De-bonding of FRS Linings. E.S. Bernard and S.G. Reid

Transcription

1 De-bonding of FRS Linings E.S. Bernard and S.G. Reid

2 Characteristics of FRS Linings in Sydney Most FRS linings are arch-shaped but may include regions of flat-roof. The ground surface is generally smooth before spraying The bolts are installed before the lining with spider plates forming a shear connection (at best) Bolt spacing lies in the range mm The lining is often applied well after excavation, so convergence is negligible and no compressive stress is likely in the lining Bond between lining and ground occasionally measured, seldom better than 1.5 MPa soon after construction. Long term bond strength is unclear.

3 Analysis of Failure Modes for FRS Linings Conventional analysis of loads on Fibre Reinforced Shotcrete linings only considers external loads Experience in Sydney has indicated that environmental loading can also be severe, due to: 1. Drying shrinkage 2. De-bonding from ground 3. Creep 4. Corrosion of steel fibres and loss of continuity at cracks An RMS-funded study has involved an analytical assessment of cracking and delamination of FRS linings and its effect on stability and load resistance.

4 Stability Analysis of FRS Lining Lining Geometry For the present analysis, lining is assumed to form an arch and be of uniform thickness. For the purpose of instability analysis, only a region comprising a single span between bolts is considered.

5 Design Capacity of Linings Lining Mechanics Properties of Sydney Linings: Linings are slender and intended to provide local support between bolts Some tunnels include very shallow curvature in the roof, and linings are mm thick. Drying shrinkage is high, about 1200 microstrain at 90 days (unrestrained), even higher in the long term Bond strength to ground about MPa in short term but unknown in longer term Evidence from older tunnels indicates widespread delamination and cracking For the purpose of analysis, geometric imperfections have been ignored. Water percolation is a common problem, cracks are clearly wide enough to let water through so they are probably through-cracks, not flexural cracks.

6 Design Capacity of Linings Cracking and Moment Resistance Conventional models of FRS lining capacity ignore in-plane compression and the effect this has on moment resistance and ductility in FRS. Moment capacity is taken to be solely attributable to fibres. Shrinkage and creep can reduce the advantage provided by in-plane action by dissipating the magnitude of in-plane compression. Conventional lining model Real lining moment capacity Conventional yield line or elastic model assumes moment capacity at hinges is fixed and independent of axial compression In reality, formation of cracks leads to an increase in axial compression which greatly increases moment resistance through Voussoir action A 1 mm wide crack will generate a mean strain of 0.001/m, which equates to 30 MPa compression

7 Design Capacity of Linings Moment Resistance

8 Design Capacity of Linings Moment Resistance An in-plane compressive stress field will increase the ductility of a lining, so it can conservatively be assumed that a strain-softening lining will act as an elastic-perfectly plastic hinge. A strain-softening FRS will become a deflection-hardening lining in rotation under compressive loading

9 Effects of In-plane Restraint Tests on mm lining with 1.2x1.2 m bolts showed very high flexural ductility compared to unrestrained panels of the same thickness and composition

10 75 mm thick macro-synthetic FRS panel mm specimens Load (N) Deflection (mm) ASTM C-1550 round panels were tested as comparison specimens to determine typical toughness and post-cracking load carrying capacity for 9 kg of BC48.

11 75 mm thick macro-synthetic FRS lining In-plane restraint is critical to achieving high lining ductility when delamination occurs mm specimens Load (kn) Deflection (mm) Despite using a strain-softening FRS, the capacity of the restrained panel was strainhardening up to central deflections of about 50 mm and 9 tonnes with BC48 fibres.

12 Unrestrained Drying Shrinkage The unrestrained drying shrinkage of shotcrete is very high compared to normal cast concrete and is increased by the addition of set accelerator. Results using 40 MPa Sydney sprayed shotcrete mixture indicates results that are very similar to data from Goodier et al (2009) indicating long-term shrinkage ~1400 microstrain.

13 Long-term Performance Cracking and De-bonding Drying shrinkage and other factors will generate cracks in a lining that will steadily widen with the passage of time. Analysis of bridge toppings show that these cracks can lead to delamination. Two scenarios exist as starting points: Cracking in the Absence of de-bonding De-bonding in the absence of cracking Measurements of linings and test panels have indicated that cracks are frequent and steadily increase in width with age Field assessments of linings in Sydney have indicated large areas of de-bonded shotcrete may exist in many tunnels

14 Long-term Performance De-bonding Before Cracking For an uncracked curved lining subject to a net tensile stress (because shrinkage is greater than compression due to possible convergence), the tension in the lining will induce tension at the ground/lining boundary. In this analysis, it is assumed that tensile creep is insufficient to relieve the magnitude of tension in the lining, thus the tension at the boundary will be sustained for a substantial period of time. Tension may cause areas of poor bond to fail leading to a spreading region of de-bonded lining. This is presently a hypothetical failure mode.

15 Long-term Performance Cracking and De-bonding Very little analysis of delamination has been undertaken for FRS linings, but a large amount of analysis and research has been done for whitetop road pavements and bridge re-surfacing. This work indicates that (at least) partial delamination is inevitable. Gravity makes things worse! Cracking in a partially de-bonded lining Cracking in a substantially de-bonded lining Analysis indicates peeling stresses at boundaries of the crack are very high Shrinkage and creep of de-bonded lining may lead to very wide cracks and possible snap-through of suspended lining for shallow curvature linings

16 Long-term Performance De-bonding and Instability If the radius of the lining is very large, and it is slender, there is a risk that shrinkage and creep can lead to unstable behaviour in which the lining snaps-through and collapses (if reinforcement is insufficient). This type of behaviour is more likely in flat-roof excavations and if gravity loading is imposed in the form of loose ground. Slippage at the lining-ground boundary adjacent to the support will greatly increase the likelihood of unstable snap-through behaviour. Only high-tenacity reinforcement that can support load over wide cracks can then prevent the lining from collapsing.

17 Long-term Performance De-bonding Peeling stress at lining-ground boundary is greatly increased by the onset of cracking. Differential shrinkage thermal cycling, and curling of the lining will lead to intense peeling stresses and enlargement of delaminated zone Analysis of de-lamination of bridge re-surfacing in RILEM Report 193 indicates that magnitude of tension generated at peeling front is MPa, which far exceeds the capacity of any cementitious lining. Growth of the delaminated area is inevitable.

18 Long-term Performance Formation of a Mechanism As the size of the de-bonded area grows, the moment at the intact boundary increases and it may crack to form a potential mechanism. Potential secondary crack As the delaminated zone increases in size, the moment due to self-weight induced in the lining increases leading to cracking at the boundaries.

19 Long-term Performance Shrinkage and Creep As the size of the de-bonded area grows, the magnitude of the out-ofplane moment and in-plane compression forces increase. This leads to compressive and flexural creep of the de-bonded lining, which combined with drying shrinkage, will reduce its length. Shortening of the components of the mechanism will cause the central hinge point to fall in elevation which will further increase the in-plane compression and creep effects. This vicious cycle might continue until snap-through of the mechanism, but only for shallow arches or flat and slender linings.

20 Long-term Stability Shrinkage and Creep Tests by Goodier et al (2009) and Bernard (2017) indicate that unrestrained drying shrinkage of shotcrete at 180 days is at least 1200 microstrain. Tests by Bernard (2017) indicate that specific creep strain at 90 days for shotcrete is about microstrain/mpa. For a lining under 10 MPa mean compressive stress, the sum of elastic and creep strains will therefore be at least 1000 microstrain For a 1000 mm long delaminated segment, unrestrained shrinkage will amount to about 1.2 mm, and both elastic and creep strains will amount to 1.0 mm shortening, or about 2.2 mm in total over only 90 days! This model ignores local crushing at hinges which will further increase apparent shortening

21 Long-term Stability Lateral Restraint For a delaminated arch section to fall out of a lining requires a concurrent loss of restraint in the lateral direction. Since bolts are also spaced about mm apart in the longitudinal direction of the tunnel, there would need to be either a very long region of delamination, or large shrinkage cracks that release a delaminated section from restraint. Cracking can lead to delamination in any direction, and there is no compression of the lining in the longitudinal direction, so there exists the possibility that shrinkage cracking can lead to substantial cracks across the span of a tunnel.

22 Long-term Stability Restraint from Reinforcement Testing and long-term monitoring of bridge over-lays has indicated that fibre reinforcement can hold cracks closed and reduce delamination, but only if corrosion is prevented Corrosion testing of SFRS by Nordstrom (2017) in Sweden and SFRC by Bernard (2015) in Sydney has indicated 50-60% loss of section capacity in only a few years of environmental exposure for cracks >0.15 mm Inspection of cracks in Lane Cove Tunnel has indicated complete loss of steel fibre section capacity after 8 years for cracks with percolating water Ground water in Sydney is simply too corrosive for steel fibres to survive if moisture penetrates the crack Lane Cove Tunnel lining collapse in 2015 shows this problem is not a hypothetical problem but a real problem

23 Cracks and Corrosion Sydney Exposure Corrosion and embrittlement occur concurrently in steel FRC but generally not in macro-synthetic FRC. Embrittlement is the more serious and insidious process since even well protected SFRC will suffer this problem. Macro-synthetic FRC offers better performance with aging due to absence of both embrittlement and corrosion.

24 Results of Analysis De-bonding Take R =14850 mm, t = 75 mm, shrinkage = microstrain (increasing with age), and creep varies with age. Estimate the tensile stress at ground/lining boundary before and after cracking. Parameter 90 days 1 year 5 year Circumferential Tension, Max tensile stress σi (MPa) Radial Stress at lining-rock Boundary (MPa) Immediately after cracking, the stresses at the lining-rock interface are as follows: Stress 90 days 1 year 5 year Normal peeling stress at boundary (MPa) Interface shear stress (MPa) Principal tensile stress (MPa)

25 Results of Analysis De-bonded Region Consider effect of gravity loading and shrinkage on capacity of a large de-bonded region to support itself. Cracking is very likely! Segment with span = 0.14 rad (2000 mm) 90 days 1 year 5 year M/Mcr at crown M/Mcr at supports Tensile stress/strength at support interface with rock Segment with span = 0.18 rad 90 days 1 year 5 year M/Mcr at crown M/Mcr at supports Tensile stress/strength at support interface with rock Segment with span = 0.22 rad 90 days 1 year 5 year M/Mcr at crown M/Mcr at supports Tensile stress/strength at support interface with rock

26 Results of Analysis Debonded and Cracked Consider a cracked and de-bonded region of lining. Cracked lining may form a stable arch, but corrosion can reduce residual capacity. Segment with span = 0.10 rad (1480 mm) 90 days 1 year 5 year M/Mcr at supports Tensile stress/strength at support interface with rock Segment with span = 0.15 rad 90 days 1 year 5 year M/Mcr at supports Tensile stress/strength at support interface with rock Segment with span = 0.18 rad 90 days 1 year 5 year M/Mcr at supports Tensile stress/strength at support interface with rock

27 Results of Analysis Crack Widths Consider a cracked and de-bonded region of lining of mm radius and 125 mm thickness. Crack widths can become very large due to sagging leading to corrosion of fibres if water is present. Segment with = 0.15 rad and 125 mm thickness 90 days 1 year 5 year M/M cr at supports Tension/adhesion at support interface with rock Crack width at centre (mm) Crown displacement (mm) These results are supported by field measurements made by Bryne et al (2010) in Sweden showing that cracks of over a millimetre are possible in delaminated areas and shotcrete over drains, mainly due to shrinkage.

28 Results of Analysis Debonded and Cracked Consider sensitivity to magnitude of drying shrinkage, R = mm, span=1850 mm and thickness = 125 mm. Stress ratio under Self-weight only εcs, b 90 days 1 year 5 year M/Mcr at ends Stress ratio under Self-weight only εcs, b 90 days 1 year 5 year Tensile stress/strength at support Crack widths under Self-weight only εcs, b 90 days 1 year 5 year Crack widths (mm)

29 Conclusions High shrinkage appears to make cracking and debonding inevitable for slender shotcrete linings Field inspections of aged FRS linings in Sydney and elsewhere have confirmed that cracks can be very wide over debonded areas leading to rapid corrosion of steel reinforcement when percolating water is present Instability of a debonded lining is more likely for slender linings associated with flat roofs or large radius arches.