International Conference on Tunnel Boring Machines in Difficult Grounds (TBM DiGs) Singapore, 18–20 November 2015

EXAMINATION OF EXCAVATION CHAMBER PRESSURE ON A 17.5 M DIAMETER EARTH PRESSURE BALANCE TUNNEL BORING MACHINE

Kamyar Mosavat1, Mike Mooney2

1Application Engineer/Geotechnical, The Robbins Company, Seattle, WA, USA 2Professor & Grewcock Chair of Underground Construction & Tunneling, Colorado School of Mines, Golden, Colorado, USA, mooney@mines.edu ABSTRACT: This paper examines excavation chamber pressure behavior within a 17.5 m diameter earth pressure balance tunnel boring machine (EPBM) used on the Alaskan Way viaduct replacement tunnel project in Seattle, Washington, USA. The study examines behavior during the first 150 rings of tunneling (10% of the project) through till and till-like deposits, granular soils, and cohesive silts and clays. Machine data, including excavation chamber pressures, screw conveyor pressures, soil conditioning inputs, and key operating parameters such as thrust, cutterhead and screw conveyor torque, cutterhead and screw conveyor rotation speeds, etc., were studied in detail to determine what parameters influenced chamber pressures and how. The results of detailed EPBM data analysis supported with field lab test results from muck testing produced a number of key findings. Excavation chamber pressures measured by 12 pressure sensors varied up to 3 to 3.5 bar from crown to invert. Chamber pressures varied during ring mining and standstill, and the responses from different heights in the chamber were synchronous. Chamber pressure variations during excavation were influenced by changes in volumetric flow rates into the chamber via the cutterhead and out of the chamber via the screw conveyor. Changes in gradient both locally and globally provide information about muck density under pressure and whether the chamber material is locally being compressed and decompressed. Horizontal differences in chamber pressure were evident throughout mining and standstill. When cutterhead rotation was clockwise, left side chamber pressures were higher, and when cutterhead rotation was counterclockwise, right side chamber pressures were higher. The fluctuation in these horizontal differences was influenced by many parameters including a possible compressed air gap at the crown, steel/muck adhesion, and conditioning. KEYWORDS: Earth pressure balance; tunnel boring machine; apparent density; soil conditioning; excavation chamber pressure 1. INTRODUCTION

A 17.5 m diameter, 99.4 m long, 7000 ton earth pressure balance (EPB) TBM is being used to excavate the Alaskan Way Viaduct tunnel in Seattle, USA. Large diameter pressurized face TBM tunnelling in this environment presents a number of unique challenges, including almost continuously changing mixed face conditions and a significant increase in lateral earth and pore water pressure from crown to invert. Maintaining adequate face support via chamber-filled earth pressure balance is critical to tunnel face stability and ground/structure deformation control. This paper examines the chamber pressure during the first 140 rings of excavation of the tunnel, where essentially no ground deformation was observed (Cording et al. 2015). The tunnel cover varies from 3 to 21m over this 280m and the soils encountered were highly variable. The magnitude of chamber pressure and its changes during excavation and ring build is presented, as well as the observed vertical and horizontal chamber pressure gradients.

2. BACKGROUND

2.1 Geology

The SR99 tunnel alignment is located in the Puget Sound area of Seattle, where the geological condition is complex due to glacially influenced sedimentary deposits (See Figure 1). During several glacier advances, a new deposit of sedimentary layered over previous material including glaciolacustrine clays and silt, glacial outwash sands and gravels, glacial till and till-like soils. Figure 1 illustrates the tunnel alignment and geological

layering based on the geotechnical baseline report. Approximately 60% of the tunnel is in granular soil and 30% in full face cohesive clay and silt. This study focuses on the initial 10% of the alignment that is in a combination of soils and layers (See figure 1b).

Figure 1: Project geology; (a) SR99 tunnel gelogy and tunnel alignment; (b) Geology at the face including grain size distribution result of 10% of the project; (c) GDR borehole data illustration

2.2 Excavation Chamber Pressure

The cutterhead, designed with 37% opening, is equipped with a significant number of cutting tools, including 20 double disc cutters, 32 scraper bits, 260 cutter bits, 87 precutting bits, 69 replaceable precutting bits, 12 trim bits, 14 wear detection bits 45 emergency bits and 2 copy cutters. 39 injection ports are available for foam and polymer conditioning, including 22 ports on the cutterhead, 8 ports in the bulkhead, 3 ports along the first screw conveyor and six ports along the second screw conveyor. A significant number of high water pressure ports are equipped on the cutterhead, bulkhead, spokes and the mixing arm to prevent possible clogging (see figure 2).

Figure 2: Excavation chamber; (a) Cutterhead assembling process; (b) Mixing chamber with center agitator in red, static bars and cutterhead legs all attached to bulkhead

The excavation chamber was outfitted with twelve earth pressure sensors (EPS) at six elevations as shown

in Figures 2b and 3. The EPSs are mounted in the plane with the bulkhead and oriented to measure horizontal total stress. The top EPSs (at locations L1 and R1) are 2.55m from the crown and the bottom EPSs (L6 and R6) are 1.55m from the invert (See figure 3). EPSs are paired at each of the six elevations, with notation L and R denoting left and right, respectively as looking forward.

Figure 3: Excavation chamber earth pressure sensors: (a) EPS horizontal distances, (b) EPS vertical distances (in m).

EPS data reflects the total lateral earth pressure p ( xp where x is oriented in the direction of tunneling) measured in the chamber that is theoretically equal to pore fluid pressure u plus lateral effective earth pressure

x in the excavation chamber (Equation 1), and related to the vertical effective stress z through the coefficient of lateral earth pressure K of the chamber soil (Equation 2). The soil is assumed to be partially saturated due to the presence of foam-induced air bubbles, and therefore, pore fluid pressure u theoretically reflects a combination of pore air and pore water pressures. Assuming that the matric suction is negligible, u reflects the pore air pressure.

up xx (1)

CUTTERHEAD PEDESTAL

CENTER AGITATOR

STATIC MIXING

PRESSURE SENSORS

Main Bearing

(a) (b)

uKp zx (2)

The vertical gradient of the total lateral chamber pressure, denoted here with , is reflected in Equation

(3), where K is the coefficient of lateral earth pressure, f is the unit weight of the chamber fluid (weighted

average of air and water), and is the buoyant unit weight of the chamber soil ( f , where is the

total unit weight of the chamber soil). Assuming K is constant with depth, the resultant vertical chamber

pressure gradient is the sum of K and f . Equation (3) can also be expressed in terms of total muck unit

weight and fluid unit weight f as shown in Equation (4). Inspection of Equations (3) and (4) reveals three

unknown parameters: K, f and or .

fzx K

dz

du

dz

dK

dz

d

(3)

KKKdz

dfff

x 1 (4)

It is difficult to infer magnitudes of these individual parameters from chamber pressure gradients without

making assumptions. When considering K, for example, the scraping and ingestion process destroys soil fabric and therefore K0 conditions no longer prevail. There is no literature on the topic of K resulting from this process. If the chamber soil is well-conditioned such that the effective stress is relatively low and the shear strength is negligible, then K = 1 (fluid-like) is a reasonable assumption. In this case, the chamber pressure gradient equals per Equation (4). To this end, the chamber pressure gradient can be used as a quality control measure of apparent conditioned soil density or muck density (Bezuijen et al. 2005, Bezuijen and Talmon 2012). If the soil is not properly conditioned (locally or overall), K ≠ 1 and likely K < 1. In this case, the gradient underestimates per Equation (4). Finally, Equation (3) illustrates a lower bound chamber pressure

gradient. If the chamber pressure is well-conditioned with foam such that K = 1 and the water content is

relatively low, f can be considered negligible and the gradient equals the bouyant unit weight of a loosely

bound soil. Values of depend on the grain size distribution in granular soils but can be as low as 10 kN/m3.

3. RESULTS

3.1 EC Pressure during Ring Advance

Relevant TBM data from a typical ring advance (ring 77) is presented in Figure 4. The soil was primarily comprised of till deposits and cohesive silts and clays. A sample from ring 77 revealed 12% gravel, 42% sand and 46% fines. Figures 4a and 4b show key operational parameters including advance rate (AR), cutterhead rotation speed ), center agitator rotation speed ( ), thrust force (F), cutterhead torque (TCH), screw conveyor 1 torque (TSC1), and screw conveyor rotation speeds ( and . Excavation of the 2m long ring took 75 min. and included two restarts, at 38 min. and at 54 min., the second of which was to reverse the cutterhead rotation direction. As illustrated in Figures 4a and 4b, the TBM advanced steadily at 30-35 mm/min under a thrust force of 100-125 MN and cutterhead torque of 25-29 MN-m. The cutterhead was rotated nominally at 1.0 rpm and ribbon screw conveyors at 2-5 rpm. Figure 4c shows the conditioning efforts, including 40% foam injection per volume of excavated soil and 15% water injection ratio, both through the cutterhead into the formation soil, and 20% bentonite injection ratio into the excavation chamber. The foam injection ratio reflects magnitudes at atmospheric conditions; under the 1-4 bar pressures in the formation and chamber, the foam injection ratio is significantly less due to the compression of air (see Mori et al., 2015), e.g., 5-20%. Figure 4d shows the three guillotine gate openings (1500 mm maximum) along the ribbon screw. Details regarding the screw conveyor and its operation can be found in Mosavat (2015).

Figure 4: Ring advance 77 operational parameter data, excavation chamber pressures and screw conveyor pressures

Excavation chamber (EC) and screw conveyor (SC) pressures are presented in Figures 4e and 4f, respectively. SC pressures are shown for completeness but are not the focus of this paper. It is worth mentioning that SC pressures dissipate along the screw conveyor as would be expected, from 3 bar (location 2 near the excavation chamber to 0.4 bar near the belt conveyor. EC pressure varies from 1.2 bar at the uppermost L1 and R1 sensors (2.55 m below crown) to 4.2 bar at the lowermost L6 and R6 sensors (1.55 m above invert). EC pressures at all levels vary considerably and synchronously throughout the excavation process (t < 1:17) and behave smoothly during stand still/ring build (t > 1:17). Leftside and rightside EC pressures, shown in different shades of the same color, are different at all levels. During the early excavation period (t < 0:35), the rightside EC pressures are greater than leftside EC pressures and the cutterhead is rotating counterclockwise (negative in Figure 4a). Upon cutterhead reversal to clockwise rotation (t > 0:55), leftside EC pressures exceed rightside EC pressures at most elevations. During ring build (t > 1:17), leftside and rightside EC pressures tend toward equilibration at most but not all levels. Leftside vs. rightside EC pressure differences have been identified in previous studies (Bezuijen et al. 2005). There are likely a number of reasons for this difference, including variations in chamber soil density caused by rotation-induced compression/decompression, sidewall friction and a nonuniform muck surface near the chamber crown in the presence of an air bubble. The significant variation in EC pressure from crown to invert also results in much different conditioned soil behavior, i.e., higher shearing resistance and density near the invert.

3.2 EC Chamber Pressure Gradient

EC pressure distributions at four discrete times during and after ring 77 excavation are shown in Figure 5. Left and rightside EC pressures at each level are presented as are the leftside and rightside vertical pressure gradients and , respectively. Shown for comparison are the geostatic groundwater pressure and geostatic total lateral earth pressure assuming active earth pressure conditions (considering full overburden vertical effective stress). The latter is a commonly used chamber pressure for pressurized face tunneling. The groundwater elevation was assumed from borehole data in the geotechnical data report. Figure 5a shows EC pressure gradients at t = 0:25, during excavation. The EC pressures are greater than the geostatic groundwater pressure and at or lower than total lateral earth pressure assuming active conditions. While some variation in gradient with depth is evident (particularly at the extreme top and bottom, the average is 0.18 bar/m (18 kN/m3) and is 0.14 bar/m (14 kN/m3) between sensor levels 2 and 5. It is conceivable that clockwise cutterhead motion loosens through lifting the right side chamber soil and densifies through compression the leftside chamber soil. If K = 1 for this soil (unknown), these gradients would reflect the total unit weight of the conditioned chamber soil. The local gradients between sensor pairs 1 and 2 exhibit interesting data. According to the data, the L1 and R1 EC pressures are the same; however, is much lower than , 0.10 bar/m vs. 0.20 bar/m, respectively. These data may suggest that an air bubble is present at the crown and an unlevel chamber soil surface near the crown (lower on the leftside).

Figure 5b presents the EC pressures at t = 1:15 near the end of excavation. These EC pressures remain above geostatic groundwater pressure and below total lateral pressure assuming active conditions. Leftside EC pressures are greater than rightside EC pressures, consistent with clockwise cutterhead rotation. Interestingly, the gradients are similar, = = 0.19 bar/m (between sensor levels 2 and 5) and there appears to be no influence of cutterhead rotation on gradient. These would equate to high total densities if the gradient was interpreted as the apparent density. Local gradients near the top sensors reveal behavior consistent with that observed in Figure 5a suggesting an air bubble and an uneven muck surface.

EC pressures at two different times after excavation, during and after ring build, are presented in Figures 5c and 5d. Here, left and rightside EC pressures have equilibrated supporting the argument that the mechanical action of the cutterhead and perhaps center agitator are the primary cause of the difference. The distributions with depth suggest two zones for both t = 1:20 and 2:30: the lower portion of the chamber soil exhibits a higher gradient ( = = 0.20 bar/m) while the upper portion of the chamber soil exhibits a gradient equal to 0.13 bar/m. The pressure levels during standstill are lower than those during excavation (see Figure 4 as well). The EC pressures are lower than the estimated geostatic groundwater pressure assuming no tunneling-induced excess pore water pressure. Because EC pressure should equilibrate with groundwater pressure and in fact be greater, we assume the estimated groundwater pressure is overestimated here.

Figure 5: Left and rightside excavation chamber pressures and gradients for ring advance 77, (a) early during mining, (b) just before end of mining, (c) standstill immediately after mining, and (d) well after mining

3.3 EC Chamber Pressure Gradient along Alignment

In a manner similar to that described above, the vertical pressure gradient was estimated along ring advances 58-147 (see Figure 6c). Here, the gradients and were estimated during ring build (after excavation) as single scalar values per ring advance using linear regression between EPS 1 to 6 (as illustrated above in Figure 5). These averaged gradients do not reflect the local gradients observed; however, they do provide a global measure of chamber behavior for trend analysis). Also presented in Figure 5b are muck unit weight test results determined by cylinder testing of grab samples taken from the conveyor belt (tests conducted under atmospheric conditions). Figure 6a presents the geological profile together with grain size information collected through sieve analysis of conveyor samples. Figure 6b presents foam, water, bentonite and polymer injection ratios along the alignment. Both the magnitudes of EC pressure gradient and the change with ring advance are worth noting. Magnitudes of and were found to be lower than belt sample unit weight results for a considerable portion of the alignment (up to ring 110), where after, the gradient matches the belt sample unit weights reasonably well. However, the unit weight under 1-4 bar chamber pressures should be higher than belt scale samples tested under atmospheric conditions. This gradient-based underestimation of

unit weight prior to ring 110 is consistent with a reasonable scenario of K < 1. In addition, magnitudes of belt scale unit weight and gradients vary along the alignment, gradually decreasing from ring 58 to 102, then increasing to ring 130 and remaining constant thereafter. The transition to higher gradient and belt sample unit weight is coincident with the geological transition from cohesive soil and till into predominantly granular soil. There is some correlation of gradient with conditioning volumes, i.e., lower gradients correspond higher foam and bentonite injection.

Figure 6: a) tunnel geology b) Soil conditioning c) chamber pressure gradient 3.3 EC Chamber Pressure Variations An analysis was performed to relate EC pressure fluctuations observed during excavation (see Figure 4e) to volumetric flow rate changes of muck through the excavation chamber. The volumetric flow rate of material into the excavation chamber Qin is comprised of formation soil ingested through the cutterhead plus injection of foam, bentonite and polymers directly into the chamber. The outgoing volumetric flow rate Qout is based on screw conveyor discharge. Both of these quantities can be readily calculated from TBM data. A positive change in chamber volumetric flow rate outin QQQ results from the chamber ingesting more soil and conditioning

volume than the screw conveyor is discharging. The opposite is true for a negative Q . A comparison of L4

and R4 EC pressures and Q is shown in Figure 7 for ring advance 77. Here, Qin and Qout were calculated

using 5 second increment data averaged over 1 minute. Inspection of Figure 7 shows periods of negative and positive Q that is to be expected, with the observed fluctuations in AR, and . It is also clear that,

for the most part, areas of positive Q correlate with increases in EC pressure (e.g., t = 0:05, 0:47, 1:00) and

areas of negative Q correlate with decreases in EC pressure (e.g., t = 0:07, 0:12, 0.27, etc.). This relationship

is logical in that an increase in Q implies that the EC material is compressing (to accommodate the increase).

The compression of chamber material increases the chamber pressure. The behavior is similare in reverse in that a decrease in Q (either through a net decrease in inflow or net increase in volumetric outflow) leads to

muck decompression and a decrease in chamber pressure.

Figure 7. Excavation chamber pressure and change in volumetric flow rate during excavation of ring advance 77

A closer look at the EC pressure vs. Q relationship for ten time intervals during ring advance 77 is shown

in Figure 8. The majority of these highlighted intervals exhibit the characteristic pressure-volume behavior described above. The relationship is not perfect. For example, the increase in EC pressure during interval 2 is consistent with a reported zero change in Q . This perhaps suggests some calibration is required in terms of

calculating Qin and Qout. Nevertheless, the overall consistency in the relationship is clear. It is logical that the magnitude of Q and rate of change of Q would be proportional to the resulting

change in EC pressure. A further analysis of the ten time increments in Figure 8 are plotted in Figure 9. Here, the Q response has been integrated over its time increment to produce a volumetric change V that reflects the change in muck volume per increment of time. Note that this does not suggest the volume of the excavation chamber is changing; it is not. Each V is plotted vs. the EC pressure change P over the same increment. The results presented in Figure 9 reveal a linear relationship between V and EC pressure change P , illustrating that the magnitude of observed P is proportional to the magnitude of V . The slope of this relationship relates to compressibility of the conditioned chamber soil (in units of volume/pressure), an important measure of the quality of conditioned soil (Mori et al. 2015). The development and use of this relationship as a surrogate measure of compressibility has the potential to provide real time assessment of soil conditioning. Here, it is important to note that this is simplified analysis in that only one pressure sensor reading was used and the estimates of Qin and Qout have uncertainty. To this end, the magnitude of the slope as a quantitative measure of compressibility requires further development. Nevertheless, the linearity and general trend capture the

observed behavior. An analysis of this type for the rest of the tunnel rings is underway and beyond the scope of this paper.

Figure 8. Comparison of EC pressure change with volumetric flow rate change for ring 77

Figure 9. Observed relationship between change in muck volume and EC pressure for ring 77

4. CONCLUSIONS

A study of EPB excavation chamber pressure was performed on early data from the SR 99 viaduct replacement tunneling project in Seattle, WA. Twelve excavation chamber (EC) pressure sensors at six elevations provided rich data throughout ring excavation and standstill/ring build. EC pressures varied by 3-4 bar from crown to invert. Leftside and rightside EC pressures were noticeably different during excavation, due primarily to the mechanical action of the cutterhead and perhaps the center agitator. Specifically, EC pressures were greater on the side where the cutterhead was moving upward. Leftside and rightside EC pressure tended to equilibrate during standstill. An examination of the distribution of the EC pressure vertically revealed locally fluctuating pressure gradients and global trends during excavation and ring build. The gradient varied locally, from the uppermost sensors indicating a potential air bubble, to the bottommost sensors indicating greater compression under much higher pressure conditions than the top of the EC. Globally, the EC chamber pressure gradient varied along the first 150 rings, tending to change with geology and soil conditioning. Increases in observed EC pressure (in ring 77) were found to occur simultaneously with decreases in volumetric flow rate changes through the EC, and vice versa. A relationship between EC pressure and volumetric flow rate change into and out of the EC was developed. The resulting linear relationship relates to the compressibility of the EC material and can serve as an important real time indication of conditioning efficacy. ACKNOWLEDGEMENTS The authors would like to thank Seattle Tunnel Partners for their support throughout this study. The study would not have been possible without their assistance. REFERENCES Mosavat, K. (2015). Examination of excavation chamber pressure behavior on a 17.5 m diameter earth pressure

balance tunnel boring machine. Master’s thesis, Colorado School of Mines. Cording, E., Nakagawa, J., Painter, C., McCain, J., Vazquez, J. (2015). Controlling ground movement on the

SR 99 Alaskan way viaduct replacement tunnel. Proc. Rapid Excavation and Tunneling Conference, New Orleans, LA.

Bezuijen, A., Talmon A.M., (2014). Soil pressures at the cutting wheel and the pressure bulkhead of an EPB-shield. Proc. Geotechnical Aspects of Underground Construction in Soft Ground, Amsterdam, NL.

Bezuijen A., Talmon A.M., Joustra, J.F.W., Grote B. (2005). Pressure gradients and muck properties at the face of an EPB. Proc. Intl. Symp. Geotechnical Aspects of Underground Construction in Soft Ground, Amsterdam, NL.

Mori, L., Wu, Y., Cha, M. and Mooney, M.A. (2015). Measuring the compressibility and shear strength of conditioned sand under pressure. Proc. Rapid Excavation and Tunneling Conference, New Orleans, LA.