National Technical University of Athens (NTUA)
Πειραματική διερεύνηση της ροής σε ανοικτό αγωγό με προσομοίωση βλάστησης
Abstract
dc:descriptionIn this doctoral thesis, the behavior of the flow field in an open channel in the presence of vegetation was studied. The effect of vegetation on river flow was previously considered a nuisance due to the resistance it offered to flow so that it was often removed to avoid reduced channel capacity and higher risk for flooding. In modern times, however, the important environmental benefits offered by vegetation are recognized, such as increasing bank stability, reducing erosion and water turbidity, and providing a wide range of services to ecosystems such as shelter for aquatic life, nutrient uptake and oxygen production that improve water quality. In recent years, much research has been conducted to understand and quantify the physical processes that govern the complex interaction between water flow and vegetation. These interactions depend on environmental factors and plant characteristics such as relative submergence, flexibility, size and density of vegetation. Early studies attempted to quantify the effect of vegetation on channel roughness either by incorporating it into the resistance coefficients such as the Manning coefficient n, or by properly modifying the velocity distribution over a smooth flat plate given by the Prandtl – von Karman relation (external law of the wall). More recent research suggests that the vertical velocity profile often has a characteristic S-shape and the distribution over the canopy follows approximately a logarithmic law. As more sophisticated numerical tools are developed to model the complex flow–vegetation interaction processes, the need for detailed experimental data becomes increasingly important, both for a better understanding of the underlying phenomena and for comparison and verification. The present work aims to contribute to filling this gap by providing detailed measurements and comprehensive comparison of key flow field characteristics, i.e. velocity profiles, turbulent intensities and turbulent shear stresses, for three types of artificial vegetation elements with different geometries and characteristics such as stiffness and porosity (two rigid and one semi-rigid), all with similar densities. After an extensive literature review, the theoretical background for the present work is briefly presented, based on the theory of turbulent flow and boundary layer, followed by a discussion of flow stratification in the presence of vegetation and for the shear stresses that develop in the various flow layers. A laboratory channel 16 m long and 0.50 m wide with a slope of 0.001 was used for the experiments (Figure 1). Vegetation elements were placed in the middle part of the channel, on an 8 m long perforated pseudo-bottom extending 5 m from the beginning of the channel to 3 m from its end. Upstream and downstream of this perforated bottom normal non-perforated bottoms were fitted to produce a straight bed. Three types of elements were used (Figure 2), mimicking submerged small plants with rigid stems and dense or sparse foliage or no foliage: (a) Simple rods 0.5 cm in diameter and 4 cm high, b) Compound elements consisting of 3.0 cm diameter plastic spheres fixed over thin rods 0.8 cm in diameter and 5 cm high, c) Compound elements consisting of flexible needles arranged axisymmetrically over the same rods. For all element types, two placement patterns were tested (Figure 3a, b), in a parallel and staggered 10x10 cm grid, yielding respective densities of 100 and 200 stems/m2. The flow rate was measured via a Venturi meter and a differential manometer installed in the laboratory supply line with an accuracy of ± 0.5 l/s, while the flow depth was recorded with a point gauge along the axis of the channel with an accuracy of ± 0.2 cm. The depth was controlled by a sluice gate located at the end of the channel so that it remained constant over the measurement area. The flow rate Q = 41 l/s and the flow depth H = 25 cm were chosen after preliminary experiments and were constant throughout the experiments. A no-vegetation experiment for the same flow rate and depth was also performed for comparison. Velocity measurements were taken via a 3D Acoustic Doppler Velocimeter (ADV LabVer. 2.7 Probe N0187 Nortec AS) with an accuracy of ± 0.01 cm/s, in 9 vertical lines at selected locations within the vegetation array (Figure 3c). The ADV instrument operates at a specific frequency of 25 Hz and each measurement lasted 2 minutes. A limited number of measurements were also made at selected locations downstream of the vegetated area in an effort to evaluate the extent of the influence of vegetation on the flow field downstream of it. Detailed data of all experiments are presented at the Appendix. The main experimental results are shown in dimensionless form in the following figures: (a) in Figure 4 vertical distributions of velocities u along the main flow direction, (b) in Figure 5 vertical distributions of turbulent shear stresses τ_xz/ρ=-(u'w') ̅ and (c) in Figure 6 vertical distributions of turbulence intensity, at all measurement locations for each experiment. In all cases the vertical coordinate z is normalized with the respective vegetation element height h; in Figs 4 and 5, the local velocities and shear stresses are normalized with the respective value at the top of the element, whereas in Fig.6 the turbulence intensity uRMS (√((u΄^2 ) ̅ )) is normalized with the local value of velocity u. As seen in Fig.4, there is in almost all cases a significant decrease in velocity (compared to the case without vegetation) which extends well above the top of the elements. The decrease in velocity is clearly more pronounced for the compound elements, with peaks at the height of the bulky part of the element. In general, the minimum velocities appear somewhat higher in the experiments with the rigid foliage (around the level of the center of the sphere), while in the experiments with the semi-rigid elements they are at about the base of the flexible foliage. In all cases the observed velocity reduction within the vegetation appears to be compensated for in the area above the elements. In the same figure the results reported by Dunn et al (1996) are shown for comparison. These are spatial average values for a staggered array of simple rods, somewhat similar to the present Exp2, but with different spacing and dimensions of the elements. It is noticed that for this experiment, excluding the positions A and C which are very close to the elements, the profiles in the other measurement positions are in good agreement with Dunn et al’s, the latter ascending more steeply above the top of vegetation due to shorter distance to the free surface. On the contrary, in other experiments considerable differences are observed, highlighting the importance of element geometry and location in the array. Regarding the turbulent stresses τxz, much higher values are observed compared to the unvegetated case, generally increasing upwards but also having a decreasing or stabilizing trend near the free surface. Of particular note is the highly nonuniform distribution occurring at most measurement sites within the vegetation array, suggesting the development of intense shear layers. Furthermore, there are no significant differences in the shear stress distributions between the measurement locations in each experiment. Shear stresses tend to reach their maximum value mostly near the top of the array and slowly decrease with height above the top; below the top they are generally smaller. This behavior is opposite to the case of unvegetated flow, where shear stresses are greatest near the bottom and gradually decrease with height. Turbulence intensity is higher compared to the unvegetated case, within and above the vegetation array. Also, above the vegetation array, the intensities appear for all cases to converge, to be almost constant and to approach the distribution for the case without vegetation but never reaching it, while a slight decreasing trend is evident towards the free surface. On the contrary, one may observe a remarkable increase in intensity at the level of the massive part of the vegetation elements and towards the top in all cases. This is obviously due to the sharp reduction of the respective mean velocity at this level. Close to the elements, a strong variation is observed, with a maximum near or even above the top of the elements. In general, the turbulence intensity depends on the size and geometry of the vegetation elements and the density of the array, but most importantly on the proximity to the vegetation element. Figure 7 illustrates a comparison of the velocity profiles of the present experiments with those of Li et al. (2014), where compound elements relatively similar to the present ones but flexible, resembling the species Hydrilla verticillate, were used at three densities (50, 70 and 90 stems/m2). Considering the differences in elements dimensions, flexibility and density, the agreement for the experiments with the compound elements is satisfactory. Figure 7.Comparison of dimensionless velocity profiles for all experiments with those of Li et al. (2014) in position B The main conclusions from the experimental measurements may be summarized as follows: a. There are considerable differences in the local flow conditions depending on the location within the vegetation array, the proximity to elements and the type of element. This implies that data obtained at a single location or by spatial averaging are inadequate to describe the true nature of flow within the vegetation. b. Overall, the effect of foliage is very important, yielding considerably higher velocity reduction, higher shear stresses and turbulence intensity compared to simple stems without foliage. However, the type of foliage, i.e., rigid or semi-flexible of about similar size, appears of minor significance. c. For all types of elements, the effect of vegetation on the flow parameters measured increases with its density. d. In general, severe reduction of velocity occurs within the canopy below the top of the elements, which is more pronounced for the compound elements compared to simple stems. The minimum velocities are observed at the level of the foliage, i.e., at the upper (bulkier) part and tend to zero at certain locations off the elements alignment .e. The turbulent shear stresses are mostly higher above the top of the canopy than below. Within the canopy they are comparable to the no-vegetation case in the experiments with simple stems, but smaller for the compound elements. f. The relative turbulence intensity in the case of simple stems is somewhat higher than without vegetation and decreases gradually with distance from the bottom whereas in the presence of foliage it is much higher within the canopy, due to severe reduction of local mean velocity in that area. g. The presence of vegetation affects the flow field downstream, and the effect is higher for the elements with foliage and for the denser pattern. The distance affected is at least twice the vegetation width and needs further investigation. For a quantitative assessment of the velocity field a general overview of velocity distribution models in vegetated flow was made. Considerable research has been carried out previously but a universally accepted model is still unattainable. There are generally two types of approaches, two-layer and three-layer. In the former, the top of the canopy clearly defines an interface between two layers, above and within the vegetation, while in the latter the velocity profile can be roughly divided into three hydrodynamic zones: one within the canopy, one above the canopy, and an intermediate transition. More specifically, two models for the velocity distribution above vegetation are considered, by Raupach (1994) and Nepf (2012b), based on essentially the same logarithmic equation (1) with differences in the way the main parameters of the models are calculated. In particular, Raupach employed Equation (1) for the velocity distribution above terrestrial canopies, considering the height h of vegetation elements and the frontal density λ as independent variables. u/u* = 1/κ ln((z − d)/z0) (1)He found that the ratio of the shear velocity u* to the velocity uh at the top of the canopy may be expressed as:γ = uh/u* = exp(cλγ/2)(CS +CRλ)−0.5 (2)where c, CS and CR are numerical constants with proposed values c = 0.5, CS = 0.003 and CR = 0.3. Equation (2) requires an iterative solution to obtain uh/u*. Furthermore, he proposed the following equations for the zero-plane displacement height d and the roughness height z0:z0/h = (1 − d/h)exp(−κuh/u* − h)(3)1 − d/h = [1 − exp(−(cd12λ)0.5)]/(cd12λ)0.5(4)where κ = 0.4 is the von Karman constant, and h and cd1 are numerical constants, with proposed values h = 0.193 and cd1 = 7.5.The approach of Raupach was implemented in the context of the present study as follows: First, for each set of experiments, the values of uh/u*, d and z0 were evaluated based on Equations (2), (3) and (4) according to the respective value of λ shown in Table 1. Then, the logarithmic law (Equation (1)) was fitted to the velocity measurements for z > h at each measurement position. Thus, the local value of u* was obtained as the slope of the fitting line in a semi-log plot. In general, a very good fit was observed, thus confirming the validity of Equation (1). In fact, in several cases, its applicability appeared to extend below the top of the canopy. The values of u* for all experiments and positions, as well as those of d and z0, are listed in Table 2. One may notice that: (a) for any given position, the values of u* are considerably larger for the compound elements compared to the simple rods as well as for the staggered (denser) pattern compared to the parallel one; (b) the values of u* for the two types of compound elements are about the same for the sparse (parallel) placement pattern, although the frontal density λ is appreciably different; (c) the values of d and z0 are also larger for the compound elements, but they differ appreciably between the two types of compound elements; and (d) for a given experiment, three groups of nearly identical u* values at positions along lines parallel to the elements’ alignment may be identified, i.e., at (A,B,C), (D,E,F) and (G,H,I), suggesting a lateral differentiation of shear offered by the canopy to the upper non-vegetated layer. Subsequently, the application of a unified distribution over the entire water column was attempted in order to reveal the extent of the influence of element geometry at different locations within the vegetation. To this end, a relatively simple model was chosen, presented by Carollo et al. (2002). Although for unfoliated vegetation this model provides a good fit with measurements throughout the water column where the velocity decreases smoothly towards the bottom, except at locations close to the elements where a sharp decrease occurs, this is not the case for vegetation with foliage. In this case the distribution generally presents a minimum value at the leaf level and is impossible to be described by a unified model, pointing out the need for the creation of a complex velocity distribution model based on the stratification of the flow. In summary, from the preceding analysis on velocity distribution of vegetated flows, the following main conclusions are drawn:1. The logarithmic distribution originally developed by Raupach (1994) for flow over terrestrial canopies is capable of well describing the velocity profile in the layer above the aquatic vegetation, with proper determination of its parameters. These were found to depend on the type and density of vegetation elements and also on the relative location with respect to the elements’ alignment.2. For vegetation without foliage, i.e., simple stems, the velocity generally decreases gradually towards the bottom, except at locations in close proximity to the elements where an abrupt decrease occurs. Thus, excluding those locations, a unified velocity distribution, such as the one proposed by Carollo et al. (2002), provides good fit with the measurements over the entire water column.3. For vegetation with foliage, the velocity distribution below the top of the canopy depends strongly on the relative location in the vegetation array and to a lesser extent on the pattern/density. At several locations, the distribution exhibits a minimum value at the level of the foliage, with the velocity reduction being more severe in between the elements’ alignment.4. The type of foliage of similar size, i.e., dense/rigid or sparse/flexible, appears to be of secondary importance concerning main features of the velocity distribution, such as the maximum velocity defect at the foliage level and the shear velocity offered to the upper free layer. Considering the above, a new model of velocity distribution in presence of compound vegetation was deemed necessary. Two new analytical models are presented: (a) for vegetation with spherical foliage, developed in the context of research collaboration by Chinese researchers based on preliminary experiments of this work, and (b) a more general, simplified model aimed at being applicable to any vegetation geometry. In the first one the water column is divided into three layers (stem, canopy and surface layer, as shown by the blue lines in Figure 8), and the lower two layers into two sublayers each. Different analytical expressions are used for the velocity distribution in each layer/sublayer, as described in Wang et al. (2023). Good agreement of this model is found with preliminary results for spherical foliage, however there are considerable limitations for its wider application as it refers to a specific geometry and the values of several parameters were determined to achieve best fit with the experimental data. Figure 8.Sketch of vegetation with compound elements and layers of analytical models In the second model, a similar procedure was followed, separating the flow into three zones, as shown by the red lines in Figure 8, applying a combination of existing theories. Near the bed, (Zone 1) a constant velocity is assumed, derived from the simplified momentum equation in the flow direction, which expresses the balance between the gravity component and the vegetation drag, ignoring the Reynolds stress, in accordance with Huai et al. (2009), Nepf (2012b) and given as: (5)where d0 is the stem diameter, m is the vegetation density (number of vegetation elements per unit bottom area), Cd is the drag coefficient, S0 is the slope of the channel and u0 is the velocity. In the surface flow zone (Zone 3), where the velocity increases up to the free surface, the logarithmic law (equation 1) is implemented in combination with Raupach's theory (1994) for determining the parameters d and z0 by means of equations (2), (3) and (4). The logarithmic velocity distribution seems to be the most suitable at least for the unimpeded flow layer, having been used by many researchers, such as Klopstra et al. (1997), Shi & Hughes (2002), Ghisalberti & Nepf (2002). An additional advantage of this distribution is its validity well inside the vegetation, i.e. below its top. In the intermediate zone (Zone 2), where the velocity decreases until its minimum located behind the bulkier part of the vegetation (foliage), a series solution is adopted following Wang et al. (2023), herein using the first three terms: (6) (7)where k1 denotes the modified von Karman coefficient and z1 the vertical coordinate (see Figure 8), which starts at the interface of Zones 1 and 2. This interface depends on the height of Zone 1 which is less than the height of the cylindrical element (stem) due to the intrusion of vortices from the foliage zone which cause a reduction in speed. According to Nepf (2012b) the penetration length scale δe (eddy penetration depth) in submerged simple cylindrical vegetation is: (8)Obviously, Eq. (8) is applicable only for Experiments 1 and 2 of this thesis. In fact, it yields large values of δe exceeding the height of the rods (0.4 m and 0.2 m, respectively), confirming the proposal of Nepf that for sparse vegetation the logarithmic velocity profile spans practically the entire flow depth. For vegetation with complex geometry a modified form is suggested: (9)where the factor ω depends on the vegetation geometry. For the present experiments, the value ω= 0.027 was chosen, which best matches the velocity profiles of the experiments and is close to literature values (Wang et al. 2023).The difference from the layer division of the previous model is that the thickness of Zones 2 and 3 is not predetermined: the zone of decreasing velocity is assumed to extend higher into the foliage and the zone of unimpeded flow lower into the vegetation, until the respective distributions meet defining the velocity minimum. It should be noted that in the case of simple vegetation elements, such as in the present experiments 1 and 2, it is assumed that zone 2 does not exist. The agreement of the new analytical model with the experimental data is presented below in Figure 9 and is deemed satisfactory.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Mavrommatis, Aristotelis
- Μαυρομμάτης, Αριστοτέλης
Subjects
dc:subject × 7Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/55975