Site mobilization in the lush green, steep sided Emolweni Valley some 16km inland from Durban for two TBM bored tunnels for the Inanda-Wiggins Phase II project in South Africa began in January 1990. In October 1990 the ATB 35 HA System Demag TBM manufactured by Voest Alpine at its Zeltweg factory in Austria for the project arrived by ship at Durban and was transported by road to site. After four in assembly, the machine with its 250m-Iong trailing back-up designed by ROWA of Switzerland, was launched from the west portal of the Clermont Tunnel to drive the first 1:700 uphill tunnel toward the Aller Valley.
For the first 2,000m of the 5.5km long drive, the TBM excavated medium to coarse-grained sandstone of the Natal Group, which has an unconfined compressive strength of between 150-250MPa and a high quartz content. At tunnel alignment, the competent horizontally bedded sandstone has an RMR (Rock Mass Rating) of good to fair and is dry despite being well beneath the ground water table.
Of the five rock support classes specified by Keeve Steyn, Engineer of the project for Owner Umgeni Water, this section required classes mainly II and III. These include 1.5m-Iong resin anchored, fully cement-grouted rockbolts tensioned to 5 tonne and occasional welded wire mesh over the crown where required in faulted zones. This is in addition to a uniform 50mm thick layer of shotcrete applied throughout the tunnel to protect the joints and discontinuities from excessive flushing-out of their fill material during the service life of the pressurised aqueduct. Relief holes of 450mm long x 35mm diameter drilled into the rock at 1.5m centres avoided buildup of water pressure behind the shotcrete lining.
Through this first 2,000m section, the 3.5m o.d. hard rock TBM achieved a maximum advance rate of 39m/day working on a 24h/day, five and a half day/week production cycle with routine maintenance carried out on Saturdays and Sundays. TBM utilization in the 11 months from launch to chainage 2,000m averaged 18% with regripping time 3%, and cutter changes about 20%. Average TBM advance through the first 2,000m section was 9.5m/day. During this period, two major modifications were performed on the cutterhead, taking up some eight weeks in total.
Since then, the TBM has entered a known fault zone of heavily fractured and jointed sandstone in which high water ingress rates were anticipated by Keeve Steyn. The geological report states that, where the tunnel passes below deeply incised valleys ,with cover reduced from the 300m maximum (180m average) to between 60m and 90m, there is the possibility of tapping surface run-off water during wet periods via open fractures extending from the surface down to tunnel level. The report anticipated water ingress in excess of 2 litre/sec/10m of tunnel length in the fault zones.
When TunnelTalk was on site in late November 1991, the TBM was at chainage 2,400m, about half way through the drive, and 40m into a third very wet and faulted zone. Total water ingress of about 50 litre/sec measured at the portal and as much as 10 litre/sec over the first 3m from and including the face, to the TBM’s tight fitting steel dust shield, was making advance difficult.
The principal problem was that fines produced by the cutting action of the TBM was forming a heavy slurry with the water ingress making it difficult to lift in the muck buckets on the cutterhead for successful deposit on the overhead primary conveyor. The slurry was being washed out around the edge of the TBM’s dust shield and had to be manually shovelled into buckets and lifted up on to an intermediate conveyor for removal. Excessive build-up of the slurry material underneath the TBM rapidly causes a damming effect of the water which then threatened to inundate the lower two of the four electric motors, which are water resistant, not waterproof. In such cases, the TBM must stop and adequate steps undertaken to control the situation.
In addition to creating a difficult working environment, the water was also making muck conveying difficult, particularly on the primary conveyor where it was washing over the sides of the belt, and on the steeper transfer conveyor on which it tended to fallbackwards. Water was also damming in two areas back down the tunnel where deviations in level had created two deep troughs. Pumps in these troughs had to operate continuously to control the water levels and allow safe passage of the muck trains, each comprising 12 x 4.75m3 Hunslet-Hudson skips hauled by 22 tonne, 135kW Hunslet-Hudson diesel driven locos.
Additional problems caused by the build up of silt included:
Anticipation of significant water ingress by Keeve Steyn was based on both extensive site investigation and on prior experience of tunnelling in the same area. Localized ingress of up to 12.5 litres/sec was experienced during excavation of the three tunnels constructed between 1982 and 1984 during Phase 1 of the Inanda-Wiggins project. The Reservoir Hills and University Tunnels, of 2.6km and 1km long respectively, were excavated by drill+blast while the 3.3km long x 3m diameter Sherwood Tunnel was excavated by the Clifford Harris/Marti JV using a 3m diameter Robbins TBM dressed with 13in disc cutters. Excavation records of the three tunnels were made available to bidders for the current contract(3,4). The current TBM operation is the fourth undertaken in the civil tunnelling industry in South Africa, with the Sherwood TBM drive having been the third.
Additional site investigation of the Phase II tunnels was undertaken in the late 1980s to augment and up-date information gathered during studies for the full length of the aqueduct under Phase I of the project. Total site investigation for Phase Il constitutes some 3.7% of the Rand105 million (approximately US$40 million at the time) awarded contract price, including contingency and escalation.
The geological report and the Engineer’s interpretation were part of the contract documents and included bill items for pumping of up to 200 litre/sec. The bill of quantities and specification also included various items to deal with water ingress.
Throughout tunnel progress, a 57mm diameter x 30m long probe with a minimum 6m overlap was maintained to warn of wet conditions ahead. There were also measurement items for the installation, where necessary, of plastic sheeting and half section ducting to channel residual overhead water ingress down to the invert drainage channels to facilitate the application of shotcrete.
The primary specified strategy against excessive water ingress was cement grout injection. This however was the most expensive option both in materials and in TBM standing time. It was also the beginning of efforts to render the tunnel watertight - a provision that was not required by the Client.
Decisions about when, and for how long, grouting should be carried out before TBM excavation would resume required clear lines of communication between the Engineer and the Contractor to arrive at mutual agreement.
Significant water ingress of up to 25 litre/sec across 2.5m to 3m of tunnel was encountered at the end of September 1991 at about chainage 2,010m. Grout injection of 30 x 50kg packets of cement, or 1.5 tonne, in one day was said to have cut the water flow to about only 20 litre/sec. This first encounter with significant water ingress also damaged three of the TBM’s four electric motors. These had to be taken out and sent to Johannesburg for overhaul causing a total TBM stoppage of about two weeks. A second 3m long section of water bearing faulted ground was encountered at chainage 2,418m and again grout injection was required. It was practically impossible to judge the length of these fault zones between sections of dry competent sandstone other than referring back to the site investigation report and core samples, and doing additional exploratory drilling ahead of the face as contemplated in the contract.
The Contractor’s tender was based on using grout injection where necessary to control water ingress to no more than a “residual flow”. As such, the JV decided to subcontract the grouting operation. This proved unsatisfactory. Instead of requiring the subcontractor on a few rare occasions, call out has been more frequent and urgent than expected. Mobilization of the subcontractor can take up to 36 hours depending on the time and day of the week when the call goes out. To rectify the situation, the Contractor intends to establish his own grouting system on the TBM backup. In the meantime the tunnelling crews attempt to improve the working environment by erecting the specified plastic sheeting, held in place with sealed expansion type bolts, to channel water to the drainage channel in the invert segment.
Shotcreting had been suspended and was to be applied as a secondary operation, rather than concurrently with TBM driving as planned. In the meantime, the Engineer is carrying out detailed examinations of the rock to review the need for the shotcrete lining, other than as required for immediate support.
The tunnel contract was a vital element in supplementing and securing water supply for the greater Durban area, which at the time, was recognised as one of the fastest growing cities in the world. Since repeal of population controls in South Africa, the population in Urngeni Water’s service area is projected to increase from the current five million in 1991 to a maximum 15 million by the year 2025. Demand for water is expected to increase to between 1,200 and 1,700 mega litres/day as compared to 547 mega litres/day in 1985. Most of the current water requirement in the Durban metropolitan region, some 7m3/sec of the total 14m3/sec demand, is supplied from the Nagle Dam and fed by gravity to the Durban Heights treatment plant through a series of four older aqueducts of tunnels and siphons. To augment this supply, a master plan was designed in the early 1970s to build the Inanda Dam on the Umgeni River and convey a further 4m3/s of water via gravity through 16km of tunnels and 4km of siphons to the Wiggins water treatment works in Durban. Construction of the dam, by South Africa’s Department of Water Affairs, was delayed for several years due to problems in relocating people in the inundation area, but was finally completed and impounded in 1988. In the meantime, due to the urgent need for additional water, Phase I of the aqueduct was implemented and commissioned in 1984. This involves lifting water through the Clermont pump station on the Umgeni River to a balancing reservoir in the Aller Valley and conveying it to the Wiggins waterworks via the Reservoir Hills, University and Sherwood tunnels and their interconnecting siphons.
It was the devastating floods of 1987 that exposed the vulnerability of the Nagle Dam aqueducts and concentrated attention on completing the second phase of the InandaWiggins masterplan to increase Durban’s water supply and secure it against future flood damage or possible sabotage.
Umgeni Water is a self financing parastatal organisation established by the South African Government in 1974 to take over the control and management of the Umgeni River and its 7,000km2 catchment and service area. To realize the current Phase II construction of the Inanda-Wiggins masterplan, it raised some Rand150 million on the financial markets. A prerogative of the funding arrangements stipulated a local currency contract with payments by the Client only in South African Rand. In 1979, the Durban office of consulting engineer Keeve Steyn was engaged to design and supervise construction of Phase I. Then in 1987 it was again selected as the designer and Engineer for Phase II.
The contract documents are based on the FIDIC form of contract but incorporating the 24 contractual practices recommended by SANCOT (the South African National Committee on Tunnelling) and developed from propositions adopted by the ITA (the International Tunnelling Association). These accommodate an equitable sharing of risk between the Client and the Contractor on tunnelling projects, and the Client carries the risk of changed ground conditions. The site investigation report, including the Engineer’s interpretation, which in this case represents R3.5 million or 3.7% of the awarded contract price, is an integral part of the contract documents, The intent of this form of tunnelling contract is to avoid traditional "Clause 12" claim situations and substitute timely resolution of measurement issues during the course of the contract. Negotiations are between the on-site authoritative representatives of the Client and Contractor wherever possible.
With this contractual approach established, Umgeni Water invited contractors from the domestic and international civil construction industry to prequalify for the contract. Of the 31 responses, 28 were selected. These then formed nine JVs of which eight submitted a tender by the deadline of September 1989. Given the 17 volumes of the contract document, the selected JVs were given eight weeks to digest the geotechnical report (12 volumes), and a further ten weeks to prepare their tenders.
The contract made provision to tender for either or both tunnels and to submit proposals for either or both a drill+blast or TBM operation. It envisaged two independent slightly uphill drives. One being the 5.5km long drive uphill from the Emolweni portal to the Inanda Dam intake and the other, 5.5km uphill from the Aller Valley to the Emolweni Valley with the two connected by the 320m-long Emolweni Valley siphon.
Drill+blast tenders received proved less competitive with the TBM proposals. A significant reason for the pricing difference was that an in-situ concrete lining specified as the finish of a drill+blast operation was eliminated and substituted with the shotcrete lining on a TBM operation. The three lowest-priced TBM options also anticipated substantially higher progress rates than drill+blast. To match projected TBM progress rates, drill+blast would have to advance from more than one portal.
The lowest tender, at Rand105 million, including contingency and escalation, (approximately US$40 million) was submitted by the JV of Murray & Roberts, the largest construction group in South Africa, and Porr International, its Austrian JV partner. The closest three competitors were the Shaft Sinkers/Seli JV at Rand115 million and a contract period of 45 months; the Ilbau/Torno/Strabag/Goldstein JV at Rand119 million and 36 months; and the Philip Holzmann/Basil Read JV at Rand124 million over 47 months (similar to a drill+blast price submitted by Basil Read Mining on its own). The highest TBM bid at R153 million came from the Concor/Marti/Hochtief JV.
In December 1989, the contract was awarded to the lowest bidder. A major saving reflected in the tender price from Murray & Roberts/Porr was its proposal to drive both tunnels from a single site set-up in the Emolweni Valley rather than from two sites as envisaged in the tender documents. It also based its tender on using one TBM to complete both minimum 3.4m diameter tunnels (a total of 10.6km) in the contract period of 46 months which requires an average TBM advance rate of 625m/month, excluding start up time. For Murray & Roberts, this is its first tunnelling project and although it brings some tunnelling experience to the job from its halfowned subsidiary RUC (which tendered in competition to its parent company), Porr of Austria is recognised as the partner with the technical tunnelling expertise. Porr has used TBMs most recently on the 6.7km x 4.75m diameter tunnel for the Alberschwende hydropower project in Austria from 1989 to 1991 and the 4.7km x 3.6m diameter tunnel for the Wald hydropower project in Austria from 1986 to 1989.
Establishment of the one central site includes a 250m-long, maximum 25m-high, temporary bridge across the valley to facilitate movement of materials from the stock yards to each portal as well as providing an effective location for the muck dumping tippler. The tippler, designed by ROWA of Switzerland, will empty two 4.75m3 muck skips simultaneously and empties a full 12-car train in five minutes.
In February 1990, the Rand9 million order for the 3.5m diameter Demag system TBM was placed with Voest AJpine in Austria. The machine weighs 150 tonne and has an installed power of 930kW. The cutterhead is dressed with 25 x 17in disc cutters and has a single speed rotation of 12.5 rev/min. Maximum thrust capacity is 6280kN and maximum cutterhead torque is 612kNm. To date, penetration rates of up to 50mm/min have been recorded through sections of the 155-250MPa sandstone of the Clermont drive. With a single pair of grippers, the TBM can negotiate curves of 150m radius while leaving sufficient space within the tight 3.5m diameter to accommodate supplementary equipment for probing and rock bolting close to the face.
Ten months after placing the order, the TBM was on site. Mobilization of the site, prior to launch of the TBM, included drill+blast portal establishment and short lengths of drill+blast from each of the four portals. At each portal, the tunnel will be lined with steel liners encased in 500mm thick insitu concrete to contain the maximum working surge pressures of up to 2.5 bar under shallow cover in the portal areas and under the minimal side cover in the deeply incised valleys. A plastic membrane installed behind insitu concrete lining will prevent water egress in areas of the TBM tunnel where the aqueduct water pressure is greater than the surrounding ground water pressure. The portal drill+blast sections vary from about 25m to 70m at the dam intake portal where the ground is very fractured with medium to widely spaced jointing.
Despite being potentially the most difficult, the JV decided to complete the 5.5km-Iong Clermont drive first. Now starting at the Emolweni Valley, the Murray & Roberts/Porr tender expected to drive a down hill TBM drive to the Aller Valley. However, to conform with good tunnelling practice, the client agreed to change the gradient to a slightly uphill drive. Being a pressure system with the water travelling at about 1m/sec through the maximum 25m head over 20km, such a change was acceptable.
After making a positive start in December 1990, several matters of concern soon became evident. First was the higher than expected abrasivity of the sandstone. Penetration required more revolutions of the cutterhead and wear and performance of the cutters was irregular. Cutter loading was on average about 20-22 tonne/disc with shock peaks two to three times higher. In addition, the TBM has suffered excessive wear of the rotating cutterhead body and on the edges of the muck buckets.
After completing 140m, a three to four week downtime was used to repair, modify and strengthen the body of the cutterhead with face welding and also to weld more plate to the side shovels of the buckets to improve their efficiency. Cutter changing was also consuming more than 50% of available boring time.
A further major concern was the fact that a cubby had to be blasted into the side of the tunnel to change the five gauge cutters to gain access to their securing bolts. Given the high cutter wear, cubbies were having to be blasted every 10m or so. To improve the situation, the mountings of the gauge cutters had to be modified which required a second TBM downtime of about four weeks at the end of June 1991. Specific cutter efficiency was improved by changing the angle of the gauge cutter attack from 45 to 65 degrees.
Dust created in the dry sandstone and by simultaneous shotcreting with the dry method also created problems. The main tunnel ventilation is a forced system with the three stage axial flow ventilation fans delivering up to 300m3/min through 800mm diameter flexible ventilation ducting. The mouth of the ducting is about 25m from the face. The TBM is also fitted with its own localized ventilation system which extracts air from in front of the heavy, tight fitting steel dust shield and passes it through a cyclonic de-duster before being exhausted about 25m behind the face. However, this had trouble coping with the two dust sources. On occasions, work had to stop due to the high silica content reading in the dust, both from the shotcrete and the sandstone. Such conditions contravene South Africa’s tough health and safety directives of the Mines & Works Act. Dust has also interfered with the efficiency of the laser beam of the guidance system. Tolerance on line and level in this positive flow pressurized tunnel is not as critical as it would be on a non pressurised gravity-fed tunnel. However a tolerance of +300mm on line and +50mm on level has been specified to avoid potential air traps in the crown or siltation in the invert. Line and level has generally been within these specifications except in one area where obstruction of the laser and a start-up inaccuracy after the modifications caused the TBM to dive by 800mm over about 25m. Overcompensation then resulted in a rise of +500mm. For a 200m length, the tunnel is out of alignment which will have to be corrected subsequently. Meanwhile, the deviation will continue to cause a major trap for out flowing ground water and silt.
Due to earlier problems and the recent slow progress through areas of high water ingress, the contract was running about seven months behind the original programme to early December 1991. Since then the contract has hit other time delaying problems. Nagging problems with the electric motors since the first encounter with high water ingress required that all four main motors and the spare be sent back to the Siemens workshops in Germany for modification and repair. IT was said that the specialized technical expertise required to maintain a civil engineering TBM operation is lacking in South Africa.
Removal of the motors in mid-December 1991 presented the opportunity to investigate a regular loss of oil pressure in the sealing system of the main bearing. This revealed that waterborne silt contamination in the flushing system of the cutterhead had caused a partial failure of the seal which had to be replaced. This involved detaching the cutterhead and backing up the body of the machine to replace the 2-tonne steel component. Total downtime of at least seven weeks (including the customary three week shutdown of the construction industry over the Christmas period in South Africa) was envisaged. At the time of writing, the TBM was expected to resume boring by 27 January 1992 with completion of the first drive anticipated for June/July 1992.
Despite these delays, the JV continued its endeavours to complete the contract by the scheduled October 1993 date, plus extensions of time presently under negotiation. With completion of the current Clermont Tunnel, the worst will be over. The 5.5km-Iong Emolweni tunnel drive is not expected to be as difficult as the Clermont drive. Water ingress is not expected to present as great a problem and the granite and granitic gneiss is judged to be less abrasive, more brittle and therefore more favourable for TBM excavation.
In addition, steps are being taken to make up time and improve progress. The JV is considering increasing total work time by 25%, utilizing the full seven-day week and a more suitable shift system. It will also take delivery of a new more robust cutterhead which will have one additional disc cutter and a new cutter configuration. These design modifications are expected to improve cutter performance and increase penetration and progress rates.
In the meantime, all involved strive together to complete the Clermont drive under very taxing circumstances. This contract is proving a test case for future application of geological information, design specifications and contractual documentation to TBM tunnelling.
Thanks are extended to Mike Trissler and Bruce Cameron of Umgeni Water; Heinz Haesloop, James McKelvey, Ed Schultz, Scott Taylor and Tina Helin of Keeve Steyn; Brian Bruce and Philip Stewart of Murray & Roberts; and Sepp Sengstschmid oj Porr International who, when interviewed, provided informationfjor the compilation of this report.