You are on page 1of 13
general, erosion rate of uncompacted cohesive mud increases with bed shear stress, once the critical shear strength is exceeded. The nature of entrainment also depends on the state and history of consolidation, which wil determine whether the clay is eroded as individual floccules, o fragments of consolidated mud. Also, the direct fluid stress may be augmented by impacts on the bed of grains already in motion, Bed load and suspended load I Mechanics of bed load transport Sediment grains with diameters greater than approx- imately 0.1 mm (sand and gravel) generally move close to the boundary (within ten grain diameters) by rolling, sliding, and saltating (jumping). These ftins move more slowly than the surrounding fluid because of their intermittent collisions with the bed and each other. These grains form the bed load. The continuing movement of bedttoacl grains requites the existence of an upward dispersive force (Ragnold 1966, 1973) that must be exactly balanced by afains for steady bedoad transport. The wo ways. imersed weight of the moving fof dispersing grains normal to the flow boundary are by collisions between the grains and bed, and by fluid lit, as discussed below. Saltation Gumping) is the dominant mode of load transport, with rolling andl sliding. only ‘occurring near the threshold of entrainment and. between individual sa Bridge & City of rolling and sliding grains increases over an regular bed, the grains are frequently launched upwards as they are forced over protruding, im tion jumps (suimmary in Dominic 1984). As the downstream velo- mobile grains, thus initiating saltations. In fact, the upward component of a centrifugal force associ ated with the forced curved path of the moving grain temporarily exceeds the immersed weight of the grain, Bagnokd (1966, 1973) suggested that salting grains rebound elastically at impact with the bed and disturb bed grains, whereas others have denied this possibility because of the effects of viscous damping of grain motions near the bed (Gordon et al. 1972; Abbott & Francis 1977). Recent ‘experimental work suggests that elastic rebound is possible for relatively arge sand and gravel grains at moderate transport rates (Nifio & Garcia. 1998: by Schmeekde ef al, 2001), At low transport rates, ly to be ons with other moving grains (Leeder 1979a, b). However, at higher sediment transport rates (greater concentration saltations will be more interrupted by collisions between moving grains contribute to the upward dispersive stress. Both turbulent and non-turbulent fluid lift forces ‘can act on bedoad grains (summary in Bridge & Dominic 1984). Non-turbulent lift forces are pres sures arising from 2 net relative velocity between of beddoad grains), and these collisions will ‘grains and surrounding fuid. They are caused by asymmetrical flow earbed grains, the reduced grain velocity relative tothe velocity gradi ent in the surrounding fluid (shear drift), and the effects of grain rota Lif). Te is very difficult to specify theit individual contribution to net lift particularly when complex grain collisions ‘occur within the bed load, and when turbulent lift in many cases these lift forces on Magnus forces are dominan As turbulence originates very close to the bed of 4 stream, bed load grains are undoubtedly affected by turbulent lift as well as by drag. As nearbed tur nerease with bed shear stress brent fluctuations ‘or shear velocity), gran saltations are expected 0 he modified more and more by turbulence as bed transport rate increase Gillert 1914; Luque & Van Beek 1976; Abbott & Francis 1977; ridge & Dominie 1984; Bridge & Bennett 1992: Nilo et al. 1994; Niflo & Garcia 1994, 1998a, b). Turbulence affects different sized. rains in differen ment transport rate of gravel, most of the move ment occurs during sweeps, and (0 a lesser extent during outward intesactions (c.g, Drake ef a, 1988; Thome er al, 1989; Williams ef al, 1989a,b; Williams 990, 1996) nteractions have lide effect on comsesediment movement because even the Largest values of turbulent lit are ‘much less than the immersed weight of the gravel grains. Furthermore, the mean grain size of the bed load increases with turbulent shea example, at low sedi ic Ways. Fo Bjections and inward ated with sweeps. With smaller (actually lighter) ‘much more influence on bedload movement (e.g Sutherland 1967; Grass 1974; Bridge & Bennett 1992; Gareia ef af. 1996; Cao 1997; Sechet & Le Gucnnee 1999), ger bed shear stresses, ejections have Models of saltation trajectories and the thickness of the bed-load zone The mean height of the becload (saltation) zone js an important quantity because i controls the roughness height, &,, Furthermore, bed load trans port theories depend critically on knowing the dis lance trom the bed of effective fluid thrust on bedload geains, which is related to saltation heigl Many approaches to determining the mean of the bedtload zone are semi:theoretical (empir ical, predicting that saltation height increases with sediment size and some measure of sediment trans port stage (reviews by Bridge & Dominic 1984 Bridge & Bennett 1992). In general, saltation height is proportional to grain diameter, There is uncer tainty, however, as to whether saltation height is proportional to bed shear stress of shear velocity ‘or both, depending on sediment transport stage. Theoretical models of saltation trajectories have also been used to determine saltation height (e.g, Reizes 1978; Murphy & Hooshiari 1982; Wiberg & Smith 1985; Sekine & Kikkawa 1992; Lee & Hsu 1994, 1996; Nifio & Garcia 1994, 1998a, b; Hu & Hui 1996; Lee et al. 2000). These models attempt (0 analyze the various non-turbulent and turbulent uid forces acting on single saltating grains. How ever, itis very difficult 10 assess the infitence of ‘usbulent fluid motions on saltating grains, and the effects on grain trajectories of impacts among ‘rains are not considered, Therefore, most theorct: ical models do not perform very well when com ppared with natural data Mechanics of suspended-load transport Smaller grains (clay, silt and very fine sand) gener y travel within the flow, supported by fluid tur bbulence directed upward against their tendency to settle. These grains, the suspended load, travel at approximately the sme speed as the surrounding fluid because they are not being decelerated by intermittent collisions with the bed, Maintenance of a suspendedsediment load. requires the exis ence of 4 net upward-directed fhuid stress to bal ance the immersed weight of the suspended grains, which can only arise if the vertical component of turbulence is anisotropic Bagnold 1966), This anisotropy can apparently be associated with the bursting phenomenon, whereby sediment is sus pended mainly in vigorous ejections, and returns to the bed under the influence of gravity and down: wardirected turbulent motions, This idea is sup. ported by experimental observations of discrete clouds of suspended sediment associated wit ejections, and the wavy trajectories of suspended grains (eg. Grass 1974, 1983; Sumer & Ogu 1978; Sumer & Deigaard 1981; Garcia et al. 1996; Nino & Garcia 1996; Cao 1997), However, sediment is also suspended in vortices generated by eddy shedding from the separated boundary layer in the trough regions of ripples and dunes (e.g, Coleman 1969; Jackson 1976c; Muller & Gyr 1982 1986; Tkeda & Aseada 1983; Rood & Hickin 1989; Soulsby et al. 1991; Lapointe 1992; Onslow et al 1993; Kostaschuk & Church 1993; Williams 1996; Kostaschuk & Villard 1996a, b, 1999). These tices are much larger ‘cous sublayer. in ejection from a vis the case of dunes, these vortices ccan reach the water surface as boils or kolks (Matthes 1947; Jackson 1976c: Babakaiff & Hickin 1996). As long as a sediment grain experiences more upwarddirected turbulent motions than the combination of downward-irected turbulent ‘motions and gravitational setting, ie will remain in suspension Two criteria for the suspension of sediment are ems! > G25) fms! ems? > V 6.26) The first criterion indicates that as long as time in excess of V,, sediment can be moved upwards ‘within the flow, even if the sediment eventually returns to the bed. The second criterion indic ates that for a gra time averaged upward-directed turbulent v experienced by the grain must exceed the down- upwarddirected turbulent velocities are n to remain in suspension the locity warddirected velocity plus the setting velocity of the grain. The relationship between the time averaged upward and downward directed velocities ’ depends on the coefficient of anisotropy, an (eqn 3.63). If an can be assumed 10 seach a maximum of between 0.1 and 0.2 near the bed Cbut not in the viscous sublayen), the suspen fon criteria in equations 5.25 and 3.26, forthe case and the overall rm: of neat bed conditions, become aemso’>V,, where aist2015 G27 where 618041009 G28) Furthermore, it is well known that rmso'/U/,= 1 near the bed (see Chapter 3), so that the suspension criteria for near-bed conditions are al,>V, and bU.>¥ G29) Suspension criteria such as these have been sug: ested by many workers (e.g, Lane & Kalinske 1939; Bagaold 1966; Middleton 1976; Engelund & Fredsoc 1982b; Van Rijn 1984b), Also, as the shear velocity and the settling velocity can be related to the dimensionless bed shear stress and the grain Reynolds aumber, the threshold of suspension can be represented on the threshold of entrainment dia sgsams (Fig. 3.7), As the bed shear stcess varies in time and space, sediment with a particular setting Velocity (or grain size) may travel as both bed load ad sediment load. This is particularly true of the sand sizes coarser than about 0.1 man, This is called intermittently suspended load. The wash load is commonly distinguished from suspended bed- material load. Wash load is very fine grained and susperteled even at low flow velo- city. Furthermore, the volume concenteation of wash loud does not vary much with distance from the bed, whereas the volume concentration of sus pended bed:material load decreases markedly with distance from the bed. Whereas the suspended bed: ‘material loa originates from the bed, the wash load ‘can come from bank erosion ancl overland flow Perhaps, the two different suspension criteria above effectively distinguish wash load (eqn 3,26) from suspended bed: material load (eqn 3.25), ant transport on flow Vertical profiles of time-averaged flow velocity and eddy viscosity in sedimer laden flows In the case of uniform, turbulent boundary layers lover flat beds with sediment transport, itis gener ally recognized that the moving sediment modifies flow characteristics. Many scientists in the 1940s to 1960s (reviewed in ASCE 1975) suggested that the presence of sediment in the flow caused a reduction in x, €, and , but an increase in nearbed velocity gradient and stream-wise turbulence intens- ity, The main effect of sediment on turbulence was expected to be near the bed where sediment concentration is highest. These conclusions hav been substantially confirmed by recent studies (e yn 1988, 1991, 1992; Nouh 1989; Wang & Larsen 1994; Wang et al. 2001). Furthermore, even weak hedioad transport affects nearbed turbulence with x’being reduced from its clear water value 10 around 0.3 (eg. Gust & Southard 1983; Bennett & Bridge 1995a, b; Nikora & Goring 2000), ‘The early approaches to this problem involved the assumption that the law of the wall is applic able to the full flow depth but with « reduced from ts clear water value of 0.41 (e.g. Vanoni 1946; Finstein & Chien 1955; review in ASCE 1975). These approaches have been criticized on the grounds that the law ofthe wa flow depth (e.g. Coleman 1981, 1986; Gelfenbaum & Smith 1986). Some recent altemative approache ‘does not apply to the whole retain a constant clear water x value of 0.41 and ‘nile the idea that the density stratification brought about by an upward decrease in suspended sediment concentration suppresses turbulent mix ing, as represented by a flux Richardson num. ber. One of these approiches involves simply adding a flux Richardson suumber term to the law of the wall (takura & Kishi 1980; Adams & Weatherly 19814, b), ie 2 ooh + aki] G30) Ri=y/L ap Bis caress : us i 6.32) gCo7p— where £ is the Monin - Obukhov length, Kis the flux Richardson number, C is ve jon of suspended sediment, &=5~7 and «= 0.4 in addition, the characteristics of the fluid are molified as: 4 = P+ (O- px. Gan Hy = HCL + 2.56 1526c G35) a= BaP 36 where wand v are the molecular and kinematic fluid viscosity and the subscript mr is the modified parameter, Another approach utilizes the velocity defect wake law for most of the boundary layer in sedimentaden flows, with Coles's wake coefficient increasing with suspended sediment concent lion as represented by a (Coleman 1981, 1986). ¥ use an empirical expression for the vertical vast Richardson number nother approach is to ation of e in clear-water flow and modify this in the presence of suspended sediment using an empisical cortection factor that involves a flux Richardson number (e.g, Smith & MeLesn 1977; Gelfenbauim & Smith 1986; McLean 1992) ~ L3syRi) 63D ao ~ pr[acray| oT paar? a3 where y= 10 and ¢ is determined in eqn 3.43. Finally, Umeyama and Gerritson (1992) account for the effects of suspended sediment by modifying the mixing length, using I= y-ray 39 where Bisa function of suspended sediment con centration C Velocity-profile laws for sedimentialen flows that incorporate a Richardson number term co account for the fects of high sediment concen sation, and tain a clearwater value of = 0.41, do not agree well with recent data collected by Bennett ef al. (1998) (Fig. 3.10). Ie appears that the law of the wall applies to most of the flow depth in sedimenciransporting flows, but that x takes values between about 0.28 and 0.34 irrespective Of sediment transport rate (e.g. Gust & Southard 1983; Bennett & Bridge 1995a, b; Bennett ef al 1998; Nikora & Goring 2000; Wang et af. 2001) Furthermore, the effect of the bedlox and low: relief bed waves (several grain diameters high) on nominally planar beds isto increase the equivalen sane roughness pared with sedimentirce flows, Thus &,/D ranges from 2 to 17 (Bridge & Hennet 1992; Bennett ef al, 1998; Wiberg & Rubin 1989), Effects of suspended sediment on turbulence intensity and length scale The nature of the interaction between sediment ‘gruins and fie turtles Stokes number: has been reited to the St= tH, G40 where f, is the particle response time and fis a scale (e.g. Hetstoni 1989; ghobashi 1994; Crowe ef al, 1996). If St<< 1 the particles follow the motion of the turbulen eddies very closely (relative velocity tends to zero), uid turbulence tin whereas for $1>>1 the grains do not respond Closely to turbulence and the relative velocity of the grains and the fluid is large. Some workers Ce. Eighobashi 1994) have suggested that for 10-* < C << 10°, suspended grains increase turbulence inten sity if Sr>1, but decrease turbulence intensity if St. For C> 10°, the effects on turbulence ‘modulation of geain-; account of. Furthermore, direct numerical simula tions indicate that suspended grains catse incr in the turbulent energy of the smallscale eddies, but decreased turbulent energy of the larg scale eddies (Squires & Eaton 1990; Elghobashi & rhs s caused by the effectof grain of samples vertical profiles of timeaverage ow laden water ove aa uppers plane bed. tn, curves numbered an 2 are for the Law of the mest 20% of he boundry core Boundary layer, respectively For curve equivalent sad ronghness/mean gr etect wake lw eae revel Geophysical Unio In contrast to this view, others believe that the main control on modulation of turbulence intensity is the ratio of grain size to turbulence microscale (DA), with increased intensity for DH, > 0.1 and decreased intensity for D/I, <0.1 (Gore and Crowe 1989, 1991; Crowe ef al. 1996; Crowe 2000). It is reasoned that the increase in turbulence intensity is associated with vortex shedding from the wake shat exists when grains and fluid are in relative motion. at sufficiently high grain Reynolds num: pers (Hetsroni 1989; Gore & Crowe 1991; Yuan & Michaelides 1992), Grains smaller than the most energetic eddieshave part o ferred to them, thereby reducing turbulence intens- the eddy energy trans- ity. This group of workers believes that the Stokes number does not describe turbulence modulation Alternat sate that itis properly, and that € is only importa determines the degree of modulation. ively, Elghobashi and Truesdell (1993) possible to get turbulenc of Djf, much lower than 0.1 enhancement at values and that wakes and vortex shedding are not necessity (0 get turbu lence enhancement, Also, Best (1998) found that grain spin can reduce the size of turbulent wakes, but still enhance turbulence Another group of workers maintains that turbu lence modulation in boundary-layer flows arises from the way particles interact with the coherent structures in the wall region (Rashidi ef af. 1990, ri ef al, 1995a, b). Relative) he frequency of bursting letsroni 1993: Kaf but decrease the ejection velocities because of the relative veld ity ofthe fluid and grains. Turbulence intensity increasesas the concentration of relatively large grains increases, but the opposite happens for small grains Recent experiments with both sedimentiaden and sediment free planar sind beds indicate that mobile-bed turbulence intensities are greater than those for sediment free fixed beds (Best et al. 199 Bennett e¢ al. Stokes numbers and ratios of grain size/tutbuslence microscale sug gest that turbule decreased in the presence of sediment for most of the flow depth. As the maximum grain Reynolds number based on the expected of fluid and sediment wa: 1998). However ive velocity about 5, vortex shed: dling from grains in suspension is not expected to be important for enhancing turbulence intensity Instead, turbulence intensity was increased due to ‘enhanced wake formation and eddy shedding (.e bursting) from the lee side of nearbed grains and low-relief bed waves, The decrease in mean mixing length (¢ddy size) and win sediment.aden flows re: ative to clearwater flows was duc to the effects of srin inertia during the relative motion of grains and fii trate (capacity) “The sedimen + rate isthe amount (weight mass, or volume) of sediment that can be moved past a given width of flow in a given time. It is very important to understand sediment transport rate because this controls the development of bed forms such as ripples and dune: sion and deposition, and the dispersal of physically transported pollutants, There is a vast and com: plicated literaure on the subject (reviews in Graf 1971; ASCE 1975; Yalin 1977; Chang 1988; Van 1993; and many others) and only the basics are pre- Bed-load transport rate Bedload transport rate can be expressed as 1, Wa, B41 a which 4, is the bedload transport rate as immersed weight passed per unit width, W is the immersed weight of becHoad grains over unit bed area, and @, is the mean bed:load grain velocity The contacts between bedltoad grains and the bed result in resistance to forward moti Ther the fiuic must exert a mean downstream force on the grains to maintain their steady motion, For bed load transport on a bed of slope angle B, the force balance parallel to the bed can be expressed as 1+ Win B= r+ 4, a2 ‘where ¢, isthe bed shear stress applied to the mov ing grains and immobile bed (but excluding form drag due to bed forms and banks), W sin f is the downslope wei ight component of the moving grains per unit hed area (with sin B positive ifthe bed slopes down flow), Tis the shear resistance due to the mowing bed load, and +, is the residual shear stress carried by the immobile bed Bed-load shear resistance can be expressed as T= (Weos B-f,) tan @ G43) ‘where Fis the net fiuid lift on bedt-oad grains per unit bed! area, taken as resulting clominantly from anisotropic turbulence, and given by R= webudy,* 64D as discussed above, and tan cris the dynamic friction ficient, From experiments on sheati in dy grain flows and coaxial drums, where volu ntrations are approximately 0.5, an er commonly lies bewween 0.4 and 0.75 (review in Bridge & Dominic 1984; Bridge & Bennett 1992), Tan ais not strongly dependent on shear rate, but decreases with increasing grain concentration. At low grain concentrations, tan a may be 0.75 t0 1 Values of tan cr for single saltating grains can be for grain flows e.g. 0.2 00.4: Abbott & Francis 1977; Nino & Garcia 19988), The residual fuid stress 1, must be equal wo the ‘considerably less than tho bed shear stress at the threshold of motion +, according to Bagnold (1956, 1966). Thus, once the applied fluid stress exceeds the value necessaty fo enteainment, grains will be entrained until the bed load resistance caused by their motion reduces the applied fluid stress to the threshold value of ¢, Therefore, the immersed weight of grains moving perunit bed area s given, from eqns 3.42t0 3.44, 38 W G45) san feos pl — (bU./¥,)"1~ sin f As the threshold of suspension is reached, the teem, in square brackets approches zero, indicating that the weight of beet-load grains approaches infinity In reality, this situation could not arise because high grain concentrations in the bed load! layer woul ‘occlude the bed from fluid drag and lf, thus limit ing the weight of bedload grains (e.g. Bagnold 1966). Furthermore, high concentrations of near sins modify tusbulence characteristics above the bed. In many natural flows, te bedslope is ‘mall enough such that sin tends (0 zero, and the term in square brackets tends to 1, s0 that 3.46) ‘The mean velocity of the average bed-load grain isgiven by = yy en suhere tty isthe time averaged flow velocity at the level of effective drag on the grain, and wis the re ative velocity of the fluid and the grain, The relative velocity is that at which the fuid drag plus down- nis in equilib- sum with the bedtload shear resistance: slope weight component on the B+ Fgsin B= Fgtan acos B= (bULIY,)) Using the general drag equation for Fz, and the fact that Fis equal to the fluid dra settling velocity, results in ata geain's terminal tug = Veta o(cos B- (U/¥,)?)sin BP? G49 Finally, as ,/U, =4, the mean velocity of bedYoad ‘gttins is given by v,can axccos 8 = aU, — Vzétan o> Furthermore, as the grain velocity must be zero at the threshold of grain motion, this equation may also be written as 4, = ath 1, fan ean a") G52) In order to determine the value of a in eqns 3.50 t0 3.52, it is necessary to have knowledge of the height above the bed level of the effective drag on the gains (that varies with the saltation height) and wr reansrort_ 61 details of the nearbed velocity profile. There have been numerous attempts to relate saltation height to some measure of flow stage (See above and review in Bridge & Bennett 1992). The flow velocity at the height of effective drag depends on whether the flow is hydraulically smooth, transitional, or rough. For transitional and rough flows, @ takes & value of about 8 to 12 (Bridge & Dominic 1984). Also, at the threshold of suspension, the grain velocity approximately equals the fluid velocity and 4, = al., which is another suspension criterion. Thus, the final approximate bed-load transport equation can be written as e G53) Equations of this form have been developed, using similar principles, by & number of workers (eg, MeverPeter & Muller 1948; Ashida & Michine 973; Bagnold 1973 Beck 1976; Engelund & Fredsoc 1976, 1982b; details in Bridge & Dominic 1984) to agree very well with experimental data for flows Luque & Van mis equation has been shown cover plane beds (Fig. 3.11). For low sediment transport rates (owerstage plane beds), a/tan ois approximately 10, but is approximately 17 for high sediment transport rates (uppers At high sediment t entrainment stresses and shear velocities are small plane beds), insport rates, the critical ‘enough relative to the applied bed shear stress und shear velocity that they can be ignored, and implying that becHload transport rate is propor tional to the 1.5 power of bed shear stress and the flows, and shear velocity is proportional to depth cube of shear velocity ‘pad in uniform averaged flow velocity, U, bedload transport rate can be expressed approximately as x pedst 55) where the proportionality depends on the flow resistance (4c. elative roughness) and the sediment transport stage Fig 2.11 Compan planed ‘wlan of bed toad anspor eqn 5.58 with ta of Gier(1914), 1970), Also showa ae (a) plot of eqn 3.53 wit 1Y7and 6, = 0.045, and lines representing data tied by (>) Meyer Peter ad Maller (1948), ce) Wil. (1966), (€) Ashita and Michive (1973), and (6) Luque and Van Beek (1976). After Bride & Donnie 1984.) © by the The bed-load transport rate integrated across the ‘width of a uniform flow is ye PaSwATT = pgs B56) where the overbar denotes widthaveraged flow depth and velocity, and Q is water discharge. The term on the righthand side is the rate of energy ture of the stream per unit stream length. In fact, a river can be thought of as a se transporting machine that expends energy to over: ‘come the frictional resistan GBagnold 1966). Widthrintegrated bed-oad trans: port rate has been expr e to sediment motion essed asan empirical power function of water discharge (or discharge in excess of that required for entrainment). However, in view of eqn 3.53, the proportionality coefficient is likely to vary widely depending on factors such as relat ive roughness (depth and width relative to bound: ary roughness) and sedimenttranspost stage. It is therefore not surprising that such empirical power functions are not generally applicable For beds covered with bed forms such as ripples and dunes, the bedoad transport equation can be applied in o method isto determ 1 of two ways. The most common ne the average bed shear stress ‘vera number of bed forms, and to semove the part of the total average stress that is due to form drag in the lee of the bed forms, because its ineffective in moving bedJoad grains (e.g. Engelund & Hansen 1972; Engelund & Fredsoe 19824, b; Nelson & Smith 19894; discussed in detail in Chapter 4). A second approach is to calculate bedload transport sate integrated over the back ofa typical bed form from knowledge of the local fluid drag and lift in this zone. This is not done routinely because itis very dificult to predict the neared fluid motions fon the backs of bed forms. As seen in Chapter 4 there are alternative ways of calculating bedload asport sates associated with ripples and dunes, ural river channels with bars have bed and bank slopes transverse to the mean flow direction, siving rise to a transverse component of bedtoad transport. Analyses of bedload transport on trans verse bed slopes are given, for example, by Ikeda (1982), Parker and Andrews (1986), Sekine and Parker (1992), Kovacs and Parker (1994), and Talmon et al. (1995) (see Chapter 5) The bedoad transport theory developed above is for mean flow and sediment characteristics, but it can be adapted ‘ment and with temporal variations in bed shear stress (c.g. Bridge & Bennett 1992). In this case, itis necessary to specify the proportion of the different or use with heterogencous sedi grain fractions available in the bed for transport and the thresholds of entrainment and suspension for each of the grain fractions. Its also necessary to specify the proportion of time that a particular bed shear stress acts upon the bed load, requiring a frequency distribution of bed shear sires. Bridge & Bennett's (1992) bedload transport model for heterogeneous sediment has been shown to agree To00 Observed ig Ne" eee Fig. 3:12 (@) Comparison of mixed ‘Wiliams (1970) data are for lower and the field daa of Milnous (1973) and Leopold and Emmet (197, feaimeat avaible for bedlond transport must be defined very car American Geophysical Unio swell with natural data (Pig. 3.12). Models such as this (eg. Wiberg & Smith 1989; Parker 1990) are essential for quantitative understanding of sedi ment sorting during erosion, tion, as discussed below ransport, and deposi Suspended-sedtment transport rate suspended-sediment teansport rate ata point in the ow is commonly expressed as i=uc asp where u, is the average speed of the sediment approximately the same as the fluid velocity) and C is the volume concentration of suspended sediment. The units of f, ae therefore volume of sediment transported per unit cross-sectional area ‘normal to the flow direction per unit time. The hecond equation of ridge ant Henne (1992) wit per stage plane beds observed i labor reduced to account forthe effets of beorms sch as dn iment rranseort 63 fl satura data, Giber 1914) nd ony Names, The bed load transport rates for are predicted wellby the equation if measured bed ly. Detain eidge and Henne (1992). © by che vertical variation of the velocity of suspencled sedi ‘ment and fluid can be calculated using an appropri ate velocity profile law (c.g. the law of the wall) Calculation of the vertical variation of C in steady uniform water flows is traditionally based on the palance of downward settling of grains and their upward diffusion in turbulent eddies (Schmidt 1925, (O'Brien 1935; Rouse 1937; Vanoni 1946; Einstein & Chien 1955), Le Ve + e,acidy =0 G58) where the first term is the rate of settling of « par ticular volume of grains per unit volume of fluid the second term i rate of turbulent diffusion of sediment per unit volume, and ¢, is the diffusivity of suspended sediment (equivalent to a kinematic edidy viscosity) This balance can also be written for grain fractions. In order (o determine C Fig. 3.12 (cont) by Compatison of mixedsize bed lead equation of ri Urge 1995). The Wilcock and Southard data at any height above the bed, alculate the vertical variation in ¢,, Assuming that 6, = fe, where is close to 1, and that the law of the he flow depth, sesuls in the well known Rouse equation ‘wall extends through a5 where C, is the value of C at y=a, This equa tion predicts that C decreases continuously and smoothly with distance from the bed, as would be expected in view of the fact that turbulence intensity also decreases with distance from the bed. The distribution of suspended sediment becomes ‘mote uniform throughout the flow depth as expon- cent © decreases; that is, as settling velocity (grain size) decreases and/or as shear velocity (near bed turbulence intensity) increases (Fig. 3.13). For example, a decrease ia water temperature causes an increase in fluid viscosity, which may in turn ‘cause a decrease in settling velocity and in increase Fig. 2.13 (2) Variation of eave concentston of suspended sediment with according fo the Rouse eq 3.59 (ater ANCE 1975). The siduton of suspended sediment becomes more unorm locity (neared turbulence intensity) incresses. Reprinted by pi less than observed in the upper part of the ow concentration, The Rouse equation has been shown to agree with measured in suspended sedimes suspended sediment concentrations in the lower part overestimated in the upper parts (Fig. 3.13). This {stream flows but concentrations tend to be discrepancy is addressed below. Some of the assumptions used in developing the equation have been criticized (akura & 980; Adams & Weatherly 198ta, b; Coleman Parker & Coleman 1986; Gelfenbaum & 986; MeLean 1992). In particular (the law Of the wall may not be applicable throughout the Rouse Kishi 981 Smith flow depth, there being a wake term: (i) the effects Of suspended sediment on the vertical profiles of flow velocity, Reynolds stress, and eddy’ viscosity and be represented using Richardson number; Gii) for high concentrations, must be considered, the influence of sedimentladen fluid density and viscosity on grain setting velocity must be consid: cred; and civ) all grain fractions in the suspended: sediment load must be considered to calculate view of the concentration adequately. However, discussion above, criticisms (@ and Gi) may not be justified, and the effect of suspended sediment © adore shear ren the Rowse flow parameters may be a reduced value of « Chiu ef af, (2000) have circum vented perceived dificulties with the use ofthe law of the wall by using Perhaps the most serious concer with the Rouse cernative velocity profile laws. equation is that the sediment diffusivity has rarely been calculated directly using quantitative observa tions of the motion of sediment in turbulent eddies. Thus, there is doubt about the vertical variation of e,and (e.g. ASCE 1975; Van Rijn 1984b). There are two main approaches to calculation of &,, The first, most common method involves assuming that the diffusionsetting balance (equation 3.58) is cor fect, such that ¢, can be calculated using meas ured values of ¥,. C, and dCdy (e.g. Chien 1956; Coleman 1970; Van Rin 1984b), This led to the conclusion that ¢, is generally greater than &, such that B> 1. mate empirical equation for n Rijn (1984b) defined a very approxi 142,109, O.1< Vy 1 3.00) Van Rijo’s explanation for > 1 is that suspended sediment grains are subjected to centrifugal forces that throw diem to the outside of eddies. This approach also led to the idea that ¢, does not vary parabolically with y/dl as € does, but may approach a constant value above y/d = 0.5 (Coleman 1970; Van Rijn 19846). I € equation (3.58) leads 4s constant, integeation of = exp] —“# G6 A second method for calculating ¢, used in modeling two-phase flows, is to use dt approach anil 10 that for calculating ei 3.62 ara where all values apply to sediment particles only (eg. Chung et at. 1986; Rizk & Elghobashi 1989). Experimental data on turbulent motions of sus pended sediment are very rare, but Bennett et a (4998) have shown that values of &, calculated ues of € suggesting that suspended sediment follows the ‘mean fluid turbulence fairly closely. In contrast, val ues of €, calculated using the diffusion settling bal: to ein the upper half of ow depth, because the sediment concentration is larger than expected. (One reason for this discrepancy cou definition of ¢, using eqn 3.58, such that sediment rains that are suspended to the higher the low are not associated with mean turbulence characteristics, but with turbulent eddies with the greatest turbulence intensities and mixing lengths, I eddy viscosities are calculated using eqn 3.62 only for relatively high magnitude ejection events, these are comparable to those calculated using the diffusion-ettling balance in the upper half of the flow, suggesting that sediment is not suspended. by average turbulent eddies higher in the flow Gennett et al. 1998). Thus, the Rouse equation, based on average turbulence characteris only be stricly applicable for the lower parts of the flow Application of the Rouse 4 requires calculation of C, and a, and there are var fous methods for doing this (e.g. Takura & Kishi using eqn 3.62 are generally similar to ance tend to be significantly overestimated rel 1 be errors in sels in jon in practice 1980; Van Rijn 198s; Soulsby & Wainright 1987, McLean 1992; Van Niekerk et al, 1992; Zyseeman & Predsoe 1994; Cao 1999), Normally, , iscaleulated ai a position within or at the top of the bedload ken as half becwave height or equivalent roughness ayer, If bed waves are present, @ can be height CVan Rijn 1984b). ‘The mean volume con centration of grains in the bedload layer, C, is the volume of grains per unit bed area divided by the mean thickness of the bedoad layer, ¥,. given using Bridge & Dominic's (1984) model as Gn 6.63) @ pay, wane The mean grain concentration can be assumed to ‘occur at a distance fzom the bed of half the thick: ness of the bedload layer, of half the equivalent roughness high An alternative to the coavection-diffusion approach to calculating suspended sediment con: mn in water flow is that based on the mechanics of two-phase flows (Drew 1975, 1983 McTigue 1981; Cao ef af, 1995; Greimann ef al. 1999; Crowe 2000; Greimann & Holly 2001). T twophase flow approach is useful for simulating actions between grains, and the elfeets of grain inertia, Interactions hetween particles lead t0 ar increase in the ability of the flow to suspend part {cles Particle intertetions are only important near the bed where local grain concentrations exceed about 0.1. Also, particle inertia is only important if Sr>> 1 (arge grains near the bed), Unfortunately be applied easily co real rivers (Greimann & Holly 2001). In order to calculate the suspended:sediment proach isnot fully developed, and transport rate integrated across the width ofa river channel, itis necessiry to define vertical profiles of flow velocity and suspended'-sediment concentra tion across the channel width. Alternatively, width: integrated suspendedt-sediment transport rates have been expressed using empirical power functions of water discharge. However, such power functions are very inaccurate because they are case depend: hharge and suspended: sediment discharge (especially the wash load) do not necessarily vary in time and space congruently as discussed below, More details of suspended: sediment loads of rivers can be obtaitied from Petts and Calow (1996), Evaluation of sediment transport rate models There are many different models for sediment transport rate in steady, uniform flows, So for bed load only, whereas others predict bed load and suspended load either separately or together (otal load). Important differences in these models. depend on whether or not they consider indivi dual sediment size - density fractions, the sediment available in the bed, and the effects of turbulent variations in fluid drag and lift on bed load; the way in which flow over bedforms such as ripples and dunes is treated; the description of the vertical ariation in flow velocity, turbulence intensity, and diffusivities of fluid’ momentum and. suspended sediment; and how moving sediment interacts with and modifies the water flow. Most sediment trans ort models are both theoretical and empirical, but itis desirable to minimize the use of empirical tors, as these limit the general applicability of a maclel, There have been many compari performances of the various sediment transport ns of the modlels (€ 8, Geaf 1971; White ef at, 1973; ASCE 1975; Yalin. 1977; Brownlie 1981; Chang. 1988 Gomer & Church 198! be treated! with caution if the size or quality of the empirical database is lacking Stich comparisons shomld Variation of sediment transport rate intime and space Sediment transport rate varies over a large range of time and space scales, associated, for example, with fluid turbulence, the migration of bed waves such as ripples, dunes and bars, the break-up of armored of partially cemented beds, bank slumps, seasonal floods, flow diversions, and earthquake. induced changes in sediment supply (e.g. Gomez et al, 1989; Hoey 1992; Bennett & Bridge 1995. Nicholas ef al. 1995; Wathen & Hoey 1998). The ‘one-dimensional convection-diffusion equation forsuspendedsediment concentration in unsteady non-uniform flow is: B= eX aye 6H ‘x Oy ay "Oe The first two terms of this equation are zero for steady, uniform flow, such that this equation pecomes the sume as eqn 3.58 after integration. There is not a simple solution to the unsteady, non uniform equatior A characteristic feature of unsteady, non-uniform flows is that sediment transport rate commonly varies incongruently with temporal and spatial changes in water discharge, i, there is alag or hys: teresis. This may be due to variations in available sediment supply, or due to the lag of sediment and water flow behind a faster moving flood wave. Wash load is never charge, and commonly reaches a peak during ris ing discharge because of an overland flow source The lag of bedoad transport rate behind a spatial change in flow characteristics is very stall (order of grain diameters); however, for suspended sedi correlated with water dis ‘ment this spatial lag is on the order o flow depth these variations are discussed in detail in later chapters Sorting and abrasion of grains during transport mhe fact that bed load and suspended load texvel at different speeds and in different parts of the flow leads to the possibility of sorting, of physical separation of grains with different settling velocit: {es (Le. grain size, shape, and density), as shown in Fig, 3.14 ¢based on Middleton 1976; Bridge 1981) Figure 3.15 indicates approximately the cange of grain sizes (for a given shape and density) that ‘can be transported as bed load for a particular bed shear stress, Actually, Fig. 3.15 represents the ‘thresholds of entrainment and suspension for mean flow conditions and for mean sediment size, The actual grain size range in the bed load depends ‘on what is available for transport, the threshold Of entrainment or suspension for individual grain fractions in a distribution, and turbulent fluctu ations in bed shear stress (Bridge & Bennett 1992), Nevertheless, Fig, 3.15 explains in a general way why intermittently suspended load exists, and why

You might also like