For many years partnering has been a well-established principle for promoting a “best for project” culture within a construction contract. This endeavours to remove the adversarial nature of classical “engineer’s design” contracts by openly discussing issues which may impact on project performance and look to engineer ways to mitigate them within the terms of the contract... Read more
The recommendation of the joint code of practice for tunnel works promulgated by the International Tunnelling Societies and Insurers requires that active risk management form an integral part of every tunnel project. Tunnelling hazards and risks should be constantly viewed during the course of a project with effective systems put in place to do so. These systems should ensure effective communication of risk mitigation actions... Read more
The objective of all tunnel projects is to deliver the completed tunnel in time and in budget within acceptable environmental impact guidelines. This requires a complex interaction between a large number of people and between people and machines. Systems make all this work. Good systems can lead a project to its objective whilst poor systems can hinder... Read more
Most often geological site investigation for tunneling projects does not include directional cored bore holes covering the whole future tunnel alignment. In that case a need arises to test the rock mass condition ahead of the tunnel by short probes together with tunnel advance. In Hong Kong tunneling practice this need is recognized particularly in case of undersea tunnels and specified by the Client in conditions of contract... Read more
The flow of water into tunnels and the lowering of ground water levels is a transient process governed by the permeability of the ground, the storage of the various reservoirs and the available recharge. The resulting settlement is a function not only of the compressibility of the deposits but also of their ability to drain. This paper draws on a large database of information both from Hong Kong and worldwide to examine the transient behavior of the ground during drawdown... Read more
The engineering community has successfully completed many exceptionally challenging construction projects. Unfortunately, history has shown that on occasion political, time and monetary pressures have exceeded those of the water and ground, sometimes leading to failure. Authorities have attempted to mitigate these risks through the implementation of a variety of independent design checkers and verifiers and through the provision of supervisory teams on site... Read more
For many years partnering has been a well-established principle for promoting a “best for project” culture within a construction contract. This endeavours to remove the adversarial nature of classical “engineer’s design” contracts by openly discussing issues which may impact on project performance and look to engineer ways to mitigate them within the terms of the contract. These discussions rely on information and often the data available to client, engineer and contractor is incomplete, different or ‘selectively’ chosen to support a particular point of view. On many contracts clients have implemented project wide systems either independent of the main contractor or as part of the contractor’s specification and these have seldom provided the common information base desired in order to facilitate the partnering approach. The Powergrid Cables tunnels in Singapore represent the first time that all contractual parties have contributed to the implementation of project wide data management systems. The paper will describe the systems; their technical capabilities within TBM process and geotechnical control, their implementation and the contractual framework origins within which they have been delivered.
Data system, Partnering risk
The North-South, East-West and Jurong Cable tunnels currently under construction by SP Powergrid in Singapore are the second set of deep tunnels constructed by Singapore Power. The first phase of tunnels faced a number of challenges many relating to tunnel depths in excess of 40m below the water table. The most acute of these challenges involved flowing sand conditions which led to a stoppage of several months on one TBM drive.
In planning phase 2 of the scheme the designers were also obliged to set the tunnels deep to avoid other infrastructure such as metro, sewage, drainage culverts and road tunnels. At this depth, the tunnels would be passing though old alluvium, decomposed granite soils, metasediments of the Jurong Formation and Bukit Timah Granite with a high proportion in mixed face. Occasional exotic conditions such as lavas, eutaxites and ignimbrites in addition to volcanic breccias would be expected. Faulted ground was expected and in some cases is suspected to be linked to hydrothermal water. The water table is high and the near surface soils including peat, alluvial clays and residual soils are sensitive to water drawdown.
Since 2004 the practice in Singapore has been to promote independence of the parties within a construction project. The contractor’s design responsibility is normally limited to the temporary works elements.
In this structure the parties are clearly divided and are less able to influence each other. However, the flow of information between the groups may be hampered by the lack of incentives to cooperate. The concern is that each party will only do their minimum scope of works. The instrumentation contractor is only required to measure instruments and has no obligation to follow the works going on beneath his feet. He is not obliged to comment on where the tunnel is or on any underground changes which may affect the instruments. The instrumentation contractor many be required to provide a database but only for real time information and with no information concerning the progress of the works. Conversely, the main contractor cannot influence the positioning or type of monitoring undertaken and receives mostly paper reports which make it difficult to interpret what is occurring. Table 1 shows the main divisions in Singapore Contracts.
|Contractor||Qualified Person Design (QPD)
Specialist Tunnel Instrumentation and GI
Qualified Person Supervision
|Authority||Building Construction Authority|
The Code of Practice for Risk Management of Tunnel Works (ITIG 2006) recognises that risk management should be practiced by all parties of a project and across all contracts. There should be adequate communication of information and involvement of all parties on a frequent basis. Risks should be identified and evaluated and where possible removed, otherwise mitigated. The process should be formalised with drivers assigned to control it. The drivers should produce a live register of risks and constantly evaluate and update this in light of the information gleaned. It is the responsibilities of all parties to contribute to ensure this happens and that the risk platform is current and communicated.
Risk management is not about a deliverable, it is about a process. The live register of risks might seem like a report but a report is only one very time consuming way of communicating information. The report is not the register and can never be live if it must be constantly updated and re-issued.
This highlights a perennial problem in engineering construction, the formal submission. Formal submission takes time and must be signed off by several parties. The only formal documents are printed ones. Within this landscape a risk register would never keep track of the changes in construction and its information would never be communicated to the people who need to be informed.
The requirement for constant review and updating involves reviewing all the progress and production results in light of changes to construction methodologies, incorporating new ground investigation and instrumentation information as soon as it arrives and information on observations around the work site in the form of memos, photos. Using normal data management (eg spreadsheets, email) would tie the risk manager to his/her desk for such an amount of time that they would become less effective at managing risk. There must be other ways to do this.
The issues described above were recognised at an early stage by both the Powergrid directorate and their consultants. Previous cable tunnel projects which had encountered significant settlements during construction stage triggered the need for comprehensive instrumentation data management systems. It was hoped this would allow all parties (client, consultants & contractors) to access realtime monitoring data as well as the construction activities carried out at site through a common platform that provide interactive functions to generate graphical plots linking the monitoring data and the construction activities. Based on the cause and effect information, the system will allow the project team to predict, forecast and use the system as the risk management tool for lookahead construction activities.
In common with Singapore government projects the instrumentation contracts were directly let under the owner and supervised by the QPS. Centralised collection of these results and site observations from the QPS teams would be straight forward but the collection of tunnelling information would need the cooperation of the contractors.
The Client negotiated with all six contractors on the project to request them to contribute to a centralised system for the management of risk as part of the partnering process. The cost of the system was equally shared by all seven parties including client and contractors. Whilst still seen by the contractors as an owner’s initiative, the requirement to have a financial stake motivated contractors to use the system for their benefit.
The system scope and delivery was managed by a steering committee with representative from all the financial contributors. In addition to providing the system the IDMS consultant also provided staff to maintain the data and provide a line of communication for users. The team was managed by an experienced instrumentation and geotechnical engineer.
The scope for the IDMS comprised the following base requirements:
The Maxwell Geosystems MissionOS platform was selected as the solution. This existing system was implemented quickly and was ready to receive data within 2 weeks and fully operational within 3 months.
The system operates on a dual level with raw data at administrator level separated from processed data public level. At administrator level the staff can see all the data including data in error and have the option to quarantine data for later correction and adjustment. Staging servers sit on sites collecting site data performing preprocessing and sending to FTP servers and then to Linux/Apache web servers for processing and updating MySQL databases. Background server administrator level databases are both physically separated and platform separated from the web LAMP environment. This makes the systems extremely robust and protected from single point of failure.
A fundamental requirement for the system is that quality data be delivered quickly to the end user. The objective is to achieve data delivery before the daily tunnelling review meetings which normally take place at 8am. The only way to achieve this reliably is to automate as many data processes as possible on the server side. It was also decided wherever possible to take raw data into the system rather than processed data. This enabled data to be delivered to the MissionOS more quickly from the field and also for independent processing to be carried out. In this way all carries, resetting and other adjustments to the data were completely transparent to the party needing to use the data. Examples of automated data processing include:
The IDMS brings together information from investigation, design, production and monitoring. The system enabled all users to see the current status of their part of the project and some part of adjacent contracts where there was an interface.
Each element of data can be queried and plotted both singly and in groups and can be combined with other elements for plotting either of common axes or against each other. Information can be plotted as plan and sectional map views with overlain contours, as schematic user defined sections, as graphs against time or as X vs Y. All graphs are highly customisable. With all data spatially related analyses can be undertaken in terms of proximity to construction faces as the graph in Figure 3.5 shows. These spatial relations are calculated on the fly and are useful tools for analysing settlement development around the tunnels in varying ground conditions.
The utilisation of the system by the user base is shown in figure 3.6.
There are many challenges to driving IDMS utilisation. There is an in built inertia which resists the move away from systems that have been in place for many years even if extremely inefficient. There is a reluctance to let go of the ‘ownership’ of data and a general desire to be innovative. It is a popular misconception that a system will take the ability to innovate away from the Engineer. Despite this resistance it is only a matter of time before expert systems are the norm. The high level steering group has driven the utilisation of the system at all levels. The systems are used at instrument review meetings and to produce reports across the project. To build on this a number of features have also been implemented to make use of the systems desirable.
Surprisingly the power of the system has proven a drawback since with wider data coverage and more analysis options comes an increased requirement for training and expertise, regularly refreshed. For live presentation purposes it was felt that users preferred to be able to pre-prepare reports for display rather than try and use the system live. Rather than have these as static reports the system provider has implemented an interactive canvas sheet system where the definition of each graph, map and section created interactively can be saved and used as components on a canvas. These can then be combined with fields and tables of data to produce content rich reports. These reports can be reproduced at any later stage and contain the latest data.
Canvas reports can also be easily configured to be regularly updated and become control screens. Figure 3.5 shows a control screen summarising live TBM, geotechnical and instrumentation parameters for the control of settlements due to tunnelling.
In addition to basic measured parameter the MissionOS IDMS provided options to combine parameters either to derive other parameters ie two utility settlement points to define a tilt, or to display parameters in such a way to enable interrelations to be visualised. This was most commonly applied to measuring structural distortion of utilities, temporary works such as shaft linings and affected structures along the alignment.
As part of the risk mitigation details of the buildings were digitised to the system by the contractors, relevant details input and documents attached. Important details such as the maximum strain levels of structures in various modes are added and since the hazards are geographically referenced to the systems they are referenced back to TBM data logging and other instruments. Building groups were also created to produce reports focussing on the key building parameters.
Conventional AAA schemes rely on email and sms to deliver notifications. Feedback as a result of these notifications is normally by email and as such becomes immediately defuse. The IDMS has instigated an adaptation of secure social media blogging to enable feedback to be delivered to the relevant group. Each alarm starts a thread and responses relating to that thread are added by members of the user group. Photos and other attachments can be added. Whilst this was welcomed by the consultant and client, the Contractors were less willing to use the blog mainly due to the fear that uncontrolled commentary may have contractual implications.
A common hurdle to adoption of the systems is the complaint that entering the data involves additional work and this is particularly true in the case of progress reporting. To surmount this a simple template was prepared for as set of key items extracted from the construction programme focussing on key items to be tracked. The site engineer or inspector only had to input the date and time and the chainage for the activities in progress on a daily basis and email it to a dedicated email address where it would be picked up and processed automatically. This proved to be considerably less time consuming that their regular site processes.
A simple but appreciated modification is the ability to add site photos to items of construction and to locations in the map. By geographically locating the photos they are immediately set in context with the data to which they are linked which increases their value to the user.
The management of risk on tunnelling projects is the responsibility of all parties. In order to manage risk effectively the large quantities of data collected must be audited and assimilated rapidly with other data feeds to assist engineers to study both the effects of tunnelling and the potential causes. With many millions of records and gigabytes of data this can only realistically be achieved using systems. Careful consideration for the system delivery method is essential to ensure that it is utilised by all parties. Incorporating the systems within partnering agreements has been an effective way to achieve this.
The recommendation of the joint code of practice for tunnel works promulgated by the International Tunnelling Societies and Insurers requires that active risk management form an integral part of every tunnel project. Tunnelling hazards and risks should be constantly viewed during the course of a project with effective systems put in place to do so. These systems should ensure effective communication of risk mitigation actions. Where the response values are reached or abnormal response is observed, then the pre-defined contingency or remedial action needs to be taken in a timely manner.
Geotechnical risk management process should not be isolated from other project activities. The communication of risk information and consultation with the project participants are two-way processes that should be proactively undertaken and should continue throughout the duration of the project. Projects are embracing this approach but are finding in practice that that this is somewhat harder to achieve due in part to geographical and functional separation of teams, short time span and the sheer volume of information to be processed. This paper demonstrates how intelligent flexible computer systems were used to assist this process on the highly sensitive Klang Valley Metro in Kuala Lumpur.
Data system, risk, tunnel
The underground portion of the Klang Valley Metro in Kuala Lumpur extends from North Semantan Portal to South Maluri Portal. The tunnels start in the meta-sediments of the Kenny Hill formation and transitions into the karstic Kuala Lumpur Limestone in front of Pavillion in Bukit Bintang. Compared with the SMART tunnel the KVMRT tunnels are much deeper. The tunnels have either a stacked or parallel configuration with cover of about 20m and 30m respectively Both slurry and earth pressure balance TBMs have been used with slurry preferred in the limestone and EPBM in the Kenny Hill.
There are many geotechnical risks in the project. The Limestone is known to be karstic with cavities that are either empty or infilled. Both EPB and VD machines were used in the main geological formation with extensive instrumentation onboard had to be monitored real time so that measures can be adopted promptly to evade potential problems. In addition to the instrumentation data from the TBMs thousands of instruments were installed on the surface to measure the impact of tunneling and the construction of deep station boxes to the surroundings, many of which were read in real time. The sheer quantity of data is in itself a risk since staff need to assimilate the monitored data quickly and compare with the expected geology and surface response. Thus a centralised instrumentation data management system (IDMS) was implemented. This paper focusses on the contribution of the IDMS system to the safe completion of the works to date.
The project owner MRT Corporation of Malaysia awarded the USD 3.27 Billion contract to MMC-Gamuda KVMRT-T SDN BHD in April 2012 on a design and build basis. Whilst the main works were contractors design, the owner’s engineer had specified strict control criteria for the works and procedures to be followed should tolerance levels be exceeded. The Gamuda technical organisation comprised divisions looking after geotechnical works, design planning and programming and production. Instrumentation contracts were let under Gamuda and looked after by the geotechnical team which also implemented the IDMS system.
IDMS systems had been successfully implemented on the SMART tunnel and in view the heavy levels of instrumentation a similar approach was considered imperative and thus the services of the Maxwell GeoSystems’ MissionOS system were procured to draw in both manual and automatic real time data and communicate changes in the data to the various project teams.
As of the 2014 some 9500 instruments have been logged and of these almost 1200 instruments are real time resulting in 15 million records of data amounting to 100GB of data. Such a volume would be impossible to manage without IDMS especially as the protection of surrounding infrastructure is one of the key factors.
At the inception of the project the planning and production team considered the need for proper feedback of ground and groundwater movement data to the designers and tunnel and station managers so that the effect of ground conditions on tunneling processes and station box excavations be better understood. For this to be effective audited data had to be presented to the user quickly. It was also hoped that the IDMS would become a back bone to the risk management process.
During the tunneling works the TBMs could advance as much as ten rings in a day therefore it was vitally important that data could be delivered to the IDMS within one day of being recorded. This too was crucial during the station box excavation. Uploading processes in the IDMS were automated and data could be filtered processed review, audited and published within one hour of receipt.
With rapid delivery of data, quality can fall and any way to simplify the data management will potentially lead to improvements.
The project is extremely sensitive and the consequences of false alarms are damaging. As such it is important that all data is verified before being published to the public domain. Rapid verification and confirmation of results is an advantage particularly if coupled with independent processing.
In order to convince staff to move away from the use of spreadsheets for data analysis the system would need to enable the user to do more with the data more quickly than they would in normal circumstances. If coupled with fully customizable reporting users would accept the concept of centralized data resources.
The system was procured on a supply and maintain basis with the instrumentation and TBM contractors required to feed data to it. In the early stages it became clear that there was a need for an additional data auditor in order to achieve reliable timely data. Inputs (data Sources) to include:
Instrumentation data was sent from the subcontractor to the MissionOS IDMS by FTP. Automated email collection and manual file upload was also made available. A requirement of the IDMS was that data be presented in raw format and processed independently as a check. A format was developed composing of a data file and a setup file. On revision a new setup file was to be provided with the new instrument settings. In reality data would often arrive before its setup file or revision file and therefore systems were put in place to recognise jumps and spikes in the data which would require further information from the contractor. The IDMS was configured to alert the administrator when such data had been received such that timely action could be taken.
Many IDMS data management processes were completely automated including: correction for inclinometer pipe top changes, correction for MPBX or piezometer pipe top level changes, corrections for benchmark movements and instrument temperature correction amongst others.
Before alarms from real time data systems were sent out care needed to be taken to recognise and quarantine values exceeding credible thresholds or potential spikes. Instruments fluctuated due to natural temperature variations and, in the case of piezometers, due to seasonal changes and it was important to avoid unnecessary alarms.
All data was quality assured by running weekly data audits in excel. The resulting files were sent to originators for correction/completion and the files uploaded to update the database. Each audit file was saved as a record of the state of the database at that time.
The main use of the system was directed at instrumentation data management and response. Since the IDMS was a client requirement the initial focus was on fulfilling this requirement. On completion of this exercise it was further developed to address contract reporting and this was well received as a tool to produce the reports.
Few on the production team used the system since it was seen as a duplication of the proprietary process control tools on each TBM. The intention of the system however was not to duplicate or replace the TBM control system used for driving the TBM but to provide a wider forensic capability to investigate ground movements, TBM performance and, through back analysis, arrive at predictions of future best practice. The open customisable reporting system and the fact that data could be combined in plan, section and graph enabled the relationships between tunnel driving parameters and ground/building response to be investigated.
The system provided tools to generate a variety of custom reports using an interactive means of saving report definitions. These report canvases could be pre-prepared and brought up at any time which made them ideal for daily PTT (Permit to Tunnel) Risk Assessment reviews.
Combinations of any parameters could be plotted with one common axis. For example a long section plot of settlement against chainage could be plotted on top of a plot of face pressure.
The first variable density TBM drive (Klados et al. 2015) was driven from Cochrane shaft towards Pasar Rakyat at a depth of about 30m through KL limestone. The subsoil profile developed from boreholes and geophysical studies indicate that this section was typical karst with potential cavities and intrusions into the TBM path.
The ground ahead of this first section had been cleared therefore this section gave a good opportunity to test the capabilities of the variable density machine. The machine performed well in consistent limestone up to chainage 8180 but here hit a karstic feature venting slurry to the surface.
After several rings (Figure 6.2) the ground was more competent and normal tunnelling resumed. This event acted as a valuable dry run where the capabilities of the VD tunnel boring machine was tested as a prelude for similar cases where such karst intersections would not be within open ground.
The TBM also hit a karstic feature at Maluri about 50m away after passing the LRT structure. As suspected a karst feature was encountered as evidenced by the excavated volume parameter which indicates negative values for three to four rings coupled with sudden drop of cutterhead torque and shield articulation contact force.
In EPB with liquid mucking mode, careful control of the flow of feed and slurry pump is also vital to maintain the face pressure. In poor and unfavourable ground condition, HD injection rate will be increased to prevent bentonite loss and stabilise the face pressure. Control of the advance speed and screw rotation speed helped to maintain face pressure at a roughly uniform level at the face.
When there is an instrument breach, the Main Contractors requires to produce an AAA report upon advised by the Supervising Consultant. The general content of the report is stated in Table 1.
|2||Details of construction activities|
|3||Result of inspection|
|5||Summarise the results of adjacent instruments which may be affected|
|6||Review of subsequent monitoring|
The report is prepared with input from the contractor’s instrumentation team, construction team, design & technical team and supervising consultant to conclude the actions/recommendations to be taken at site due to this instrument breach. This report will be circulated within a stipulated time so that timely and appropriate actions can be taken. Due to the large scale of this project and instrumentation numbers installed, a good database management system is indeed need to ensure these reports can be easily created, generated and distributed with easy access to the system.
Using blogging technology the system supplier was able to automate almost 90% of the report and furthermore keep track of the timing of responses. The time requirement for generating this kind of report reduced as all related parties involved in contributing to the report can access and register their input via a web-based portal at their convenience. The system also prompts the users to input their comments when needed.
The implementation of live reports online means that content can continue to be added as instruments worsen or additional information is provided without the need for duplicate report revisions.
One of the key factors when determining the success or failure of systems is to appreciate the motivation for using it. A system imposed on a contractor through the Particular Specification will get less support than a system implemented by the Contractors themselves. A system put in place by the owner may not get buy in from the Contractor which is an absolute necessity if the data is to be current and of high quality. Some considerable success has been gained when implementing systems as part of partnering approaches.
Cloud based systems enable widely separated working teams to collaborate and share information however it is human nature for groups often to work in isolation and sometimes to compete. Driving projects with wide technical standards and platforms is a full time business and can only succeed if championed by motivated staff at high level.
The enormous quantities of data collected are always going to challenge networks and it is important that systems are designed for low bandwidth environments. This involves pre-processing of high volume data, distribution of processing between browser and server and dynamic analysis of bandwidth to assess whether to send data and plot locally or plot on the server and download image. The KVMRT MISSIONOS system is capable of plotting a map of all 9000 instruments and tunnel progress within 5 seconds and a graph of 50 rings of TBM data within 15 seconds over a mobile phone connection.
The objective of all tunnel projects is to deliver the completed tunnel in time and in budget within acceptable environmental impact guidelines. This requires a complex interaction between a large number of people and between people and machines. Systems make all this work. Good systems can lead a project to its objective whilst poor systems can hinder. Important characteristics of all systems are their cost and capabilities, time for implementation, available inputs and outputs, user friendliness and ease of communication. Robust systems require built in redundancy and disaster recovery measures to account for whatever fate throws your way. Really useful systems integrate across multi-disciplines to bind the team together and focus resources. The paper will review human and computer systems from various points of view and look at the options for integrating these into various types of projects.
One of the often ignored characteristics of construction is that in many cases each project starts from scratch and once finally delivered loses almost all of its assets to other projects and companies. This is unavoidable since no company can afford to keep the workforce if no follow up project exists. As a result, systems that have evolved on one project are not always transferred to other projects. Even when successive projects occur the same teams are seldom involved or are not in place when systems are set up.
Systems are essential parts of modern construction. They provide the communications and the checks and balances. They introduce rigor into daily activities and control day to day risk. If set up effectively, systems provide efficiencies which translate into time and cost savings.
In his 2009 Terzhagi lecture Alan Powderham stressed the importance of observational engineering in driving both innovation and safety. All systems can be divided into components of procedure and feedback. Procedure is the series of systematic steps required to undertake a task and the feedback deals with the way results are reported back to enable the procedure to be assessed and modified. Human systems work well on the procedural level provided that sufficient training is given but fail in the feedback where they are often limited by a number of factors:
The application of standards for data and reporting can go some way to deal with the issues of consistency but then these must be controlled adding an additional overhead to the system. Machine based systems can provide improvement in each of these areas.
Table 3.1 shows the typical contractual and organisational hierarchy of a project and the locations where systems will exist. Even in this simple Owner – Engineer – Contractor matrix the number of potential systems in place is large, and often in excess of a hundred systems can co-exist. Those shaded in grey in Table 3.1 are transferable in that they are corporate systems which are part of an entity’s quality assurance documentation and in many cases part of a software system built into the core of the company management.
At the discipline and task level, systems are largely brought to the project by individuals and are commonly based on that individual’s experience of a particular type of construction. Whilst this experience is valuable it may also be somewhat prejudiced to a certain set of conditions which may not apply in the new role. Such new systems will require some effort to initiate and maintain especially with teams unfamiliar with the methods. Often the architect of the systems does not get the required quality of input because of this initial unfamiliarity. If the team subsequently transfers en-masse from one job to another the system will evolve but unfortunately this is seldom the case.
|Types of systems||Owner||Engineer||Contractor||Subcontractor|
|Report to executive||✓||✓||✓||✓|
|Discipline Group||Geology Geotechnical||✓||✓||✓|
|Instrumentation and Monitoring||✓||✓||✓|
|Daily working methods||✓||✓||✓||✓|
|Information analysis methods||✓||✓||✓||✓|
Consider the systems in place along each stage of a Project Delivery Cycle. In addition to the variety introduced by the various corporate boundaries rigid contractual boundaries also ensure that a wide variety of different systems are used at various stages of construction projects. Very little live factual data is transferred across the contract boundaries. In most cases deliverables are PDF reports and CAD files and the ownership of the data remains with the party undertaking the contract.
Table 4.1 shows a listing of the various systems which may be utilised during the lifetime of a project. Many of these are based on IT systems but there is an array of forms and formats used. Typically systems are initiated at the construction stage and can vary from document management to instrumentation management systems to full data management systems.
The main driver of instrumentation systems is the sheer quantity of information which puts it beyond the capability of conventional spreadsheet management. Real time monitoring and alarms also require systems.
The key requirements of instrumentation management systems:
Whilst the instrumentation systems contain information about progress and instrumentation it does not contain the technical or job specific data collected as the job is progressed. Full TDMS systems have been implemented on seven tunnel contracts over the last 15 fifteen years for the Hong Kong Government
The TDMS systems expand on the instrumentation systems through the addition of:
|Project Management|| |
|Planning|| || |
|FEASIBILITY||Feasibility|| || |
|INVESTIGATION||Investigate|| || |
|Contract delivery|| |
|Commercial resolution|| |
It is clear that in most projects there is no strategy for live information transfer across contract or organizational boundaries other than in the form of reports and drawings. This is a considerable limitation if the information is to be used effectively in later stages.
A possible reason for non transfer across contract boundaries is the perception that non paper data results in questionable liability. Those who define deliverables within the project delivery cycle may consider only reviewed and properly signed off reports and drawings to be worthy of handing to downstream service providers. This approach is not always applied. In the case of the Channel Tunnel Rail Link a database of over 3000 boreholes and trial pits was made available to Contractors during the bidding process to facilitate access to information. It can be argued that by making available the factual data on which interpretations are based improves the knowledge base of future project contributors.
The organizational barriers are a function of the project procurement method adopted. These are summarised in Table 4.1 below.
|Project Procurement Method||Effectiveness of Systems|
|The Private Project||Clearly defined separate roles and responsibilities – systems unlikely to be unified|
|The Project Delivery Partner||Ultimate owner my require systems for his own monitoring but otherwise as with engineer’s design|
|The Public Private Partnership||Teams joined by financial shareholding in the scheme. Information systems may be shared until things go wrong|
|The BOOT or BOT Scheme||As with PPP joint financial commitment may engender information sharing.|
|The Engineer’s Design||Clear contractual demarcations mean information systems are only shared if specified|
|The Design and Construct||Designer and contractor united and there is a high possibility of combined systems. Unless specified there may be no obligation to provide information to the Owner|
|Partnering||True partnering requires certain system to be shared eg. risk and progress/performance monitoring. However if the partnering has little financial incentive attached it is rare for parties to unify systems.|
|Alliancing||One party delivering the project with high reliance on independent verifiers for safety. The optimum environment for unified systems.|
Most systems are implemented at construction stage and since construction projects do not have the time to engage in systems development the systems have to be in place almost immediately. This is especially true when the need for systems only becomes apparent later in a project when the existing human systems are unable to cope and the lack of information becomes a significant risk. Whilst the systems need not be the complete the architecture must be defined such that they can be augmented later.
System implementation has been driven by subcontractor, contractor, designer/consultant and owner within a variety of contractual settings. Observations on the different levels of success are documented below.
Generally the subcontractor’s scope is limited and the subcontractor is interested only in parts of the system which are critical to his scope. Focus is applied because the systems are mission critical.
If the systems are implemented by the contractor for his immediate benefit then focus will be applied. If data is to be published widely then the data entered to the system will be limited only to what the Contractor is contractually obligated to provide. If the Contractor has other systems then he is normally reluctant to integrate these into one environment if significant additional up front expense is incurred.
The designer/consultant faces the most design risk on the project and therefore requires the feedback on project performance particularly for instrumentation. In this respect the focus is most acute when systems are implemented by the designer.
In Engineer’s design contracts the resident site staff are tasked with tracking the progress and technical details of the project to ensure the job is built in accordance with the specification and contract terms. Systems implemented through the engineers site staff will be of high quality but will be limited if contractual provision is not made to require the Contractor to pass information.
Since the owner can specify anything of the contractor and designer he is in the best position to implement the systems on the project. Since he is not the designer or contractor it is often difficult for him to define the requirements of the various parties and provide a useful environment for all.
The previous review has indicated that none of the existing forms of application of systems is entirely satisfactory. Information does not remain live across a project’s lifespan and cannot pass organizational or contract boundaries except as “dead” PDF reports. Information provision is limited by contract obligation and in many cases information is deemed to have market value to the various organizations and is guarded. The designers who really need the information often do not get it and project wide systems are normally only put in place at construction stage.
The following extracts from the Nicol Highway Joint Committee of Enquiry interim report 2004 reflect how important breaking down these barriers is to safe completion of the project. The committee state that:
“There is a need to integrate information from the various instruments and to relate the crucial information to what is happening on the worksite, as well as the quality of each of the elements in the construction. These two requirements must be present in all relevant projects.”
“A consistent supply and collation of up-to-date and accurate monitoring information is essential. There is a need to ensure this. Its correct and timely interpretation, including comparisons between predicted and actual values, is crucial for safety. Monitoring at critical locations as construction progresses is important. This will allow adverse trends to be detected early.”
It is clear from these two statements that construction control systems require the integration of all data including construction information, instrumentation results, ground conditions both assumed and encountered, design assumptions, method and prediction.
Owners have the power to implement such systems across a project but not necessarily the skills since this will require a balance of IT capability and engineering knowledge. Design consultants have the understanding but will normally consider this to detract from services they would historically provide themselves.
The management of information flow across these boundaries is clearly in the hands of the owner. By implementing management systems for the project at implementation stage, the Owner can streamline the many processes of investigation, design and construction and manage risk and increase safety. To do this the Owner must be prepared to publish data in forms other than PDF reports and drawings.
One example of an owner led initiative is the implementation of an independent monitoring consultant role by the Mass Transit Railway Corporation for its Regional Express Line Project (XRL). The XRL comprises a signature 8 hectare station with platforms underground and 26km of high speed underground alignment up to mainland China. This project is subvented to the Mass Transit Railway Corporation by the Hong Kong Government and as part of the subvention a number of independent consultants are put in place to check on the technical and financial details of the subvention.
One of these is the Independent Monitoring Consultant or IMC. The IMC’s role is to provide a review of the adequacy of geotechnical monitoring for the project, to independently measure a proportion of instruments and to set up systems for the collection and processing of data from all parties. The IMC is required to report on alarms and abnormal trends and to check that movements are in line with those expected. The systems provided by the IMC are intended for the use of all parties.
An extension of the Independent Monitoring Consultancy successfully applied for MTR is the Independent Information Consultant. The IIC is a company spanning both engineering and information technology and able to define the requirements of the information system for the project lifetime at the outset.
Early and independent implementation of systems by engineers with knowledge of all stages enables the owner to dismantle the contract and organizational boundaries which are a hindrance to effective management of the project. Only the owner can do this.
The system is The MissionOS and acts as the single repository for published data on a project. In such a content and context rich environment data is displayed as designed ie: maps are displayed as maps, geographical data is displayed and accessible as live data for graphing and analysis, documents which cross reference to data and geographical objects are available through the map and data links. Normalization of all the data via a single structure and interface enables engineers to explore relationships and even to define rules which can be implemented programmatically to extract information or highlight anomalies.
The IIC spans all the stages of a project and is initiated as soon as the project feasibility is to be investigated. A key part of the IIC concept is that all consultancy and construction contracts include a clause defining an interface between the consultant/contractor and the IIC, a description of the data to be published and the form in which it is to be provided.
The IIC role does not negate the Contractor or Consultant’s ability to utilize their own systems to undertake their work. It actually facilitates this by making all historical information available. Their only obligation is to publish data and results back to the IIC.
The backbone of this expert system is the project programme and the project risk register both of which should be live documents (BTS 2003, GEO 2009). Each separate job on the programme is linked via its spatial coordinates to the map. The programme determines the expected progress for that job and the system tracks the progress and any commentary.
Key events eg. stoppages etc can be added to the programme at any time and the programme can be updated easily from any standard programming software. The risk register is also linked to the system by jobtype, location and jobtitle and rules are set within the register to define where and when risk profiles may change. For example:
|The Statistic||Cutter change interventions are running at a frequency of 1 per 40m within ground of the current type and cutters have not been changed for 30m.|
|The Monitoring||The TBM is approaching a change from Grade V residual soil to colluvium.|
|The Hazard||The ground type is not self supporting and cannot take compressed air|
|Susceptibility||Cutterhead interventions are highly susceptible to the hazard|
|The consequence||Collapse during cutterhead intervention and potential settlement of overlying utilities.|
As the probability of a required cutterhead intervention rises along with probability of the ground hazard the increase in overall risk of instability during cutterhead intervention triggers an alarm. The mitigation measure for this alarm will be shown.
On a project with many headings it is potentially difficult for a risk manager to have all of the details all of the time. Risk registers may be many thousands of records long and contain the combined experience and expertise of many engineers and geologists. The risk engineer may not be expert in all of the areas that are covered by the register. Having increasing risks highlighted by such a system is of benefit to all parties and for greatest impact the warnings can be directed at the risk owners for immediate action. Such warnings are also published to the weblog such that action and follow up is tracked.
Current systems for tunnel information management and communication seldom satisfy the requirements of all parties. Contractual and organisational barriers to information flow increase project risk, cause frustration and waste time. A kaleidoscope of high powered technical systems are in use but despite this PDF is the common data standard. This is not acceptable. Insurers, lawyers and international institutions have agreed that data from all stages of a project must be integrated, shared and communicated rapidly in order to effectively manage risk. The paper has described features of systems which are available to do this. The paper has reviewed the delivery mechanism and has concluded that the systems are best delivered by the owner using an independent specialised body combining geotechnical and tunnel engineering and IT.
Most often geological site investigation for tunneling projects does not include directional cored bore holes covering the whole future tunnel alignment. In that case a need arises to test the rock mass condition ahead of the tunnel by short probes together with tunnel advance. In Hong Kong tunneling practice this need is recognized particularly in case of undersea tunnels and specified by the Client in conditions of contract. However, the understanding of probing ahead is limited to visual logging of color of flush, grade of chippings and penetration rate with the use of stop watch by the geologist during percussion drilling. Visual logging proved to be efficient and accurate in detecting extreme rock mass condition e.g. completely decomposed rock. Yet from the other hand it appeared to create a significant safety hazards for the logging geologist who works close to maneuvering booms and rotating string at noise level from hammers above 150 dB. In such hard working condition the quality of the records can also be affected. It is presented in this paper an interpretation of Jumbo percussion probing to detect adverse rock mass condition ahead of the tunnels as an alternative to visual logging.
Analysis was carried out on data recorded by AMV Jumbo two boom machine: P — penetration rate (m/min), HP — hammer pressure (bar), FP — feed pressure (bar) and RP — rotation pressure (bar). Selected cases were used to test the developed data processing algorithm in addition to the automated interpretation provided by Bever Control software. The results of the tested algorithm show the adverse rock condition can be inferred from the percussion drill logs with confidence, hence the presence of the geologist at the face when drilling could be eliminated.
A significant advantage of this system was the ability to interpret not just probe holes but also the drill holes belonging to regular grouting rounds. As such, probing on HATS2A was undertaken as part of regular grouting works and did not cause any additional delay to the production cycle. The statistical interpretation methodology was largely based on works by Schunnesson (1997). In his approach the penetration and rotation pressure can be assumed as dependable variables while the rock resistance, hammer pressure and feed pressure as well as drill string length are the independent variables. By removing hammer, feed and length effect on penetration rate the remaining variations can be inferred to correspond to changes of rock resistance.
The sample that will provide a reference during further processing needs to include a representative of all expected Penetration PR hence the probes have to intersect all possible weathering grades, intensity of fracturing and rock lithological types. These extreme rates do not need to appear in the Combined Sample in large numbers yet they need to be present e.g. a thin seam of completely decomposed rock within predominantly fresh to highly weathered granite provides full range of penetration rate in the sample. In that sense completeness of the Combined Sample can be verified by geological mapping of the tunnel section from which the probes were selected. Practically it may not be possible to assemble a sample without a histogram bias towards higher or lower penetration rates. This effect can be mitigated when processing the data. Typical distribution of pairs (PR vs HP) in the Combined Sample is presented on the Fig.1A and Fig.1B.
Each string of dots on Fig. 1A represents P vs HP dependency when drilling through uniform material of particular strength. The highest string of dots represents the strongest material while the lowest string of dots represents the weakest material. The Combined Samples A and B on Fig. 1B are limited to a small “clouds” of all possible readings demarcated as grey area. Both include points corresponding to the weakest and the strongest material yet each shows significant bias of the distribution of P values, A towards the weaker material, B towards the stronger material.
Filter criteria are determined by analysis of graphic presentation: Penetration P, Hammer Pressure HP and Feed Pressure FP versus probe hole length L of all probes included in the Combined Sample. The selection of filter minimum and maximum threshold values can be also aided by analysis of histograms of P, HP and FP.
Filtering criteria are used as an input to Excel custom filter option and applied both to combined sample and particular probe log being investigated.
Filtered data of the Combined Sample is used to obtain regression equation representing dependency between penetration and probe length. This dependency is machinery specific and is well researched. It is related to increasing weight of the drilling string and increasing friction area between the string and the probe hole wall. The linear regression represents the best fit for the effect of drill string length on Penetration Rate. The coefficient of determination R2 is low and generally irrelevant due to large variation of the Penetration related to other than length factors.
The factors obtained from Combined Sample linear regression are used to correct the data of the string length effect both in the Combined Sample and in the particular probe log being investigated. The correction is independent to the intercept of the linear regression, hence there is no distortion of the data related to Combined Sample not covering uniformly the different rock condition e.g. harder rock is represented by far more numerous readings than the weaker rock.
The dependency between Penetration and Hammer pressure and Penetration and Feed pressure is determined by arranging the P and HP as well as P and FP pairs from the Combined Sample and obtaining linear regression. Hammer effect is usually the strongest and needs to be removed from variations of P. For the purpose of the interpretation the linear regression is assumed to adequately represent the relationship. The coefficient of determination R2 of the linear trend line is very low and is inapplicable in general due to large variations of penetration related to variables other than Hammer pressure. The trend can be used to correct the Penetration values by subtracting the portion explained by the regression i.e. the difference between the regression value at particular Hammer pressure and any reference value e.g. average Penetration Pav =(Pmax-Pmin)/2. The selected reference level, as well as the intercept of the regression have no impact on the final result of the interpretation as they are reduced during final scaling of the Penetration to the 0 — 1 a range when using the equation (1). The residual Penetration rates can be understood as deviation from Pav (or other reference value) related to variations of rock quality. This regression trend does not represent rock mass of average quality (due to the likely bias of the Combined Sample) so the residual Penetration can not be used directly for evaluating rock quality. It can be considered as an interim parameter only in the processing.
The correction of Hammer pressure effect is carried out both on the Combined Sample and on particular probe log being investigated.
Scaling of the corrected Penetration (often found as normalization) can be performed by application of a formula (1) to L and HP corrected log of particular probe under investigation.
The Pmax and Pmin for scaling purpose are derived from the Combined Sample after L and HP effect correction. Scaling the corrected penetration to the standardized range e.g. 0 — 1 provides an easy tool for interpreting the meaning of the parameter in terms of rock mass condition. Assuming the Combined Sample includes the whole range of expected Penetration rates, Pmax then Piscaled = 1 represents the worst condition (completely decomposed rock or extremely highly fractured rock with thick clay coating on joints) while the Pmin and Piscaled = 0 correspond to the best rock condition (fresh an hard rock with few joints and scarce or no infill on joints). Using the scaled parameter instead of the absolute value of Penetration allows also for comparing the logs and correlating the 0 — 1 range with Q-value or RMR scale, as well as with Rock Grade scale.
A minor fault encountered in tunnel N of HATS 2A (Sandy Bay to Cyberport) (Fig. 2) was used to test the efficiency of interpreting algorithm.
There were two distinctly different rock types covered by the probes that should appear in the interpreted probing results. The rock within the fault was extremely highly fractured, grade I to III, fine grained tuff with abundant unconsolidated clay coating on joints. Total clay content was evaluated as much as up to 50%. The surrounding rock was grade I, fine grained tuff, with scan line estimated RQD around 80%. Discontinuous decomposed rock coating was present on a small percentage of the joints.
From the positions of the probes against the fault zone (Fig. 3) it was inferred that probe 12 was fully within the stronger rock mass while probe 14 encountered fault zone material along part of its length.
The Combined Sample consisted of data of all 17 grout holes drilled at chainage 1032 of Tunnel N. Filtering criteria — the threshold maximum and minimum values of P, HP and FP — were selected by analysis of raw data graphs of the probes and histograms. Example HP graph of probe 12 shows minor fluctuations of working HP within 100 — 110 bars (Fig. 4) and peaks related to rod changing, hence data containing HP values below 100 bar could be safely removed from Combined Sample as well as from particular logs. The example log of penetration from probe 12 and 14 (Fig. 5) indicates the penetration (and conjoined HP and FP) lower than 0.5 m/min could be filtered out from the Combined Sample and from logs. The upper range filter was assumed as 5 m/min as the upward peaks were related rather to probe cleaning. The histogram plot of penetration from the Combined Sample and from probe 12 (Fig. 6) shows the expected different distribution. Note that the extreme values of penetration are still present in the Combined Sample.
Penetration logs of probes 12 and 14 were corrected for the effect of increasing drill string length (Fig. 7) and subsequently for the effect of Hammer pressure variations (Fig. 8). The regression line shows Length correction significant as it may reach up to 0.7 m/min — 17% of the whole range of the filtered penetration 0.5 to 4.5 m/min. The P vs HP linear regression indicates the possible Hammer correction up to 1.5 m/min (37%) within the range of working hammer pressure. The final scaled penetration parameter is independent from intercept of the linear trend.
The corrected values of penetration in the Combined Sample were used for selection of Pmax and Pmin values for further scaling the P record of particular probes 12 and 14 into a range 0 — 1. The obtained parameter after applying 6 point moving average (together 180 mm) is presented on Fig. 9. and proves distinct difference between probes that coincides with their spatial position: 12 within surrounding competent rock, 14 within fractured an partially weathered rock of shear zone.
There are several sources of error in the proposed processing of the data to obtain scaled Penetration as an indicator of rock properties. These are e.g.:
However, from the objective point of view, which was detecting the extremely adverse rock condition, these simplifications and assumptions showed to be insignificant. The tested algorithm produces a parameter sensitive enough to detect the potentially hazardous rock condition. The completeness of the initial Combined Sample can be verified and amended during the project. Data from other project in similar rock condition can also be used to form the Sample.
The advantages of applying interpretative probing based on automated logs of drilling parameters are:
The flow of water into tunnels and the lowering of ground water levels is a transient process governed by the permeability of the ground, the storage of the various reservoirs and the available recharge. The resulting settlement is a function not only of the compressibility of the deposits but also of their ability to drain. This paper draws on a large database of information both from Hong Kong and worldwide to examine the transient behavior of the ground during drawdown and reviews the effectiveness of surface recharge systems.
Under drainage is the process by which a tunnel (sink) drains water from an aquifer which has limited immediate recharge. This can be as a consequence of an impermeable upper layer, an aquiclude, and/or due to lateral recharge being restricted.
The net result of under-drainage is a lowering of groundwater pressure in materials, some of which may be compressible. This paper looks beyond the conventional steady state view of under-drainage and addresses transient characteristics. It summarises the authors’ experience of rocks exposed in an uninterrupted transect across 20km of Hong Kong during mining of the Strategic Sewage Disposal Scheme Phase 1.
Hong Kong is a mountainous region with a thick mantle of residual soil and saprolite, draped in complex sediments comprising intercalations of terrestrial colluvium and alluvium with marine sedimentary deposits. Reference to Fookes (2007) illustrations on geomorphology and Leeder (2011) on sedimentary process show immediately the difficulties faced when attempting to assess the impact of a sink installed within this ground. Imagine a thick sequence of marine deposits overlaying the model shown in Figure 1 and it is clear how inadequate conventional analyses are. Predominantly models are two dimensional or if three dimensional they are so simplistic as to be misleading.
The following provides a list of the typical characteristics of materials and their impact on the analysis of ground water flow:
Water: Tests carried out during SSDS Phase 1 shows that three types of ground water exists in Hong Kong. Freshwater from rainfall percolating downwards, seawater percolating sideways as the freshwater table varies and deep groundwater which has been in place for millions of years and whose chemistry changes in response to hydrothermal incursions from depth, through diffusion and through in-situ reaction with the ground. Seawater and fresh water may percolate downwards only after this ancient water is displaced. The characteristics of seawater and freshwater are well known but the deeper water is characterized by very high biochemical oxygen demand, high Fe content and conductivity, the natural consequence of which is to promote corrosion and rapid deposition of salts, in particular Ferric Hydroxide on exposure to air.
Rock: May be massive or closely jointed but back analysis shows that bulk rock permeabilities in Granites and Tuff vary from 5 x 10-8 m/sec to 5 x 10-7 m/sec with an average of 1 x 10-7 m/sec being equivalent to steady state tunnel inflow of 1 litres/minute/metre of tunnel at 100m of head. The porosity of igneous rocks is considerably less than 1% (except for some rare tuff breccias) and increases with weathering. Flow tends to occur along discontinuities which may be cooling joints, un-roofing joints, tectonic joints or volcano-sedimentary features such as tuff breccias. Typically. rapidly cooled igneous/volcanic rock tends to be more closely jointed and more permeable than granites cooled gradually. Those rich in volatiles such as Rhyolites may also have open structures and close jointing and as a result are difficult to drill. The actual permeability in rock depends as much on the size of the discontinuities as on their connectedness and this is most influenced by the extent to which mineral deposition has or has not taken place. Deposition of quartz, calcite, chlorite, various iron compounds and clays minerals including the ubiquitous kaolin are hard to predict and occur in several phases. Faults may not always be the main conduits so often assumed since they are frequently observed to possess significant clay in the gouge and decomposition and therefore annealing on one or other side of the fault plane. These may sometimes behave as dams and it may be that the inrush of water experienced on hitting faults is a function of the high head and not necessarily of high permeability.
Decomposed rock and Saprolite: For the purpose of this classification we group all decomposed rock from grade IV to VI within this category and as such the permeability can vary typically from 1 x 10-7 to 5 x 10-6m/sec. Porosity varies from <5% to up to 15%. Flow in decomposed rock is governed by relict fissures and soil pipes which may or may not be connected since the arrangement of soil pipes may also be governed by the terrain of the decomposed rock surface.
Colluvium: The distribution of colluvium is dependent on the palaeogeomorphology of the decomposed rock/rock surface even where hidden by other sediment. Colluvium may be intercalated with alluvium at the margins of upstanding areas of rock. Colluvium can be highly heterogenous and comprise transported boulders and/or chaotic slide materials on slopes and/or debris flow materials which are strongly channelized in their upper reaches but are spread thinly over wide areas at their distal ends. The key characteristic of colluvium is that it is discontinuous and is unlikely to behave either as an aquiclude or as a significant pathway for water migration.
Alluvium: Alluvium is an often mis-used catch all phrase and is often considered to be a single material for the purpose of hydrological assessment. In reality this is a complex arrangement of gravels, sands, silts and clays which form an alluvial plain. Permeabilities can range from 1 x 10-5 m/sec in gravels to as low as 1 x 10-10 m/sec in consolidated clays. Porosity can also vary from 20% to 40%. Of all materials it is the alluvium which most influences the movement of groundwater during under drainage.
Marine deposits: Marine deposits are deposited over the alluvium during sea level rise and may vary from well winnowed medium sands to silts and clays depending on the energy of the marine environment in which they are deposited. Small amounts of clay and silt can significantly affect the permeability and particle size distribution curves should be used to extend in situ and lab permeability tests in order to estimate relative permeability.
Fill:It is often assumed that fill is highly permeable, however this will very much depend on the type of fill, marine sand or CDG, and level of compaction. As a result permeability can vary from 1 x 10-6 to 1 x 10 -5 m/sec. The vertical permeability of the fill will also largely be controlled bv layers of re-precipitated mud which may have accumulated at the base of the reclamation and within the reclamation during storm periods. It is often assumed that dredged channels for seawalls are always pathways for recharge whereas in reality the foundation alluvium material may not be particularly permeable even discounting the potential for considerable thickness of re-precipitated mud.
Artificial materials: The presence of linear construction projects, which often comprise kilometers of diaphragm wall, will severely affect the natural drainage paths within the soil strata. Assumptions of lateral drainage, particularly from hillside catchments, will be incorrect. The effect of retaining systems on natural drainage has been seen on countless projects, where installation of D-walls has resulted in elevated water levels and significant (up to 5m) differences in piezometers from one side of an excavation to another. Nevertheless in most designs water pressure is assumed to exert evenly on both sides of an excavation.
The conclusion appears to be that it is too complex to model. Certainly any simple model cannot be considered to be in any way to represent what will happen except if that model is constructed on a small scale and a lot of GI data is available to confirm that the materials between source and sink are hydrogeologically consistent. On a basin scale analyses will be grossly misleading. There is still room for modeling but only as a test of sensitivity. Particular factors to identify are: What are the key drainage materials? What is the recovery time likely to be for various assumptions of recharge boundary,aquiclude permeability and thickness? What is the specific storage in these key materials and how will they be affected by depressurization?
Once these key permeable horizons have been identified it is necessary to try and identify where they are situated, the connectivity between them, and the potential rock conduit (if the tunnel is in rock) and to areas which may be sensitive to depressurization.
A concept that is common in geology is the facies map. Used frequently in petroleum and groundwater exploration this maps units not by their geological unit or formation name but by material or origin. A map therefore can represent the distribution of material type at a particular time, depth or relating to a particular stratigraphic event ie an unconformity.
Such maps are easily prepared from GI logs by picking a chronostratigraphical boundary and mapping the distribution of material types at that boundary. Key horizons are: the material at the base of the alluvium which is unconformable on the decomposed rock and the alluvial material immediately below the base of the marine deposit which sits unconformably on the alluvium. These two horizons will determine the connection to the rock/decomposed rock flow paths and the locations where depressurization will affect the compressible marine deposit. Facies maps provide an assessment on whether the assumptions used in numerical modeling are likely to be correct. Consider the example in Figure 2. For any arbitrary section, is it sensible to assume all the flow is in the line of the section? Is it sensible to assume that the materials identified are in homogenous layers or will flow tend to be concentrated on particular pathways? In reality fluvial systems are far too complex to be modeled on a catchment scale by simple layer cake models. It is far better to identify a realistic hydrogeological model using basic geological tools and relationships and use this as a basis for simpler experiments to investigate the potential sphere of influence of a groundwater sink and the likely consequences.
When water is encountered in a tunnel the flow path is immediately determined by the high permeability geological units. The speed of reaction is determined by the porosity of the medium. In rock, highly connected and conductive pathways can produce drawdowns in piezometers several hundred metres away whilst rock piezometers several 10’s of metres away remain unchanged. Rock fracture porosity is so low that a small quantity of water extracted will have a large effect.
The effect will be dependent of how deep within the rock the piezometer tip is placed ie away from recharge from nearby more permeable horizons, and how close this is to the tunnel sink. Commonly these preferential water pathways are associated with deep weathering and some recharge will be drawn from overlying saprolitic soils producing depressurization and consolidation. Whilst compressibility is relatively low, these layers can be thick and consolidation is rapid, typically completed within 2-3 months.
The most important pathway for depressurization is that which connects fissure flow in the rock to highly permeable gravel and cobble horizons in the lower alluvium. From here connectivity is assured since the deposits originated in a fluvial/alluvial environment and are probably still operating as conduits for ground water drainage.
For open tunnels in rock the first interception with water is through probe holes. These will produce immediate drawdowns in some rock piezometers which will recover once grouted. Drawdown in overlying compressible deposits will not occur if the holes are allowed to drain for periods of only a few hours. Longer periods of drainage may start to produce drawdowns which will start consolidation at rates governed by the permeability of the consolidating material. The amount of consolidation will depend on the length of time the change in stress operates. This is termed drawdown days and is determined by the length of time the sink is in operation and the rate at which the ground will recharge once the sink is removed Figure 4.
In the example in figure 3, the rate of discharge is related to the permeability of the aquifer tapped by the sink and in this case the initial rate of drawdown is chosen to be 2m/day and it is stopped after 2 days. The rate of recharge is determined by the surrounding permeabilities at the recharge boundaries and the specific storage of the aquifer. It is set at 2/3 of the drawdown rate in this example which was typical of the recovery of piezometers in SSDS after transient depression due to probe hole drilling. Note that actual recharge rates may be many times lower than this value.
Settlement can be calculated by integrating the curve for the number of drawdown days and applying simple one-dimensional consolidation theory the result of which are shown in Figure 4. Table 1 summarises some typical soil compressibility for Hong Kong materials. Those for materials with high permeabilities are taken from the results of field observations including both drawdown related consolidation and rebound. It is noticeable sthat the results of back calculation of Mv and Cv values for decomposed soils suggest significantly higher compressibility and lower coefficients of consolidation for decomposed soils particularly within fault zones.
|HDG||CDG/CDV||Colluvium||Alluvial Sand||Alluvial Clay||Marine Sand||Marine Clay|
The values of Mv are based on back analysis of field observations combined with laboratory test data from Hong Kong. Values for compressibility of sands are based on observations of comparative density from SPT data. The values back analyses for CDG and CDV are relatively high and are in contrast to their higher stiffness. The observation of rebound on water level recovery suggests that the observed consolidation includes significant shrinkage. In the short term, CDV, CDG, confined loose sands particularly those with clay intercalations and re-deposited clays pose the most significant threat of short term settlement (Figure 5).
The following example is taken from SSDS tunnel the details of which are published in Territory Development Department (2000). Tunnel C was excavated at -90mPD between Tseung Kwan O and Kwun Tong across Junk Bay. Several instrumented reclamations were nearby as shown in Figure 6. In this example water inflow increased markedly from chainage 700 to 800 and was constant up to chainage 1200 (shown in red).
The graphs in Figure 7 show the change in piezometer head against radial distance to tunnel face which occurred as a result of mining in this section. In rock there is a clear drawdown in only one of the piezometers and this is the nearest to the tunnel. A steep drawdown curve can be seen in some piezometers but others show no response despite being relatively close. This is due to the control of rock structure and connectivity on transient water flow. In the decomposed rock (CDV) a wider drawdown curve can be identified with affects seen as far as 400m from the tunnel face but still several piezometers are unaffected as would be expected with the decomposed rock inheriting many of the characteristics of the rock.
The alluvium by contrast shows no recognizable drawdown curve but affects are observed as far as 700m from the tunnel source. Many piezometers are unaffected and there is no correlation with radial distance. Since the marine deposit also relies on the alluvium for its connection to the sink it is no surprise that the marine deposit shows the same relationship.
The results suggest that the use of simple layer cake homogenous hydrogeological models is not valid for transient water inflow studies. Their use should be restricted for assessment of steady state water balance analyses only and not used for assessment of the likely lateral extent of water drawdown effects. For this more regional hydrogeological drainage models should be constructed.
Control measures implemented within deep tunnelling projects are typically limited to pre-grouting at pre-determined intervals, based on simple models of rock permeability combined with remediation grouting when inflows continue and / or exceed the rate of allowable discharge. During shallow tunnel construction, for instance by cut and cover methods, groundwater is typically controlled by implementation of a cut off (diaphragm walls, secant pile walls, grout curtains etc) and a recharge system, which is often a series of recharge wells places at pre-determined spacings with little understanding of the complexity of the hydrogeological conditions.
The difficulties in creating a successful recharge system are well recorded, notably in CIRIA C515 (Preene et al., 2000). The primary consideration is whether compressible materials that are of concern can be recharged directly or whether recharge should be targeted at surrounding, more permeable facies which will prevent under-drainage in the first place. The concept that low permeability materials will not readily accept recharge is logical yet apparently overlooked in the majority of recharge system designs.
In addition to targeting recharge wells to intercept suitable facies, issues of groundwater chemistry and suspended solid content within the recharge water often leads to clogging and bio-fouling of the wells with the net result that a systematic programme of maintenance is essential and many contingency wells must be installed, in addition to the original design number, in order to maintain the design quantity of wells in operation while maintenance is ongoing. Recharging too close to the sink may also result in excessive feedback, where additional grouting / pumping solutions are then required within the tunnel to deal with the extra inflows.
Recharge schemes fail due to a limited understanding of the hydrogeological complexities of the surrounding environment and the assumption that an holistic recharge of that environment is necessary and can be achieved. In addition to the recommendation that facies maps are adopted during the planning and design stages to model groundwater movement those receptors sensitive to groundwater drawdown in the surrounding environment should be recognised and targeted for analysis. Where groundwater drawdown is controlled by fissure flow and palaeo-channels, and not simply primary permeability of overlying units, drawdown may not occur above or adjacent to tunnel alignments but also at some lateral distance from the tunnel. Having undertaken a facies mapping exercise and recognized the potential flow paths across the wider area surrounding the alignments, control of groundwater around specific sensitive receptors should then be considered in addition to any cut-off, grouting or recharge adopted at the tunnel site.
Sensitive receivers may vary from concern for surface habitats, including protection of river systems’ base flow, particular ecosystems or crops, to excessive settlement, distortion, damage or collapse of slopes, utilities, infrastructure and buildings. In addition to consideration of annual mean rainfall, protection of surface habitats could include the use of shallow trench recharge systems or sprinkler systems. Targeted protection of slopes, utilities, infrastructure and buildings may rely on a combination of local groundwater cut-off and recharge systems around the receptor of concern, taking into account the underlying hydrogeological conditions, foundation type and particular risk of that receiver. The authors are aware of at least one current project in Hong Kong where an existing tunnel is being protected in this way while the construction of new tunnel proceeds beneath it.
In most tunnel projects where water is strictly controlled, it is the transient behavior of ground water which is of key concern. In this regard it is the higher permeability materials which are most important and in particular those which are compressible and have restricted recharge. These include CDG/CDV deposits in fault zones, sand pockets in alluvial systems and some re-deposited clays including those in seawalls and beneath dredged reclamations.
In the transient case detailed engineering geological and hydrogeological subsurface mapping is recommended ahead of numerical modeling. In complex hydrogeological settings modeling should be aimed at sensitivity testing and identifying the likely relative contributions of recharge and discharge in certain key areas.
Groundwater control measures must take into account the complexity of the model produced and target recharge at appropriately receptive facies in areas where key sensitive receptors have been identified.
The engineering community has successfully completed many exceptionally challenging construction projects. Unfortunately, history has shown that on occasion political, time and monetary pressures have exceeded those of the water and ground, sometimes leading to failure. Authorities have attempted to mitigate these risks through the implementation of a variety of independent design checkers and verifiers and through the provision of supervisory teams on site. These organizational systems have resulted in improvements but a common complaint is that the monitoring information is received too late and in forms which are not readily analysed or checked by the engineers. For the first time a role has been provided for an independent professional body to check, audit and deliver project monitoring data to the project stakeholders. This is in its second year on the Express Rail Link (XRL) in Hong Kong and the paper will report on its method of implementation, benefits to the project and provide guidance for those considering the management of project risk on future projects.
The Mass Transit Railway Corporation (MTRC) have developed the concept of the Independent Monitoring Consultancy as part of a project wide risk management initiative. This paper describes the origins of the role, its original concept and the way in which the role has evolved to become a key part of the MTRC systems. The key elements of construction monitoring can be divided into:
In all construction cases design and prediction is provided by the design engineer with review from the owner and associated regulatory bodies. The designer will design instrumentation to suit the monitoring of these parameters and to enable sufficient data to be collected to feedback information of the performance of the design. This is then passed to the site teams for implementation. At this time a variety of bodies become involved in the project delivery process. These are indicated in Figure 1.
The communications and interactions between these various bodies are complex but at this stage the construction is live and decisions will need to be made within a tight time framework. The ability of the management to react to change will be governed by the systems in place.
With such a complex interaction, management of projects face several challenges. Much effort is put into ensuring that the objectives of all the team members are aligned towards common safety, technical and commercial goals by the use of partnering and in some cases alliancing. Historically little focus has been placed on the management of information flow between the various parties to the project.
Ultimately deviations from the prediction for the works lead to technical and commercial conflict and parties often justify their positions by “cherry picking” information to suit a particular argument. Neither party has all the information. In some cases two sets of records exist and this is counter-productive.
In fact many of the issues boil down to information.
Change is a natural part of construction and the management of change should be embraced within the project management scheme. Since this is an expected occurrence the management structure should be geared to respond proactively rather than defensively. The objective should be to maximise the amount of time spent engineering rather than operating a computer.
Better information sharing and communication is required and the delivery of “agreed” factual information should be speeded up. There needs to be better cross pollination with information from other teams and facilitation of back analysis and comparison against design. Risk management should be integrated into the systems rather than being separated. Better communication, agreeing, sharing and condensing of data leads to smaller more effective teams. Efficiencies relieve time and monetary pressure which impact on quality.
In Hong Kong all construction instrumentation is carried out by the Contractor using specialist subcontractors. An estimate of 1% of construction cost is normally set aside in budgets for instrumentation of ground engineering projects within the urban environment. This should be considered as a “lower bound” and in some very complex high risk projects up to 5% has been set aside.
Monitoring within Asia tends to be seen as an imposition by the owner/designer on the Contractor to safeguard against failure or damage to the environment. Such a policing approach has not engendered buy in from Contractors and as such many will opt for the cheapest solution. If results are inconclusive or instruments fail this is seen as removing restrictions from the Contractor’s working environment. A change in attitude to one where the monitoring is considered a help would require that the contractor take some benefit from the monitoring particularly if it were to show that movements are better than predicted. Such observational engineering requires careful application but if applied would give incentive to produce quality reliable information.
In Hong Kong instrumentation Contractors are normally subcontractors of the main contractor. If time or monetary pressure is felt by the main Contractor this is normally passed on to the Instrumentation Contractor. This is particularly risky if the Contractor’s instrumentation subcontractor is also constructing temporary works. The MTRC have addressed such a conflict by requiring the instrumentation contractor to be independent of the geotechnical works.
In Singapore, all instrumentation for Government works are contracted directly to the owner. This removes any potential pressure the Contractor may bring to bear but also removes any direct involvement of the instrumentation contractor in the construction process thereby breaking the feedback loop.
When the KCRC and the MTRC merged in 2009 all rail ownership and rail project delivery was brought under one roof. Since this was a part public company a certain level of review was required for Government projects to be sub vented to the MTRC for development. Independent monitoring of environmental compliance has been there for some time but additional areas covering technical monitoring, finance and design were added.
Initially the independent monitoring of the West Island Line was issued as a works contract given the high proportion of measurement over consultancy services. The second independent monitoring contract was for the Regional Express Line and by this time it was issued as a Consultancy reflecting the increased focus on engineering services. The consultancy comprised:
Low prices put pressure on quality. In some cases where resources are just not available data may be extrapolated or fudged. Often there is a focus on providing good looking graphs rather than truly representing the data. The advent of independent physical monitoring helps to ensure that the instrumentation and survey personnel provide the required frequency of measurement and that the measurements are undertaken with the required levels of accuracy.
The provision of a central unified system for the publishing of data to all the project members is the foundation of a new method of construction risk management. Provided by an independent third party, this system acts as the published repository for construction data which is accessible over the web to all. Secure layers are set such that on multi-contract projects parties can only see information relevant to their contract and to the contracts adjacent.
The setup of the system is designed to ensure the maximum independence of the data and speed of processing. Key aspects are:
These capabilities significantly extend those required by the original MTRC specification. This highlights the difficulties common with specifying a system implementation. Unless there is in depth knowledge of what can be done the Engineer tasked with specifying a capability has no knowledge of how difficult or much time this will take to implement, particularly when dealing with a general software house. In this case the supplier reacted to the intentions of the MTRC and provided them with achievable solutions.
Whilst the public portal is most users interface to the data this is only the tip of the iceberg. Behind the scenes systems are designed to cope with imperfect and incomplete data so that data can be presented in a timely fashion. Such systems include:
Key aspects of the physical monitoring undertaken are that the results reduce and not increase uncertainty. To ensure that monitoring is effective it is important for the project team to appreciate:
As a specialist in this field the Independent Monitoring Consultant provides useful advice. A further important area of independent review is the choice of alarm levels. These are normally linked to design predictions and tolerances and it should be clearly identified what is the basis for the alarm eg: tolerance of a building or structure or predicted movements of a piece of temporary works and this indicated on any AAA report. In some cases it may be neither. Most of these alarms are based on the primary parameter – settlement, deflection, load, draw down. In many cases this is not the crucial parameter. The advantage of systems is their ability to track other derived parameters. In most cases this is distortion and resulting tension.
The use of systems greatly assists the on-going review of project performance. The relationships between changes in instrumentation and changes in the works progress can be identified easily and subtle geographical and temporal relationships revealed though combining data together and even animating. The web access allows experienced engineers to view the data from remote sites and provide feedback based on accumulated knowledge some of which may be directly related to strata into which the project is being constructed.
Whilst the IMC service is focused mainly on the instrumentation results, the holistic web management of data from constructions is already a reality. Total Data Management Systems (TDMS) are already in place on projects for Drainage Services Department and include all aspects of production and technical data. These are in the process of being combined with commercial management, programme management and risk management systems to provide a single project resource for information. The ability of the systems to enhance communication and facilitate decision making may support future use of observational engineering but in the meantime the availability of data in structured system guarantees its availability for the engineering of the future.
The provision of independent monitoring, processing and reporting of ground and structure movements linked to the provision of truly independent advice has improved the risk profile of the MTR projects leading to a trend of lower insurance premiums.