Scour countermeasures at bridge piers and abutments

Transcription

1 Scour countermeasures at bridge piers and abutments A. H. Cardoso 1 Contents of the presentation Introduction Main features of the flow field at bridge piers and abutments Countermeasures for local scour at bridge piers Introduction Design of armouring countermeasures Notes on flow altering devices Experimental study on the effectiveness of slots, bedsills and combinations of slot plus bedsill Riprap mattresses as a countermeasure against scour at bridge abutments. 2

2 Introduction Is scouring a real problem? in the USA: 383 bridges have been destroyed or damaged between 1964 and 1972; 73 bridges were destroyed in Pennsylvania; Virginia; West Virginia in 1985; 17 bridges were destroyed in NY and in N Eng. in 1987; in Portugal: Penacova; Alva; Gafanha; EntreosRios, 21 (56 casualties). 3 4

3 What are the open issues? Available methods (scour depth prediction; design of protection solutions) are often not satisfactory; this is particularly true for abutments. Why? The flow field is highly 3dimensional at piers and abutments; The sediment transport phenomenon is complex in the scour hole. Other aspects? Local phenomena may be superimposed to riverbed degradation as well as to contraction scouring (due to the increase of U). 5 Main features of flow field around bridge piers and abutments The presence of obstacles implies the flow stagnation close to the walls pressure increase (kinetic energy potential energy). bow wave y y h p = (u 2 )/2 Descending flow u h Pressure increases are bigger at the free surface than close to the bottom. 6

4 The local change of the pressure field originates: bow wave; descending flow (which triggers the scouring process); flow separation. bow wave descending flow main vortex stagnation point 7 The combined action of the deflected descending flow and the separated flow creates: the horseshoe vortex (in the case of piers); the main vortex (in the case of abutments). Separation also occurs at the lateral walls of obstacles, inducing wake vortices (rotating at successively alternate senses). For bridge piers: bow wave pier wake vortices approaching flow horseshoe vortex scour hole descending flow 8

5 The flow structure around abutments is very similar to the flow structure around piers. main vortex secondary vortex The horseshoe vortex (or the main vortex) carries the bed material downstream as bed load. Wake vortices pick up sand particles from the bottom; they transport the particles downstream in suspension. 9 Countermeasures for local scour at bridge piers Introduction Types of countermeasures: Armouring countermeasures: act as barriers withstanding the elevated shear stresses that occur around bridge piers or abutments; riprap mattresses; gabions and Reno mattresses; artificial riprap; cabletied blocks; concretefilled bags and mats. Flow altering devices (for piers): act to reduce the strength of the main features of the flow field around piers (horseshow vortex; downflow; wake vortices). sacrificial piles; collars; flow deflecting vanes; permeable sheet piles; suction applied to pier; slot in pier. 1

6 In practice, two different sediment transport conditions may be observed clearwater flow ( < c or U < U c ); livebed flow ( > c or U > U c ). Under clearwater flow, failure mechanisms of armouring countermeasures are: shear failure; winnowing failure; edge failure. Under livebed, failures mechanisms are as under clearwater plus: bedform undermining; degradation failure. 11 Occurrence domains of failure mechanisms (excluding degradation). u * /u *c sand Riprap blocks of any kind.35 for piers; variable for abutments u * /u *c blocks 12

7 shear failure armour blocks do not withstand the local hydrodynamic forces and are entrained by the flow. remediation sufficiently heavy blocks. winnowing failure the finer underlying bed material is eroded through the voids between the blocks under the action of turbulence and seepage flows. remediation sufficiently thick mattress; underlying filter. 13 edge failure armour blocks fall into the scour hole that develops in the periphery of the armour mattress. remediation sufficiently wide mattresses. 14

8 Bedform undermining armour blocks are undermined and settle with the migration past the pier/abutment of the trough of large dunes. remediation place the mattress sufficiently below the original sand bed. 15 Degradation failure armour blocks are undermined and settle due to the general erosion of the river bed. remediation construction of bed sills or check dams (one or more) immediately downstream of the bridge site. 16

9 To be reminded (above all): 1. Completely avoid scouring is practically impossible; 2. Armouring countermeasures are to be considered only temporarily effective; a. At the edge of mattresses, scour holes typically develop, in any case: scour holes are attenuated and displaced from the pier/abutment; b. Maintenance and monitoring are recommended; Remote monitoring is an issue. c. Preferably, new bridges should be conceived to withstand scouring. 3. Flow altering devices may be useful to reduce scour at bridge piers Research is required in this front. 17 Design of armouring countermeasures Main issues Questions to be answered (assuming that degradation scour does not occur or is mitigated): 1. How big should blocks be to face shear failure?; 2. How thick should the mattresses be to face winnowing?; 3. How wide should mattresses be to face edge failure?; 4. How deep should mattresses be placed to face bedform undermining? 18

10 Riprap mattresses How big should blocks be to face shear failure? There are many formulae available in the literature: Isbash 1938 Inglish 1942 Blench 1957 SousaPinto 1959 Maza and Sanchez 1964 Nicollet & Ramette 1971 Neil 1973 Bonasoundas 1973 Quazi & Peterson 1973 Posey 1974 Hjorth 1975 Breusers et al Farraday & Charlton 1983 Worman 1989 Parola & Jones 1989 Breusers & Raudkivi 1991 Parola 1993, 1995 Austroads 1994 Richardson & Davis 1995 Chiew 1995 Fotherby 1995 Croad 1997 Lagasse et al Lauchlan 1999 Fotherby & Ruff 1999 Choi et al I have used the equation of Parola & Jones 1989, as recommended by Parker et al. 1998: D r5 2 2 U K f 2,89 s 1g U approach flow velocity; K f pier shape factor; s density of the blocks. I also use those of Bonasoundas 1973, Quazi & Peterson 1973 and Breusers & Raudikiwi 1991, which lead to central predictions, for U 5 ms 1. 2,5 D 5,85F 2 r r D r 5 6 3,3U 4U h s 1 1, 25 D r5 h,278f s 3 rc F rc critical approach Froude number, defined with U c = 2U; h approach flow depth. 2

11 Melville & Coleman 2 suggest the use of the equation of Parola and Lauchlan Assuming that D r5 is known, the riprap gradation curve can be given by (Parker et al. 1998):» 1% finer than 1,5 D r5» 8% finer than 1,25 D r5» 5% finer than 1, D r5» 2% finer than,6 D r5 How thick should the mattresses be to face winnowing? For piers, settling can be observed for mattresses of up to t = 12D r5!!!!!!!!!!! thick (Nanyang Technological University). 2D r5 t 3D r5, placed on geotechnical filter (Terzaghi criteria; usually difficult to built) or on geotextile filter (properly designed and dully attached to the pier). 21 How wide should the mattresses be to face edge failure? According to Parker et al. 1998: Flow D h h2 1 filter riprap B 1 = 4D/cos; B 2 = 3D/cos; h 1 = 1.5D/cos; h 2 = D/cos. B 2 B 1 For cylindrical piers, adapt accordingly (3D B 1 4D; B 2 =.75B 1 ). 22

12 How deep should mattresses be placed to face bedform undermining? only the study of Lauchlan 1999 could be found in the literature. D h Y 2.75 r5 r 1.2.3S f 1 Fr h Sf safety factor (minimum of 1.1). h approach flow velocity 23 Gabions and Reno mattresses Preliminaries: Gabions and Reno mattresses should only be used in sandy bottom rivers with small bed load discharge (to reduce abrasion effects); They allow the reduction of block sizes (compared to riprap); There is not enough experience on the durability of this solution. How big should blocks be to face shear failure? How thick should the mattresses be? Blocks should be able to withstand a velocity 2U U cr 3.5U (after Hancu 1971; after Ciew 1995); U approach flow velocity; U cr design velocity. 24

13 Gabions and Reno mattresses without bitumen (Agostini et al.) Gabions and Reno mattresses with bitumen (Agostini et al.) 25 How wide should mattresses be to face edge failure? How deep should the mattresses be placed to face bedform undermining? Use the same criteria as for riprap; Place the mattresses on adequate filter. Artificial riprap What is this? + 26

14 Apart from in Japan, there are very few examples of the use of artificial riprap as a bridge pier countermeasure. Artificial riprap is an alternative to riprap where it does not exist with proper dimensions or it is very costly. Design criteria are the same as for riprap with the exception of block sizing (in the case of toskanes and dollos). Ruff and Fotherby 1995 suggested the equation:.5 D.255 D e Fr p U t Fr h s 1 h gh where: D e = equivalent spherical diameter of toskane/dollo; D p = projected pier width; U t = 1.5C l C s C h U (C l = coefficient of pier location; C s = coefficient of pier shape; C h = coefficient for the level of the top surface of the toskane layer). 27 Cabletied blocks System typically consisting of concrete blocks or slabs interconnected with steel cables. The blocks may be unstable by themselves but the mat is capable of withstanding large forces. The solution has already been used in the USA. Weight of blocks per square meter: s W 2 U s 1 Height of blocks: h b W g 1 s 2 void fraction in the mat 28

15 Concretefilled bags and mats Concrete or groutfilled bags are sacks that are filled with concrete and stacked to form an armour layer. Typically, the mat is strengthened with steel cables. The solution applies in sandy rivers only and in absolutely exceptional circumstances, due to the lack of angularity of blocks. Each pillow should be D 5 = 1.2D r5 (Parker et al. 1998). For the rest (thickness, area to be covered, depth of installation, filter), the recommendations for riprap will probably apply. 29 Notes on flow altering devices Sacrificial piles Piles placed upstream of the bridge pier for the purpose of protecting it from scour, by deflecting the highvelocity flow and creating a wake region behind them. Effectiveness of sacrificial piles depends on number of piles; their protrusion from the bed, geometrical arrangement, approach flow angle,, flow intensity, U/U c. The results presented hereafter were obtained by Hadfield

16 The best configurations (see below), extending up to 4% of the flow depth, reduce scour up to the percentages indicated in the table (in parenthesis for rectangular pier). U/U c Few field applications of sacrificial piles are know; Sacrificial piles are recommended only when flow remains aligned and the flow intensity (sediment transport rate) is relatively small. 31 Collars Collars are thin horizontal plates attached to the piers assumed to shield the sediment bed from downflow and horseshoe vortex. Effectiveness of collars depends on collar dimension (diameter); position of the collar relative to the bed For cylindrical piers, D collar = 2D p, under clear water, the collar effectiveness is 5% when the collar is.2h below the bed. It reduces to 2% when it is flush with the bed (Chiew 1992). For cylindrical piers, D collar = 3D p, under clear water, the effectiveness increases to 5% when it is flush with the bed (May et al. 22). Collars have only been tested for clearwater; they should not be considered for use under livebed. 32

17 Flow deflecting vanes These vanes are similar to Iowa vanes. They form arrays of vertical plates installed upstream of the pier. The most efficient, seem to interact with the sediment bed rather than with the approach flow. According to studies carried out at the University of Auckland, their best configuration is as follows: Effectiveness can be as much as 5% for relative submergence of 5/6 (protrusion of h /6), plate lengths of 1.5D and alignment angle of 3º, under livebed conditions. Promising countermeasure; deserves further investigation. 33 Permeable sheet piles Permeable sheet piles are based on the principle of permeable dikes in rivers and snow fences, supposedly inducing deposition. The best configuration seems to be as follows: Two 3D long panels forming an arrow pointing upstream; placed at an angle of 9º; apex at a distance of 4D; Panels made of 3D x.25d slabs, covering 5% of the area; Top of the panels protruding 1D above the bed and forming a 1º upward angle downstream; two triangular panels on the first and third slabs extending 3D downstream. Effectiveness 45% for cylindrical piers; 3% for rectangular piers, under live bed conditions. (Parker et al. 1998). 34

18 Suction applied to the pier The technique involves the removal of fluid from the surface of the pier by internal suction. The only known study (Rooney & Machemehl 1997) claims that it is possible to eliminate scouring. This technique seems quite unrealistic due to the need of driving the pump at field installations and the potential clogging of the holes. 35 Experimental study on the effectiveness of slots, bedsills and combinations of slot plus bedsill Carmelo GRIMALDI, 25 36

19 Literature review On slots Effectiveness depends on: slot length, l s ; slot width, w s ; sinking depth, z s ; skew angle,. For z s, Tanaka and Yano (1967) + Chiew (1992) suggest slot effectiveness of 15% 3%, depending on z s, l s and w s. Performance increases as z s. According to Kumar et al. (1999), the best effectiveness ( 3%) is achieved when l s > h, for z s. For l s h, Heidarpour (22) has shown that the lower effectiveness is achieved when the slot is placed near the water surface. No field applications are known. On bedsills Bedsills are regularly used to mitigate bed degradation mechanism). (5th failure 37 Experimental study Reported experiments 15 tests on the effectiveness of isolated slots. l s = h ; w s =.2D p. Variables: sinking depth, z s ; pier diameter, D p. 15 tests on the effectiveness of bedsills. Variables: distance between the pier and the sill, L s ; pier diameter, D p. 3 tests on the effectiveness of combined bedsill + slot. 38

20 Experimental Setup Tests were carried out at UBI and LNEC Flume dimensions: Length = 12.7 m Width =.8 m Depth =.7 m 1 recess box: Length = 2.5 m Width =.8 m Depth =.35 m Flume dimensions: Length = 4.7 m Width = 2 m Depth = 1. m 2 recess boxes: Length = 5 m Width = 2 m Depth =.35 m 39 Design of the tests: UD p / 7 (Franzetti et al., 1994); h /D p 2 (Laursen and Toch, 1956; Breusers et al., 1977) B/D p 1 (Laursen and Toch, 1956) D p /D 5 5 (Ettema, 198; Chiew, 1984; Breusers and Raudkivi, 1991) Cylindrical piers with rectangular slot, aligned with the flow direction; Uniform flow at the condition of beginning of sediment motion (U U c ). Two sands: Material Sand #1 (UBI) Sand #2 (LNEC) D 5 (mm) 1.28,7 D s r Measuring equipment: Point gauges installed on rolling bridges Leica Reflectorless Total Station TCR37 scour depths freesurface levels Topographical surveys 4

21 Results on isolated slots Test b (mm) z s /h () T (min) d se (mm) A e (m 2 ) V e (m 3 ) r de () r Ae () r Ve () r r r de Ae Ve d d d se Ae Ae % Ae Ve Ve % V se se % z s /h = 1/3 provided satisfactory scour reductions in all cases (Best performances of 3% of d se ). e A A1 A2 A3 A4 A5 A6 B B1 B2 B3 B /6 1/3 1/2 2/3 1 1/6 1/ The slot acts from the beginning of the tests, sucking downflow and weakening horseshoe vortex. C C1 C2 C3 C /6 1/ Results on isolated bedsills Bedsills placed close downstream the pier reduce scouring by cutting the lower part of the wake vortices (inside the scour hole). Bed sills do not act immediately at the beginning of the experiments. The smaller the distance between the pier and the bedsill, the larger the effectiveness is. Best effectiveness of 25% of d se. Test A A1 A2 A3 A4 B B1 B2 B3 B4 C C1 C2 C3 C4 b (mm) L/b () T (min) d se (mm) A e (m 2 ) V e (m 3 ) r de () r Ae () r Ve ()

22 Results on combined bedsill+slot The best configuration of a given isolated countermeasure was chosen to be combined with the other best one. Effectiveness increased to become of the order of 4% to 5%. Test A7 B5 C5 b (mm) z s /h () 1/3 1/3 1/3 L/b () d se (mm) r de () Riprap mattresses as a countermeasure against scour at bridge abutments Cristina FAEL, 27 Tests on clearwater, corresponding to the most usual flow conditions in flood plains, where abutments tend to be more frequent. Tests for verticalwall abutments, supposed the most unfavourable. 44

23 Literature review Riprap stone sizes (to face shear failure) General formula for verticalwall abutments (after PaganOrtiz (1991); Atayee et al. (1993), Austroad (1994); Richardson & Davis (1995): Dr5 C n Fr h s 1 where D r5 = median riprap stone diameter; h = flow depth; s = specific gravity of blocks; Fr = Froude number; C, n = coefficients. Mattress thickness (to face winnowing) With a suitably graded filter or filter cloth placed underneath according to NZ Ministry of Works and Development, t according Lagasse et al t.5d or t D 1 r5 r1 2D r5 45 Layout of riprap mattresses (to face edge failure) According to Richardson & Davies 1995 w 2h According to Eve & Melville 2 w D r5 B h h B L According to Melville & Coleman 2 w H / Vd se 1. 5d se 46

24 Experimental Setup Tests were carried out at UBI. Flume dimensions: Length = 3 m Width = 4 m Depth = 1. m Recess box: Length = 3 m Width = 4 m Depth =.6 m 47 Verticalwall abutments were placed on the bottom of the recess, at its mid crosssection, protruding at right angle from the glass wall. Abutment dimensions: Width =.14 m Length =.3 m.51 m.72 m.93 m 1.13 m L/d The recess was practically filled with natural quartz sand; riprap stones were placed around the abutments, on top or embedded in the quartz sand. 48

25 2 types of sand + 3 types of riprap stones were used in the tests Material D 15.9 (mm) D 5 (mm) D 84.1 (mm) D s r Sand # Sand #2.64, Riprap # Riprap # Riprap # Measuring equipment Point gauge installed on a rolling bridge Video camera installed inside the abutment scour depth freesurface level visualization of riprap instability 49 1st set of experiments (15 tests on riprap stone size) t = 3D r5 D r5 = 3.59 mm; D r5 = 7.48 mm; D r5 = mm on a filter tests started with a low flow velocity. the velocity was successively increased while the flow depth was kept constant until riprap stones began to move close to the abutment. 5

26 Evaluation of existing contributions F r of the approach flow F r of the contracted crosssection D r5 d.15 L/d=7.8 L/d=6 L /d =9.3 L/d=4.1 L/d=2.3 D r5 d d h.5.5. F r Austroads (1994) Present study. F r PagánOrtiz 1991 Atayee et al Richardson e Davis 1995 Present study None of the equations adequately fits the experimental data. D r5 depends both on F r and L/h (L/d). 51 Critical value, I c, of the approach flow intensity, I c = (U/U c ) s, below which scour does not show up, for a given value of L/h : I c decreases with L/h. 1. I c Sand Riprap #1 Riprap #2 Riprap #3 It seems that I c increases with h /D r5. An envelope curve was established that ensures the stability of riprap stones..5. Hager and Oliveto 22 Envelope curve h h Fael et al L I c 1 5 h 52 L h

27 2nd set of experiments (42 tests on the minimum mattress thickness to face winnowing) t = 1D r5 ; 2D r5. 2D r5 D r5 = 7.48 mm; D r5 = mm sand #1 and sand #2 without filter Velocities were kept equal to 9% of those inducing scour at the abutments, as defined in the first set of tests. 53 h se h 1..8 riprap #3; sand #1 N = 1 N = 2 N = 3 Increase of layer thickness decreases scour depth; t = 3D r5 seems to be enough to stop scour if riprap acts as granular filter h se h.8 h h L h.6.4 riprap #3; sand #2 L/h = 9.42 L/h = 7.75 Scour is negligible for N > 6 when it does not act as filter, but it is still present for N = N riprap mattresses should preferably lay on a filter; 54

28 3rd set of experiments (14 tests on the minimum plan dimensions of mats so as to avoid edge failure) 2.46 L/h 9.42 t = 2D r5 = 31 mm riprap #3 w =? on filter cloth U Uc of sand 55 dse d unprotected bed protected bed Riprap mattresses reduce the equilibrium depth of the associated scour hole. Reduction is not very significant, particularly, for short abutments. L d Flow direction dx a 4 d h w wu w abutment 35 a =3.6º+2.9º w dy 3 a =3.6º a =3.6º2.98º w d L d is practically constant irrespective of L/h. 56

29 Variation of w/h with L/h w h 3. w u abutm ent w w w w Nonfailure Richardson and Davis 1995 Failure h h w d L h w increases with L/h. The influence of h /D r5 cannot be addressed since it was kept 12/ Richardson & Davis s (1995) equation leads to safe predictions of w. 3 5 h A better predictor of w is suggested: L 2 w h 57 END 58