Settlement of Funing Natural Clays under 14-Year Embankment Loads : A Case Study

Ph.D. Candidate, Institute of Geotechnical Engineering, School of Transportation, Southeast University, Nanjing 210096, China Professor, Ph.D., College of Civil Engineering, Fuzhou University, Fuzhou 350108, China Associate Professor, Ph.D., Institute of Geotechnical Engineering, School of Transportation, Southeast University, Nanjing 210096, China Professor, Ph.D., Institute of Geotechnical Engineering, School of Transportation, Southeast University, Nanjing 210096, China


Introduction
It has been well recognized that the constructions of embankments over the soft natural clay are often beset by settlement problems at various degrees [1][2][3][4].Settlement analysis of soft ground with natural clays under embankment loads inevitably involves three issues: calculation model; determination of vertical superimposed stress in subsoil (σ s ′ ); and measurements of calculation parameters.
e layerwise summation method is generally adopted as a simple way of calculating embankment load induced settlements with the one-dimensional assumption [3][4][5][6].e soil profile is subdivided into layers, and the mean values of effective stress and soil compression index at each corresponding layer are used for settlement calculations.e Osterberg method is often adopted as a conventional way of determining the vertical superimposed stress (σ s ′ ) in subsoils [1,7].e Osterberg method was established based on the assumptions of instantaneous loading from embankment constructions, perfectly flexible loaded area along the base of embankments, and linear elastic theory of soils [8,9].On the contrary, the compression parameters are generally obtained from the experimental results of one-dimensional incremental load consolidation tests performed on undisturbed specimens sampled from different soil layers below ground surface [1,5].It has been well documented that the laboratory observations of compressibility may be significantly affected by sample disturbance [10][11][12][13], consolidation stress paths [14][15][16][17], testing methods [18][19][20], soil microstructure [21,22], etc.
Note that the probable errors caused by the assumptions in calculating model, the determinations of σ s ′ , and the observations in compression parameters should be in balance for obtaining the settlements with an accepted accuracy.Several researchers reported that the degrees of accuracy in the calculated results vary from good agreement [23][24][25] to poor agreement with the field observations [26][27][28][29].A case study is a powerful way of understanding the key factors responsible for difference between calculated and observed settlements for natural clays induced by embankment loads.Note that case studies have been little reported in China on embankment load-induced settlements over soft natural clays.
e first-phase engineering project of Huaihe River Waterway had been constructed for three years from 1999 to 2002.e planning on the second-phase project of Huaihe River Waterway provides us an opportunity to investigate the settlement behavior of natural clays under a 14-year embankment load. is study obtained undisturbed samples at different depths below ground surface from two holes at the central point of the embankment and out of the embankment, respectively.e physical properties of samples from two holes are compared to identify that the samples out of the embankment can be used to measure compressibility of natural clays before embankment constructions.en, the settlement under embankment loads is analyzed using the Osterberg method.e calculated settlement is compared with the observation to check the behavior of σ s ′ determined by the Osterberg method in this case, and a modified way suggested by Wang et al. [30] is adopted for improving the accuracy of calculations on σ s ′ .e values of σ s ′ obtained by the Osterberg method and the modified way are compared with the yield stresses by one-dimensional consolidation tests.Finally, the most probable factors of controlling the calculations of settlements are discussed.

Project Background and Measured Settlement
Figure 1 shows the location of the embankment investigated which locates at Huiahe River Waterway.e Huaihe River is a flood-prone river in China that originates from Tongbai Mountain, flows through Henan and Anhui Provinces, and pools up into the Hongze Lake in Jiangsu Province.e Huaihe River Waterway was designed to link up Hongze Lake with Yellow Sea for enhancing the flood discharge of Huaihe River and Hongze Lake.e first-phase waterway project was constructed from 1999 to 2002 for increasing the floor control standard of the Hongze Lake from once every 50 years to once every 100 years.
e soft deposit in the first-phase of the waterway project has a length of about 52 km.Based on the experience and stability analysis, the embankments over soft deposit were stage-constructed without any reinforcement of soft foundation.To monitor the magnitude and the rate of the soft foundation settlement, the surface settlement gauges were installed during the construction.Figure 2 shows the development of surface settlement at the centerline of embankment with elapsed time during and after construction, based on the observations reported by JPEIRI [32].It can be seen that the surface settlement at the end of construction is 1.35 m.Note some settlement gauges were destroyed or damaged caused by the flooding emergency in July 2003.e reconstruction of the monitor system was conducted in August 2005.e observations were continued to December of 2013.As shown in Figure 2, the maximum postconstruction settlement under the embankment load is recorded as 1.7 m in December 2013 [32].Figure 3  Nowadays, the State Council of China [33] has approved of the project for further reducing the frequency of major downstream flooding of Hongze Lake to once every 300 years.
e second-phase of Huaihe River Waterway project will broaden and deepen the entire channel and strengthen the existing embankments on the basis of firstphase project [31].

Undisturbed Samples of Soft Deposit and Physical Properties
Based on the information of site investigation reported by JPEIRI [34,35], a typical subsoil profile can be illustrated in Figure 4(a).e subsoils beneath the embankment can be divided to eight soil layers.e top layer is a thin crust (TC) of approximately 1.0 m thick; below the TC, there are four soft clay layers (labeled as SC-1, SC-2, SC-3, and SC-4) identified by detailed physical tests described later; their average thicknesses are 3.5 m, 6.4 m, 6.0 m, and 7.4 m, respectively.ese soft deposits were laid down during the quaternary period [36].Next are three layers of stiff sand clay (SSC-1, SSC-2, SSC-3) extending down to 37.9 m depth, and their N values of standard penetration test (SPT) are 8, 45, and 50, respectively.e planning on the second-phase project of Huaihe River Waterway calls for further detail site investigations by performing laboratory tests on high-quality undisturbed samples.Taking this opportunity, we obtained the undisturbed samples of soft natural clays beneath the embankment constructed at K85.5 section of the first-phase engineering project.For comparing the physical and mechanical behaviors of undisturbed samples with and without embankment loads, the undisturbed samples of soft natural clays under the ground surface out of the embankment (50 m distance from the centerline) were also sampled, as shown in Figure 4(b).
A thin-wall free-piston sampler was used to obtain two holes of undisturbed samples at the depths of 3 m, 6 m, 9 m, 12 m, and 15 m below ground surface.e hole under the center of embankment was termed as Em borehole, and the hole out of the embankment is termed as Gr borehole, as shown in Figure 4(b).e number after the borehole represents the depth of the samples.For example, the Em_U9m represents the undisturbed sample obtained from a depth of 9 m at the embankment borehole.

Advances in Civil Engineering
For evaluating the sample quality, the method suggested by Lacasse et al. [10] was adopted with measuring volumetric strain (ε v0 ) at the e ective overburden stress (σ v0 ′ ).e ε v0 was calculated as follows: ε v0 (e 0 − e v0 )/(1 + e 0 ) × 100%.e term e 0 and e v0 represent the initial void ratio and the void ratio under σ v0 ′ , respectively.e values of σ v0 ′ were calculated by c′H, where c′ is the e ective unit weight of soil and H is thickness of the soil layer.As suggested by Lacasse et al. [10], the sample quality can be classi ed as good or fair corresponding to the values of ε v0 within 0-2% or 2-4%, respectively.Figure 5 shows the sample qualities of the undisturbed samples from Gr borehole and Em borehole, and the most values of ε v0 varied from 1.4% to 4%.It can be concluded that the qualities of investigated samples are good or fair.Note that embankment loads were not considered in calculating σ v0 ′ for undisturbed samples from the Em borehole.Embankment loads will induce vertical superimposed stress in subsoils and decrease the values of e 0 .On the contrary, the embankment loads will cause destructuration of natural soft deposits [13], consequently resulting in the decrease in the values of e v0 .More studies are required to investigate the balance among the variation in e 0 and e v0 , the destructuration, and the vertical superimposed stress for evaluating sample quality of natural clays subjected to embankment loads.
Table 1 shows some physical properties of the samples investigated.e liquid limits (w L ) were measured using the Casagrande method according to Head [18].e plastic limit tests were also conducted in accordance with rolling thread suggested by Head [18].It can be seen that the investigated clays have a wide spectrum of w L , ranging from 50% to 81%.  Figure 6 shows the plasticity chart, indicating that all samples lie above the A-line de ned by the Uni ed Soil Classi cation System [37].e Funing soft clays can be classi ed as high plasticity clay (CH).
From Figure 6, it can be also seen that the investigated clays can be divided into four groups with the values of w L : (1) w L 50%-51%; (2) w L 58%-60%; (3) w L 68%-70%; and (4) w L 80%-81%.It is interesting to note that the w L value of Gr_U3m is approximately equal to that of Em_U6m.at is, the sample of Em_U6m with embankment loads is identical to that of Gr_U3m without embankment loads, with considering the total settlement of 3.05 m during and after the construction of embankment.Meanwhile, the value of w L for Gr_U6m is almost exactly the same as that of the Em_U9m.Furthermore, the w L values of Gr_U12/15m are in good agreement with the values for Em_U12/15m.Detail information on embankment load-induced changes in subsoil layers can be seen in the later section.
On the contrary, Figure 7 shows the particle size distributions curves for all tested soil samples, which were measured by hydrometer analysis as described by Head [18].It is interesting to note that there are four type of patterns in 4 Advances in Civil Engineering the plot, corresponding to w L 50%-51%; w L 58%-60%; w L 68%-70%; and w L 80%-81%, respectively.It can be seen that the grade curve of Gr_U3m is almost identical to the Em_U6m.For the Gr_U6m, its grade distribution is consistent with the Em_U9m.e grade distribution curves of Gr_U12/15m are almost the same as those of the sample of Em_U12/15m.Hence, the Em_U6m, Em_U9m, and Em_U12/15m can be considered belonging to the SC-1 layer (1.0 m-4.5 m), SC-2 layer (4.5 m-10.4 m), and SC-3 layer (10.4 m-16.9 m) in preconstruction stage, respectively.Note that the samples of Em_U3m are the embankment ll materials [31].e above physical properties of the Atterberg limits and particle size distributions indicate that the foundations can be considered as horizontally homogeneous before embankment was constructed.at is, the samples out of the embankment can be used to measure compressibility of natural clays for calculating embankment load induced settlement.

Assessment on Embankment Settlements
It is recognized that the natural sedimentary clays are generally subjected to soil structure e ects developed during the depositional and the postdepositional processes [1,[38][39][40][41][42][43], resulting in the vertical yield stress (σ p ′ ) often being larger than the e ective overburden pressure.e soil structure results relatively the small deformation of natural clays under e ective vertical stress up to the consolidation yield stress. e compressibility of clays increases dramatically when the applied load exceeds the yield stress [39,[44][45][46][47].
e settlement of natural sedimentary clays under σ v ′ can be calculated using the following equations: where the e ective vertical stress (σ v ′ ) in subsoil can be calculated as σ v0 ′ + σ s ′ .σ s ′ represents vertical superimposed stress and C r represents the compression index at the preyield zone when σ v0 ′ ≤ σ v ′ ≤ σ p ′ and is expressed as C r Δe/log(σ v ′ /σ v0 ′ ).C c is the compression index at the postyield state when σ v ′ > σ p ′ and is obtained by C c Δe/log(σ v ′ /σ p ′ ).e compression parameters (C r , C c , σ p ′ ) were determined by one-dimensional incremental load consolidation tests on undisturbed natural clay obtained from ground borehole.All the specimens had a diameter of 61.8 mm and an initial height of 40 mm.e loading steps ranging from   Advances in Civil Engineering 6.25 kPa to 1600 kPa by doubling the load for each increment and the duration of every loading increment was about 3 days, following Zeng et al. [48].Detailed test program is listed in Table 1.
It can be seen from Figure 8 that the compression curves of undisturbed samples show a typical inverse 'S' shape as a result of the e ects of soil structure.e relationship between σ p ′ and σ v0 ′ is shown in Figure 9. e consolidation yield stress (σ p ′ ) for all the investigated specimens was determined by the Casagrande method.σ p ′ values were all larger than σ v0 ′ values due to the soil structure during depositional and postdepositional processes [40].e values of C r , C c , and σ s ′ are shown in Table 2.Note the values of σ s ′ in Table 2 were calculated by the Osterberg method.
Figure 10 shows the comparisons between the calculated settlements and eld measurements.It can be seen that the calculated settlement is approximately 14% larger than the eld after construction measurement of 1.7 m.Note that the Osterberg method of determining σ s ′ is based on the assumption of instantaneous loading, perfectly exible loaded area along the base, and soil elasticity.ese assumptions are far from real condition of embankment [9,49].
e assumptions in the Osterberg method results in oversimplifying actual embankment condition, leading to the signi cant miscalculations of embankment load induced settlements, as reported by Wang et al. [30].

Vertical Superimposed Stress in Subsoil
In the calculation of embankment load induced settlements, quantitatively evaluating the contact stress along the base (σ b ′ ) and vertical superimposed stress in subsoil (σ s ′ ) is great importance [8].Wang et al. [30] proposed a modi ed way of determining σ s ′ for overcoming the assumptions of instantaneous loading, perfectly exible loaded area along the base, and soil elasticity in traditional approach using the geometric parameters of embankment and the friction angle of ll material (ϕ).Moreover, a reduction coe cient of 0.85 is suggested for considering the e ect of elastoplastic nature w L = 80-81% Em_U3m 6 Advances in Civil Engineering of subsoil.Note the geometric parameters in Figure 4 and ϕ 30 ° [31] were adopted for analysis in this study.e typical comparisons of predicted contact stresses along base between the modi ed method and traditional method are shown in Figure 11.It can be seen that the contact stresses predicted by modi ed method are close to a bell-shaped distribution that yields lower stresses at the central portion of embankment and higher stresses near outer edge by comparison with the traditional method.Figure 12 depicts the typical comparison of vertical superimposed stresses at embankment boreholes.It can be observed that the modi ed method yields lower stress than the traditional Osterberg method, up to approximately 24 kPa at the centerline of the embankment.ese discrepancies are attributed to the e ects of the contact stress along the base and elastoplastic behavior of subsoil on load transfer [30].
e calculated settlements with the modi ed method of determining σ s ′ by Wang et al. [30] are presented in Figure 10.It can be seen that the predicted settlements using the modi ed method of determining σ s ′ are consistent with the eld measurements.Figure 13 presents the varied depths of the soil strati cation beneath embankment determined by the calculated settlements with the modi ed method in Table 2 and recorded construction settlements.It is encouraged to nd that the varied depths of soil strati cation are consistent with the results of liquid limit and particle size distributions tests.
Note the values of σ s ′ obtained by the Osterberg method and the modi ed way are compared with the yield stress determined by one-dimensional incremental load consolidation tests on undisturbed samples of natural clay obtained from embankment borehole, as shown in Table 2. Figure 14 depicts that the predicted e ective stress using the modi ed method yields slightly higher than the yield stresses. is result is attributed to the uncompleted primary consolidation in the soft foundation.e calculated results with the Osterberg method are also shown in the same gure, again indicating that the Osterberg method may signi cantly overestimate the values of σ s ′ .

Key Factors on Settlement Calculations
For the classical layerwise summation method, the embankment behaves like a sample tested in an oedometer condition, which neglects lateral displacements.e e ect of accuracy of the vertical superimposed stress (σ s ′ ) on settlements is evaluated as shown in Figure 14, but the predicted settlements with the modi ed method still slightly smaller than the eld measurements, as shown in Figure 10. is discrepancy would be induced by the neglect of lateral displacements, the e ect of strain rate, the sample disturbance, and the stress paths on compression behavior [1].

Advances in Civil Engineering
In practice, the eld consolidation strains are generally higher than the values obtained from conventional consolidation tests, and the rate e ects are associated with the C c /(1 + e 0 ) [50].Leroueil et al. [1] reported that the e ect of strain rate is small for the clay with low C c /(1 + e 0 ) values between the 0.2-0.25.
e values of C c /(1 + e 0 ) for investigated natural Funing soft clay in Table 2 (vary from 0.19 to 0.28) reveals a minor e ect of strain rate in this study.
On the contrary, the natural Funing clay has a yield stress ratio greater than 1.0 (Figure 9) that attributed to the e soil structure of natural clay can be easily disturbed during sampling and handing.e sample disturbance reduces the consolidation yield stress and the compressibility of soil in the postyield state that results in the underestimation of the settlements [1,13].
e following two factors may be responsible for the discrepancy on settlements using traditional method: the neglect of lateral displacements and the e ect of consolidation stress path on mechanical behaviors of natural clays.In practice, the neglect of lateral displacements resulted in an underestimation of the settlements induced by the embankment load and the compression parameters calibrated from oedometer tests result in overestimation of settlements [1,16,20].As shown in Figure 10, the former factor has a much signi cant in uence on the calculation of settlement than the latter factor in this work.

Conclusions
is study performs a case study on the settlement behavior of soft natural clays induced by a 14-year embankment load.
e predicted settlements obtained by di erent methods are compared with eld observations.e main conclusions are summarized as follows: (1) e settlement predicted by combining layerwise summation method of settlement calculation model, the modi ed method of determining vertical superimposed stress suggested by Wang et al. [30], and compression parameters measured from undisturbed samples obtained by a thin-wall freepiston sampler is in agreement with the eld observation (2) e di erence in predicted settlement between traditional method and the modi ed approach can be attributed to the overestimated vertical superimposed stress in subsoil determined by the Osterberg method (3) e agreement in settlement between predicted value and eld observation is a result from the balance among the errors in calculation model, vertical superimposed stress, and compression parameters Advances in Civil Engineering

Figure 3 :Figure 1 :
Figure 3: Development of settlement rate with elapsed time during and after construction.

Figure 2 :
Figure 2: Development of surface settlement with elapsed time during and after construction.

Figure 4 :
Figure 4: Test site N88K: (a) cross section of embankment and soil pro le before construction; (b) borehole location.

Figure 10 :
Figure 10: Comparisons of settlements between the calculated and the measured values.

Figure 11 :
Figure 11: Contact stress along the base.

Notations C c :Figure 14 :
Figure 14: Comparisons between the calculated e ective stress and the yield stress.

Table 1 :
e oedometer test for undisturbed samples.

Table 2 :
e calculated results.Obtained from the site survey (Jiangsu Province Engineering Investigation and Research Institute, 2010, 2013).