International Society for Soil Mechanics and Geotechnical Engineering

INTERNATIONAL SOCIETY FOR SOIL MECHANICS AND GEOTECHNICAL ENGINEERING

This paper was downloaded from the Online Library of the International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE). The library is available here: https://www.issmge.org/publications/online-library

This is an open-access database that archives thousands of papers published under the Auspices of the ISSMGE and maintained by the Innovation and Development Committee of ISSMGE. Geotechnical Aspects of Underground Construction in Soft Ground – Viggiani (ed) © 2012 Taylor & Francis Group, London, ISBN 978-0-415-68367-8

Dewatering tests results for underground C Line stations construction

Massimo Grisolia & Giuseppe Iorio Department of Civil, Architectural and Environmental Engineering, University of Rome “Sapienza”, Italy Antonio Zechini RomaMetropolitane S.r.l., Rome, Italy

ABSTRACT: The new C Line of Rome Underground -T4, T5 and T6A lots- mainly runs into volcanic deposits deriving from the Colli Albani apparatus, widely extended in south-eastern areas of Rome. The local hydrogeological framework is very complex due to the large variations of permeability in function of granulometry, cementation processes and secondary fracturing of pyroclastic strata. The bottom of the excavation of the underground stations lies 25/30 meters below the groundwater level. In order to allow dry conditions and to prevent bottom heave during excavation, the groundwater level is lowered by deep wells systems. Some dewatering field tests showed how the local stratigraphy strongly affects the dewatering efficiency. Steady state filtration FEM analyses allowed to compare permeability of soil with Lefranc in situ tests results and to verify the efficiency of grouted plugs.

1 INTRODUCTION stiff overconsolidated clay of Pliocene age (APL unit) covered by a layer of fluvio-palustrine very The new C Line of Rome Underground runs NW dense sandy gravels of Pleistocene age (SG) fol- to SE across the city centre, with a total length of lowed by either medium stiff clayey silts and dense more than 25 km and 30 stations. At present, the sandy silts (Paleotevere units AR, ST) (Figs. 1 T4, T5 and T6 A lots, about 10 km twin running and 2). tunnels and eleven stations, are in advanced work- These deposits are covered by a Middle to Upper ing state of progress. Pleistocene piroclastics volcanic soils deriving from Due to the presence of a densely built urban the Colli Albani apparatus, widely extended into environment, the selected solution for the construc- the southeast of Rome. tion of the stations consists of a “cut and cover” The pyroclastic sequence, linked to the past excavation retained by multi-propped diaphragm explosive volcanic activities, is characterized by old walls to minimise settlements and prevent damage altered tuffs (lithoid and pseudo-lithoid tuff T1-T2 to the structures. and clayely tuff TA), “pozzolane” (Pozzolane The excavation depth of the stations are about Rosse PR and Pozzolane Nere PN) and lithoid 25 to 30 m, with a maximum of 20–25 m below tuff (Tufo Lionato). Frequent and relatively deep the water table. In order to assure dry conditions ancient ditches filled by alluvial silty-clayey and and hydraulic heave stability, deep wells systems sandy deposits (Alluvioni dei fossi unit, LSO) cut have been designed to lower groundwater level and into the pyroclastic deposits. piezometric head during excavation. A layer of made ground (R) of varying thick- Dewatering systems have been preliminary ness covers the stratigraphic sequence and the nat- tested by long term pumping field tests. ural soil profile everywhere. The paper shows the result of numerical analy- The local hydrogeological framework is very ses carried out to interpret dewatering field tests complex and characterized by a double groundwa- made for three different stations of T5 lot, namely ter system. The upper main aquifer is mainly repre- Teano, Mirti and Gardenie. sented by the “pozzolane sequences” (PR/PN). The lower aquifer consists of the Pleistocene deposits (SG) which overlain Pliocene marine claystone 2 HYDRO GEOLOGICAL AND bedrock (APL). The pyroclastic deposits (poz- GEOTECHNICAL CONDITIONS zolane, lithoid and pseudo-lithoid tuff and clayely tuff) show large variations in permeability in func- The geological sequence along the new C Line tion of primary porosity, compaction processes, under construction consists of a base deposit of sealing, and secondary fracturing. Permeability

619 Figure 1. Geological profile from S. Giovanni to Malatesta stations (adapted from Metro C, Progetto costruttivo).

Figure 2. Geological profile from Teano to Alessandrino stations (adapted from Metro C, Progetto costruttivo).

Middle/Upper Pleistocene

VS - Silty sand, sandy silt VS Pleistocene

TL - Lithoid tuff ST ST - Silty sand, sandy silt TL Holocene R - Man-made ground TT TT - Pyroclastic Silty sand, sandy AR AR -Clayey silt and silty clay R silt SG SG - Sandy gravel LSO - Alluvial clayey silty and PN/PR PN / PR - Pyroclastic Sandy silt LSO sandy silt deposits black or red Old althered tuff TA Pliocene T2 TA - clayey tuff TA T1/T2 - lithoid and pseudo-lithoid APL - Silty clay T1 tuff APL

T3 T3 - Lithoid tuff

Lefranc Tests Lefranc Tests Permeability (m/s) Test number -9 -8 -7 -6 -5 -4 -3 -2 10 10 10 10 10 10 10 10 0 20406080100

R R LSO LSO

PN-PR PN-PR TT TT

TA TA

T1-T2 T1-T2 ST ST AR AR SG SG

APL APL

Figure 3. Lefranc permeability tests results.

620 values for each soils type have been determined by level was 14 m above the bottom of the excavation. Lefranc tests at variable head (Fig. 3) and by some Dewatering system was constituted by n. 12, preliminary pumping tests. Lefranc tests indicated 400 mm diameter wells penetrated 26 m into the a wide range of permeability, with values from volcanics deposits (“TA” and “T1-T2”) with slot- 7.5 ⋅ 10−9 m/s up to 3.6 ⋅ 10−4 m/s for pozzolane PR/ ted screen extending from the water table surface PN, 4.9 ⋅ 10−8 m/s to 1.0 ⋅ 10−4 m/s for the lithoid and (assumed +23.5 m O.D.) to the bottom (Figure 4a). pseudo-lithoid tuff (T1-T2), and 1.3 ⋅ 10−8 m/s to The total pumping rate was approximately 17 l/s 1.7 ⋅ 10−5 m/s for the Pleistocene deposits ST. and the established pore water pressure distribution inside the ground allowed dry conditions during excavation. Side effects on external water table 3 GEOTECHNICAL AND were very limited (maximum lowering of 0.20 m). ENVIRONMENTAL CONDITIONS 4.1.2 Mirti station The construction of C Line stations requires deep The station consists of a rectangular box retained excavations under high hydraulic head. The main by diaphragm walls 65 m long and 37 m wide. geotechnical and environmental problems related The maximum excavation depth is about 34 m to the water seepage are connected with: from the ground surface. In this case the required drawdown of the water table in the volcanic soils 1. lowering the groundwater level in order to allow is about 18 m. The dewatering system consists of the excavation in dry conditions; 14 400 mm diameter wells placed inside the station 2. stability of the bottom line during excavation; box, pumping from the volcanics complex “TA” 3. seepage effect on the stability of diaphragm and “T1-T2”, (Fig. 4b). During long term tests the walls; achieved drawdown was enough to allow dry con- 4. ground settlements induced by water table ditions during excavation. The total pumping rate, lowering. approximately 70 1/s, was much more than design prevision, and a water pressure distribution with depth, different from an hydrostatic trend (Fig. 4b) 3.1 Dewatering field tests was observed. The lowering of external water table Dewatering field tests were systematically carried in the volcanic deposits was quite limited (0.7 m). out, consisting of: 4.1.3 Gardenie station − step drawdown and constant rate discharge tests The station, retained by diaphragm walls, is about (24h) carried out in the first installed well; 114 m long and 28 m wide. The maximum exca- − long term (8–21 day) dewatering test by pump- vation depth is about 27 m from the ground level. ing all the wells installed into each station. Dewatering system was constituted by 14 400 mm Monitoring activities included the measure- diameter wells penetrated 34 m into the volcanic ments of daily pumping rate and the hydraulic deposits “TA” and “T1-T2” (Fig. 4c). During long head for each well. Measurements were taken term tests the water table drawdown was only of inside the stations by electrical piezometers and about 2.0–6.0 m at a total pumping rate of approx- outside by Casagrande piezometers installed along imately 70 1/s. The external groundwater level low- the perimeter of excavation. The instruments lay- ering was about of 1.0 m. out is schematically shown in Figure 4.

5 SEEPAGE FLOW ANALYSIS 4 DEWATERING FIELD TEST RESULTS Although the geological pattern of the three sites The results of the dewatering field tests, in terms were similar, the performances of the dewatering of total pumping rate, drawdown achieved and tests were very different. pore pressure distribution inside the excavation In the case of Teano station, water flow mainly areas, are presented for three deep stations (Teano, derived from the deeper aquifer located in ST Mirti and Gardenie. (Fig. 5). The upper volcanic deposits (PR) and fluvio-palustrine (ST) were not acting as a single 4.1.1 Teano station hydrogeological unit since the piezometric level The excavation consists of an elongated box about recorded in ST stratum was around 12–14 m lower 140 m long and from 14 to 28 m wide which accom- than the level measured into the upper volcanic modates the station and the “Manufatto Diramazi- deposits. The low value of pumping rate (<1.0 l/s one C1”. The maximum excavation depth is about for each well) indicates that the deeper aquifer is 29 m from the ground level. Natural groundwater fed by sandy gravels of Pleistocene deposits (SG),

621 a) TEANO STATION

Teano Station - Corpo stazione Dewatering field test results

Hydraulic head (m) 010203040

+ 38.19 m 40

+ 32.84 m Dewatered ground water level (m) + 28.89 m 30 Design value (m) + 24.30 m Outside ground water level (m)

PR 20

10 T2

TA Level (metersa.s.l.) T1 0 - 3.00 m TA

- 8.00 m -10 ST

STa -20

b) MIRTI STATION c) GARDENIE STATION

Mirti Station Dewatering field test results Gardenie Station Dewatering field test results Hydraulic head (m) Hydraulic head (m) 010203040 0 10203040 + 40.40 m 40 40 + 34.91 m

+ 31.80 m + 30.16 m 30 Dewatered ground water level (m) Design value (m) 30 Dewatered ground water level (m) TT Design value (m) Outside ground water level (m) PN TT Outside ground water level (m) 20 20

PR PR 10 10 TA + 5.65 m Level (meters a.s.l.) (meters Level TA T2 a.s.l.) (meters Level 0 0 -3.80 m T2

TA - 7.20 m TA

-10 T1 T1 -10

- 14.20 m TA -14.60 m TA

-20 ST -20

Figure 4. Teano, Mirti and Gardenie Station—Planimetry of well and piezometer arrangement during dewatering field test (Courtesy of Metro C S.p.A.) and pore pressure during dewatering field test. and it is semi-confined by the low permeable altered Into Gardenie and Mirti stations, the hydraulic pseudo-lithoid tuff (T2) and clayey tuff (TA) lay- seepage scheme was very different from Teano con- ers. The hydraulic disconnection between ST and figuration and far from design forecasts. In Mirti overlying pozzolane PR caused no relevant effects station, seepage flow is mainly horizontal and on shallow aquifer. derives from the aquifer located in old altered tuffs

622 Teano Mirti Gardenie

TT PR Slotted Slotted screen screen PR Slotted TA PR screen Bottom Bottom excavation excavation T2 Bottom excavation TA TA TA T2 T1 T2 TA TA TA T1 T1 ST TA TA

Figure 5. Teano, Mirti and Gardenie: different seepage flow.

TA-T1 characterised by high permeability. The Mirti Station Pore pressure measures in presence of grouted soil plug overlying low permeability pseudo-lithoid tuff T2, in which piezometric pressure drop is concentrated, Hydraulic head (m) determines an hydraulic disconnection between TA 010203040 + 40.40 m and overlying pozzolane PR. As for Teano Station, 40 + 34.91 m no significant effects on shallow aquifer located in Piezometer measures (m) + 30.16 m the overlying pozzolane PR was recorded. 30 Design value (m) + 25.00 m At Gardenie station, seepage flow derives Excavation depth (Oct 2010) directly from the shallow aquifer located in vol- PR 20 canic deposits. Piezometric measures in the upper + 14.00 m level of the old altered tuffs TA indicates a similar + 8.65 m 10 Base of the excavation

piezometric level as in the overlying pozzolane PR. a.s.l.) ( meter Level 0 Continuous pumping modified the shallow aquifer -3.80 m located in the PR pozzolane PR. Bottom plug -10 When the flow of water occurs upwards from -12.50 m -14.60 m TA the Pleistocene deposits (ST), the pumping rates -20 were low (Teano, Q < 1.0 l/s for each well). When the seepage flow directly derives from the shallow Gardenie Station aquifer located in volcanic deposits the quantity Pore pressure measures in presence of grouted soil plug and velocity of water dramatically increases. Hydraulic head (m) 010203040

+ 31.80 m Piezometer measures (m) 6 GROUTED BOTTOM PLUGS 30 Design value (m)

+ 25.15 m

In the case of Mirti and Gardenie stations, the quan- + 19.80 m 20 Excavation depth (Oct 2010) tity of pumped water was quite greater than the PR 10 design assessments. To reduce the local permeabil- TA + 5.65 m Base of the excavation ity of granular and fractured tuffs and minimise the + 5.15 m Level (meters a.s.l.) (meters Level groundwater inflow, an horizontal grout curtain was Bottom plug 0 realised by a Multiple-Packer Sleeved Pipe (MPSP) - 7.20 m -10 injection system. Preliminary pumping tests showed - 13.70 m - 14.20 m STa how much the grout injections reduced the water ST -20 inflow. Figure 6 shows the profile of the hydraulic head obtained from the piezometers measurements at Figure 6. Mirti and Gardenie: pore pressure measures the intermediate excavation phase of October 2010. in presence of the grouted bottom plug.

623 In Mirti station the total pumping rate 7.1 Analysis results (approximately 8 1/s) ensure safe and workable con- Pore water pressure distributions obtained with ditions throughout constructions phases. External FEM filtration analyses inside the excavation are water table lowering in the volcanic deposits are shown in Figure 7. quite limited and less than 0.3 m. In Table 1 are presented the permeability values In Gardenie station the mean total pump- obtained by finite element analyses calibrated on ing rate is approximately 40 1/s and significantly pumping tests. For each station the main k values lower than that measured at the initial conditions for each soil layer are compared with the range of (>70 l/s). The leakage through the grout plug allows value obtained by in situ Lefranc tests (Fig. 3). excavation to proceed in dry conditions. Pore water Analyses results indicate a wide range of per- pressure distribution inside the excavation is in line meability values for the same soil in the different with the design forecasts. The effects on the exter- stations: from 1.0 ⋅ 10−6 m/s to 5.0 ⋅ 10−5 m/s for poz- nal water table lowering in the volcanic deposits zolane PR/PN, 1.0 ⋅ 10−7 m/s to 5.0 ⋅ 10−3 m/s for are quite limited (≈0.8 m). the lithoid and pseudo-lithoid tuff (T1-T2), and 2.4 ⋅ 10−6 m/s to 5.0 ⋅ 10−4 m/s for clayey tuff (TA). Great discrepancy appears between such 7 NUMERICAL ANALYSES results and Lefranc in situ tests mainly for the lithoid and pseudo-lithoid tuff (T1-T2) and clayey A series of simplified two-dimensional steady tuff (TA). In particular, from Gardenie station state flow FEM analyses using PLAXFLOW tests: layers T1-T2: k = 5.0⋅10−3 m/s; Lefranc v1.6 code were carried out, in order to define ft range: from 4.2 ⋅ 10−7 m/s to 9.5 ⋅ 10−4 m/s; layer mean permeability of volcanic deposits in Teano, Mirti and Gardenie stations. Permeability values were routinely calibrated to obtain pore water Mirti Station pressure distributions similar to those meas- Pore pressure measures in presence of grouted soil plug ured. To simulate dewatering, it was necessary to define a geological cross section, physical proper- Hydraulic head (m) 01020304050 + 40.40 m ties of soils and hydraulic boundary conditions. 40 The reference stratigraphies of every stations + 34.91 m Piezometer measures (m) + 30.16 m were chosen similar to those already showed in 30 Design value (m) FEM Filtration Analysis Figure 4. In the analysis it was assumed an iso- + 25.00 m Excavation depth (Oct 2010) tropic and constant value of permeability k for PR 20 each stratum. Diaphragm walls were simulated + 14.00 m with “screen elements” across which flow does + 8.65 m 10 Base of the excavation Bottom plug not occur. a.s.l.) ( meter Level 0 To calibrate the permeability for each soil lay- -3.80 m ers it was necessary to force a specified outflow for -10 -12.50 m each well. This was defined by an equivalent dis- -14.60 m TA − charge flow rate q per unit width determined by -20

Gardenie Station Q Pore pressure measures in presence of grouted soil plug q− ()= m naa ⋅ BL Hydraulic head (m) dr l ()station 01020304050 40

Where Q is the steady state global mean pumping + 31.80 m Piezometer measures (m) m 30 Design value (m) rate on entire station, B and L are the characteris- + 25.15 m FEM Filtration analysis tics dimensions of the station and al is the number + 19.80 m 20 Excavation depth (Oct 2010) of wells on the station alignment. PR Once defined the mean permeability of volcanic 10 TA + 5.65 m Base of the excavation deposits k , in the cases of Mirti and Gardenie + 5.15 m

ft a.s.l.) (meters Level 0 stations it was possible to deduce the loss of per- Bottom plug - 7.20 m meability of grouted soils kgrout after the treatment -10 expressed by a “Reduction Permeability Factor” - 13.70 m - 14.20 m STa F ST R: -20

k Figure 7. Mirti and Gardenie: pore pressure distri- ft bution inside the excavation obtained from numerical FR = kgrout analyses.

624 Table 1. Permeability reduction factor values calculated in back analyses.

Permeability lefranc tests Range (m/s) Permeability Reduction FEM analysis permeability Layer From To field tests (m/s) factor

Teano station PR 5,80E-07 6,30E-05 5,00E-06 TA 3,00E-06 2,80E-04 3,00E-04 T2 6,00E-08 3,70E-04 9,00E-04 TA 3,00E-06 2,80E-04 5,00E-05 T1 6,00E-08 3,70E-04 8,50E-06 TA 3,00E-06 2,80E-04 2,40E-06 ST – – 2,00E-05 STa – – 5,00E-08 Mirti station PR 5,20E-08 8,30E-05 1,00E-06 – PR 5,20E-08 8,30E-05 1,00E-06 1 TA 2,90E-08 2,80E-04 1,00E-04 10 T2 2,90E-08 1,30E-05 3,00E-07 1 TA 2,90E-08 2,80E-04 5,00E-04 300 T1 2,90E-08 1,30E-05 5,00E-05 20 TA 2,90E-08 2,80E-04 5,00E-04 250 Gardenie station PR 2,10E-07 4,50E-05 5,00E-05 – TA 1,40E-06 2,10E-04 5,00E-04 – TA 1,40E-06 2,10E-04 5,00E-04 5 T2 4,20E-07 9,50E-04 5,00E-03 700 TA 1,40E-06 2,10E-04 9,00E-05 5 T1 4,20E-07 9,50E-04 5,00E-03 700 TA 1,40E-06 2,10E-04 5,00E-04 2000

−5 −4 TA: kft = 9.0 ⋅ 10 ÷ 5.0 ⋅ 10 m/s; Lefranc range: In pyroclastic soils, Lefranc tests, performed on from 1.4 ⋅ 10−6 m/s to 2.1 ⋅ 10−4 m/s. From Mirti sta- a limited section of the boreholes, may be not sig- −4 −4 tion tests: layer TA: kft = 1.0 ⋅ 10 ÷ 5.0 ⋅ 10 m/s; nificant due to local variation of the geotechnical Lefranc range: from 2.9⋅10−8 m/s to 2.8 ⋅ 10−4 m/s. properties. Grouting with MPSP system reduced the per- MPSP grouting system was appropriate to con- meability of almost two-three order of magnitude trol permeability of soils for complex fractured mainly for very fractured lithoid tuffs (T1-T2, Gar- lithoid tuffs and coarse clayey tuffs strata. denie) and very coarse clayey tuff (TA, Mirti and Gardenie). The efficiency of this consolidation REFERENCES techniques appears lower in presence of lithoid tuffs (T2, at Mirti station) and fine grained clayey Cashman, P.M. & Preene, M.. Groundwater Lowering tuff. in Construction. A practical guide. Taylor & Francis Group ISBN 0-419-21110-1. Corazza, A., Giordano, G. & De Rita, D. 2006. Hydroge- 8 CONCLUSIONS ology of the city of Rome. In Heiken (eds), Tuffs—their proprierties, uses, hydrology and resources. Geological The monitoring of dewatering field tests was use- Society of America Special Paper 408: 113–118. ful to better understand the groundwater flow in Iorio G. (2010). Problematiche geotecniche connesse con a typical complex geological sequence of Rome la realizzazione di stazioni metropolitane. PhD Thesis, Sapienza University of Rome (in italian). subsoil. Linney, L.F. & Withers, A.D. 1998. Dewatering the Numerical FEM analyses calibrated on pump- Thanet beds in SE London: three case histories. Quar- ing tests results allowed to estimate the average terly Journal of Engineering Geology, 31: 115–122. permeability values of pyroclastic deposits better Metro C. Progetto costruttivo—Metropolitana di Roma than classic Lefranc variable head tests results. Linea C—Tratte T4-T5.

625