Organising Committee:
(Chair) Mr Rama Venkta, Land Transport Authority

Dr Jeyatharan Kumarasamy, Land Transport Authority
Ms Jee Yi Yng, AECOM Singapore Pte Ltd
 

Technical Committee:
Dr Chew Soon Hoe, National University of Singapore

Mr Chow Weng Lee, Sterling C&S

Prof Chu Jian, Nanyang Technological University

Dr Goh Kok Hun, Urban Redevelopment Authority

Mr Khor Eng Leong, Shanghai Tunnel Engineering Co (Singapore) Pte Ltd

Dr Nick Mace, Mott MacDonald

Mr Massimo Marotta, Land Transport Authority

Mr Nick Shirlaw, WSP

Dr Oskar Sigl, Geoconsult Asia Singapore Pte Ltd

Dr Wen Dazhi, Changi Airport Group

Dr Wong Kai Sin, Singapore Institute of Technology

PARKROYAL on Beach Road, Singapore

Address: 7500 Beach Road, Singapore 199591

Open in Maps

PARKROYAL on Beach Road is about a 20-minute drive from Singapore Changi Airport. Nestled by Singapore's city centre, PARKROYAL on Beach Road, Singapore lets you relax in comfort. Located in the culturally-rich district of Kampong Glam, we are also in close proximity to business areas and shopping malls. Offering seven stylish and functional spaces in the heart of the city, PARKROYAL on Beach Road’s professional event venues have everything guests need to host a seamless meeting or event in Singapore, from modern audio-visual equipment to a wide array of food and beverage options.

• 20-minute drive to Changi International Airport

• 5-minute walk to Nicoll Highway MRT station

• 6-minute drive to the Singapore Sports Hub and Marina Bay Financial Centre

• 8-minute walk to Bugis Junction Shopping Centre and Bugis MRT station

• 10-minute drive to the Central Business District and Shenton Way financial hub

 Click here to be directed to the online registration: Register now

All sponsorship packages are sold out

As of 14 July 2023

13 PDU and 12 STU has been confirmed for Underground Singapore

Session 1-1: Tunnelling Key performance Indicators (KPIs)

Theoretical Basis and Practical Application of Excavation Management System for Tunnelling with EPB TBM

Veenuganan Sivalingam

Question 1 (Jefry Santhosam George Jaya Kumar): What is the efficiency of the belt volume scanner?

Response: Belt volume scanner efficiency can be achieved by number of adjustments and improvement to the current practices, suggested in the paper in detail. First to understand how it works from the shape of the belt till shape of the muck and what the input in the PLC to display are. For example, maintaining the datum level of belt and identifying ground change to condition the muck to get that consistency of muck which not only helps to maintain the plug but also improve the belt scanner values measured. Idlers on the belt and supports of the scanner at specified height from the belt shall be studied to make fine adjustment suit to the site conditions. Please also refer to the answer of some of the questions below.

 

Question 2 (John Poh): What are some of the refinements/adjustments made and so far, how often do you need to make the adjustment to reconcile?

Response: Depend on the results and operation, there are several refinements and overcoming limitations suggested in the paper. Occasions where load cells or scanner show erroneous values due to various reasons, handling the muck in the chamber as well as behind the TBM. As explained in the above question on improving scanner efficiency, number of adjustments in the configuration and input to the program required, such as any inclination of the belt to the scanner will impact the result, scanner reads usually at 100Hz over the mining duration integrating the readings to give the total flow. As you may study the behaviour of the measuring device, improvised method of calibration and continuous improvement needed, which in between “you could do frequent as possible and as & when required”. Reasonable judgement required.

 

Question 3 (Abdul Hafiz Gani Shaikh): There were 5 modes of muck measurement. With so much information getting recorded for each advance Which was the most reliable method of measurement?

Response: I assume, the question is referring to the large diameter excavation management system presented. Mass measurements seems more reliable than volume. The reasons are, Volume measurement is heavily influenced by empirical bulk factor and to some extent affected by the conditioning foam which gives a bulk often challenging to determine. Therefore those 2 belt scales are crucial at the back up gantry, and the 3rd on surface belt can be a verification. However, volume measurement such as scanner will give you a real time measurement and muck pit shall be a verification method although every modes of reconciliation method has its own limitations. Further, it is important to keep the monitoring and assessment to every certain stroke (Eg.300mm) is better way to control. Please refer the paper for practices to improve the results.

  

Question 4 (Jefry Santhosam George Jaya Kumar):  In the event of contradicting values from all muck reconciliation method, what will be the governing method of reconciliation?

Response: First to identify which is your primary reconciliation, secondary reconciliation and tertiary reconciliation, where primary means of not only the accuracy but also real time monitoring so that we can respond before the ground response. In the event they show different values, before conclude, a holistic assessment should be done, which includes, to the behaviour of load cell or scanner or a skip for standard tunnels, and any ground changes should be identified immediately to detect the erroneous reading. It is advisable to monitor every certain stroke to monitor the values and necessary action should be taken. This is exactly why the physical verification methods are required so that can help to verify the values and any calibration required.

 

Question 5 (Bennett Chang):  Were the 5 modes of excavation data managed collectively in real-time?

Response: Not all are real time. Example, for volume, muck pit measurement can only be measured either on surface or shaft bottom (if you are using temporary muck bins) but definitely before the next advance, similarly for weight lifting of the skips by gantry crane is measured on the shaft bottom. But there are alternate ways to make real time such as load cells on the flat car are suggested in the paper, which can give a real time solution.

 

Question 6 (Low Voon Fui): Can you share how does the data acquired in excavation monitoring system benefits us in surface settlement or the built tunnel

Response: During the tunnelling process to control the volume loss it is absolutely necessary to control the excavation values, analysing the data during tunnelling and apply necessary changes will assure the settlements are within desired limit. Failing to do so, can be either immediate settlement/sinkholes or post tunnelling settlement, which is due to consolidation/pore pressure loss or even a left in void propagate to surface after certain time. On the other hand, data acquired in the similar ground condition will benefit you to define input values for future projects with similar ground conditions, such as tunnelling bulk factor, co-relating in-situ density, face pressure application to Kallang formation to control the settlement. Example, for marine clay, a plastic behaviour if the face pressure is too high of the overburden will result in heave and lower bound may result in settlement. Excavation management is not just the volume and mass but also as an entire operation. Leaving air pockets after build the tunnel also can cause post settlement.

 

Question 7 (Lim Ken Chai): TBM KPI is useful for over-excav control. Is there any new studies look into hydogeology where high hydraulic conductivity will affect the data?

Response: Agreed, EPB TBM tunnelling is an open discharge system whether conveyor or skips unlike slurry network you can calculate the deviation flow in-terms of taking ground water. Start from degree of saturation it will affect the data. High hydraulic conductivity will affect the data if the muck handling and operation is not defined to address the flowing ground could lead to gushing through screw conveyor. Not only TBM KPI changes required but also conditioning of the muck, as maintaining plug in the screw is an immediate concern to an EPB TBM, secondly the excavation management system reading and reconciliation. It is best to address the concerns by provision to the TBM design, otherwise the risk has to be addressed at the operational stage, as adjusting KPI and conditioning these difficult ground, for example, it is necessary to add water absorbing polymers to condition the watery soil as it will directly affect the data behind the TBM.

It can be a long explanation to this particular question, but yes, the studies were conducted over several TBM drives to this particular ground condition and both ways concerns have been addressed, in terms of TBM design, PLC control to identify pressure gradient, screw front and rear pressure differences, as well as operational efficiency. Suggest to read last section 6.0 of the paper, briefly discussing about TBM operational control on excavation management. 

Slurry TBM Performance in Volcanoclastic Rocks of Jurong Formation – Experience in DTSS2

Aung Ko Ko Soe

 

Question 1 (Nick Shirlaw): During free air interventions in rock, did you find a tendency for the cutterhead to drop due to loss of the buoyancy force?

Response: From TBM monitoring data, we did not notice this tendency or any notable changes in cutterhead/shields' attitude/behaviour during free-air cutter head interventions in rock. It is our understanding that the buoyant/uplift force will drop and resulting vertical force will change when the slurry fluid is emptied in the chamber in free-air CHI with no or minimal water ingress, in rock. During the stoppage for CHI, thrust force also drops. It is therefore possible that the contribution to vertical turning moment by change in buoyancy/uplift force may closely follow the path of horizontal thrust moment, in rock. However, we did not notice this tendency, and we did not study the magnitude and their contribution to shield behaviour.

 

Question 2 (Tai Joe Shi): Dr Aung, is the embedded limestone layer between the Jurong formation considered highly karst?

Response: We did not observe or encounter karst features in limestone unit in Jurong Formation during excavation and tunnelling, although some core loss and highly fractured limestones were identified during soil investigation works. Limestone layers were encountered in few DTSS2 excavation and tunnelling sites but no karstic features such as solution cavities/voids were encountered nor reported. 

Muck Reconciliation Practice in DTSS2 Link Sewer Pipe Jacking Works 

Aung Ko Ko Soe

 

Question 1 (Ben Tan): How is the monitoring of the muck reconciliation done during pipe-jacking works. How soon is it uploaded to the STEM system?

Response: Most of the time, it is done manually, based on the muck discharge setting. For slurry system, there are two types of setup in hydraulic spoil handling, with desander/treatment plant, and without desander/treatment plant only having discharge tank(s). They spoil volume in the pit and suspended solid volume in the slurry are measured manually using mass balance solid volume or total volume approaches, at the end of the shift or the pipe or as and when required. As for the setup with muck discharge tank, change in slurry level/volume in the tank are measured manually and it is done more frequently because it is more straightforward. Real time monitoring can not be done as no flow meters (more accurate with density meters) and logging software are not utilised in almost all jacking sites (not part of the contract requirement). The records are required to upload to STEMS as soon as verification are done – within 24 hour. 

Technical Trial for Free-Air Cutterhead Intervention in Hard Cemented Alluvium Soils in Contract T316

Paul Sweeney

 

Question 1 (Grace C Hangadi): To Paul, what are the lower boundary for safe free-air CHI in OA in terms of OA class or SPT-N, tunnel depth, and permeability?

Response: Pending response

Question 2 (Nick Osborne): For last speaker. It looks like free air interventions we're a success, is there a recommendation for all free air interventions in OA moving forward.

Response: Pending response

 

Session 1-2: Tunnelling through Challenging Ground Conditions

Design and Construction Challenges of Pipe Box Tunnel at Shenton Way

Pan How Yen

Question 1 (Michael Williamson): Were HDD bhs considered for verification/risk mitigation

Response: Horizontal Boreholes were carried out to verify the soil conditions.

Question 2 (L Tee): Hi Mr Pan, in your opinion and site observations, how can we improve for the Horizontal Grouting. Thx

Response: Based on the site observation and findings, the following considerations will help the process.

-           Carry out trial test in horizontal manner for the to-be-grouted soil at/near to the job site, (if job site allows)

-           Soil profile is changing along the tunnel mostly. Single set of grouting parameters is not sufficient. Comprehensive soil investigation is needed before determine the grouting parameters.

-           HJGP to be done from both side of the shaft (if site allows), thus a shorter length of HJGP to minimise the potential deflection of drill rod.

-           The inherent geological risk such as left in bakau piles (broken) and construction waste largely affecting the grouting coverage in this job site.

-           The horizontal grouting column formed mostly in non-circular shape. The grouting points shall be planned carefully if full coverage grouting is required.

-           Contingency measure shall be in placed if the grouting quality not satisfactory after exposed the excavated slope.

Please refer to the paper for more details.

 

Question 3 (Massimo Marotta): For HJGP, was the angle of the columns horizontal or slightly downward to prevent grout back flow? 

Response: Contractor has done the grouting columns horizontally. However, with the inherent geological condition throughout the entire length, the drill rod can be deflected in any directions. Thus, the continuous pile formation is difficult to achieve at this job site.

Question 4 (Chao, Hsiao-chou): Is there any specific type of instrument, other than settlement mark, you were using for measuring ground settlement? Thank you.

Response: Utility settlement markers were installed during trial trenches. It was used to measure the ground settlement at slightly deeper position as compared to surface settlement markers.

Question 5 (Diwakar Velu): Was the HGP done post jacking. Was there any impact to jacked pipes. Would doing GI before jacking help to reduce overall movements of all works. Tks

Response: HJGP was done after completed the pipe roof. Based on the observations, an immediate ground heave was measured. However, the movement had gradually settled towards its initial movement after HJGP passed the relevant location. If site allows, ground improvement to be carried out vertically before pipe roof could help on the overall movements, provided that the grouting also cover the pipe roof position sufficiently.
 
At this job site, numerous of utilities have limited us to do any vertical grouting along the Shenton Way road.

Design and modelling challenges during design of underpinning base slab of existing MRT station

Ashokvardhan Somasundaram

Question 1 (Lim Heng Thong): To 2nd speaker. How do you determine the percentage of stiffness to reduce?

Response: Using Plaxis 3D model simulation, we computed the stiffness reduction based on force and settlement changes before and after mining

Question 2 (L Tee): Hi Mr Ashok, how is the comparison of the forces from the Plaxis 3D slab modelled as slab compared to the slab in SAP2000? Thx

Response: The magnitudes were comparable. Overall, there was a difference of 10-15%

Question 3 (Tan): For the underpinning of orchard MRT station, how was the base slab strengthened for bending and shear respectively?

Response: Strengthening carried out by thickening the slab locally for bending and shear. Some location DFRP was used for bending only. 

Undercrossing of Changi Airport runway 2, Taxiway and Ground treated with Prefabricated Vertical Drains

Johnny Lim Chia Yong

Question 1 (Ashley Mascarenhas): Was it necessary to use special cutting tools to cut PVD’s? As while cutting GFRP bars thru soft eyes we use the same tools during mining.

Response: Based on our study and knowledge of the type of PVD envisaged along our alignment, the material properties of the PVD there and typical GFRP reinforcements used for launch shaft construction are different. The GFRP reinforcement tend to be brittle  which are suitable to be cut by the rolling disc on the TBM. On the other hand, the PVDs were envisaged to consist of a inner plastic drainage wrapped by a flexible geotextile like material. Both materials in the PVDs are expected to be “softer” and more flexible which may not be readily cut by the rolling disc. Secondly, the other consideration in this project context was the subject part of alignment with PVDs is located underneath airport operational taxiway. Hence, it was more prudent that we had adopt a more proactive approach to have specialised cutters to manage the PVDs.

 

Question 2 (Ben Tan): To Johnny: what were the advance rates, torque speed adopted for the taxiway undercrossing and what type of conditioners were used for F2 & F1 soils?

Response: Please find the operational parameters applied at taxiway

Thrust Force < 20,000 kN

Cutter Torque < 2,500 kNm

Mining Speed < 25mm/min

Cutterhead Rotation Speed : 2.0 rpm

Penetration Rate : 12.5mm/rev

Backfill Grout Volume : 110%

Conditioners

Old Alluvium and Marine Clay – Foam

Fluvial Sand – water absorbing Polymer, Mapedrill M1 from Mapei

 

Question 3 (L Tee): Hi Mr Lim, do you encounter higher tunnel overcutting due to presence of PVD. Thx

Response: No. The muck reconciliation and parameters like backfill grout volume + pressure did not point to presence of overcut. The PVD tools at gauge profile/ location were installed to have same cutting profile as normal tools.

 

Question 4 (Lim Ken Chai): Hi Johnny: are there advanced grouting probes ahead of the tunnel face?

Response: We have probe drilling equipment standby but probing in front of the face was not necessary hence was not carried out.  

Session 2-1: Design and Construction of ERSS 1

Parametric Design Study of Hybrid Steel Fibre Reinforced Concrete in Singapore Cut-and-Cover MRT Structures

Ang Wei Jian

Question 1 (John Poh): a) What was the dosage of steel fibre used in the study?  b) What was the max span of the sections analysed?  c) Any cost benefit analysis done?

Response: Reply to 1a –The study considered fR,1 values of 3 and 5 MPa. The steel fibre dosage to achieve this fR,1 value is dependent on the material selection and concrete mix.
Reply to 1b –The maximum span of the section analysed was 24m.
Reply to 1c –The cost of hybrid SFRC structure is dependent on the prevailing steel rebar and steel fibre price, location and type of structures application and the overall construction methodologies. High-level hybrid-SFRC CBA was carried out for this study and the CBA is favourable when hybrid-SFRC is adopted on slab.

Question 2 (Anon): steel fibre vs rebar - is there any economical benefit in combining steel fibre with rebar?

Response: See Reply to 1c

Question 3 (Podianko Surya): what is your view on testing a vertically cast test beam to verify the effect of fibre orientation?

Response: A vertically-cast test beam would only be able to capture the effect of aspect ratio on fibre orientation. It would not be able to assess the impact on fibre orientation due to workmanship and the flow of SFRC through rebar layers. Moreover, the size and boundary condition of the test beam and the actual structure will be very different. Thus, the resulting fibre orientation factor obtained from such a vertically-cast test beam would not be representative of an actual structure cast on site. We would suggest doing a mock-up (that is representative of the actual hybrid SFRC structure) to back-calculate the fibre orientation factor for design.

Question 4 (Ben Tan): has you looked into the nf value for dwalls for which concreting is carried out using tremie method which may affect the final orientation of fibres.

Response: We adopted the recommendation from SS674 National Foreword, and used an orientation factor ηf of 0.5 for non-horizontally cast members i.e. DWall. We do agree that there is much higher uncertainty associated with DWall concreting i.e. casting with bentonite slurry against soil, and this might affect the overall strength of SFRC. 

Question 5 (Jason): a) is the "ideal" orientation for steel fibres on d wall vertically oriented?

b) Does more steel fibre have to be dosed due to the not-ideal orient?

Response: Reply to 5a – For a vertical DWall subject to lateral loading, the principal direction of the tensile stress would be vertically along the Dwall element. Ideally, the steel fibres should be oriented vertically, in the direction of the tensile stress to bridge across potential cracks and to resist tension more effectively.
Reply to 5b – The SFRC strength would have to be discounted due to its unfavourable fibre orientation as per SS674. While higher steel fibre dosage can be considered to increase the SFRC strength, it is important to note that this could result in steel fibres clumping together (i.e. fibre balling) that affects the workability and concrete quality. "

Question 6 (Sofren Leo Suhaendi): for clarification, does the rebar saving mentioned consider the quantity of steel fibres added to the concrete mix ?

Response: The steel savings mentioned in the study are calculated solely based on the reduction in steel rebar contents. The quantity of steel fibres required is not considered. 

Question 7 (Chow Wei Mun): in your recommendation, what would be the alpha,cc for steel-SFRC design?

Response: For hybrid-SFRC design, if the minimum reinforcement areas based on SS EN 1992 are satisfied, the sections can be considered as reinforced and the αcc can be taken as 0.85 (as per NA to SS EN 1992). In this study, the minimum reinforcement areas were met so αcc can be taken as 0.85 for design.

Question 8 (Kuok Siang): Is the design strain of the reinforcing steel at maximum load different between 500MPa and 600MPa rebar?

Response: For SLS design, the working stress is much smaller than the rebar yield stress. Hence, for the same set of loading, the design stress in 500MPa and 600MPa rebar would be similar. Correspondingly, the strain in 500MPa and 600MPa rebar would be similar given that Modulus of Elasticity remains unchanged at 200GPa (as per SS EN 1992). For ULS design, ductile mode of failure is preferred – this is also the design assumption made in this study. This means that rebar would have yielded at ULS, and attained the maximum yield stress. The design strain at ULS would be larger than 0.217% and 0.261% for 500MPa and 600MPa rebar respectively. The actual strain difference between 500MPa and 600MPa rebar is not critical as both sets of rebars would have already yielded at ULS.

Question 9 (James Stabler):  What happens if some production tests for tensile strength don’t meet the specified tensile strength for the concrete.  How to rectify?

Response: As the trial mix design has been tested and approved before production, the SFRC design mix is unlikely to be the cause of the issue, but likely on the concrete production control. To minimise such occurrence (where production tests fail to meet the QAQC acceptance criteria), it is important that a strict regime on the quality control for concrete production e.g. careful material storage, material handling, material delivery and accurate batching process is maintained.

Use of Observational Method for Thomson East Coast Line Project

Chepurthy Veeresh

 

Question 1 (Lim Ken Chai): Hi Vereesh, may I check what's the day strength eg 30days or 60days taken for the ground improvement?

Response: Strength is tested after 28 days and maximum value is 2300 kPa which is more than 600kpa design strength specified

 

Question 2 (Podianko Surya): To Dr. Veeresh, do you think is it better to prepare a few designs in advance say in case GI&IM is better than expected, so that this can avoid multiple amendment

Response: Agree, this is what current BCA’s OM approach encourages, but from builder point of view, it may be additional design cost which he may not be able to implement if the performance is not favourable.

 

Question 3 (Nicholas Mace): could the vertical stiffness of the OA be a significant parameter since the wall movement may be influenced by rotation of the GI block?

Response: Yes, it could be stiffness of OA, however in this submission we only changed GI parameters to predict conservative wall deflection without matching actual performance in site

 

Question 4 (Idan): when the GI trimmed, it'll be "more" cantilever, the tension will be developed within GI. Does it mean acceptable to exempt the Clause of no tension in LTA CDC?

Response: GI design still follows LTA CDC ie no tension is allowed

 

Question 5 (Massimo Marotta): any view about having an "overdesign" in first place later optimised by OM, versus having an "Optimized design" first with contingency measures by OM?

Response: This is not the case of over design, GI parameters achieved much better and are nearly 4 times of original design specified. Hence we have taken advantage of better parameters to implement OM approach

 

Question 6 (J. Jiang): With the marine clay layer below FEL, was there any ecc problem with the constructed piles after reaching FEL at the center part without GI?

Response: To address this issue we had specified inclinometers in foundation piles, during excavation we did not encounter any excessive ecc in piles

 

Question 7 (Thuva): could you suggest from industry point of view -' "how we can fasttrack and implement this at site and what needs to be changed in authority req"

Response: This is where we expect BCA to consider allowing OM approach C (Ipso Tempore) so that industry will be encouraged to optimise in case the performance is much better than initial design. 

 

Contract T311 Bedok South Station and Tunnels – ERSS Design and Performance

Darryl Lai Hao Wen

 

Question 1 (Wong Chin Yik):  What was the consideration to determine toe of sheetpile wall to act as a water cutoff?

Response: The design of the sheet pile walls was based primarily on the toe stability. In the design, the toe level based on toe stability was sufficient for groundwater cut-off to the 1st tier slope/berm. The excavation at the passive side considered the groundwater level to be at excavation level.

 

Question 2 (Nicholas Mace): were there any observable groundwater flows from the OA?

Response: Due to the relatively low permeability of the OA soils found around the area, there were no observable groundwater flows from the OA. The monitored groundwater drawdown was generally -2m from the baseline, and the building settlement markers and tilt meters were all within the alert level.

Furthermore, seepage analysis was carried out, making use of the sheet pile wall as a groundwater cut-off (see first question set). 

 

Question 3 (Jason): At the t311 project, why did the DSM terminate near/at the same height of the 3:1 slope top? Was the resultant movement expected, or addtl penetration omitted

Response: The DSM was intended to treat the soft marine clay, and the termination criteria is at the base of the marine clay, which can be found near the bottom of the 1st tier slope (top of the 2nd tier slope). The resultant movement was expected. No addition penetration was omitted.

 

Question 4 (Darrell Choong): To third presenter-Darryl, I noticed there are sheet piles on the slope for water cut off. What is the length of the pile to allow this cut off?

Response:  same reply as Wong Chin Yik

 

Question 5 (Alex Yang): What is the water drawdown near surrounding building?

Response: same reply as Nicholas Mace

 

Question 6 (Inbaraj): for the open cut excavation what was the water table drawdown consideration during the design stage?

Response: same reply as Nicholas Mace

NSC Contract N102 Sequential Excavation Above Existing DTL 2 & 3 Tunnels in Soft Marine Clay

Prabu Ramasamy

 

Question 1 (Wong Chin Yik): Was NSF considered to the short non-load bearing DWall?

Response: The short non-load bearing Dwall loads are designed to be transferred to deep foundation element which consists of bored piles and deep load bearing Dwall panels. NSF is mobilised under long term when there is any consolidation effect within the soft marine clay layer. Hence the possible NSF along non-load bearing wall was considered in term of loading to be ultimately carried by permanent deep foundation elements.

 

Question 2 (Anon): what is the motivation behind considering 10% stiffness reduction in longitudinal direction for sensitivity study?

Response: It is a sensitive studies where there is no any specific guidelines. It is actually 90% reduction in stiffness along the longitudinal direction of tunnel lining considered for the sensitive study. The longitudinal direction stiffness is noticed not to govern the tunnel movement since the differences from sensitive study is less than 1mm.

 

Question 3 (Darrell Choong): kingppst load transmit into GI block, what is the tensile strains in GI and heaving issues?

Response: The temporary kingpost embedded into temporary piles within the GI block assist in eliminating excessive heaving during excavation unloading exercise. The mobilised stress within the GI block is checked in the Plaxis analysis where the GI layer assessed to be in compression stress without any tension.

Session 1-3: Large Tunnels & Innovations in TBM Technology

Moving Total Station: Taking TBM Guidance to the Next Level

George Wei

Question 1 (Edi Hartono): What is the accuracy of the moving total station method that you presented?

Response: thanks a lot for your questions! A very good and at the same time complicated question!
 
Accuracy is very complex topic, there are many factors which will influence the accuracy, like the instrument accuracy, reference line accuracy, tunnel environment!
 
When we talk about moving station, there is also fixed station! For moving station, the total station is moving together with the machine, and fixed station means the total station is fixed on the tunnel wall, from time to time, the total station needs to be moved forward manually!
 
As I have explained at the conference, our moving station system operates with two steps, when the machine stops, the total station will do the measurement like the fixed station works (first gets its own coordinates from the 3 known prisms)! The fixed station has a system accuracy as 10mm (this error will not accumulate).
 
When the machine starts to advance, the total station will stop to work as it , and another system called Sensor fusion (the values from various sensors) will be collected for getting the TBM position, will be automatically started, and from now on, based on the measured results from the total station, the sensor fusion will take over the responsibility to guide the machine. When machine stops again, the total station will measure the target again, and the position from the total station will be compared with Sensor fusion result and see if there is difference!
 
Generally speaking, the result of the moving station and the fixed station may have some small differences, but there is not much accuracy loss!
 
The two steps I talked about, like the example that I gave at the conference: II want to travel from my home to a hotel in Singapore, I take the flight between Beijing and Singapore and, I take cars from my home to the airport and from the airport to a hotel in Singapore! The similarity is that the total station will be responsible for the big part, as a control means, and the Sensor fusion will be responsible for guidance of a local/short distance.

Question 2 (Wes Chua): Who provide the X, Y Z data of prisms. 2. Who Install the prisms 3. What is the ideal distance between prisms and total station to TBM

Response: your questions are quite relevant questions and very practical. Please see my answers to your questions below!
 
For fixed station system, total station is fixed on the tunnel wall, and a refence prism is needed for the total station to measure the target installed on the machine. This prism is prepared by the jobsite.
 
For moving station, total station is installed on one of the backups of the machine, the moving station needs 3 known fixed prisms in order to know its own position, the 3 prisms are prepared by the jobsite surveyors from the beginning, these 3 prisms must be measured manually by the jobsite surveyors at the beginning! When the total station is moving forward, the total station could not see the last prism anymore, then the prism will be moved forward to a new position (5 minutes of physical work), then we press a button in our so called Mobile App” installed in the mobile phone, the system will automatically search and locate and measure the new prism, so we have 3 known prisms again! At the beginning the coordinates of the 3 prisms must be done manually! The coordinates of the 3 prism need to be manually checked periodically
 
in the beginning, the VMT staff and jobsite surveyors could install the prisms together, and later, the jobsite surveyor could do it himself, or anybody responsible at the jobsite could do the job, it is not necessary to be surveyor, the prims installation is quite easy, for segment lining, glue would be used, and for open face machines, a screw is needed, all is quite easy! And it could be finished within 5 minutes, and it will not influence the advance and other tunnel work!
 
The distance (Distance 1) between the total station and target installed on the machine body of the TBM, the distance (distance 2) between the total station and the prisms, and also the distances (distance 3 arrangements between the reference prisms, means lot！
 
In order to achieve a better accuracy, based on the actual situation of line of sight on the machine, Distance1 and Distance 2 shall be proportional, it shall be avoided that distance 2 is much shorter than distance 1. Generally speaking, distance 1 shall be as short as possible.
 
For distances 3, the distance between the prisms shall be proportional too, it shall be avoided that the distance between first and the second prism is much different to the distance between the second and the third prism! The 3 prisms shall not be installed on the same line. Generally speaking, the distance between the 3 prisms shall be as long as possible!

Application of Belt Conveyor System for Tunnelling Soft Ground using Large Diameter of EPB TBM

Pierre-Alain Scherwey

Question 1 (Grace C Hangadi): What is the main consideration for belt storage arrangement?

Response: In Singapore with only small space on side the available space is the main driver for the belt storage unit selection and capacity
If we have 50 meter long space. We could install a belt storage of 450m capacity
If we don’t find 50 meter of space, generally we use a vertical belt storage unit

Preliminary Study of Utilizing Machine Learning to Predict Cutter Wear in Tunnel Boring Machines

Zhang Yunhuo

Question 1(Nick Shirlaw): Would it be better to use contact force on the cutterhead rather than total force as input to the analysis?

Response: In the preliminary analysis, it was observed that total thrust force exhibited an impactful influence on TBM cutter wear. This finding might be attributed to the wider range of values associated with total thrust force compared to cutter contact forces, which have a narrower range. We also recognise that contact force is a crucial factor in cutter wear and that, in future studies with larger datasets, we can explore the use of cutter force as one of the input parameters.

Question 2 (Bennett Chang): To Dr Zhang, did it take long and a lot of data for the model to be adequately trained?

Response: In this study, the numerical prediction model utilized 676 data points, while the classification model utilized 921 data points. Given that the dataset size remained below one thousand, the training process proved to be efficient without consuming an excessive amount of time. A portion of the time was attributed to the task of fine-tuning optimal hyperparameters for the KNN and DT algorithm. This involved multiple iterative steps, but even so, each process was completed in a reasonably short span of approximately 10-15 minutes when leveraging the Scikit-learn library in Python. This processing time is expected to increase when using more datapoints in future or experimenting a more complex algorithm such as the neural network. Of course, this timeframe does not account for the time spent on data pre-processing, which is an essential step in preparing the original dataset for input into the machine learning algorithms.

Question 3 (Kok Hun): To Dr Zhang - would it be possible to identify TBM parameters with stronger influence on the ML prediction model as a quick indication on cutter tool wear?

Response: The current study did not delve into identifying the stronger influences since it already focused on key operation parameters, which did not consist of a very long training process. However, in future studies, by incorporating a broader range of parameter features and employing more computationally intensive algorithms, it is possible to run a correlation analysis on which combination of parameters have stronger influence to allow a quicker indication on cutter wear.

Question 4 (Grace C Hangadi): Are the geological parameters (e.g. abrasivity, UCS) considered in the machine learning? Could the finding be applied to another drive in BTG?

Response: In this study, the UCS and abrasivity parameters were not considered, mainly because the TBM encountered softer soils for which UCS and abrasivity values were unavailable. The primary objective of the study was to assess whether machine learning could accurately estimate cutter wear based solely on the readily available TBM operational parameters stored and recorded in LTA's tunnelling cloud database.
The study has included the application of the trained model to another TBM drive within the same project, also in the BTG formation. It yielded promising outcomes as shown in Section 4.2 of the paper. 

Watercare’s Central Interceptor project – Designing the longest bored wastewater tunnel in New Zealand

Sofren Leo Suhaendi

 

Question 1 (L Tee): To speaker from Arup, how are the NZ tunnel lining design requirements compared to PUB or LTA design requirements? Thx

Response: In general, some similarity of design requirements can be observed between Central Interceptor and PUB tunnel projects (i.e. durability, loading, etc). Specifically, there are also some differences that can be observed between CI and PUB tunnel projects (i.e. surcharge, seismic, etc.)

Question 2 (Inbaraj): To Leo, could you pls tell us bit more, how did you apply ground relaxation for the lining design?

Response: The ground relaxation for the lining design was firstly verified using finite element model analysis to ensure the relevancy of its application. Once verified, the ground loading in the analytical method can be optimised, accordingly

Question 3 (Inbaraj): To Leo, how the CHIs works are managed with the face pressure of 8.7Bar?

Response: For clarification, 8.7 bar is the maximum hydrostatic pressure on the lining in the project. During tunnelling works, the face pressure and compressed air works applied are never more than 3 bars.

Question 4 (Chepurthy Veeresh): Dr Leo What is the estimated ground relaxation from Plaxis 3D and how it is compared with Terzaghi's method

Response: From Plaxis 3D, we can verify ground relaxation of more than 90%. Comparing to Terzaghi’s method, it is found to be better (or similar at certain case).

Session 1-4: Lessons Learnt in Tunnelling

Designing the First Single Bore SFRC Transit Tunnel in Eastern Canada

Jasper Jiang

Question 1 (Nick Shirlaw):  We're the furnace fire tests with the segment under load or not?

Response: Pending response

Question 2 (Colin Bullock): did you compare FLAC3D analysis with empirical methods such as Muir Wood for lining analysis and Peck for surface settlement?

Response: Pending response

Question 3 (W T Lai): Ms Jiang, noted the elimination of the additional surcharge due to shallow overburden for large TBM, any measures adopted to control the grd movement on site?

Response: Pending response

Undercrossing Tuas Basin DTSS Phase 2 Contract –T- 08 Subsea Tunnel

Mahesh Chander Joshi

Question 1 (Kok Hun): For 2nd speaker, can you clarify if there is extra provision for primary (segmental) lining against seawater, in terms of gasket, concrete mix, rebar specs?

Response: The gaskets were tested for sea water exposure. The Extrados of wegment were installed with Waterproofing epoxy coating.

 
Other requirements pertaining to the segment design were similar to those of other tunnels in Singapore (for example LTA 's M&W specs).

Question 2 (Ben Tan): can you briefly share the flood protection plans for the undersea crossing? What additional measures are needed as compared to in-land tunnelling?

Response: For the flood protection the both tunnels were driven in upward gradient this brought the low point to the launching shaft, where two different sources of power were arranged for the two different dewatering pumps, with one pump in service and other in standby. While working In the excavation chamber under free air a sudden inrush of water (potentially resulting from was-hout of joint filler in fractured rock under sea) could be disastrous. To overcome this risk of flooding in the excavation chamber, the CHI interventions were carried out under hyperbaric conditions, even if there was not much concern of the settlement on sea bed/ surface.

Question 3 (Edi Hartono): Hi Mr Joshi, does the invert of the tunnel eventually gets covered with HDPE lining?

Response: No, the bottom 30 degrees remains without HDPE lining.

Question 4 (DC): What were the steps taken to manage the slurry quality supporting the TBM excavation face during the prolonged TBM stoppage in the Covid period

Response: An essential services team of Engineers and supervisors was deployed during the Covid period to maintain the machine operation-ready and this team maintained the required face support pressure on daily basis, carried out circulation of slurry in the Slurry system and cutter rotation (at very slow speed) to avoid sedimentation at excavation chamber every alternate days for all 4 TBSMs.

Construction of 45m long cross passage in rock under the live CCL tunnel in Thomson Line Contract 212

Mohiadeen

 

Question 1 (Ben Tan): for CP21 was there a pump sump within the CP? how long did it take in total to complete the 45m long CP?

Response: There was no pump sump in CP21 and two inline pump sumps were present at both bounds of bored tunnel.
It took almost 4 months for the excavation works.
It took almost 4 months for the excavation works.
 

Question 2 (Diwaker Velu): -For D&B, how was the vibration effects for the CP works  ; what would be the measures to control such impacts if works are  at shallow depths. Tks

Response: CP itself does not have any adverse effects as temporary lining were designed for the blasting works by the designers, but to minimise impact on the bored tunnel segments blasting was only carried out after the collar portion of CP.
There are several factors were considered to control the impacts due to blasting by conducting trial blasting in CP19 to determine the actual ground condition that transfers the vibration and adjusting the powder factor, setting of delay timings in the blasting design.

Overcoming Challenges during Bored Tunnelling in Contract T216 of Thomson-East Coast Line, Singapore

Sahabadeen Mohamed Mohideen

 

Question 1(Nick Shirlaw):  A sinkhole in the cemetery was in the newspaper. How many sinkholes overall?

Response: Pending response

 

Question 2 (DC):  With the uneven topography, what is your advice to consider when designing or monitoring the excavation face support pressure during TBM tunnelling

Response: Pending response

 

Question 3 (Lim Ken Chai): hi! Pls share your experiences for mixed-face tunnelling where the softer face is connected to an artesian pressure. Thanks.

Response: Pending response

Session 2-2: Design and Construction of ERSS 2

Effective Load Transfer and Overcrossing Tunnel

Krishnagopalan Nadarajah

Question 1 (Nicholas Mace): we can emphasise that explicit gaps were conservatively modelled in the analyses at the D-wall joints, leading to the stress concentrations

Response: Agree, there is no physical gap between diaphragm wall panels. Diaphragm wall panels are installed against each other with water stop at interface. Friction between panels is conservatively ignored and mass concrete part of the panels at interface is also conservatively ignored. Both contribution of interface friction and the mass concrete will help to reduce the concentration on stresses at the panel and slab joint corners, as the contact areas increases

Question 2 (L Tee): Hi 1st speaker, is the couples joint used for the shear bar? Thx
Response: Coupler connecters are used for shear bars. The way of construction demands the use of couplers. Usage of Couplers have better track record, and it is common in underground construction.

Design Considerations and Construction Monitoring for a 30m Height Open Cut Slopes in Bukit Timah Formation Soil

Edward Lim

Question 1 (MW): For Speaker 2 - Did the discharged ground water recirculated back to the original ground?
Response: Pending response

Question 2 (Podianko Surya): Edward, as groundwater is critical, does analysis case for GW flow during heavy rainfall event analysed? What was the return period of storm considered?
Response: Pending response

Question 3 (Sje Ting Tan): 2nd presenter - what is the maximum water drawdown at the top of the slope, any settlement observed and any concern from NPark on the water drawdown?
Response: Pending response

Question 4 (Ben Tan): Edward: the rise and fall of gwt presented is in area with the sheetpile erss or without "obstruction"? Any difference in both scenario?
Response: Pending response

Design of NSC Tunnel on Pile-Raft Foundation above Under-running MRT Tunnels and its associated Impact

Won Mon Thi

 

Question 1 (Chao, Hsiao-chou): Did the EL beam sensors installed on both sides of the tracks? Were the deformation and convergence of tunnel lining measured?

Response: EL beams are installed across the track at 2.5m spacings to measure tract twist. A series of EL beams are also installed along the track (perpendicular to those installed across track) to measure dip or peak. Versine can also be calculated based on the readings from the above arrangement.

 

Deformation and convergence of the tunnel lining are measured by automatic tunnel monitoring system (ATMS) in which 5 numbers of 3D prisms on a ring are installed at 3m spacings. 

 

Question 2 (Lim Heng Thong): What was the actual reason that the tunnel box had to be extended? Were there any change of instrumentation readings when relief wells were activated?

Response: The tunnel box was not actually extended but it was just an indicative scheme to show BCA the nearest location to install DFE for south bound and justify that DFE is not feasible for south bound. The original scheme with raft foundation at south bound is granted approval.

 

The current progress at the presented design zone is just D-wall installation. Excavation to strut S1 level is started at adjacent zones with relief wells activated but there is no appreciable change in ground water table outside excavation since the excavation is still shallow.

 

Question 3 (L Tee): do you also consider the sub grade modulus from the Plaxis 3D with actual loads, settlement and bearing stress estimated on the raft plate. Thx

Response: In view of the length of the interfacing portion which is 800mm, only 2D analysis is performed. As such, subgrade modulus for raft is based on 2D only.

 

 

Design Aspects and field observations of excavation works using HAT-type sheet piles and tubular struts with ground improvement layer

Yandi Layadi Lay

 

Question 1 (Ben Tan): any lessons learnt in terms of additional QA/QC tests on the GI before mass excavation commences?

Response: A couple of suggestions that can be considered:

1) On top of the common field trial to test the GI mixing and operating parameters, a field trial adjacent to the completed sheet pile wall is recommended to demonstrate feasiblity of achieving a good contact.

2) Higher intensity of coring to be done at the targeted interface zone instead of just random locations.

3) Another check can be carried out again when excavation has reached final excavation level before struts are removed.

 

 

Question 2 (Michelle Lew): Was site investigation carried out to check the gap btw WSM & ERSS wall?

Response: Some coring tests at selected locations were done to check the gap between WSM and ERSS wall and proved satisfactory.

 

Question 3 (K Barthi): To the last presenter: would you able to provide who is the strutting contractor with tubular struts.

Response: The strutting contractor was Yongnam.

 

Question 4 (G): For Yandi - how could design w GI in the future take into consideration such gaps?

Response: Sensitivity analysis could be done to understand the impact of such gap and mitigation measures could be prepared such as having concrete plug in between the sheet pile and base slab. More attention should be given to the grout column arrangement near the interface for example smaller diameter columns may be needed at those locations.

 

Question 5 (TMM Naing): Is there any strut monitoring instruments like strain gauge and loadcell during excavation? Why no perimeter grout at base slab level?

Response: Yes, there are strain gauges and load cells placed on the struts. Concrete plug between sheet pile and base slab was not required based on our analysis.

 

Question 6 (Lau LS): What is the reason of choosing Hat-type sheet pile wall over the usual soldier pile wall?

Response: The installation time for HAT type sheet pile was considered to be faster than that of soldier pile which requires installation of the pile and lagging. HAT type also offers higher bending stiffness and capacity.

 

Question 7 (Podianko Surya): To 4th speaker, considering the GI quality is key, would you recommend tighter QC for the GI?

Response: In my personal opinion, higher intensity of coring to be done at the targetted interface zone instead of just random locations will be useful.

 

Question 8 (Kuok Siang): Last speaker: wall movement at the GI location show movement during strut removal compared to FEL, potentially suggesting post-yield behavior in the GI?

Response: The presence of gap between sheet pile and GI causes the measured movement at the GI elevation to keep increasing as the excavation progresses and subsequently at strut removal, contrary to what is expected if the GI was effective. It is possible that at the removal of S1 stage, the GI has gone into post-yield behaviour since the axial strain at the GI elevation was around ~0.6%, which exceeds the typical recommended value of 0.4% in SLS.

 

Question 9 (W T Lai): how the gap without GI to get the approval?

Response:  Before the mass excavation commences, coring tests at selected location were done on the GI and proved satisfactory. It was only during after the removal of the last strut S1 that the inclinometer had a sudden jump in reading breaching the Work Suspension Level. At that point in time, the remaining works at this section is the casting of the remaining portion of permanent wall.

 

Question 10 (Jorgin Tan): To 4th speaker, is there a possibility that the quality of backfill compaction could be a reason as there is a change in analysis trend before/ after s1 removal

Response: Based on our sensitivity analysis, reducing the backfill strength & stiffness still does not provide match of wall deflection shape with the inclinometer measurements. In addition, we understand that the backfill was compacted using a mini compacter on site. Hence, the change in analysis trend before/after S1 removal is unlikely due to quality of backfill.

Session 2-3: Geology and Site Investigation

Assessment of Geological Strength Index (GSI) for sedimentary rock formation in Singapore

Kar Winn

Question 1 (Zhang Yunhuo): 1) Lab testing of UCS could vary a lot. Any advise how to identify the intact rock UCS?

Response: To choose better quality of core samples (if possible by geologist and geotechnical engineer) without any joint or fracture. However, some sedimentary and metamorphic rocks got inherited bedding and foliated planes and difficult to choose intact samples.
 
Advise to carry out UCS tests with stress and strain graphs. If the samples failed at very low strain level (e.g 1-2%), the results are not reliable.

Question 2 (L Tee): Hi Dr Kar Winn, is there general relation of UCS with rock modulus and is rock modulus increase with depth too. Thx

Response: Naturally UCS directly related to modulus. In poor rock mass (GSI value very low with many discontinuities), it may not correct.

Question 3 (LC Lee): Can GSI be determined from the rock core extracted from the SI borehole or must it be determined from exposed rock surface?

Response: GSI / Q / RMR should be determined on the exposed rock mass faces in tunnel / caverns after each run of excavation. Those results determined from core samples should take as an indicative values because the rock mass condition can widely vary between the adjacent boreholes.

Deep Basement Construction at a Site with Adverse Geological Features

Edi Hartono

Question 1 (Yandi): A concern of blow up was said thus some piezometeres inside the excavation. Could you pls share any of those readings? any mitigation measures?

Response: During excavation, the piezometers registered a reduction of pore pressure as expected.

As a mitigation measure against unexpected rise of pore pressure, relief wells are to be provided within the excavation area.

Question 2 (Ben Tan): how has the fault zone affected the dwall construction? Any additional measures taken?

Response: The project covered in this paper does not experience any major issues during construction. The additional probing reduces the risk of encountering adverse ground condition during trenching works.

 
To reduce the possibility of water seepage during excavation, the joints and base of D-wall were grouted. The bentonite stability check should be done accordingly. As a good practice, it is recommended to ensure that rebar couplers do not protrude out of the rebar cage.

Question 3 (Yandi Lay): Would you please elaborate the motivation behind the geological mapping during excavation? is that part of some Observational method of the ERSS?

Response: The motivation is to understand the ground condition in relation to the expected soil condition. The project does not adopt any observational method/approach for the ERSS.

Question 4 (Lim Ken Chai): Hi Hartono! Very insightful to the Labrador S4/S3. For Dwall toe grouting, was there any studies or field indication to determine the direction of water flow

Response: There were no studies carried out in determining the direction of water flow. During the project’s excavation stage, we generally encounter a dry excavation condition. We suspect that the joint and base grouting done was effective in reducing water seepage.

Question 5 (Kiefer Chiam): Can you share any challenges faced during dwall installation, especially at fault zone? What is water drawdown recorded by instrumentation?

Response: The project covered in this paper does not experience any major issues during Dwall installation. The water drawdown is managed through the activation of recharge well.

Question 6 (AKK): Does this adverse ground condition have any impact on original design/construction plan? If yes, did the contractor claim and was it successful?

Response: The potentially adverse ground condition was identified prior to the design stage.

Question 7 (Chepurthy Veeresh): Was there any impact or movement observed for adjacent tunnels due to fault zone? what mitigation measures are in place?

Response: Recorded movement to the adjacent tunnels due to the ERSS and excavation works were lesser than anticipated from impact assessment analysis. Less than 5mm of tunnel movement is recorded.

 
Adopted mitigation measures include (1) Early installation of ATMS along the influence line. (2) Regular tunnel access to monitor the general condition of tunnel. (3) Reduction of Dwall panel length from 6m to 3m to reduce the likelihood of trench collapse. (4) Adoption of top-down construction with basement RC slabs as ERSS method. (5) Continuous monitoring of nearby structures.

Question 8 (L Tee): Hi Hartono, how is the permeability for the fault zones. Thx

Response: We adopted a post-grouting permeability acceptance criterion of 1 lugeon prior to commencement of excavation works.

Development best practice for 3D geological modelling on Singapore’s Infrastructure Projects

Andrew Forsythe

Question 1 (Podianko Surya): as different risk for different structure,do you see it necessary for multidisc review of the 3d profile,and how to make it friendly to layperson?

Response: I am not sure I fully understand the question but I assume you mean a multi-discipline review – whereby we include all disciplines who may be impacted by the ground. If so then yes I fully support this approach. It is always key that an engineering geological model considers all aspects of the engineering that will be impacted by the ground. Models should never be purely geological when they are intended for the design of infrastructure. They need to address the key geological and geophysical features that will interact and impact the proposed structures. As such I recommend that the modeller holds an initial review with the wider engineering team to make sure they understand the concerns from each relevant discipline. A follow up review (or even multiple reviews if it is a highly complex model) should be held as the model is progressed. 
 
The relevant attendees for these reviews will depend on the project, the ground and the stage of works. But they might include structural engineers, utility engineers, contractors involved in the construction, client representatives, stakeholders including those from neighbouring infrastructure, etc. Often though the design team will have a good understanding of the external parties concerns so the initial reviews may focus on a core team from the designer. Then results may be shared in summary reviews with external parties.
 
In terms of ensuring the geology and the geological risks are well communicated to the layperson – yes this is a key issue. Thankfully 3D modelling software makes the geological profile very easy to communicate but the modeller must ensure that the key features and risks (e.g. a fault or a high permeability zone) are clear and highlighted so that the layperson can understand where the risks lie. Often the modeller will have multiple versions of the same underlying model. One might be the master – considered to be the most accurate and detailed and one might be simplified to highlight the more critical issues – for sharing with a layperson. There may be others too – such as that used for the 3D hydrogeological model in MODFLOW.    

Question 2 (Daphne Sun): how are the multiple sand lenses modelled in the Santa Clara Valley leapfrog model?

Response: The stratigraphy comprised a fluvial sequence of repeated sand and gravel channel deposits within a silty flood plain. The sequence was quite deep due to the gradual infilling of the basin and this meant there were hundreds of channels, lenses and layers of deposition.
 
The deposits were generally horizontal in deposition and remained horizontal but the channels had complex geometries as they were laterally continuous in all directions.
This made the geology too complex to model using a traditional deposit based approach where we map out each layer.
 
Instead we used Leapfrog’s “Intrusion” method, modelling the sand and gravel “intruding” into the silt and clay of the flood plain. This is not really an accurate deposition model but the sand and gravel does effectively intrude into the flood plain deposits as the river cuts through it. So we felt this was an acceptable workaround.
 
To ensure that the sand and gravel deposits were thin but laterally extensive lense shapes (rather than blobs) we adjusted the ellipsoid ratio parameters and added a general trend of elongation towards San Francisco Bay (to where it drained). The specific parameters took some trial and error because the method is only an approximation of the reality but it was true to the data and was in line with our conceptual model so we were happy.
 
Modelling such complex lenses should be treated with caution though and even though the model was in line with our conceptual model we recognised that we didn’t have enough data to truly model the lens geometries and may have missed features. As such we had to include some carefully worded disclaimers on our sections and provide a geological modelling report which gave a full description of the method applied and the justifications for doing so. 

Question 3 (Lim Ken Chai): Hi Andrew! For your last slide, could you share with us on the NSC various contract hydraulic conductivity, if applicable. Thank you.

Response: The hydraulic conductivity properties of the various units encountered along the NSC contracts MM is involved in were not included inside the geological model as unit wide parameters. However, we did include the test results from the boreholes so we could see the higher permeability results and areas locally. The range of results was really too large to answer here though and isn’t really related to my paper or presentation. However, I think I can say that we had very high permeabilities locally within certain zones of F1 and some of the more fractured zones of the Bukit Timah Granite – as would be expected.

  

Question 4 (Ben Tan): how much of the borehole data can be included in the geological model? how can the model be exported/used as an input for plaxis?

Response: We can include all the borehole data in the model. If it is included in the AGS file then it is very easy to add. Even data such as the dates of drilling, the name of the driller etc. In terms of log data we include all descriptions, SPTs, rock core data, etc. If we don’t have the AGS file it usually means we need to digitise it so we are usually more selective about what we include – e.g. the key descriptors, SPTs and RQDs.

Once the data is in we generally need to develop our own interpreted model stratigraphy and consider if we want to build any numeric models based on the test data. So in summary all borehole data can be included within the model but we are selective about which data is relied upon for the profile interpretation.

 

In regard to exporting a model for importing to Plaxis – yes it can be done. For 3D Plaxis the whole surfaces can be extracted. For 2D Plaxis design sections are cut and the 2D profiles extracted directly from the model. However, typically Plaxis prefers the surfaces and profiles to be a little more simplified than that which is developed in software like Leapfrog. As such we typically have an intermediate stage using software such as CAD or Civil 3D to prepare the sections and surfaces for import into Plaxis. And even then the Plaxis modeller may still need to do some edits within Plaxis itself. We are currently working on ways to streamline this flow but there is still work to be done. Although also worth noting that Leapfrog and Plaxis are both under Bentley now – so potentially later software revisions will allow for easier transition between the two softwares. 

 

Question 5 (KC): Hi Andrew, the 3D ground modeling is based on limited SI data done. Do you think good quality geophysical survey data is a must to create good quality model?

Response: Geophysics is not a “must” to create a good quality model but is highly beneficial and in most cases will lead to a better-quality model. The objectives of the model and the ground conditions present will inform which methods would provide the most benefit. But generally when trying to model the profile, particularly the rockhead, seismic refraction from the surface is very useful. Crosshole (seismic) tomography is even better where you have localised features you want to investigate – but you need a closely spaced (<30m centre to centre) grid of boreholes to get the best out of that method. We have found this particularly useful in investigating faults, voids, and conditions under rivers.

 

Most geological modelling software will have a plug in to allow you to import the geophysical data (e.g. SEG-Y format) but this is sometimes a premium feature. We have used this recently in Leapfrog but in the past we also included such data by reformatting it outside the modelling software to bring it in as simple numeric data in csv format, or even just importing the interpreted 2D sections from the seismic refraction.

The only issue with geophysics is that it does need to be treated with caution. There will usually be some discrepancies between the borehole data and the geophysics. In such cases we typically give more confidence to the borehole data but use the geophysics to better inform our interpretation.   

 

Question 6 (Yasunori Shoji): 3rd presenter: To ensure the accuracy of DGM, how do you think about using geophysical and geotechnical investigation data as reference?

Response: Yes. For geophysics see my above answer to Q4. For geotechnical test data, yes particularly on complex models we try to include whichever test data we think will help us better understand the geotechnical conditions in a 3D spatial context. This includes lab test data as well as in-situ test data and logging parameters. For example we might include oedometer data so we can see which areas and units (and sub-zones of those units) might be more compressible. We’ll do the same with permeability tests, strength tests, etc. Often we use this to better understand the stratigraphy but even when the particular tests don’t inform the geological profile it is often useful to see them in relation to the 3D model – as it is a very strong visual tool and often clearer than when presented on a chart or 2D plan or section.

 

Question 7 (Colin Bullock): obviously depending on the geology but can you recommend a maximum borehole spacing for model creation?

Response: As highlighted in the question the suggested spacing required to generate a 3D geological model is dependent on the anticipated consistency of the geology. It is also dependent on the sensitivity of the structure proposed that will impact and be impacted by the ground it will be constructed on or in.

However, I would generally suggest the following:

1. Develop conceptual model of the ground based on existing SI data and desk study information.

     a. Critical to answer how complex is the ground? What ground risks exist? What is relevant to the structure proposed?

    b. SI is never a replacement for desk studies and good geological understanding. The SI further feeds into that but doesn’t replace. And that geological understanding is really the most critical aspect of developing a 3D geological model.  

2. Based on the desk study and conceptual model the SI should be proposed. Ideally the SI should be phased with a preliminary investigation to build understanding of the ground. Then an more detailed investigation to constrain the uncertainty and reduce risks.

3. The extent and density of the preliminary investigation should be as detailed as the budget and programme will allow. There is no minimum or maximum spacing at this stage but the more we do at this stage the better informed the design will be and the greater opportunity for optimisation so there is a clear financial incentive to make this as detailed as possible.

    a. For tunnels the spacings could be 50-100m at this stage with LTA projects often even more closely spaced. However, it is also common to see greater spacings due to the lower budgets and short programmes allowed at preliminary stage. Such spacings leave significant gaps between and with linear infrastructure there is often a lack of data either side of the alignment to gather a three dimensional picture of the subsurface. So where possible it is recommended to include some boreholes that address the 3D aspect and if possible include geophysics to fill the gaps and also address the 3D aspect.

      b. For stations and structures with a wider / 3D footprint, it is good to get a good spread of data that extends out beyond the immediate footprint. A grid spacing of 30-60m may be a good target depending on budget and site constraints. Again it is good to supplement with geophysics where possible.

      c. The preliminary stage investigation should be sufficient to build a detailed model of the ground as it is at this stage it will have most benefit.

4. At detailed design stage the aim should be to verify the conditions identified at the preliminary stage, further constrain uncertainty and risk, and refine the model to allow for accurate design.

     a. At this stage it is usually too late to make major changes so the benefits to loading in all your investigation at this stage are lessened. Although it is still critical to ensure the final investigation is sufficiently comprehensive because even if you can’t make major changes you can still refine the design and better identify ground risks that must be mitigated to enable safe and timely completion of the project.

      b. Boreholes along LTA tunnels might target 25 m spacing and boreholes along station perimeters could be as close a 6-12m. This is very close and should enable good refinement of the model.

      c. However, it should be noted that in Singapore it is often the practice to rely on wash boring with SPT samples only every 2-3m. So you can still have significant gaps in both the horizontal and vertical axis – because critical features like F1 layers may be missed.

      d. The inadequacies of the investigation and the high complexity of the ground must always be considered in the modelling and documented to highlight the model limitations.  

 

Question 8 (LC Lee): To Andrew, is the 3D geological modelling adopted in your NSC projects also used to carry out 3D analysis for ERSS design?

Response: Yes, it is but the different contracts have different requirements – from the contractors, checkers and the approval agencies. Typically the ERSS design sections do not follow the exact geometry of the interpreted profiles. But those profiles (from the model) are used to inform the design sections. Often the profiles are simplified slightly and often adjusted to make more conservative. In some cases the ERSS design sections still rely on the worst nearby borehole. I would like to see a more uniform approach to profile selection for ERSS design and a greater reliance on the interpreted profiles where they can be well justified. This is done successfully on many ERSS designs but more work is needed. 

Question 9 (Massimo Marotta): 3d interpolation between boreholes may not be fully realistic in case of complex geological features with faults /discontinuities. Any view ?

Response: Yes – interpolation between boreholes will not accommodate discrete features such as faults or discontinuities. There are a number of ways to deal with this though. If your fault or localised fracture zone or other such feature can be mapped or otherwise constrained then you can model that as a specific feature in the model. Depending on how it is modelled it may form a clear break/boundary so that interpolation does not try to span the break. Offsets may be modelled too so that interpolation can span the break but with a set offset. The problem is often that such features are not easily mapped – particularly during early stages of the feasibility or design process. In such cases the interpolations should be treated with caution and potential zones of faulting (or other such features) should be highlighted within the model.

 

It is always key that the design teams understands that the model is only a model and not an accurate representation of reality. The model limitations must be clearly documented and communicated with the wider team. It is here that issues with complexity and uncertainty can be included.

 

In addition I strongly recommend that 2D sections which portray the potential complexity are included within the 3D model, where such features are anticipated.

Field Pumping tests for Hydrogeological Studies

Seow Zhi Yi

 

Question 1 (Peiling Teo): Hi, may I ask how did you establish the lower impermeable boundary in your pumping test analysis? Thank you

Response: Model boundary was set 80m below bottom of pumping well. Sensitivity check was carried out and using open boundary does not affect the analysis.   

 

Question 2 (Yandi Lay): how do you determine the boundary in your plaxis model where you attempted to simulate the pumping test?

Response: Horizontal model boundary was determined using sensitivity check varying from 200m to 1000m. Sources of groundwater recharge such as nearby waterbodies should be taken into consideration in this back analysis. 

 

Question 3 (Lim Chi-Sharn): Would you be able share on the measured ground or wall deflection and adjacent building movement when the 4 pumping tests were carried out?

Response: No wall deflection or building movement were caused by pumping tests as the volume of water removed is relatively small and duration of pumping is relatively short, compared to typical excavations.

 

Question 4 (Kiefer Chiam): For pumping well, what is perforated screen length used? For Location2, drawdown of 18m, any settlement marker placed around to measure the impact?

Response: Perforated screen length was based on the SI and depth of target ground type. The drawdown of 18m refers to water level in pumping well. The cone of depression induced is quite steep hence groundwater level outside of pumping well experiences little drawdown. For location 2, surrounding settlement markers did not record any impact. This is likely because the pumping period was short and recovery of water level was relatively quick. 

Session 2-4: Grouting, Ground Improvement & Impact Assessment

Pre-Excavation Grouting Works Implemented in Deep Tunnel Sewerage System Phase 2(DTSS) Project

Kyi Khin

Question 1 (Ben Tan): most of the link sewer shafts are constructed using caisson method which is risky in permeable soils. Do u think this method should be disallowed?

Response: I think it is better to check the prevailing permeability of the ground where the caisson shaft is going to construct.  And need to look at the track record of the performance of the similar shafts in the area. It is always a good practice to carry out pumping test or the permeability test, at least.

If the soil condition is not favourable, for example soft compressible soil overlying, pre-ground water control measures should be implemented before excavation to mitigate risks beforehand. Recharge wells also need to be installed in advance. But Caisson method is nothing wrong in low permeable ground area, or in the area where soft soils are not present. So, the performance of this method with pre-excavation groundwater control should be further observed and assessed.

Currently in DTSS2, we are implementing Shaft Construction review (SCR) meeting before start excavation. It is more effective way to discuss technically amongst QP, SO, Contractor and PUB on geotechnical risks to prevent and mitigate before actual construction starts.

Ground Water Control for Deep Excavation Works at Stevens MRT Station, Singapore

Sahabadeen Mohamed Mohideen

Question 1 (Nicholas Mace): Sahab - can you confirm why the relief wells were included in the design? Presumably hydraulic uplift would have occurred otherwise, i.e. the PRW are critical

Response: Pending response

Question 2 (LC Lee): To speaker 2, did the relief well contribute to the water drawdown?

Response: Pending response

Question 3 (Ben Tan): any investigations or review of the dwall construction in terms of the embedment length and quality of stop-end or joint grouting?

Response: Pending response

Question 4 (Peiling Teo): Q1. What was the pressure applied to the deep recharge wells? Q2. Where were the localized areas where significant water ingress was observed?

Response: Pending response

Challenges of Jet Grouting for Mining Works

Seow Zhi Yi

Question 1 (Nicholas Mace): can you explain how non-treatment of F2 results in groundwater flow?

Response: Water ingress is not from F2 layers, but the presence of F1 strata has resulted in groundwater flow, non-treated F2 is an indication that similar problems in F1 layers.

Question 2 (Podianko Surya): To speaker 3-what was done to rectify the very high water ingress during mining?

Response: Chemical grouting was carried out. (Please refer to UGS 2021 paper by Ivan See).

Question 3 (Ow CN): any idea how columns will be like for inclined jgp grout columns in uniform and non uniform soil?

Response: A similar trend should be expected for inclined jgp grout columns compared to vertical however ensuring overlap should be more challenging.

Question 4 (Wong Chin Yik):  Were the permeability test results of JGP satisfactory despite ingress was observed?

Response: Yes probe drilling was carried out at certain locations and it should help where it is carried out. 

Question 5 (LC Lee): To Zhi yi, did probe drilling carry out before mining works start? Will the probe drilling help to identify the quality of treated ground?

Response: Yes the permeability test results were satisfactory.