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

Modelling and monitoring of a urban underground excavation

A. Cantone Department of Civil Engineering, II University of , R. Fico Structural & Geotechnical Designer Collaborator, Naples, Italy F. Cavuoto Structural & Geotechnical Designer, Naples, Italy A. Mandolini Department of Civil Engineering, II University of Naples, Italy

ABSTRACT: The highly urbanization of the area where Montesanto Station (Naples, Italy) is located and the need to guarantee the transportation service to travellers during the works gave birth to the plan of the combined use of 3D numerical analyses and the real time monitoring of significant parameters (displacements, strain, stresses and temperature) to confirm the set of design criteria assumed and calibrate the design parameters affecting the problem faced. The 3D analyses simulated the step by step excavation predicting the stress-strain behaviour; hence the comparison of the analytical predictions with the corresponding values derived through the monitoring (238 points being monitored) allowed to calibrate the model as the excavation advanced, thus refining the analysis itself and improving the safety level.

1 INTRODUCTION adopted, an intensive monitoring activity was planned under the supervision of the Civil The Montesanto Station of Line, Engineering Department (DIC) of the Second owned by S.E.P.S.A. (“Limited Public Service University of Naples: a monitoring system based Company”), is located in the very historical hearth on the observation method was thus contrived; of Naples (Italy) and can be deemed a critical junction of the city transportation network, given the closeness both to the Montesanto Station of the cable railway line and to the Montesanto Station of Metro , a few meters far on the West side (see Fig. 1). Since 2004 expansion works are in progress at the Station in order to build, what is more, a pedestrian tunnel providing an alternative exit to travellers from the two lines currently in service (Cumana and Circumflegrea Lines). The highly urbanization of the area where the opera is located, the nature of soils crossed, and mostly the need to guarantee the transportation service to travellers during the works, gave birth to the plan of different measures to protect the existing tunnels and to soil improvements in the excavation area. The design of the technical interventions that could minimize the effects of the excavation on the existing tunnels was carried out by means of both 2D and 3D numerical analyses. In order to control the proper execution of the ongoing works and to verify the adequacy of the design solutions Figure 1. Location of Montesanto Station.

231 starting from the data collected through the been already executed to sustain the top tunnels monitoring (sidewalls displacements, stress state sidewalls, by means of 10 m long micro-piles; in the liners, etc), the numerical model selected for a 0.60 m deep parterre on piles has been also the design analyses was controlled or calibrated, constructed to sustain the rails. if necessary; in the latter case, a new verification The excavation of the descending tunnel started of the design solutions adopted would have been on Oct. 4, 2005 and was completed by Sept. 2006 achieved and more appropriate auxiliary solutions (see Fig. 3), whereas the big service tunnel was would have been set. completed by Dec. 2006. In the early 2007 the first connection portion with Metro was built; the concrete casting of the as built tunnels was 2 DESCRIPTION OF THE OPERA & WORK carried out before the excavation of the pedestrian CHRONOLOGY tunnel. However, many works slowdowns occurred mainly due to the complexity of the specific yard The 40 m long, 7 m large pedestrian tunnel will be position. perpendicular to the two upper tunnels in service In April 2008 the construction of the pedestrian (see Fig. 2a); the distance between the upper rails tunnel below the two railways started. According and the design tunnel crown will be of just 3.5 m to the design, 7 advancement steps had to be (see Fig. 2b); the connection of the pedestrian performed: a first 6 m advancing step, five further tunnel with the station outside will be guaranteed 4.5 m steps, and a final 6 m step. At each step, by escalators, therefore a 30° sloped 8 m diameter before cutting and placing the pre-support liner, the tunnel has been already constructed (see Fig. 2c), soil improvements were accomplished, while the turning into a big service tunnel (11 m diameter) permanent reinforced concrete liner was planned when going down at the design tunnel depth. to be cast in place backwards after reaching the The intervention area soil, upper than the water dig bottom, in order to speed up the advancement, table, is rather heterogeneous: the descending given the reduced room available. tunnel housing the escalators and the big connection service tunnel lie within a region made of Neapolitan yellow tuff, whereas the tunnel to build will lie in pyroclastic sand (except for a short first portion). The excavation, carried out by employing the traditional technique (by cutting and temporary supporting with steel arches and shotcrete, before the permanent concrete casting) will be executed after the accomplishment of different soil improvements: a ring of sub- horizontal metallic pipes in the crown (7° sloped, 10 m long), 5.5 m overlapped with those installed in the previous step; the excavation ground will be improved by glass-reinforced plastic nailing (20 mm diameter, 12 m long); jet-grouting sub- horizontal columns will support the pre-support bases. Moreover, consolidation interventions have

(a) (b) 2 E IN L O T

(c)

Figure 2. Pedestrian tunnel views: (a) plan; (b) section 1-1; (c) section 2-2. Figure 3. Tunnel excavation snapshots.

232 The last excavation step was completed at Dic 1, the simulation; in particular, the elastic modulus

2008. The activities were very complex, due to both of the pyroclastic sand Es was assessed; the geo- the reduced room available (Fig. 3a) and the need technical investigation lead to assume a value of to manually carry out the cutting of the sub-verti- Es ranging between 50 and 100 MPa; within this cal micro-piles consolidating the ground from the range the displacement derived with the analyses > sidewalls and the foundation slab of the upper tun- varied of about 5 mm, whereas for Es 100 MPa nels, where these were interfering with the excava- the displacement variation highly reduced. tion front (Fig. 3b). The results that were finally accounted for cor- = respond to Es 100 MPa, considered the most reli- able value according to the experimental campaign 3 ANALYSIS previously performed; the displacement of the four sidewalls relating to the Circumflegrea and the The numerical analyses were based on both two- Cumana Lines tunnels were thus derived, together dimensional and three-dimensional simulations, with the corresponding stresses; moreover, a fur- performed by means of Plaxis2D software and ther inspection was provided by the stress analysis FLAC3D software, respectively; given the com- of the temporary support of the design tunnel. plexity and the significance of the project, the The displacements of the most relevant points two-dimensional-based design assumptions were on the existing tunnels (corresponding to some confirmed and refined with the 3D analyses. In par- of the points monitored, see next section) were ticular, the FLAC3D is an explicit finite-difference derived for each of the 7 advancing steps; program for engineering mechanics computation Table 1 reports the vertical (w) and the horizontal that can simulate the behavior of three-dimensional (u) components of the estimated displacements structures built of soil, rock or other materials that relating to the four sidewalls of the two existing undergo plastic flow when their yield limits are tunnels caused by the excavation occurring beneath reached. The full dynamic equations of motion are (the values are ordered by following the excavation used, even when modelling systems are essentially advancement direction). static; this enables FLAC3D to follow physically unstable processes without numerical distress. The model assumed in the FLAC3D is consti- Table 1. Displacement of sidewalls relating to the exist- × × 3 tuted by a 72 40 50 m parallelepiped, having ing tunnels: Circumflegrea (Circ.) and Cumana (Cum.) 8 different solid groups, each with different and specific mechanical properties (see Fig. 4); besides Circ. Cum. Circ. Cum. the modelling of the different types of soil and the structural elements of the existing tunnels (tuff), w (mm) u (mm) the step by step soil improvements foregoing the tunnel excavation have been integrated as well. Step d (m) 12341234 All the mentioned groups are characterized by the 00 00000000 Mohr-Coulomb constitutive model. 1 6,0 17,6 12,9 13,0 14,9 4,2 5,4 3,9 6,3 Several parametric analyses were performed 2 10,5 21,4 13,0 13,1 15,0 4,7 5,4 3,9 6,4 in order to weigh the main variables affecting 3 15,0 21,6 13,6 13,5 15,0 4,8 5,7 3,6 6,6 4 19,5 21,6 17,2 15,4 15,3 4,6 6,1 2,7 6,8 5 24,0 21,7 19,5 19,6 15,6 4,8 5,3 2,7 6,8 6 28,5 21,7 19,9 20,5 15,6 5,0 5,0 3,1 6,8 7 34,5 21,8 20,0 20,5 18,8 5,0 5,0 3,0 7,5

Wall 1

R3 V1 PEDESTRIANR2 TUNNEL 10 11 R1 9 1 UNDER CONSTRUCTION 3 2 MONTESANTO 4 3 bis 5 6 T A STATION 7 NMEN 8 ALLIG E.L. - LINE T B REA NMEN UMFLEG IG CIRC E.L. - ALL 3

SEC.A E

NA LIN

SEC.B R(int.) A SEC.C CUM D1 MENT C D2 LLIGN D3 E.L. - A D4 V2 D5 R(int.) D6 C1 SEC.D V D7 C2 SEC.E R2 P1 C3 SEC.F T D V4 C4 R(int.) NMEN P2 C5 ALLIG P3 C6 B1 E.L. - P4 C7 B2 B3 R3 B4 R(int.) B5 R1 B6 A1 B7 A2 Wall 2 A3 A4 A5 A6 A7

Bhk Otil Hi tll t l l Figure 4. FLAC3D model view. Figure 5. Monitoring system scheme.

233 4 MONITORING AND CONTROL PLAN and to the first sidewall of the Circumflegrea Line. Figure 6 reports the displacements observed The monitoring and control plan (see Fig. 5) is along this wall from the 29th of June 2006 (zero given by two independent measurement systems, value) to the 29th of January 2009 (last available one automatic the other manual. scanning); vertical dot-dashed lines point out the The automatic system provides the control of most significant dates of the excavation history. vibrations, movements and stresses variations All the benchmarks had vertical displacements, induced in the liners by the excavation operations: with a maximum value of 24 mm at benchmark the first is based on the use of 4 speedometers laid # 6. The deformed shape of the wall clearly shows on structures placed both outside and inside the a sagging zone and a hogging zone. The maximum = × −4 station; the second is based on the employment angular distortion recorded (βmax 7.1 10 ) is of 56 electro-levels (36 horizontal and 20 vertical) lower than the threshold prescribed in the scientific to control the sidewalls movements, placed on a literature to avoid the structural damage of the type × −4 < < × −3 18 m long line along the new tunnel axis; the third of structure investigated (8 10 βadm 3 10 , system is based on the use of load cells inserted Day 2000). within the existing tunnels liners and on vibrating Figure 7 depicts the displacements gauged on wire gauges installed on both the tension and the the 1st Circumflegrea sidewall so far, confirming compression sides of the pre-support steel arches that the automatic and manual control systems are of the pedestrian tunnel. In particular, to control in good agreement. the stress state of the existing tunnels liners, 4 monitoring sections are identified, two for each tunnel, each one having 5 load cells; as for the con- 30 60 trol of the stress state in the pre-support liner of 25 (1) (2) (3) (4) 50 20 TEMP 40

the pedestrian tunnel, five sections were instru- 15 30 T [°C] 10 20 mented, each one having 6 control points, with a 5 10 total of 12 gauges per section liner. 0 w 0 -5 MIN -10 In February 2007 two cracks-gauges were -10 -20 installed as well, in order to monitor the opening w [mm] -15 -30 -20 w MAX -40 over time of a crack occurred on Vicereale Wall, -25 -50 retaining the cable railway line. -30 -60 All sensors installed converge into automatic switchboards that record the data at time intervals 07/11/05 29/06/05 03/01/08 13/05/08 31/01/09 scheduled depending on the excavation operations 18/03/06 27/07/06 05/12/06 16/04/07 25/08/07 21/09/08 8,E-04 and via modem transfer them to the Administration, (1) (4) 7,E-04 (2) (3) to Building Contractor Bureau, and to the DIC. 6,E-04 The manual system is given by a network of 5,E-04

benchmarks and optical surveys: 40 measurement MAX 4,E-04 benchmarks and 13 reference benchmarks, control- 3,E-04 led by means of topographic levelling, have been 2,E-04 1,E-04 installed to monitor the displacements of both the 0,E+00 existing tunnels sidewalls and the structures nearby the station (a residential building and the Wall 1, 07/11/05 21/09/08 31/01/09 29/06/05 05/12/06 16/04/07 25/08/07 03/01/08 13/05/08 18/03/06 see Fig. 5); plus, 34 optical surveys, inspected by 27/07/06 employing a total station, have been placed in Configurazione deformata del muro order to gather the convergence measurements in 5 the tunnel and to monitor a supporting wall that 36sib 4 5 8 0 will be directly influenced by the work in progress (1) (2) activities (“Paradise Stairs” supporting wall). -5 The monitoring plan is completed by the -10 (3)

temperature control on site by means of thermal w [mm] -15 sensors properly placed. The total number of (29/01/09) -20 points under control is 239 (see Fig. 5). (4) -25 0 5 10 15 20 25 30 35 40 45 50 55 d [m] 5 MEASURES Figure 6. Time evolution of displacements (1) work The most significant displacement measures start; (2) completion of descending tunnel; (3) completion gathered relate to the cable railway restraint wall of service tunnel; (4) pedestrian tunnel excavation start.

234 5 The maximum compression value (∼9,3 bar) was detected on cell B2 during the 3rd excavation step. 0 optical survey benchmarks It can be observed that the pressure variations -5 electro-levels seem to be strongly influenced by the temperature variations. -10 w [mm] -15 6 COMPARISON OF THEORETICAL -20 AND EXPERIMENTAL RESULTS

-25 The calibration of numerical analyses carried out to predict the displacements expected after each excavation step, strictly depended on the mechanical 07/02/06 07/10/06 05/02/07 06/06/07 05/10/07 03/06/08 02/10/08 31/01/09 10/10/05 08/06/06 03/02/08 parameters initially based on the experimental test Figure 7. Displacements of 1st sidewall. results; afterwards, with the measures progressively gathered in terms of sidewalls displacements, existing and as built liners stresses, a further 10,0 40 B2 (1) (1) 7,5 30 calibration of such parameters could be performed B3 5,0 20 in the pipeline.

B1 SECTION 1 B4 2,5 10 T [ °C] 0,0 0 After the completion of the fifth excavation Circumflegrea Line -2,5 -10 step, the comparison between the analytical and Direction of the excavation -5,0 -20 p [bar]

Δ -7,5 -30 B 2 B 4 measured displacement values could be performed, B6 -10,0 B 5 B 6 -40 -12,5 B 7 Temp. -50 as shown in Figure 9a (the zero displacement value B5 SECTION 2 B7 -15,0 -60 ? Circumflegrea Line p < 0: Decompression refers to the construction of the pedestrian tunnel -17,5 ?p > 0: Compression -70 Direction of the excavation -20,0 -80 below the two railways). It is clear that the expected displacements 10/10/08 06/03/07 18/05/07 11/10/07 23/12/07 05/03/08 17/05/08 29/07/08 22/12/08 06/03/09 30/07/07 overestimate the values measured; it is believed that Figure 8. P and T data of Circumflegrea. (1) 1st break- such disagreement is due to three main factors: through start. • the uncertainty related to the assessment of the

Elastic modulus, Es; • the model did not take into account the aforemen- As for the accuracy of the tunnels displacements tioned reinforcement of the upper sidewalls; measured, the scientific literature does not give spe- • the simulation of the progressive excavation cific indications, despite the existence of case stud- determines displacement values that do not ies relating to tunnels interested by new forthcoming take into account the time dependence; this tunnel excavations. Nevertheless, it is certainly pos- implies that the displacements expected involve sible to set a displacement threshold related to the delayed adjustment phenomena (due to the vis- structural safety of the two tunnels in service. cosity behaviour of soil) that might occur along For this purpose it was necessary to know the with time. mechanical and geometrical characteristics of the existing tunnels liners, and to assess the ini- After the completion of the (last) seventh exca- tial stress state. Hence, two different experimental vation step, the comparison between analytical and campaigns were performed along with time (June measured displacement values gave the two curves ’05 and February ’07), made of coring, single and shown in Figure 9b; a slight approach of the meas- double jack tests carried out on specimens taken ured displacements trend with respect to the theo- within the intervention area soil. retical values can be observed, related to the first The tunnels liners were found to be made of three sidewalls which were more influenced by the piperno masonry, red bricks masonry and variable excavation steps 5 to 7. thickness concrete strata. The single jack tests Finally, the comparison of analytical versus allowed measuring the service stresses, ranging displacement values measured after one year from between 0.35 and 2.40 MPa; the double jack the excavation completion was performed, as tests showed an ultimate stress value higher than depicted in Figure 9c; a further approach between 3.50 MPa. The displacement scannings were aided the two curves is observed. by stress scannings given by load cells installed on It is clear from Figure 9c that the expected two sections of the Circumflegrea tunnel astride the soil sinking still overestimate the corresponding tunnel to build (the 1st in March ’07, and the 2nd measured values, but at a lower rate with respect in June ’08); the pressure variations are reported in to Figure 9b: this confirms that time plays a Figure 8, together with the temperature variations. significant role in this process; it is believed that a

235 25 21,7 19,5 19,6 29/07/07 22/04/10 07/12/05 25/06/06 14/02/08 31/08/08 19/03/09 04/10/09 22/05/05 10/01/07 20 08/11/10 15,6 0 15 a) excavation step #5 a) sidewall #1 -2 measured settlements

w (mm) 10 expected settlements -4

-6 5 3,1 w (mm) 1,1 0,7 0,2 -8 0 -9,6 1234 -10 sidewall 0 25 b) sidewall #2 21,8 20,0 20,5 18,8 -2 20 b) excavation step #7 -4 15

measured settlements w (mm) expected settlements -6

w (mm) 10 6,2 4,6 -7,8 5 3,5 -8 0,2 0 0 -1 c) sidewall #3 1234 -2 sidewall 25 -3 21,8 20,0 20,5 -4 18,8 w (mm) 20 -5 c) one year later -6 15 measured settlements -6,8 -7 expected settlements 0,0

w (mm) 10 8,2 6,4 5,4 -0,2 d) sidewall #4 5 -0,4 1 -0,6 0 -0,8

1234 w (mm) -1,0 -1,2 sidewall -1,4 -1,4 Figure 9. Comparison between theoretical and meas- -1,6 ured vertical displacements. Figure 10. Displacements vs time behaviour. further approach will be acquired between the two trend lines along with time. model approach fits the effective general behaviour This is confirmed by displacements-to-time of the complex space investigated. trend of Sidewall 1 (see Fig. 10a), Sidewall 2 The stability verification of the 3D model (see Fig. 10b), Sidewall 3 (see Fig. 10c), and Side- corresponding to the displacements derived was wall 4 (see Fig. 10d): the sinking trend lines have supported by the verification of the stress levels been extrapolated up to one year after the last of the existing liners; in particular, the maximum measuring; the trend along with time goes to an differential vertical displacements observed horizontal asymptotic value, corresponding to the between the adjacent sidewalls after the 5th and the occurred settlement process, still ongoing at the 7th advancing simulated step was derived between present data. the two sidewalls of Circumflegrea tunnel; the It can be concluded that the expected val- maximum compressive stress value derived on the ues have to be deemed an upper bound for the relevant tuff liner was found to be 3.5 MPa, which measured data. is lower than the ultimate value derived from the Nevertheless, the trend of analytical versus experimental testing. Figures 11a and b illustrate measured displacement curves confirms that the the stress state just discussed.

236 With reference to the new pedestrian tunnel serving the Montesanto Station, it has been shown how an integrated approach between 3D numerical analyses and real time monitoring of significant parameters as displacements, strain, stresses and temperature can confirm the set of design criteria assumed for the specific project, and in the meantime can refine their use by extending its application field through the proper calibration of the manifold parameters involved in the definition of complex works interacting with earthy materials which behavior is very unclear and thus needing Figure 11a. Maximum compressive stresses of liners deep investigations to support the design. (excavation step #5).

ACKNOWLEDGMENTS

The work presented herein is part of the research agreement between Ferrosud 2 S.C.R.L. and the II University of Naples. Special acknowledgments go to Mr. A. Allagrande from S.E.P.S.A., Mr. S. Buttarelli, Mr. P. Guarino and Mr. L.Murena from Ferrosud 2, and to Mr. L. Campobasso, project manager.

REFERENCES Figure 11b. Maximum compressive stresses of liners (excavation step #7). R.W. Day—Geotechnical Engineer’s Portable Handbook— New York, 2000—McGraw Hill. Itasca Consulting Group, Inc. 2006. FLAC3D—Fast 7 CONCLUSIONS Lagrangian Analysis of Continua in 3 Dimensions, Ver. 3.1 User’s Guide. Minneapolis: Itasca. In the present paper the main issues related to the accomplishment of underground structures in deeply urbanized and highly historical and monumental areas has been illustrated.

237