A risk assessment of downdrag induced by reconsolidation of clays after upwards pipe jacking
N. A. Labanda, A. O. Sfriso, D. Tsingas, R. Aradas, M. Martini
AA risk assessment of downdrag induced by reconsolidation of clays afterupwards pipe jacking
N. A. Labanda & A. O. Sfriso
SRK Consulting, ArgentinaUniversidad de Buenos Aires, Argentina
D. Tsingas & R. Aradas
Jacobs, Argentina
M. Martini
Salini-Impregilo S.p.A., Italy
ABSTRACT: Salini-Impregilo is building part of the largest sanitary sewer system in the history of Argentinain the suburbs of Buenos Aires City, to serve a population of almost five million people. The project is an outfallTBM tunnel 12 km long, starting from a reception shaft in the river margin, and transporting the sewage 35meters below the
Rio de la Plata riverbed to the point of discharge. Within the final kilometer of the tunnel,a set of 36 standing pipes so-called risers are constructed by driving steel tubes upwards and passing throughdense sands, sandy clays and soft clays. Risers are linked-up with the launching lining segment using flangeunions.Driving of risers upwards will generate excess pore pressure and disturbance in fine soils and, once the pipeis placed in its final position, negative skin friction due to reconsolidation and creep. A risk assessment of thedowndrag is presented in this paper, based on the estimation of the force and/or displacement in the riser-tunnelunion generated by this effect. The issues of whether it is desirable to instalock the riser-tunnel union at anearly age after installation of the riser and the time lapse required to reduce negative skin friction effects arediscussed. Results are validated by comparing the model results with field measurements in prototype models.1 INTRODUCTIONThe
Matanza-Riachuelo river flows along the riparianlands between the City and the Province of BuenosAires (Argentina). It is a water stream 64 km longand 35 m wide in average that flows into the
Rio dela Plata . The Riachuelo river is considered as one ofthe most polluted water courses in the world, ruinedacross the years by the discharge of several kind ofuntreated industrial and waste water.The
Matanza-Riachuelo basin recovery project isan ambitious plan to manage the wastewater from theleft margin of the
Rio de la Plata through a sewer-age interceptor tunnel, a treatment plant and a dis-charge tunnel into the
Rio de la Plata . The projectis divided into several contracts. One of them, named
Lot 3 , holds the outfall tunnel, an EPB-TBM tunnelrunning 35 meters below the riverbed of the
Rio de laPlata . A scheme of the plant together with the designof the system at conceptual level is shown in Figure1.
Figure 1: Conceptual design of the Matanza-Riachuelo outfall.Detail of launching rings and risers in diffusor zone.
The tunnel has an internal diameter of 4.3 m and atotal length of 12 km including a 1.5 km diffusor zonewith of 34, 28 m long standpipes named ’ risers ’ thatdaylight in the riverbed. Risers are lifted from the in-terior of the tunnel by driving upwards steel tubes. Aconstruction sequence of risers installation is shownin Figure 2.Driving the risers will generate excess pore pres-sure and disturbance in the fine surrounding soils and, a r X i v : . [ phy s i c s . g e o - ph ] S e p igure 2: Risers construction stages. once the pipe is placed in its final position, nega-tive skin friction due to reconsolidation and creep ofsoft clay layers. Due to an optimization of the design,the tunnel elevation was lifted and, consequently, thelength of the risers reduced from 30.5 m to 28.0 m,thus reducing the thickness of sandy layers to becrossed. While this optimization is beneficial to re-duce the jacking forces through dense sands, it doesworsen the downdrag phenomenon due to the loss ofthe positive skin friction provided by the sands. Thiscould generate two scenarios:Scenario N ◦
1: If the riser is not bolted to the tun-nel, a vertical displacement is experienced by thepipe, slightly increasing the local hydraulic loss.Scenario N ◦
2: If the riser is bolted to the tun-nel, the negative skin friction will produce a non-negligible force acting on the crown lining seg-ment.A risk assessment using a simplified model to esti-mate displacements (scenario N ◦
1) or forces (scenarioN ◦
2) produced by negative skin friction in risers dueto consolidation and creep is presented in this paper.A brief introduction to the geotechnical site character-ization is presented in Section 2, while the constitu-tive models employed are presented in 3. The numer-ical model and its results about downdrag, structuralforces in risers and comparisons with field test resultsare shown in 4. Finally, some conclusions are drawnin Section 5.2 GEOTECHNICAL SITECHARACTERIZATIONThe mentioned project is placed below the
Rio dela Plata , where first 20 to 25 meters depth from theriverbed are composed by normally consolidated softclays and silty clays stemmed by fluvial depositions,known as the ’
Post-pampeano ’ formation. This for-mation is characterized by a liquid limit ω L varyingfrom ∼ to ∼ , plasticity index from ∼ to ∼ , s u σ (cid:48) v ∼ . | . and fine content above the (Sfriso1997). The ’ Puelchense ’ formation can be found below thepost-pampeano, between 27 to 31 meters depth fromthe riverbed, composed by a dense clean sands with anin-situ relative density above , an effective fric-tion angle in constant volume φ cv ∼ ◦ and a dila-tancy angle ψ ∼ ◦ | ◦ .Between the puelchense and post-pampeano for-mation, a transition of 5 m to 11 m thickness com-posed by interleaved layers of soft clays and mediumto dense silty sands is encountered. Figure 3 showsa detailed stratigraphy of the site, where the transi-tion layer is represented as fundamentals soil layersclassified by CPTu in-situ test and Robertson charts(Roberton 2009). It also shows the trace of the men-tioned tunnel, the position of the CPTu tests used tocalibrate the numerical models and, in red, the con-sidered risers for our analysis. Figure 3: Site stratigraphy, risers location and crown elevation.
In order to illustrate the soil classification in thetransition, Figure 4 plots the Robertson chart for thelayer in CPTu-01A and CPTu-05A. It can be seen thatthe mentioned stratum is mainly composed by nor-mally and under consolidated soils, with an erraticdistribution of soft clays and sandy soils. The under-consolidation is produced by a constant upflow com-ing from the puelchense formation, a confined aquiferwith an over pressure with respect to the river of, ap-proximately, 2 m.
Figure 4: Robertson charts for the transition for CPTu-01A andCPTu-05A.
It is clear in the site profile that, as the crown el-evation is lifted, the thickness of sandy soils acrossthe risers decreases. This is crucial for our problemecause, as is demonstrated in this paper, sandy soilsprovides a confinement pressure in the pipe diminish-ing displacement or forces in the riser-tunnel union.3 CONSTITUTIVE MODELLINGSoft clays in the riverbed of ther
Rio de la Plata are anisotropically normally consolidated soils, suffi-ciently ancient to assume that primary and secondaryconsolidation has been occurred. Nevertheless, duringrisers driving, an intensive soil remolding is carriedout in the surroundings of the pipe, till a distance of4 to 5 times its diameter. The soil disturbance, gener-ated at constant humidity, produce excess pore pres-sure erasing the stress history of the material. Oncethe driving is ended, excess pore pressure dissipationand reconsolidation in the influenced area, producedowndrag forces and, consequently, negative frictionin the riser. Soft soil creep model is used for claysto take into account the time-dependent behavior pro-duced by remolding. At the same time sandy layerslose densification, approaching the effective frictionangle to the constant volume friction angle, reducingthe dilatancy angle to ◦ and, consequently, its con-finement potential. Hardening soil small is used tosimulate the mechanical behavior of this kind of soils.Figure 5 shows experimental results of compress-ibility index as a function of the liquid limit, ob-tained by unidimensional consolidation test for thepost-pampeano formation, with a comparison withTerzaghi and Skempton correlations (Sfriso 1997).The swelling index is considered as C s ≈ | C c and the secondary compressibility index for the men-tioned formation is C α = 0 . (Ledesma 2008). C o m p r e ss i b ili t y i nde x C c Liquid limit ω L Post-pampeano formation
Sfriso 1996 Ledesma 2008 Terzaghi Skempton
Figure 5: Compressibility index C c versus liquid limit forthe post-pampeano formation. Comparisons with Terzaghi andSkempton expressions. Parameters for sandy soils are obtained from classi-cal correlations based on the standard penetration test(SPT) and the transition layer is simulated as a com-position of soft clays and sands layers. The geotech-nical profile for each riser is calibrated by comparingpore pressures calculated with the numerical modeland CPTu measurements. Table 1 summarize consti-tutive models and parameters used for our simula-tions.
Table 1: Constitutive models and parameters used.Unit Soft clay Sandy soilsModel - Soft Soil Creep HSsmallDrainage - Undrained (A) Drained γ kN/m φ (cid:48) ◦
24 32 c (cid:48) kP a ψ ◦ C c - 0.410 - C s - 0.080 - C α - 0.013 - G ref M P a - 200 γ . - - − E refur M P a - 120 E ref M P a - 40 E refoed M P a - 40 m - - 0.5 ν ur - 0.20 0.20OCR - 1.00 1.00 K nc - 0.51 0.60 R inter - 0.80 0.70 k − m/s 0.005 1.00 Figure 6 shows a comparisons betweenconsolidated-drained triaxial test of samples takenfrom the post-pampeano formation and the constitu-tive model calibration. Soft soil creep has a limitationto reproduce the mechanical behavior in low pres-sures in low strains. Nonetheless in this particularproblem, the numerical model works in large strainsand the proposed constitutive model reproducereasonably the maximum deviatoric resistance of softclays. D e v i a t o r i c S t r e ss ( k P a ) Figure 6: Constitutive model calibration and comparisons withCD triaxial test.
The analyzed phenomena is not only dependent onthe geotechnical parameters, but also has a stronglydependence with the riser-soil interaction. In thissense, it is important to define a proper friction an-gle for the interface. Several research papers has beenpublished regarding this matter for driven pile, com-bining different materials such as concrete, wood andsteel with different confining pressures. It has beenproved that the friction angle for the steel-soil inter-action is independent with the contact tension and thesaturation degree and, for smooth steel, φ (cid:48) = 24 ◦ (Po-tyondy 1961). This value is adopted for our numericalmodels. igure 7: Total Strain path in a driven pile surroundings and com-parison with a radial cavity expansion (Baligh, 1985).Figure 8: Construction stages of the finite element model. Soil stress state after pipe jacking
Some authors have proposed theoretical and experi-mental models to estimate the strain path in the sur-rounding of a penetration test. Baligh (1985) providesa framework to study driven piles, assessing the dis-turbance of soil and showing that the strain producedby a driving maneuver is comparable to a radial cav-ity expansion since 2 radii from the tip (see Figure 7).Other researchers arrived to similar conclusions (Pes-tana, Hunt, & Bray 2002, Chong 2013).An axial-symmetric model is considered for ouranalysis, where a displacement equals to 35.5 cm, theriser radius, is imposed from the rotation axis. Afterthis, a plate to model the metal sheet of the riser isused to simulate the soil-structure interaction. Figure8 shows a detail of the mentioned stages.Table 2 details characteristics of each analyzed riserand the closest CPTu test, taken as reference to vali-date the geotechnical profile. It also summarize thesandy layers thickness crossed and riser length in both
Table 2: Characteristics of analyzed risers and reference CPTutest. New elevation Former elevationRiser CPTu Sand layers Riser Sand layers RiserN ◦ thickness length thickness length[m] [m] [m] [m]1 01 2.7 28.0 5.3 30.53 01A 1.0 28.0 1.0 30.55 01B 2.7 28.0 5.2 30.59 02 3.5 28.0 6.0 30.512 03 2.8 28.0 5.3 30.516 04 1.9 28.0 4.4 30.521 05 6.0 28.0 8.5 30.523 05A 5.0 28.0 7.5 30.527 06 4.2 28.0 6.7 30.534 07 1.3 28.0 3.8 30.5Figure 9: Geotechnical profiles proposed for each considered ris-ers. Former crown elevation. considered crown elevation. This data is used latterto correlate both, displacement and structural forces,with the presence of sands.The assessed geotechnical profiles in each riser isplotted in Figure 9 for the former elevation, wheresoft clays are represented with green and sandy soilsare represented with pink. Remolded clays, with a no-ticeable creep effect, are represented with grey and itoccupies 2 diameter in riser’s surrounding.After the radial cavity expansion, the riser is in-stalled and an excess pore pressure is generated insoft clays. Figure 10 shows the total pore pressurefield in the former crown elevation, that has to be dis-sipated in the consolidation procedure. It is importantto note that models for the actual crown elevation con-siders the same geotechnical profile, but ends at 28.0m depth. Figure 10: Numerical model results of total pore pressure in eachriser. Former crown elevation. .2 Comparisons of numerical results with CPTutest
Obtained total pore pressures in our numerical mod-els are compared with CPTu reference tests performedclosest to the their future place, and the results arepresented in Figure 11. Informed pore pressures aresurveyed in the soil-riser interface.It is interesting that our proposed model, despite itssimplicity, reproduce accurately the excess pore pres-sure measured with CPTu. Whether the cone penetra-tion test is performed or the riser is driving, producea relatively similar pore pressure profile due to thestrain range at both cases are working. As is shownin Figure 6, after 12 % of axial strain, the deviatoricstress reaches a plateau and, for an undrained test, thepore pressure is nearly stabilized.After riser collocation, a consolidation stage is cal-culated considering a lifetime of 100 years. As wasdescribed in the introduction, two scenarios are stud-ied in the risk analysis: a non bolted riser is analyzedin Scenario N ◦
1, allowing a vertical degree of free-dom in the pipe to be displaced inside the tunnel. Thiscase is presented in Section 4.3. Scenario N ◦
2, pre-sented in Section 4.4, considers a vertical fixity inriser base to calculate the applied force in the launch-ing ring.4.3
Scenario 1 - Time evolution of riser downdrag
Riser vertical displacement is conceptually given by δ = δ e + δ c + δ cr , (1)where δ is the total vertical displacement, δ e is theelastic bouncing right after the driving maneuver, δ c is the consolidation component and δ cr is the creepcomponent. It is well known that elastic bouncing dis-placement in driving piles is negligible. In this way,the δ e component is not considered in our paper fo-cusing our curiosity in time-dependent terms.Riser displacements over time due to primary con-solidation δ c (solid line) and due to primary and sec-ondary δ c + δ cr (dashed line), are plotted in Figure 12for each considered riser in former crown elevation.Results obtained for the same analysis but in actualcrown elevation, are presented in Figure 13. A hypo-thetical geotechnical profile where the riser does notcross any sandy layer , named ’pure soft clay’, is in-cluded in both cases to estimate an upper bound forthe results.When sand lenses are coarser, displacement are lessthan 3 mm after 100 years in all cases with the excep-tion of riser N ◦
3, where there are not sands over thetunnel crown, neither in actual or former elevation. Inthis case, displacements in the end of the lifetime is35 mm and the 90 % of the value is reached after 3month since its installation.A completely different setting is observed for theactual crown elevation, where risers N ◦
1, 3, 5, 9, 12, -30-28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -30-28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -30-28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -30-28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -30-28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] D ep t h [ m ] -28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] -30-28-26-24-22-20-18-16-14-12-10-8-6-4-20 D ep t h [ m ] Figure 11: Comparison between CPTu test (black line) and nu-merical results obtained after a radial cavity expansion (red line).igure 12: Riser displacement due to primary and secondaryconsolidation in former crown elevation.Figure 13: Riser displacement due to primary and secondaryconsolidation in actual crown elevation.
16 and 34 have maximum displacements highers than10 mm, being a critical condition for the local hy-draulic loss. Similarly to riser N ◦ % of maximum displacements are reached af-ter 2 or 3 month since their installation.Collecting all maximum displacement obtained inboth elevations and ordering them in terms of thesandy soil thickness crossed, it can be seen in Fig-ure 14 that, while sands thickness increases over 4.0m, values are stabilized between 1 and 2 mm and thecreep effect is almost negligible because the confine-ment effect produced by coarser soils.The mentioned effect produced by sands generatesan increment in structural forces in pipes. Figure 15shows the maximum normal force reached in risers interms of the sand soil thickness crossed, for both el-evation. It is clear that, when the displacement free- Figure 14: Maximum displacements in terms of sand soil thick-ness crossed by riser in both elevations in the end of the lifetime. Figure 15: Maximum normal force in riser in terms of sand soilthickness crossed by riser in both elevations in the end of thelifetime. dom is blocked by some reason, the structural ele-ment is more stressed. When sand layers are less than3 m, the maximum normal force is almost constantbetween 900 and 1400 kN. From 3 m and up, the nor-mal compression increase dramatically, reaching val-ues nearly to 2800 kN. Different from the displace-ments, the presence of sandy soils has not a blockingeffect of creep in the structural forces.4.4
Scenario 2 - Structural forces induced byconsolidation
When risers are bolted to the launching ring, the ver-tical displacement is restricted and consolidation andcreep of soft clays generates a concentrated force intothe crown lining segment. This new scenario producea rearrangement of internal stresses in pipes.Figure 16 shows maximum normal forces for riser-tunnel union (shaded in blue) and for all pipes (shadedin red), in the end of the lifetime, in terms of crossedsand layer thickness. When sand thickness is less than3 m, the maximum normal force is placed in the riser-tunnel union and are almost three time higher thanthe maximum normal force reached in scenario N ◦ ◦ ◦ ◦ igure 16: Maximum normal forces in risers in terms of sand soilthickness crossed in both elevations in the end of the lifetime.Figure 17: Normal forces for risers N ◦ ◦ layer. If the geotechnical profile is composed mainlywith soft clays, differences between both scenariosare notorious and the maximum normal value is in thetunnel crown.Circumferential forces has their maximum valuescoincidentally with sand layers and differences be-tween scenarios are negligible no matter the presenceof sand layers, but being clear its contribution to theriser confinement.5 CONCLUSIONSThe driving procedure of piles and tubes it is stilla challenging task to scientist and engineers. In thispaper, a simple but powerful finite element modelhas been proposed to estimate the downdrag phenom-ena and the negative friction due to consolidation andcreep, calibrating the soil profile using CPTu test. Ourproposal deals with the identification and modelingof a layered stratum, characterizing the transition asa composition of simple geotechnical units and cal- Figure 18: Circumferential forces for risers N ◦ ◦ ibrated comparing the pore pressure obtained in ournumerical models with those obtained in field tests,simulating the penetration strain path as a radial cav-ity expansion.After the riser driving, a negative skin friction in thesoil-structure interface due to the consolidation andcreep is generated due to clay remolding, moving thepipe inside of the tunnel in Scenario N ◦
1. In ScenarioN ◦
2, where the riser is bolted in the launching ring,the mentioned phenomenon produce a non-negligibleconcentrated force in the crown tunnel.In both scenarios, displacements or the forces, hasa strongly dependence with the sand layer crossed bythe riser. In Scenario N ◦ ◦
3, and it was provedthat a lifting in the elevation of the tunnel implies anincreasing movement in the rest of pipes, due to a lessthickness in crossed sand layers. It was also showedthat an increment in sand layers rebounds in a con-siderable reduction of the creep component and, after60 to 90 days, the majority of the movement has beenperformed. When sand layer thickness is larger than 4m, the riser displacement is negligible.If the pipe is bolted to the launching ring, the forceexpected in the union range from 2900 kN to 1000kN, being also the maximum force within the riser ifthe sand layer thickness crossed is less than 2.5 m.While the thickness increases, the force in the tun-nel crown decrease and the maximum normal force ismoved to the transition layer.REFERENCES
Baligh, M. (1985). Strain path method.
Journal of GeotechnicalEngineering , 1108–1136.Chong, M. (2013). Soil movements due to displacement piledriving.
Case Histories in Geotechnical Engineering , 1–13.Ledesma, O. (2008).
Calibracin del Cam Clay para suelos delpost-pampeano . Buenos Aires, Argentina: Tesis de grado dengeniera Civil - Universidad de Buenos Aires.Pestana, J., E. Hunt, & J. Bray (2002). Soil deformation and ex-cess pore pressure field around a closed-ended pile.
JournalOf Geotechnical and Geoenvironmental Engineering 102 , 1–12.Potyondy, J. (1961). Skin friction between various soils and con-struction materials.
Geotechnique , 339–353.Roberton, P. (2009). Interpretation of cone penetration test - aunified approach.
Canadian Geotechnical Journal 46 , 1337–1355.Sfriso, A. (1997). Formacin post-pampeano: prediccin de sucomportamiento mecnico.