DESIGN OF AN NATM TUNNEL FOR MISSION VALLEY LIGHT RAIL - EAST EXTENSION

David Powell PhD CEng MIMM – Mott MacDonald

David Field MEng MSOA CEng MICE – Hatch Mott MacDonald

Richard Hulsen MEng CEng MICE – Mott MacDonald

DESIGN OF AN NATM TUNNEL FOR MISSION VALLEY LIGHT RAIL - EAST EXTENSION

David Powell, David Field, Richard Hulsen

The Mission Valley Light Rail will loop through San Diego State University (SDSU) and to minimize the

impact of construction most of the works will be underground. The project will require construction of a

330m long (1080 feet), 11.24m wide (36.9 feet), single twin track tunnel through Stadium Conglomerate,

a poorly cemented dense sandy gravel with cobbles. Key technical issues are discussed including the

ground conditions and predicted ground response, the use of sequential excavation methods to control

deformations and specific construction concerns such as instrumentation and monitoring. Management of

the construction process to minimize risk is particularly important in this environment and the strategy

selected to control the works is also considered.

PROJECT DESCRIPTION

The original method of construction planned for the Mission Valley Light Rail Tunnel section was a

Tunnel Boring Machine (TBM) and a precast concrete segmental lining installed behind the cutter head.

Changes to the alignment to significantly reduce the impact on University buildings also substantially

reduced the length of the tunnel section, Figure 1, and led to a re-evaluation of the proposed construction

method.

The revised alignment transitions from cut and cover construction into mined tunnel at Scripps Terrace

and passes underneath the steep escarpment adjacent to the Women’s gym with a maximum of 16.0m

(52.5 feet) cover, Figure 2. Due to a constant grade of 4.2% towards the Station, the amount of cover

along the tunnel gradually reduces from this maximum and under campus athletic facilities and important

utilities at Campanile Drive is as low as 6.0m (19.7 feet).

The ground conditions and the reduced mined tunnel length, much of which lies above the water table,

supported the use of sequential excavation methods and support techniques commonly used in the New

Austrian Tunneling Method (NATM). This approach allowed a switch from single track twin bores to a

single bore twin track configuration, Figure 3. Cost savings were realized with this change and, despite

the increase in span to 11.24m (36.87ft), the NATM approach when applied with appropriate risk

management was expected to be effective in controlling the construction risks and impact on the

University facilities.

Figure 1 – Alignment

Figure 2 - Profile & Geology

Figure 3 - Typical Cross Section

GROUND CONDITIONS AND EXPECTED GROUND BEHAVIOUR

The entire length of the NATM tunnel will be mined in a lightly cemented dense sandy cobble

conglomerate that forms part of the Stadium Conglomerate Formation. Although extensive information is

available on the geological setting around the San Diego State University Campus, there is only limited

experience of tunneling in the Stadium Conglomerate formation. It was important therefore to carefully

assess the worst credible ground conditions likely to be encountered during excavation.

Subsurface Hazards

While the Stadium Conglomerate is generally uniform in composition, due to the high energy

environment during deposition there are sufficient variations laterally and with depth to give rise to some

local concerns during tunneling. These include the presence of large boulders (or clasts), layering with

discrete contact surfaces and grading variations and the presence of groundwater conditions above the

crown for part of the tunnel length. These are discussed below.

Clasts. The Stadium Conglomerate Formation contains a significant number of clasts that will be

encountered during construction. These are usually known as “Poway Clasts” and are generally

composed either of metavolcanic or quartzitic materials with an average unconfined compressive strength

of 200MPa (29,000 psi). The table below indicates the likely quantity and grading of the clasts.

Clast Size

Range

(mm)

Baseline Average

Cumulative

Clast Volume

(% of excavated material)

Number of Clast/m

3

of Clast Material

76 to 150

12%

2,000

150 to 300

8%

230

> 300

1%

95

Table 1 - Clast Distribution in Stadium Conglomerate.

Cohesionless Material. During tunnel excavation layers or lenses of poorly graded sand dominated

matrix lenses up to 600mm (23.6 inches) in thickness are predicted to occur within the Stadium

Conglomerate. Although most of these are likely to occur above the tunnel crown, some will be present

in the crown or sidewalls of the tunnel. If encountered below the groundwater table, these layers could

lead to fast raveling and/or free flowing conditions that locally represent a serious tunneling hazard.

Groundwater Table. The initial section of the bored tunnel will lie just below the water table, Figure 2.

Observations from large diameter boreholes designed to allow inspection of the layering at depth

indicated seepage at the contact of layers, and particularly at the base of poorly graded sand layers.

Between these layers seepage was not observed and the overall interpretation from pumping tests

suggested that the zones of saturation represented semi-confined aquifers. The principal concern is the

possibility that these saturated layers have the potential to create fast raveling and/or running ground

conditions during excavation. The overall horizontal saturated hydraulic conductivity is considered to

range from 2x10

-6

to 2x10

-5

cm/s.

Ground Behavior. In order to provide some indication of the likely ground behavior, use has been made

of observations from local cuts for foundations, test pits with horizontal enlargements to monitor and

assess raveling and Terzaghi’s classification (Terzaghi 1950) for soils for tunneling. In general, the

Stadium Conglomerate falls into the category of firm to slow raveling when above the groundwater table.

Below the groundwater table it will behave in a similar manner in the short term but is expected to

degrade to fast raveling within a period of a few hours.

In terms of ground response during construction the material is extremely stiff and the design of the

primary linings is based on the ground having sufficient stand-up time for advances of up to 1.0m to be

excavated safely. The test pit excavations and the enlargements indicated that this was likely to be

several hours and adequate to allow an initial layer of shotcrete to be applied to maintain excavation

stability.

CONSTRUCTION SEQUENCE AND SUPPORT TYPES

The use of the NATM in soft ground such as Stadium Conglomerate is relatively recent in North

America although it has been used successfully in rock tunnels since the 1970’s, e.g. Washington Metro

and L.A. Metro. It is also proposed for several current high profile projects such as Seattle’s Link Light

Rail and Devil’s Slide Road Tunnel in California.

While several notable and widely reported collapses have occurred on projects where the NATM has

been used, a careful examination of the problems suggests that either the application was inappropriate

for the ground conditions or the management and control procedures employed during construction were

less than adequate (Health & Safety Executive 2000). Ensuring that the design intent is not compromised

is especially critical for shallow tunnels in soft ground in high risk urban areas where tight excavation

and support sequences are required to control settlements.

Based on the predicted ground conditions, and precedent practice elsewhere in similar tunnels, e.g.

Taiwan High Speed Rail system, three excavation and support types were selected. The expected

distribution of the types along the tunnel length was related to key constraints and issues such as:

• Variations in the amount of cover above the crown
• The likely stand-up time and other relevant characteristics such as the frequency of boulders, poorly

graded sand layers and the groundwater table

• The location of surface structures and utilities

• Estimated volume losses and predicted settlements
• The type of equipment used for this type of construction
• Interfaces with the cut and cover and SDSU station.

Support Type 1

From Station 4+591 to 4+745 the tunnel crown will be above the water table and firm to slow raveling

conditions are predicted. The sequence will use a split top heading although a full top heading can be

completed before the invert of the primary lining is constructed. Full closure of the lining is achieved

within 4.0 m (13.1 feet) of the face.

Figure 4 - Support Type 1

Support Type 2

The phreatic surface is present above the tunnel crown from Scripps Terrace portal to Station 4+591.

Where poorly graded sand-dominated matrix lenses and layers are present this could lead to either fast

raveling or flowing ground conditions. To reduce the level of risk associated with this behavior, a side

gallery and enlargement sequence has been specified to reduce the span in each of the top headings. With

the possibility of flowing conditions it is safer to fully support one sidewall of the tunnel before enlarging

to the full section. In addition, routine probe holes with slotted liners have been specified to drain the

ground ahead of the face and reduce hydrostatic pressures. Full closure of the lining is achieved within

4.0-8.0 m (13.1 - 26.2 feet) of the face.

Figure 5 - Construction Sequence 2

Support Type 3

At Campanile Drive from Station 4+745 to 4+799 the cover to the tunnel will be less than 6.0 m (19.5

feet) and it will underpass a number of utilities. The clearance between the utilities and the crown of the

tunnel is less than 4.0 m (13.1 feet) in some cases. In order to maintain a suitable cover to span ratio of

at least 1.0 for each stage of the excavation, a twin side gallery and central pillar arrangement has been

selected to tightly control surface settlements and any potential impact on the utilities. Full closure of the

lining is related to the pillar removal sequence.

Figure 6 - Construction Sequence 3

Additional Support Measures

In addition to the sequences and support requirements for each of the support types, additional support

measures have been specified. It is recognized from the site investigation information that local factors

such as clasts and poorly graded sand layers or lenses could influence the stability of the unsupported

excavation at the face. The range of measures available to the contractor, if required, include:

Temporary Face Support. A temporary shotcrete layer to support the face of the advancing excavation.

This is a prudent safety measure that protects the workforce by controlling any potential for ground loss

due to boulders or running ground as well as preventing the face from drying out.

At various stages of the work, stoppages may occur that result in the face standing for longer periods than

normally experienced during construction. To minimize the risk of instability, the bench/invert will be

completed up to the face and the tunnel face domed and supported by a temporary headwall. This same

procedure will be implemented for a change in the construction sequence.

Spiling and Canopy Tubes. Around 15% of the lenses/layers of poorly graded material are predicted to

be greater than 600 mm (23.6 inches) in thickness. Whilst these are not expected within the tunnel

excavation, they could be as much as 2.4 m (7.9 feet) in thickness and would represent a stability concern

in relation to the face and the adjacent shotcrete lining. If local instability develops, the layers will

require temporary support to the heading in advance of the excavation and either spiling or canopy tubes

have been specified for this purpose.

Probe Drilling. Probe holes drilled ahead of the face have been specified for the entire length of tunnel

to relieve any potential hydrostatic pressure ahead of the face. Installation of these holes will be

implemented after logging of the face features and detection of the potential for hydrostatic pressures.

Pressure relief holes may also be required in the temporary shotcrete face support to prevent local

instability.

Infill Shotcrete. Although clasts of up to 600 mm (23.6 inches) have been predicted for the entire length

of the tunnel, the overall percentage encountered during excavation is expected to be small, as

highlighted in Table 1. These clasts may be partially exposed in the excavated surface of the tunnel and

remain stable even after excavation with mechanical methods. The integrity of the primary lining is

unlikely to be affected by clasts intruding up to 100 mm (4 inches) providing that they do not impede the

placement of lattice girders. Where clasts exceed this dimension or are not sufficiently stable they will be

removed and the excavation made good with infill shotcrete.

DESIGN APPROACH AND METHODOLOGY

The design of the primary and secondary linings is based on the Stadium Conglomerate having sufficient

stand-up time for advances of up to 1.0 m (3.3 feet) to be excavated and supported safely. The excavation

sequences specified, Figures 4, 5 and 6, are designed to control strains in the ground so that as much as

possible of the ground load bearing capacity is used and the strains are maintained at levels that prevent

yielding.

Numerical methods of analysis (FLAC Version 3.4) were used to model the different excavation stages

and predict the performance of the linings in terms of the stresses and corresponding deformations. Three

sections along the tunnel route were chosen to represent fully the range of ground conditions that could

be encountered during construction, particularly the maximum potential hydrostatic pressures and the

maximum and minimum ground cover to the exterior of the tunnel primary lining.

The input parameters for the analyses to model ground and lining behavior are summarized in Table 2.

Throughout the mesh a dry unit weight

γd of 17.65 kN/m3, a porosity of 0.286 and a tensile strength of 0

MPa were adopted. In addition, the initial tangent modulus assumed small strain stiffness with a non-

linear reduction as the stress levels changed and strains increased. The Duncan and Chang (1970)

method was used to represent the non-linear behavior. To verify the results of the plate loading tests

carried out during the site investigations, the tests were modeled with FLAC to derive the stiffness values

indicated in Table 2.

Level

(mOD)

Effective

Internal

Angle of

Friction

φφφφ' (º)

Effective

Cohesion

Intercept

c' (kPa)

Angle of

Dilation

ψ

ψ

ψ

ψ (º)

Initial Elastic

Tangent

Modulus

Eut (MPa)

Poison

Ratio

νννν

Gl - 131.4

40

5

6

400

0.35

131.4 -

126.4

42 15 8 800

0.35

126.4 -

121.4

44 15 10

1600

0.35

121.4 -

base

44 10 10

2400

0.35

Table 2 - Soil Parameters adopted for the Analyses

Each analysis used drained parameters and the estimated non-linear stress-strain ground response was

derived from the plate loading tests and from review of the down-hole geophysical testing. In addition, a

range of Ko values (0.5-1.0) was applied to check the sensitivity of the headings and primary lining to

variations of the in situ stresses.

The following steps typically were used in the analyses to model the construction process:

• Excavation of the ground
• Establish steady state flow conditions
• Relaxation of the ground to represent deformations in advance of support

• Installation of the primary lining
• Analysis of the model and solving to equilibrium
• Installation of the secondary lining
• Analysis of the model and solving to equilibrium.

The results of the analyses were checked in accordance with ACI-318 to ensure that the design complied

with Ultimate Limit State and Serviceability Limit State requirements. In addition to static loadings

derived from the construction process, the performance of the secondary lining was checked for

compliance with the dynamic loading predicted for the ODE and MDE.

MANAGEMENT OF THE CONSTRUCTION PROCESS

Tunneling is not a deterministic process and therefore carries higher risks than most other forms of civil

construction. A recent and important development in the tunneling industry has been the introduction of

management systems that ensure the safety of personnel and the public as well as exercising control over

costs and schedule.

Recent high profile collapses where the NATM has been used, for example Heathrow Express,

highlighted the importance of implementing management procedures throughout the project process. This

failure was strongly related to poor risk management, especially quality control in forming construction

joints, and the main conclusion from the lengthy proceedings instituted by the UK’s Health and Safety

Executive was that effective site supervision is essential. This served to emphasize one of the main

concerns of soft ground NATM tunneling, best practice has to include provision for the designer to

actively follow through into the construction phase. This not only provides an owner with reassurance

that any necessary adjustments can be made before problems arise, it also promotes the development of

management and organizational systems that ensure key decisions on issues such as changes to the

excavation sequence or support are monitored progressively and kept under constant review.

Management Approach

The owner, the Metropolitan Transportation Development Board (MTDB) of San Diego, has been

proactive in requesting an Engineer’s Design Representative (EDR) to maintain continuity of the design

through the construction phase. This is an important role and management systems and procedures have

been implemented in the contract documents to allow the designer to form part of the supervision team.

What is important in this process is avoidance of the more traditional confrontational relationship

between the construction supervisor and contractor and promotion of a single team approach with the

very clear objective of safe construction.

Such an approach is important where the Designer is not the Construction Manager, as is typical in North

American contract practice. There will always be numerous assumptions built into any tunnel designs

that are not always easy to convey or obvious to the Construction Manager without the facility of regular

communications and discussions of key issues. This type of approach, often viewed as “Partnering”,

worked well on the LA Metro and will always be successful where there is a willingness for all parties to

work together even though the risks and contractual arrangements are not always fully compatible with

this objective.

A single team approach, where key engineering decisions are made through assessment and evaluation of

information on performance on a routine basis during construction has demonstrated clear benefits in

terms of reducing risks related to safety and quality control. The single team approach does not, however,

mean that lines of responsibility are blurred as each organization is required to appoint experienced staff

to understand the engineering as well as contractual risks. In this context, the EDR will represent the

Designer’s interests and has as a principal role in the interpretation of the Contract Documents for the

Construction Management and Contractor’s Teams. Experience of similar type of construction and the

response expected during the progress of the works also provides the client with the reassurance that the

project can be completed on time and budget.

Design Risk Management

The input parameters and assumptions used in the design have been derived from the Geotechnical

Interpretative and Baseline Reports (GIR & GBR). The site investigation data indicates reasonably

uniform ground conditions along the alignment although there is layering, poorly graded bands, boulders

etc., that could influence the ground response locally. Even with designs based on conservative

assumptions, there are always factors in this type of ground that are difficult to assess accurately prior to

construction, such as:

• How certain is it that the Site Investigation has identified the worst credible conditions?

• Are there unforeseen local planes of weakness or pockets of wet poorly consolidated ground that

could affect installation of support?

• Are the predicted deformations ahead of the face realistic for the support types?
• Are the predicted volume losses realistic?
• Although the ground conditions suggest that the face will be stable are there local factors that

could affect safety?

The excavation and support sequences have been selected to meet such risks and the additional support

measures are designed to provide sufficient flexibility during construction for the contractor to respond

to the actual conditions encountered. However, verification of the design performance during the early

stages of construction is essential and instrumentation and monitoring requirements are directed to

achieve this.

The assumptions made at the design stage will be updated progressively during construction through the

EDR role. Support types will be verified and sequences adjusted as necessary to suit the actual ground

conditions encountered. Notwithstanding the possibility of unforeseen conditions, having developed the

design on the basis of lower bound conditions, the verification process should be viewed positively with

the aim of benefiting the client in terms of cost and program while also ensuring that the contractor is

properly compensated.

Construction Management Procedures

Control of the construction process is related to the predicted and actual performance of the excavations.

Three principal activities are required to ensure that the control is effective:

• The use of Key Performance Indicators (KPI’s)
• The implementation of daily site meetings to review the KPI’s
• The facility of emergency Review Meetings should KPI’s exceed trigger and limit values.

Most designers dislike defining KPI’s because structurally tunnel design is not a deterministic process.

However, on projects in the UK such as Heathrow Express and the Jubilee Line Extension for London

Underground Limited these worked well and they are compatible with the aims of the GBR. For the

Mission Valley project lining and ground deformations will be monitored and trigger and limit values

will be specified for:

• In-tunnel deformation arrays
• Borehole extensometers and inclinometers
• Surface settlement monitoring points
• Piezometers

In terms of procedures, the trigger and limit values are compatible with the Serviceability and Ultimate

Limit State for the primary and secondary linings.

The daily site meetings will review the instrumentation and monitoring results and look at related issues

such as the excavation sequences, additional support measures and quality control. The KPI’s will also

be viewed in terms of the trends and the response by the site team if trigger values are exceeded will take

a balanced view of both the actual values and the trends. If there are real concerns in terms of the trends,

the contract empowers the CM to convene emergency review meetings that will consist of the EDR, CM

and Contractor’s representatives.

The instrumentation array for in-tunnel deformations, and the trigger and limit values and the

deformation trends defined for such as array are presented in Table 3.

Key Performance Indicator

Frequency of Reading

Distance from Face

Frequency

In-Tunnel Deformation

0 m to +30 m

Daily

+30 m to +60 m

Twice Weekly

> +60 m

Weekly

Inclinometers & Extensometer

-30 m to –15 m

Twice Weekly

-15 m to 0 m

Daily

0 m to +30 m

Daily

+30 m to +60 m

Weekly

> +60 m

Weekly

Settlement Monitoring Points

-30 m to 0 m

Daily

0 m to +30 m

Daily

Key Performance Indicator

Frequency of Reading

Distance from Face

Frequency

+30 m to +60 m

Twice Weekly

> +60 m

Weekly

Piezometers

General Weekly

-15 m to 0 m

Daily

0m to +30 m

Daily

Table 3 - Frequency of Key Performance Indicator (KPI) Monitoring

Quality Control Procedures

The need for quality control throughout the construction process and, most importantly, during the

installation of the initial support is fundamental. To achieve the specified shotcrete performance, Table 4,

will require clear guidance on the mix design requirements, site trials to prove the mix design and

application processes, and qualifications of key staff.

Age of Specimen

Design Strength

8 hours

4 MPa (580 psi)

24 hours

7 MPa (1015 psi)

72 hours

14 MPa (2030 psi)

28 Days

25 MPa (3750 psi)

Table 4 - Specified Shotcrete Strength Development

Specific mix design issues will include pumpability, workability and cohesiveness. Additives and

admixtures, for example silica fume, have proven benefits in achieving these requirements without

affecting either the early and long-term strength or durability characteristics. At 10% by weight of

cementitious material, silica fume, has been specified to promote:

• Increased mix cohesiveness and increased bond to substrate
• Reduced rebound

• Increased shotcrete strengths
• Reduced permeability and hence increased durability

Comprehensive site trials of both the mix and applicators using the equipment specified for the tunnel

works will be required to provide assurance that the specified mix design is workable and its short and

long term strength characteristics are achievable. A series of test panels, both vertical and overhead, and

laboratory and site testing will be used to satisfy performance requirements.

Shotcrete performance will be thoroughly monitored throughout the tunnel construction. Recent failures,

as mentioned previously, have clearly demonstrated the need for strict inspection and quality control

procedures. The intention is that the EDR will assist the Construction Manager to strengthen the

management team to ensure that the contract requirements are met.

The use of shotcrete in an underground environment calls for experienced shotcrete applicators capable

of placing well compacted, void free shotcrete. The need for experienced and qualified shotcrete

applicators has been recognized in North America with the introduction of the ACI 506.3R - Guide to

Certification of Shotcrete Nozzlemen. This will be replaced shortly by ACI C660 - Certification of

Shotcrete Nozzlemen. Moreover, the introduction of the American Shotcrete Association Nozzlemen

Certification Program in 1999 will benefit projects such as Mission Valley by setting training standards

and creating a pool of trained applicators for the industry.

REFERENCES

1. Terzaghi, K., (1950). Geologic Aspects of Soft Ground Tunneling, Chapter 2. Applied Sedimentation.

P.D.Trask (ed.), John Wiley and Sons, New York.

2. Health & Safety Executive, (2000). The collapse of NATM tunnels at Heathrow Airport. HSE Books.

ACKNOWLEDGEMENTS

The authors wish to thank Dave Ragland of MTDB for his constructive comments and encouragement in

preparation of the paper. Siegfried Fassman of BRW, the Project Manager for detailed design of the

SDSU Loop, and John Hawley, Hatch Mott MacDonald’s engineer-in-charge for the NATM section,

have provided valuable advice and assistance in developing the bored tunnel design and their guidance

and inputs during the design process are gratefully acknowledged.