Drag Coefficients of Debris Accumulations
Publication: Journal of Hydraulic Engineering
Volume 150, Issue 4
Abstract
This study experimentally investigates the hydrodynamic forces acting on wood debris accumulations. Tests were performed under steady subcritical flow conditions by letting debris elements accumulate against an obstacle. We analyze the influence of a variety of variables involved in the problem, namely the Froude number, the blockage ratio, the geometric parameters of the dam, its porosity, and the debris type. Resulting forces are expressed through a typical drag coefficient formula, where the coefficient depends on the Froude number and the blockage ratio, but is independent of the geometric characteristics of the debris accumulation. The model is validated against literature data, providing both best-fit and safe-side predictors.
Practical Applications
Wood debris can build up on bridge piers and decks, which increase the water’s force on these structures. In this paper, we offer a straightforward method to estimate these forces. We have calibrated an empirical model using data from laboratory experiments and checked its accuracy against information from existing studies. With this model, one can input the basic characteristics of the flowing water (such as its depth, width, and speed) and the size of the debris pile facing the flow. The model then provides the drag coefficient, which, in turn, allows determining the drag force acting on the debris pile.
Introduction
During flood events, floating wood debris are entrained in rivers and transported downstream (Keller and Swanson 1979; Diehl and Bryan 1993; Lyn et al. 2003; Schmocker and Hager 2011). Drift can accumulate at bridge piers or superstructure: the process is known as debris damming (Yeh et al. 2014; Stolle et al. 2018). Debris dams produce higher backwater rise (Schmockers and Hager 2013; Schalko et al. 2018), exacerbate the maximum scour depth at bridge piers (Melville and Dongol 1992; Pagliara and Carnacina 2011), and cause additional hydrodynamic forces on the structure; the combination of these effects may lead to bridge failure.
Estimation of drag forces in the presence of debris requires two modeling steps: (1) definition of the shape and extension of the accumulation in front of piers and/or deck of the bridge; and (2) calculation of the hydrodynamic force over such area. In this paper, we focus on the second step; for the accumulation geometry the reader can refer to Schmocker and Hager (2011), Gschnitzer et al. (2013), Schalko et al. (2019), Panici and de Almeida (2020), and De Cicco et al. (2020).
Some literature studies have been devoted to the determination of drag force on debris accumulations (Wellwood and Fenwick 1990; Parola et al. 2000; Panici and de Almeida 2018; Mauti et al. 2020). An operational synthesis of such studies is proposed by Cantero-Chinchilla et al. (2018). In most experimental campaigns, an idealized rigid obstacle was fixed to a bridge pier or superstructure model and only few authors used a woody debris accumulation in their test programs (Panici and de Almeida 2018; Parola et al. 2000). Most of the cited works propose to estimate forces through typical drag relations (see the next paragraph for details); drag coefficients ranging from 0.5 to 3 are indicated, but no explicit relations with control variables are provided.
In this work, we present results from 165 laboratory tests over wood accumulations aimed at defining a model for drag coefficient estimation as a function of the main geometric and kinematic control parameters. The model is validated through equivalent literature data (Panici and de Almeida 2018; Parola et al. 2000); for the sake of brevity in the following these two papers will be referred as PdA18 and Par00, respectively.
Conceptual Framework
The classic (aeronautic) approach evaluates drag forces aswhere = force; = fluid density; = projection of the obstacle over a plane orthogonal to the flow direction; and = “undisturbed” reference flow velocity. For sufficiently high Reynolds numbers (defined by the velocity and some proper length scale of the obstacle), the drag coefficient, , should only depend on the shape of the obstacle, including its roughness. In cases involving accumulations of wooden debris we may also expect some dependence on the porosity of the obstacle. For river hydraulic applications of Eq. (1) one must also consider the effects of gravity and of flow lateral confinement, through the Froude number (where is gravity and is the reference water depth) and the blockage ratio , where is the flow area related to water depth and , being the flow rate. By increasing and/or the difference between upstream and downstream water elevations also increases, affecting the corresponding pressure distribution on the obstacle, which, in turn, determines the drag force. It is well known [Chow 1973, Eqs. (17)–(27)] that ideal limiting combinations of and are defined by
(1)
(2)
Eq. (2), holding for rectangular cross sections, identifies conditions for the restricted section to become a control for the flow, with a subcrical–supercritical transition at the obstacle.
From a different perspective, for low and values the water elevations upstream and downstream the obstacle are very similar, so that depth and velocity values in either position can be equivalently assumed as reference condition in (1). However, for higher values of and the choice of reference conditions becomes not obvious. In this work we only consider subcritical flows; consequently, we follow the approach of PdA18, and adopt the downstream values as reference conditions, as they do not depend on the presence of the obstacle (and represent values without the obstacle). The choice of upstream values as reference is also possible (see, for example, Par00 and Qi et al. 2014): however, application of consequent drag models requires the additional calculation of the backwater across the obstacle, which introduces additional uncertainties.
Finally, an alternative approach to Eq. (1) involves decomposing the force into two distinct components: a hydrostatic pressure force associated with streamwise variation of flow depth, and a dynamic force associated with flow velocity (Par00; Qi et al. 2014). We also tested such approach, but we do not consider it advantageous in terms of both accuracy and simplicity of use (see analyses in the Supplemental Materials).
The conceptual framework is synthetized in the following points.
•
is expected to be a function of the shape of the accumulation; we parametrize shape through three parameters: height, ; width, ; and depth, (see next section for exact definitions).
•
may also depend on the accumulation density, ¸ defined as the ratio of the net wood volume with respect to the gross volume occupied by the accumulation.
•
may also depend on the roughness of the accumulation. We have chosen not to incorporate this parameter for two primary reasons: (1) it is not readily apparent how to quantify the roughness of a self-formed wood accumulation, nor could we manipulate it during our experimental investigations; and (2) findings from the study by Pagliara and Carnacina (2010) concerning erosion on artificially assembled accumulations suggest that “roughness (…) has only a minor impact on the erosion process for situations involving limited constriction and low flow intensities.”
•
For relatively low Fr and BR values, well below limiting conditions (2), upstream and downstream conditions are very similar so that both conditions can be equivalently chosen as reference values, while we may expect that only depends on the shape properties of the obstacle (and, if the case, on the Reynolds number, Re).
•
As the Fr–BR combination approaches the limiting curve, we may expect that the effect of such parameters on becomes more relevant.
The resulting dimensionless relation between the drag coefficient and other variables is, therefore
(3)
We organized the experimental campaign and comparisons with literature data along this conceptual model.
Experimental Setup
Experiments were conducted at the Fantoli Hydraulics Laboratory of Politecnico di Milano, in a 7.0 m long, 0.5 m wide, and 0.60 m deep plexiglass laboratory rectangular flume with zero bottom slope. Water discharge was measured by magnetic flowmeter. The downstream condition was controlled by movable bars at the end of the flume. Water depths were measured by 11 piezometers along the flume.
A rectangular frame with a nylon thread mesh inside was located at the centerline of the flume, 4.5 m downstream from the inlet section. The frame was connected with a dynamometer positioned above the flow, allowing for measurements of the horizontal component of the force on the frame. The dynamometer was calibrated at the beginning of each experimental series (details in the Supplemental Materials); the standard deviation around the calibration fitting is reported together with measured forces in the Supplemental Materials.
The net volume of the elements composing the accumulation was a control parameter measured prior to the experiment. The debris elements were introduced into the water upstream of the obstacle and allowed to accumulate against the frame (Fig. 1); any elements that escaped from the accumulation were retrieved downstream and subsequently reintroduced into the upstream flow. At first, force was measured over the empty frame, to estimate the fraction of the force due to the frame; in this way we could evaluate the net force over the accumulation only. The idea is that forces over debris accumulation do not depend on the shape of the underlying structure (pier, deck), while forces over the portion of the structure that are exposed to flow can be estimated separately.
Experimental tests were conducted using nonuniform natural debris elements, both with and without branches. Four types of debris elements were tested, with different densities, lengths, and shape, from almost cylindrical to mildly branched: such properties are indeed relevant control parameter for the formation of the accumulation (PdA18), and indeed our results suggest that they affect porosity ; however, we could not detect any correlation among and debris type: therefore, most of the experiments were carried out mixing all the debris elements. The length of sticks was between 7 and 13 cm, while their diameter varied from a few millimeters up to 10 mm.
Each test was conducted introducing debris elements into the flume at a distance 2.5 m upstream of the frame. After the debris dam had formed and the flow conditions had stabilized, water depths and drag force were recorded. The reference (downstream) water level was set as the average among three piezometers (P2-P3-P4 in the Supplemental Materials) placed within 2.05 and 1.45 m downstream of the obstacle.
Four Gopro Hero Black cameras were used to identify the geometry of the debris accumulation. Two frontal cameras were placed 1.30 m downstream of the obstacle, one above and one below the water surface. The third camera was installed looking vertically 80 cm above the debris dam. The last camera was positioned sideways, lined up with the debris jam, 80 cm away from the side of the flume. A remote control with wireless connection allowed simultaneous shooting. Prior to experiments, a squared mesh chessboard was put in water at the position of the frame for pixel-length conversion. The geometric parameters of the debris dam were computed by composing information from the four cameras. Details on image processing and consequent derivation of the geometry of the debris accumulations are provided in the Supplemental Materials. We here limit the explanation by saying that and (see Fig. 1) are evaluated as median values of the corresponding dimensions over the cross-stream projection of the submerged portion of the dam. Thickness is evaluated as the ratio of the gross submerged volume of the accumulation divided by the area of the cross-stream projection.
The experimental investigation consists of 165 tests carried out under steady and subcritical flow conditions. Tests were conducted maintaining the downstream water depth equal to approximately 0.25 m and varying the flow discharge from 15 to and the net debris volume from 145 to . As mentioned above, the Froude number was calculated with the reference (downstream) flow velocity, , and water depth, : consequently, was defined a priori on the basis of the boundary conditions (range: ). Differently, ranges for the other dimensionless variables resulted from the actual shape of the self-formed accumulation (; ; ; ; ), where the Reynolds number was calculated with the reference flow velocity, , and as length scale. The complete data set is available in the Supplemental Materials. Fig. 2 shows present investigated condition in the plane, together with literature data from PdA18 and Par00.
Results
As explained, we could only fix among the control parameters in Eq. (3). Single-parameter analysis was, consequently, not possible. We perform multivariate regressions of the data, testing various regression function models. As a result of the exercise, we could not detect any dependence of on shape parameters and . We also could not enhance any dependence on the Reynolds number, but this was expected due to its limited range of variation within our tests; however, we should anyway expect a limited effect of the viscosity for a turbulent flow around a blunt rough obstacle. Finally, we also did not detect any effect of density . This result is not aligned with findings of Par00, who indicated that “porosity of the debris was found as one of the most relevant variables affecting the overload of bridge piers due to debris accumulations.” It should be, however, noticed that Par00 used thin porous plates as idealized debris accumulations: it is reasonable that effects of porosity diminish with the thickness of the obstacle, which finally behaves as an impervious body. Such behavior is also confirmed by Pagliara and Carnacina (2010) with respect to effects of accumulation porosity on scour around piers.
As expected, we detected a positive correlation among the drag coefficient and both the Froude number and the blocking ratio; a good interpolation formula of our data issuggesting an asymptotic value for vanishing and , and increasing drag coefficient when and/or increase. Fig. 3 shows the comparison between Eq. (4) and selected subsamples of data for nearly constant values of [Fig. 3(a)] and [Fig. 3(b)]. The tendency of to increase for increasing value of can also be detected in PdA18. A direct comparison with the findings of Par00 is not possible, due to their different approach to the problem.
(4)
Fig. 4(a) shows the measured–calculated correlation for the whole experimental data set. Most of the data fall within the deviation from the perfect agreement line. Finally, Fig. 4(b) plots contour lines for according to Eq. (4) in the plane. The figure allows us to observe the sensitivity of the model to the variables. Specifically, the region enclosed by the axes and the line can be regarded as an area of insensitivity concerning variations of and . This is because the drag coefficient remains within 20% of the asymptotic limit, , in this range.
Validation
Eq. (4) has been validated against available literature data for drag forces over wood debris accumulations (559 values from PdA18 and 21 values from Par00; corresponding ranges for and in Fig. 2; complete data set in the Supplemental Materials). Fig. 5(a) shows the measured–calculated comparison for forces: overall the model correctly reproduces experimental data, with a tendency of underestimating the highest values from PdA18 and Par00. The dotted lines parallel to the perfect fit line contain 98% of the data, excluding only 1% of the point on each side. The solid line represents perfect agreement. Owing to the structure of the regression model, dashed lines would also correspond to lines of perfect agreement in plots with different values for . The dotted line corresponding to can be viewed as a (safety side) envelope of experimental data. Notice that when present data only are considered, such an envelope can be restricted to .
Some statistics for the regression are reported in Table 1. Eq. (4) reproduces data with negligible distortion on the average. Due to its structure, the model cannot explain experimental values lower than 1.5; maximum calculated values also significantly differ from experimental ones. However, the mean percentage absolute error is relatively low (16%).
Experiment/model | Drag coefficient min/average/max | Force mean percentage error | Force mean percentage absolute error |
---|---|---|---|
Experiments | — | — | |
Eq. (4) | 0.01 | 0.16 | |
0.21 |
In addition to Eq. (4) we tested the performance of a zero-variate model, namely = constant = . This assumption neglects the amplification of the drag coefficient for increasing and . Results from the exercise can be appreciated in Fig. 5(b) and the last line of Table 1. Overall, the result is acceptable, with an obvious tendency to underestimate the largest values, which can be recognized both from the plot and from the average statistics in Table 1. The mean percentage absolute error is only 30% higher than that derived from Eq. (4). Having neglected the corrective effects of and , a value of as large as 3.1 is required for a 99% envelope of the experimental data.
Discussion
From a quantitative point of view, a constant model, possibly slightly higher than 1.5, does not perform significantly worse than Eq. (4), being thus preferable for its simplicity. However, we claim that our regression model should be preferred, as it indicates the role of Froude number and blockage ratio as limiting conditions of Eq. (2) are approached.
From a different perspective it should be noticed that, in practical application, the proper definition of expected dimensions of the accumulation (and the related probability) is the most critical issue for design and/or verification scenarios (see introduction for references). It is reasonable to assume that uncertainties related to the characteristics of the accumulation are much larger than those related to the calculation of the drag force. Therefore, we propose to use Eq. (4) with , which represents a best fit estimate for forces, while safety factors may be introduced by assuming safety side evaluations of the dimension of the debris accumulation.
Conclusions
The aim of this work is to provide an operational tool to compute drag forces on (given) debris accumulations, typically over bridge piers and/or superstructures. We structured laboratory experiments within a conceptual framework, found a suitable regression of the data, and validated the result over two independent literature series. Tests were limited to subcritical flow conditions. Findings can be synthetized by the following points.
•
Force over the accumulation can be evaluated by means of a classic drag coefficient formula, where downstream quantities are conveniently used as a reference condition for a subcritical flow.
•
We could not detect any dependence of the drag coefficient on the shape of the accumulation, its porosity, and the characteristics of individual debris pieces. In this way, the accumulation is characterized by its frontal area only. All further features, which are unavoidably variable in both laboratory and real self-formed accumulations, are probably the reason for the relatively high scatter of resulting drag coefficients.
•
We could identify a positive effect of the Froude number, Fr, and blockage ratio, BR, on the drag coefficient. For the lowest values of Fr and BR a best fit for the drag coefficient is , and a safe side envelope is . For higher values of Fr and BR experimental data indicate values as high as ; however, we do not know the extent of such an increase when Fr and BR approach limiting conditions where the flow is forced to a subsupercritical transition.
•
We were able to provide a prediction model for both the best fit and envelope limits of the experimental data within a relatively large domain of the independent variables.
The main limitation of present study is that it does not provide indications for flows close to the transition from subcritical to supercritical conditions. In such a situation, however, the proposed conceptual model fails, as the flow is no longer controlled by downstream parameters. On the other side, it is not trivial to structure an experimental campaign with self-formed accumulations under such conditions.
Supplemental Materials
File (supplemental materials_jhend8.hyeng-13865_ballio.zip)
- Download
- 756.30 KB
Data Availability Statement
All data, models, and code generated or used during the study appear in the published article.
Acknowledgments
Authors acknowledge the contributions of Marco Negri and Luca Martinelli for the support in designing and running the experimental campaign. This study was carried out within the RETURN Extended Partnership and received funding from the European Union Next-GenerationEU (National Recovery and Resilience Plan–NRRP, Mission 4, Component 2, Investment 1.3–D.D. 1243 2/8/2022, PE0000005).
References
Cantero-Chinchilla, F. N., G. A. M. De Almeida, and M. Escarameia. 2018. Assessing the effects of debris accumulations at river bridges. London: Univ. of Southampton.
Chow, V. T. 1973. Open channel hydraulics. New York: McGraw Hill.
De Cicco, P. N., E. Paris, L. Solari, and V. Ruiz-Villanueva. 2020. “Bridge pier shape influence on wood accumulation: Outcomes from flume experiments and numerical modeling.” J. Flood Risk Manage. 13 (2): e12599. https://doi.org/10.1111/jfr3.12599.
Diehl, T. H., and B. A. Bryan. 1993. “Supply of large woody debris in a stream channel.” In Proc., 1989 National Conf. of Hydraulic Engineering, 1055–1060. New York: ASCE.
Gschnitzer T., B. Gems, M. Aufleger, B. Mazzorana, and F. Comiti. 2013. “Physical scale model test on bridge clogging.” In Proc., 35th IAHR World Congress, edited by H.-W. Zhaoyin and S. Jizhang, 1–11. Beijing: Tsinghua University Press.
Keller, E. A., and F. J. Swanson. 1979. “Effects of large organic material on channel form and fluvial processes.” Earth Surf. Processes 4 (4): 361–380. https://doi.org/10.1002/esp.3290040406.
Lyn, D., T. Cooper, Y. Yi, R. Sinha, and A. Rao. 2003. Debris accumulation at bridge crossing: Laboratory and field studies. Washington, DC: USDOT, Federal Highway Administration.
Mauti, G., J. Stolle, T. Takabatake, I. Nistor, N. Goseberg, and A. Mohammadian. 2020. “Experimental investigation of loading due to debris dams on structures.” J. Hydraul. Eng. 146 (5): 04020029. https://doi.org/10.1061/(ASCE)HY.1943-7900.0001731.
Melville, B. W., and D. M. Dongol. 1992. “Bridge pier scour with debris accumulation.” J. Hydraul. Eng. 118 (9): 1306–1310. https://doi.org/10.1061/(ASCE)0733-9429(1992)118:9(1306).
Pagliara, S., and I. Carnacina. 2010. “Temporal scour evolution at bridge piers: Effect of wood debris roughness and porosity.” J. Hydraul. Res. 48 (1): 3–13. https://doi.org/10.1080/00221680903568592.
Pagliara, S., and I. Carnacina. 2011. “Influence of wood debris accumulation on bridge pier scour.” J. Hydraul. Eng. 137 (2): 254–261. https://doi.org/10.1061/(ASCE)HY.1943-7900.0000289.
Panici, D., and G. A. M. de Almeida. 2018. “Formation, growth, and failure of debris jams at bridge piers.” Water Resour. Res. 54 (9): 6226–6241. https://doi.org/10.1029/2017WR022177.
Panici, D., and G. A. M. de Almeida. 2020. “Influence of pier geometry and debris characteristics on wood debris accumulations at bridge piers.” J. Hydraul. Eng. 146 (6): 04020041. https://doi.org/10.1061/(ASCE)HY.1943-7900.0001757.
Parola, A. C., C. J. Apelt, and M. A. Jempson. 2000. Debris forces on highway bridges. Washington, DC: National Cooperative Highway Research Program, Transportation Research Board.
Qi, Z. X., I. Eames, and E. R. Johnson. 2014. “Force acting on a square cylinder fixed in a free-surface channel flow.” J. Fluid Mech. 756 (Oct): 716–727. https://doi.org/10.1017/jfm.2014.455.
Schalko, I., L. Schmocker, V. Weitbrecht, and R. M. Boes. 2018. “Backwater rise due to large wood accumulations.” J. Hydraul. Eng. 144 (9): 04018056. https://doi.org/10.1061/(ASCE)HY.1943-7900.0001501.
Schalko, I., L. Schmocker, V. Weitbrecht, and R. M. Boes. 2019. “Laboratory study on wood accumulation probability at bridge piers.” J. Hydraul. Res. 58 (4): 566–581. https://doi.org/10.1080/00221686.2019.1625820.
Schmocker, L., and W. H. Hager. 2011. “Probability of drift blockage at bridge decks.” J. Hydraul. Eng. 137 (4): 470–479. https://doi.org/10.1061/(ASCE)HY.1943-7900.0000319.
Schmocker, L., and W. H. Hager. 2013. “Scale modeling of wooden debris accumulation at a debris rack.” J. Hydraul. Eng. 139 (8): 827–836. https://doi.org/10.1061/(ASCE)HY.1943-7900.0000714.
Stolle, J., T. Takabatake, I. Nistor, T. Mikami, S. Nishizaki, G. Hamano, H. Ishii, T. Shibayama, N. Goseberg, and E. Petriu. 2018. “Experimental investigation of debris damming loads under transient supercritical flow conditions.” Coastal Eng. J. 139 (Sep): 16–31. https://doi.org/10.1016/j.coastaleng.2018.04.026.
Wellwood, N., and J. Fenwick. 1990. “A flood loading methodology for bridges.” In Proc., 15th Australian Road Research Board Conf., 315–341. Melbourne, Australia: Australian Road Research Board.
Yeh, H., A. R. Barbosa, H. Ko, and J. G. Cawley. 2014. “Tsunami loadings on structures: Review and analysis.” Coastal Eng. Proc. 1 (34): 4. https://doi.org/10.9753/icce.v34.currents.4.
Information & Authors
Information
Published In
Copyright
This work is made available under the terms of the Creative Commons Attribution 4.0 International license, https://creativecommons.org/licenses/by/4.0/.
History
Received: Aug 4, 2023
Accepted: Feb 13, 2024
Published online: May 10, 2024
Published in print: Jul 1, 2024
Discussion open until: Oct 10, 2024
Authors
Metrics & Citations
Metrics
Citations
Download citation
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.