The Kishon River Drainage Authority (KRDA) has had the responsibility of dealing with the very complex flooding problems experienced in Israel’s Haifa Bay, since 1991. Yodfat Engineers Ltd were employed by KRDA as consultants and planners with full responsibility for producing a comprehensive solution to this serious and widespread flooding problem.
Extensive studies of the catchment area’s rainfall hydrographical studies and visual reporting identified the junction of the Gedora Canal with the Kishon River as one of the major causes of the flooding. The junction of the two flows occurred at a point to the south and upstream of the Histadrut Road Bridge that crossed the river in a single span. The narrowing of the banks at the bridge abutments meant that the two flows were unable to pass under the bridge in times of heavy flows without backing up and causing extensive flooding.
The project to relieve this particular point of flooding was called the ‘Gedora Canal Diversion Project’. The project was in fact only a small part of the ‘Gedora Canal Development Programme’ (1995), which in turn was part of the larger ‘Haifa Bay Flood Prevention Programme’, identified in April 1992.
In detail the ‘Diversion Project’ aimed to interrupt the Gedora Canal at a point to the north west of the existing Haifa Oil Refinery and to take the canal waters under the Histadrut Road in a new concrete culvert. The canal was then to be put in a new open channel to a point downstream where the energy balance of the currents in the Kishon River and the Gedora Canal would enable the water from the canal to enter the river, in a manner that would prevent flooding to industrial and residential areas nearby.
There were a number of proposed schemes for providing the new concrete canal culvert under the road, including diverting the road and using cut and cover, a box-slide, and box-jacking. The volume of both light and heavy traffic, using the Histadrut Road throughout the day and night, was a major influence in the local Highway Authority deciding that there should be no disruption to traffic on the road. This persuaded the KRDA to opt for a box-jack with no road diversion.
Yodfat Engineers were retained by KRDA as planning engineers for the scheme but brought in technical design experience for the tunnelling works through UK based consultant Charles Haswell and Partners.
A preliminary design was established for the culvert, which was to be 10m wide, 3m high and 56m long, and meet the requirements of 45m³/sec flow. Cover to the dual carriageway was to be just 4m. Services in the road included water, electric, telecoms and a live fuel main. A ground investigation scheme was undertaken, which included boreholes on either side of the road and trial pits. No boreholes were allowed in the centre of the road due to the possibility of disruption to the traffic.
The ground conditions were dry red-brown fat clay to 3m, then wet grey clay to 5m, divided by a very wet band of shells (kurkar) 0.2m-0.4m thick. Beneath the grey clay was a dense wet sand layer to 12m, under which was a fat clay layer to 14.5m and finally a limestone bedrock aquifer (Figure 1). From this it was clear that the face would need stabilising, particularly the wet shelly layer and the sand below, if the box jack was to be excavated without consequential settlement damage to the road. Dewatering was not allowed, due to the risk of consolidation settlement damage to the road from ground water draw down.
The contract
Armed with the outline design, the ground conditions and special contract documents, an international tender was sent out to all countries that were known to have the requisite expertise. Following much prevarication, tenders were received back, but all were non-compliant because of the dewatering clause. The lowest price received was from a German consortium, already working in Israel, but it transpired the consortium wanted to use its existing equipment and pipe-jack the crossing. This would have meant up to seven 2.1m diameter pipes under the road to take the flow. The risk of multiple settlements was too great a risk, and was again rejected as being non-compliant. Eventually after much dialogue a consortium of Swiss Engineers VSL, Ram Oceanic Engineering Ltd and Cementcal Ground and Structural Engineering submitted a late compliant bid, using grouting as its ground stabilisation method.
Once an agreement had been reached between the parties, VSL provided a design for the culvert and the ‘Autofoncage’ jacking system through Jean-Marie Beauthier (JMB Methods). In their design VSL confirmed that the culvert was to be split into two units each internally 5m wide. This was to reduce the jacking loads, as the poor ground conditions provided insufficient reaction to push a single box weighing some 3,500t. With each unit now weighing only 1,840t, the boxes would be slid forward one at a time without the risk of jacking wall failure.
The box jacking technique
The system to be used involved constructing the individual concrete box units on a very flat jacking slab and then pushing them through the ground, under the road, using heavy duty jacks acting on multi-strand cables anchored to the front of the jacking slab. As each unit moved forward the soil was removed from within the box. Each box was equipped with an angled cutting edge, set roughly to the angle of repose of the soil at the leading end, and bentonite injection holes in the roof and side-walls. Bentonite slurry was to be injected around the boxes to reduce the frictional forces between the boxes and the ground. The spread of bentonite was aided by the use of ‘Omega’ shaped channels set in to the outside of the walls and roof, which were connected directly to the inside gout holes. As an additional feature the side-walls were lined on the outside with 20mm thick HDPE sheets to reduce the side wall friction. The accuracy at which the boxes could be pushed forward depended on the flatness and alignment of the jacking slabs and these were to be set very accurately to 0.1mm
In order to construct the boxes a major cofferdam had to be erected to construct the jacking slab and the concrete boxes. This 80m long by 20m wide piled excavation was 7m deep with no internal bracing. Dewatering wells were used to assist in excavation and the walls were supported by continuous double channel wailings held by ground anchors at 2.5m centres. These anchors had a working load of 58t but were tested up to a maximum 70t. To prevent the cofferdam from filling up during construction of the jacking slab, boxes or during the actual jacking operation, a system of normal injection dewatering was installed in the bottom of the cofferdam to keep the water down below the point where it might affect the works. This level of dewatering was permitted because it was well away from the road area, but carefully placed piezometers in the ground around the cofferdam confirmed that the drawdown was only very local to the area of dewatering and was not widespread enough to affect the road or services.
The author’s involvement was to assist in checking the parameters for the cofferdam design. The particular risk was that the cofferdam base would not be stable and that the underlying artesian water pressure from below would ‘blow’ the base making working conditions impossible. Working closely with Professor Sam Freidman from the Technion University, we arrived at the design parameters for the cofferdam and identified the pile cut-off length to be 17m. We also confirmed the pile sizes and the expected wailing loads provided by the contractor. While the jacking slab design was completed by JMB it was my job to work with Haswell to check the design assumptions and to come up with our own jacking forces and loads, taking into account the expected frictional forces and the actual site conditions. We then compared these with the submitted design and, with certain modifications such as deeper ‘downstands’ on the slab and a higher backwall, confirmed the contractor’s design.
The proposed construction period was to be over the summer months of 2000 with completion in October for the start of the flooding season. Following the protracted negotiations work stated late, in mid summer, with the construction of the cofferdam.
Sinking the cofferdam went well but after seeing the ground conditions the contractor decided against grouting the tunnel as a solution to the stability. It wasn’t certain that the cement grout could be injected the full length of the tunnel. Working room was restricted at the far end of the tunnel. The sand was very compact and might need expensive chemical grouting rather the cheaper cement grouts. Micro-fine cement was suggested, but the contractor felt that it was not possible to fully grout the ground and there would be ‘holes’ which would not be grouted, posing a risk to the security of the road. The contractor again requested dewatering as an alternative. This was refused.
Soil improvement and jacking
A different method had to be found quickly and it was agreed that freezing was the only other alternative. The risk of heave was going to be high and I was involved with Haswell and Alan Auld Associates, consultants to British Drilling and Freezing (BDF), in predicting what the maximum heave would be over the road. It should be said that the predicted maximum heave was exceeded by a large margin in the final analysis.
The design for the freezing was accepted very quickly and BDF organised themselves to ship equipment to provide a -10ºC sub-surface temperature, in climate conditions reaching temperatures of +35ºC during the day. BDF planned to use three refrigeration plants each with a capacity of 175Kcal/hr (Capable of refrigerating a 20 storey office block). These machines operate by using ammonia gas in a closed circuit system, which refrigerates brine to a temperature of -38ºC. This brine is circulated through the horizontal freeze tubes and overlapping circles of ice then form around the tubes. The effectiveness of the freeze depends on the accuracy of the drill tubes and their relationship to the other tubes in the group. Variations in tube spacings can cause holes in the ice that can be soft and result in collapses.
While the equipment was on its way, the contractor started to drill the freeze tubes underneath the full length of the road. Drilling had to be very accurate and the holes were generally very straight and to the DBF specification, however drilling the long holes with water flush caused settlement of the carriageway and resurfacing had to take place to maintain the road surface.
The freezing medium was to be brine, as mentioned above, as liquid nitrogen was available but not in the quantities required to maintain the freeze. Over 75 horizontal tubes were drilled together with seven monitoring holes, the freeze started in mid July. By the end of September (8-10 weeks) the core was around -10ºC, but the outer holes were still lagging behind and some concern was expressed about the bottom right hand corner of the first box where temperatures were still positive. This corner proved to be problematical and some ground was lost due to unfrozen ground, possibly due to an off line tube. The steel cutting edge also had an over cut of about 35mm which, if not carefully trimmed, meant the box was actually splitting the ice and, where the ice surround was thin, resulted in water entering the face and melting the ice. Some localised dewatering was installed to counteract this problem.
There were also problems in that the ground was too hard for the equipment the contractor had originally provided for the grouted ground. The rotary cutter boom that could cut grouted ground up to 25MPa was finding it difficult to cut ground that was above 80MPa. Hand tools were therefore used to do much of the work. This slowed the work and the expected rate of 3m to 4m per day was now little more than 1m per day. This meant the overall programme for the boxes went from 4 weeks to 12 weeks and the final length of the second box into the exit cofferdam happened well after the rainy season had started and the final connection into the canal had to be delayed for a period of six months.
The box was finally completed in summer 2001 with the open-cut section finally connected into the Kishon River to complete the scheme.
The results of the freeze were that the road having fallen by around 50mm during the freeze tube drilling rose by up to 165mm during the freezing operations. This was almost twice as much as the predicted heave. The reasons were that a more extensive and prolonged freeze was required due to the problems with the thinness of the ice wall and with the extra length of time taken to excavate the very strong frozen layers.
The road surface was monitored using standard levelling techniques, although this was considered very dangerous because of the high levels of traffic using the roads at all times both day and night. Sophisticated total station surveying equipment was set up to real-time monitor the surface of the road, but continuous traffic vibrations made the instrument susceptible to error and this system was discontinued soon after the jacking started.
The contract was completed to everyone’s satisfaction and flooding in this particular area has not been recorded since completion of the works.
Related Files
Fig 1 – Cross section through the Cofferdam and the dewatering arrangement
Fig 2 – Cross section through the tunnel face