard deviation and the reduction of CT-value means that voxel was occupied with not soil particles but void. Similar trend was obtained from Fig. 18.
Figures 19-22 show 3-D X-ray CT images of the cavity area generated in the model ground for Cases 4, 5, 6 and 7, respectively. As for Case 6 shown in Fig. 21, the cavity area was smaller than that for Cases 4 and 5. The fine particles are rearranged due to cyclic water infiltration and drainage so that some particles were interlocked and hence, the large and continuous cavity such as shown in Figs. 19 and 20 was not generated. In similar, there was almost no cavity generation in the model ground for Case 7 as shown in Fig. 22. From the view point of the grain property for Cases 4, 5 and 6, the value of B/Dmax is summarized for Cases 4, 5, 6 and 7 in Table 3. It can be recognized that the value of B/Dmax for Case 4 is 5.9 and 100% of soil particles can pass through the defected part in Case 5. Meanwhile, 75% of soil particles with that the value of B/Dmax is 5.9 can pass through the defected part for Case 5. However, the pass percentage of soil with the value of B/Dmax more than 5.9 for Case 6 and Case 7 is less than 35%. If the soil particles with the value of B/Dmax more than 5.9 were more than at least 75% of the ground, the large and continuous cavity would be generated.
Practical imprecation
Once the underground pipe has fatal cracks such as more than crack width or hole of 30 mm, most of soil particles installed as backfilling would get into the underground pipe and the collapse of road base would be caused such as Cases 4 and 5. As for Table 3, despite that the value of
B/Dmax is 1.1, the cavity area was observed as the result of Case 6 in Fig. 21. Since this study focused the defected part on the connection part of underground pipe, the configuration of defected part was radical slit. In general, gravels with 13 mm in mean diameter were placed around the connection part of underground pipe [
2]. Meanwhile, it has reported that the defected width at the connection of the underground pipe is 10-20 mm [
13]. Case 6 or Case 7 may model the above real condition. To avoid the critical settlement of the road surface, it is important to perform the scan to the road surface to discover the aged underground pipes and replace them. Moreover, it would be effective to backfill the gravels with more than 20 mm in mean diameter on the underground pipe at the reinstallation work.
Conclusions
In this paper, the model tests were performed to evaluate the mechanism of the cavity generation in the model ground of sandy soil with regulating relative density and distribution of grain size due to water inflow and soil drainage cycle. As key parameters, the relative density and grain properties such as uniformity coefficient and the coefficient of curvature of model ground were focused in this paper. The conclusions summarized are listed as follows:
1) The uniformity coefficient and the coefficient of curvature of soil were key parameters about the cavity generation of the model ground due to water inflow and soil drainage.
2) The model ground with high relative density at least greater than 80% was not severely collapsed due to the cyclic behavior of water inflow and soil drainage.
3) The model ground with low relative density less than 60% was failed at the number of few cycles so that an efficient compaction is somehow important.
4) Fine grains were transported with seepage force at the cycle of water inflow, and the effective stress was lost around the defective model pipe; then, those soils were well drained due to the drainage cycle and eventually, the large cavity was formed in the model ground.
5) The greater ratio (B/Dmax) of crack width (B) on the defective pipe and maximum grain size (Dmax) was 5.9 and the pass percentage of soil was at least more than 65%, the worse the model ground was collapsed.
Authors and Mukunoki et al. [
10] have discussed failure mechanism of model ground with single defective part of model pipe analyzing X-ray CT images. As the future work, two issues would be concerned. One is to enlarge the dimension of model ground in order to discuss the effect of variation of defective part such as more than two holes in the model pipe and the other is a numerical simulation which can model soil erosion due to water inflow and soil drainage. The results based on both ideas would give more quantitative discussion about the failure mechanism of soil with soil-water interaction subjected to defective underground pipe.
Higher Education Press and Springer-Verlag Berlin Heidelberg
PDF
(1906KB)
