EXPERIMENTAL AND NUMERICAL INVESTIGATIONS ON WOOD ACCUMULATION AT BRIDGE PIERS WITH DIFFERENT SHAPES Dissertation submitted to and approved by the Department of Architecture, Civil Engineering and Environmental Sciences University of Braunschweig – Institute of Technology and the Department of Civil and Environmental Engineering University of Florence in candidacy for the degree of a Dottore di Ricerca in Civil and Environmental Engineering by Pina Nicoletta De Cicco born 28/11/1985 from Foggia, Italy Submitted on 28.02.2017 Oral examination on 09.05.2017 Professorial advisors Prof. Luca Solari Prof. Hans Matthias Schöniger 2017 Experimental and numerical investigations on wood accumulation at bridge piers with different shapes De Cicco Pina Nicoletta (Ph.D. Candidate) Tutors: Prof. Luca Solari Prof. Hans Matthias Schöniger Co-tutors: Prof. Enio Paris Ph.D. Virginia Ruiz-Villanueva Abstract The awareness of in-channel wood transport in rivers changed significantly with the historical contest. In 18th century logs were transported from forests to sawmills and pulp mills by the natural streamflow of the waterways. Later the waterways were substituted with the railroads and the in-channel wood became only a natural river component. In last decades, the increasing abandonment of rural lands has caused significant growth in the total volume of available wood to be transported by the flow, especially during flood events. In-channel wood has become an additional component in potential hazard evaluation, in particular when it leads to the obstruction of hydraulic structures, e.g. the bridges. Wood accumulation at bridges exerts additional forces to the structures and aggravates local scouring around piers, which may result to bridge failure. Moreover, it may considerably reduce the flow opening causing higher flow levels and inundation of nearby areas. On the other hand, increasing awareness of the importance of the ecological role of wood in rivers calls for a compromise between the preservation of river ecosystems and in-channel wood management strategies for the prevention of wood-related hazards. The present PhD research aims to enhance the knowledge on the process of interaction between wood and bridge piers. The two main objectives were first to find the wood accumulation probability (here called “blockage probability”) as a function of the bridge pier geometry (with a focus on non-standard pier shapes typical of historical cities), hydraulic conditions of the approaching flow, and wood transport regime, second to assess the capability of 2D and 3D numerical models in reproducing the interaction between wood and structures (i.e. the bridge pier). The combined experimental and numerical research approach is used. The thesis first presents a review of recent advances in research on wood accumulations at bridges that highlights main gaps in knowledge. Secondly, the formulation of the blockage probability function based on the analysis of the main variables that controls wood accumulation at a single-pier. Lastly, the experimental and numerical results are analysed and discussed. The experiments were done at the hydraulic laboratory of the Department of Civil and Environmental Engineering of the University of Firenze aimed at investigating the ii most critical conditions for wood accumulation at a single-pier for a combination of different pier shapes, wood transport regimes and Froude numbers (in sub-critical conditions). The numerical simulations were carried out both with 2D and 3D models. In particular, the 2D model was aimed at reproducing the experimental findings and to assess its strengths and weaknesses. The 3D model was applied to investigate the 3D character of the interaction between wood and bridge pier and at reproducing the 3D secondary flow field. The results showed that blockage probability at the flatter pier shape is three times greater than the triangular shaped piers, in congested wood transport regime (logs move as a single mass and are unable to move independently of each other) and at high Froude number (in this case Fr=0.5). In case of Ogival pier, zero blockage probability was found for both cases of Froude numbers. Potential flow analysis indicated that the lower curvature of the streamlines at the rounded pier favours the log sliding the pier. The comparison between experimental and 2D numerical simulations revealed strengths and weaknesses. Despite the capability of the 2D model in reproducing the log transport the discrepancy between experimental and the 2D numerical results showed the two-dimensional numerical model cannot correctly capture the complicated log travel path and thus the log blockage probability at the pier. The problem is related to inability of modelling the important secondary flows and the log-pier interactions. In the straight rectangular flume, in addition to the Reynolds number, stronger secondary currents occur due to the different roughness between the side-walls and the bed. The secondary flow strength is greater towards the walls or the centreline according to the ratio between the channel width and water depth. This means that the logs may move differently in the cross section affecting the probability to touch and stop at the central pier. Further aspects to be improved are the elastic collision between logs, the smooth surface of logs and the constant drag coefficient, the superposition of logs. The 3D confirmed the action of secondary currents on log motion. Furthermore, the 3D modelling allowed to reproduce the 3D character of the wood-pier interaction process as the logs that move along the vertical upstream face of the pier, the non-elastic collision between logs and between logs and the pier, and the skin friction of logs. Finally, one of the main novelties of the current research in the estimation of blockage probability represented by the definition of a new pier hydraulic-shape coefficient (𝑐 ) that takes into account the shape of the pier and the 2D velocity flow 𝑝𝑖𝑒𝑟 field upstream of the pier. The flatter is the pier shape, the higher is 𝑐 . The probability 𝑝𝑖𝑒𝑟 of logs blockage at the pier increases with increasing 𝑐 . These results suggest that the 𝑝𝑖𝑒𝑟 pier shape has an important effect on the flow field upstream of the pier and thus on the log motion and its blockage. The thesis was also successful in defining the joint blockage probability at a bridge pier for the prevailing variables used in the study. The concept should find applications both in research and practical situation. It is highly desirable and useful to state the total probability of log blockage at a bridge pier given a set of flow, log, and pier variables. To investigate log transport in rivers the thesis suggests a twostep approach. Step one is the use of a relevant 2D model as a good starting point that should give a general understanding of log transport combination. The main advantages are the robustness and the limited required CPU power. In the second step, a 3D log transport model could be applied to study the specific features and to explore the deviations between the two models. The latter results could sever for calibration of selected the 2D. The models used in the present study (Wood-Iber & Flow3D) showed promising results within the constraints of the model applicabilities iii Acknowledgements I wish to thank all the people, whose comments, knowledge, experience have been precious for this thesis. I am first grateful to my supervisors. The constant guide of Prof. Luca Solari during all the phases of my thesis and his sincere support and encouragement. He has been of great help, especially when “the road diverged in a wood…”. My warm thanks goes to Prof. Enio Paris who supported and believed in my research topic from the beginning, infusing his enthusiasm and knowledge. I would like to thank Prof. Matthias Schöniger, who gave me interesting inputs, helping me to see things from the “basin scale” point of view in wood transport. I would like to express my gratitude to Dr. Virginia Ruiz-Villanueva, her precious knowledge, kindness, professionality and humanity made me honoured to work with her. The time spent with her at the Dendrolab in Bern has been an extraordinary opportunity for me, so I wish to thank the dean of the Dendrolab, Markus Stoffel, who gave me this opportunity and all the guys and researchers at Dendrolab, they made me feel at home. My warm and sincere thanks to Prof. Bijan Dargahi for his guidance and generous hospitality during my stay at KTH in Stockholm, and for his great work of review of the present manuscript. I appreciated a lot his energy and love for research. I wish to thank also his wife Shahrzad for the hospitality, and for the nice shared moments drinking a good rose tea and listening to the Iranian music. A special thank goes to Prof. Hocine Oumeraci, who has been a precious guide especially during my first year at the Leichtweiss-Institut Hydraulic Engineering and Water Resources (LWI), in Braunschweig (Germany). He provided me with the first research tools and showed a continuous interest for my research. I wish to thank the entire staff of the Leichtweiss-Institut Hydraulic Engineering and Water Resources for the hospitality during my stay in Braunschweig. I would like to thank Prof. Francesco Comiti for his precious comments and suggestions in reviewing the manuscript. Special thanks go to the technicians of the Hydraulic laboratory in Firenze, Mauro Gioli and Muzio Mascherini, and to the BSc students for their support during the experimental phases. I wish to thank both the Department of Civil and Environmental Engineering (University of Firenze) and the Department of Architecture, Civil Engineering and Environmental Sciences (University of Braunschweig – Institute of Technology) for this opportunity. Finally, I would like to thank my beloved family and my sincere friends. iv v Summary List of Figures ............................................................................................................... vii List of Tables .................................................................................................................. xi List of Symbols .............................................................................................................. xii 1 Introduction ............................................................................................................ 15 1.1 Motivation ........................................................................................................ 15 1.2 Objectives and methodology ............................................................................ 17 1.3 Thesis structure ................................................................................................ 19 2 Literature review .................................................................................................... 21 2.1 Wood accumulation at bridges ......................................................................... 21 Wood jam formation ................................................................................ 21 Type, shape and geometry of wood accumulation at bridges .................. 22 2.2 Factors influencing wood accumulation at bridges .......................................... 24 2.3 Impacts of wood accumulation at bridges ........................................................ 26 2.4 Recent advances in flume experiments and numerical models on wood accumulation at bridges ....................................................................................................... 28 Flume experiments ................................................................................... 28 Numerical models .................................................................................... 34 2.5 Knowledge gaps ............................................................................................... 38 3 Theoretical background ........................................................................................ 40 3.1 The governing equations for wood transport ................................................... 40 3.2 Collision between logs ..................................................................................... 45 3.3 Wood budget .................................................................................................... 46 3.4 Blockage probability ........................................................................................ 47 4 Methodology and approach................................................................................... 51 4.1 General aspects ................................................................................................ 51 4.2 The concept of blockage probability ................................................................ 51 4.3 Experiments ..................................................................................................... 52 Setup......................................................................................................... 53 Measurements .......................................................................................... 56 Flow variables .......................................................................................... 57 Test procedure and blockage probability ................................................. 58 4.4 2-D numerical model setup .............................................................................. 58 vi Geometry and model mesh ....................................................................... 59 Boundary conditions ................................................................................. 60 Modelling the logs .................................................................................... 60 Simulations ............................................................................................... 61 4.5 3-D model: Flow-3D ........................................................................................ 62 Model setup ............................................................................................................. 62 Geometry and mesh .................................................................................. 62 Initial boundary conditions ....................................................................... 63 Modelling the logs .................................................................................... 64 Simulation procedures .............................................................................. 65 5 Results ...................................................................................................................... 67 5.1 Experiments ...................................................................................................... 67 Blockage probability B vs. ratio of log discharge Qlog to the flow discharge Qflow .................................................................................................................. 67 Blockage probability B vs. ratio of log length Llog to the pier width wp .. 68 Logs accumulation size ............................................................................ 69 Effective (EA) and Potential (PA) Accumulation .................................... 71 Blockage probability B vs. Froude number Fr ......................................... 73 5.2 2D- numerical model ........................................................................................ 75 Hydraulic- Model calibration ................................................................... 75 Log motion calibration ............................................................................. 77 Potential Accumulation (PA).................................................................... 79 Blockage probability B vs. pier hydraulic-shape coefficient 𝐶𝑝𝑖𝑒𝑟 ......... 82 5.3 Joint Blockage Probability ............................................................................... 89 5.4 3D- numerical model ........................................................................................ 91 6 Discussion ................................................................................................................ 98 6.1 Thesis assumptions ........................................................................................... 98 6.2 Hydraulic interpretation of log movement at bridge piers ............................... 98 6.3 Comparison between experimental and numerical results and limitations .... 100 6.4 The effects of secondary flows ....................................................................... 103 6.5 Comparison with previous works ................................................................... 104 6.6 Insights into the physics of the problem ......................................................... 105 7 Conclusions ........................................................................................................... 107 8 Future work .......................................................................................................... 109 References .................................................................................................................... 111 Appendix ...................................................................................................................... 119 vii List of Figures Figure 1.1 Cases of wood accumulation at bridges during a flood: A) La Spezia, Italy, 2011 (picture from Comiti F.); B) bridge failure by wood Oklahoma, USA (picture from Bradley et al., 2005); C) Pamplona, Spain, 2013 (picture courtesy of Virginia Ruiz-Villanueva); D) bridge failure by wood, New York, USA (picture from Bradley et al., 2005); E) Kyushu, Japan, 2012. ................ 16 Figure 1.2 Graphical summary of the main goals and the methodology of the current research. ........................................................................................................... 18 Figure 1.3 The thesis outline. ....................................................................................... 20 Figure 2.1 Plan view scheme of the effective opening between bridge piers (a) and of the types of wood accumulation at bridges (b). ........................................... 23 Figure 2.2 Factors influencing wood accumulation at bridges. ................................... 24 Figure 2.3 Wood transport regimes. Logs at the initial time (left) and at a later time (right). (Picture from Braudrick et al., 1997). .................................................. 25 Figure 2.4 Debris mass at the water surface on a single circular pier and effective pier diameter (Melville and Dongol, 1992) (in the figure y=h=water depth). . 26 Figure 2.5 Wood accumulation at a rounded single pier in laboratory tests from Lyn et al. (2003). ..................................................................................................... 28 Figure 2.6 Accumulation of logs at a bridge with no pier in the flume experiments from Gschnitzer et al. (2013). Picture from WWR2015 (International Conference on Wood in World Rivers, 2015). ................................................ 29 Figure 2.7 Model wood accumulation upstream of the bridge piers (on left) and ineffective flow areas for wood accumulated at the bridge pier (on right) (Source: Parola et al., 2000). ............................................................................ 34 Figure 3.1 Scheme of the forces acting on a piece of wood located in water stream .. 42 Figure 3.2 Angle of channel bed in flow direction and orientation of the log respect to the flow direction. ........................................................................................ 43 Figure 3.3 Example of the relationship between log transport regime, the dimensionless water depth h* (d=𝐷𝑙𝑜𝑔) and the dimensionless force 𝛹. Source: Haga et al., 2002. (Assuming that the density of log is equivalent to the water, hc* is 1). .......................................................................................... 44 Figure 3.4 Sketch of the joint blockage probability distribution (tot B) from the marginal probability functions (B1, B2, B3, B4) ............................................... 50 Figure 4.1 Side and planimetric view of the laboratory channel. ................................ 53 Figure 4.2 Flume cross-section with the pier located in the channel centreline (left, dimensions are in millimetres) and the notation for pier width (Wp) and length (Lp) (right). ........................................................................................... 54 Figure 4.3 Beech wooden cylindrical dowels used in flume experiments. .................. 55 Figure 4.4 Cross sections and points for surface flow velocity measurements. .......... 56 Figure 4.5 Sketch describing the labelling rational for flume tests ............................. 58 Figure 4.6 Scheme of the log that crosses different mesh elements (a) and that falls into one single element (b). (Source: Bladè et al., 2016). ................................ 59 Figure 4.7 Unstructured mesh ...................................................................................... 60 Figure 4.8 Iber screen for wood input parameters. ...................................................... 61 Figure 4.9 Model of flume and detail of the gravel bed. ............................................. 63 viii Figure 4.10 Scheme of the control upward force applied to the log to reproduce the log release according to the experimental tests. ............................................... 65 Figure 4.11 Scheme the 3D numerical tests and number of logs. ................................ 66 Figure 5.1 Blockage probability for congested and uncongested wood transport, different pier shape configurations and Froude numbers. ................................ 68 Figure 5.2 Wood accumulation at the triangular pier shape in flume experiments (a), and in the real case (b) of the Ponte Vecchio in Florence after the flood of November 2016. ............................................................................................... 68 Figure 5.3 Blocking probability versus Llog/wp for Fr=0.5 and Fr=0.3. ....................... 69 Figure 5.4 Log accumulation volume [cm3] at bridge piers for congested and uncongested wood transport, different pier shape configurations and Froude numbers. ........................................................................................................... 70 Figure 5.5 Log accumulation volume (V ) relative to the total volume of the acc released logs (V ) for congested and uncongested wood transport, different tot pier shape configurations and Froude numbers. ............................................... 71 Figure 5.6 Potential (upper panel) and Effective (lower panel) Accumulation for different pier shape configurations and Froude numbers in congested transport regime. ............................................................................................... 72 Figure 5.7 Scheme of the log movement observed in flume experiment in case of Fr=0.5 and Fr=0.3. ............................................................................................ 73 Figure 5.8 Graphical summary of the experimental results.......................................... 74 Figure 5.9 Example of the log movement in the flume predicted by the 2D numerical model Iber. ........................................................................................................ 75 Figure 5.10 Comparison between the predicted and observed values of flow depth, expressed in cm (a) and depth averaged flow velocity, expressed in m/s (b) .. 76 Figure 5.11 Measured depth averaged flow velocities and calculated with numerical model along six cross sections for Fr=0.3 (upper panel) and Fr=0.5 (lower panel). ............................................................................................................... 77 Figure 5.12 Comparison between the experimental (right) and numerical values (left) of log centre advection velocity at Fr=0.5 using two different numerical approaches. ....................................................................................................... 78 Figure 5.13 Boxplot of the velocity of log centre in flume and numerical tests for both Froude numbers obtained with the dynamic method for computing the log centre velocity and Cd=1.4. ........................................................................ 78 Figure 5.14 Comparison between the Potential Accumulation (PA) of wood at different pier shapes in flume and in numerical tests (left) for Fr=0.3 and residuals of the data (right). .............................................................................. 80 Figure 5.15 Comparison between the Potential Accumulation (PA)of wood at different pier shapes in flume and in numerical tests (left) for Fr=0.5 and residuals of the data (right). .............................................................................. 80 Figure 5.16 Boxplot of PA (Potential Accumulation) in flume experiments (observed) and numerical tests (predicted) for Fr=0.3 (a) and Fr=0.5 (b). ...... 81 Figure 5.17 Comparison between observed (upper panel) and predicted (lower panel) depth averaged velocities for different Froude number.................................... 82 Figure 5.18 Smooth Contour fill plot of flow velocity for Fr=0.5 and rectangular pier shape (R0). ........................................................................................................ 83 ix Figure 5.19 Isovels plot for Fr=0.5 and rectangular pier shape (R0) and semi-circular pier shape (R1). The dashed lines indicate the Isovels i=0 and i=0.6∙U that ∞ delimit the low flow velocity area (ALFV). ....................................................... 84 Figure 5.20 Blockage probability B versus pier hydraulic shape coefficient 𝑐𝑝𝑖𝑒𝑟 for Fr=0.5 and all log size classes. ......................................................................... 85 Figure 5.21 Blockage probability B versus pier hydraulic shape coefficient 𝑐𝑝𝑖𝑒𝑟 for Fr=0.5 and large logs. ...................................................................................... 86 Figure 5.22 Blockage probability B versus pier hydraulic shape coefficient 𝑐𝑝𝑖𝑒𝑟 for Fr=0.5 and medium logs. ................................................................................. 86 Figure 5.23 Blockage probability B versus pier hydraulic shape coefficient 𝑐𝑝𝑖𝑒𝑟 for Fr=0.5 and small logs ....................................................................................... 87 Figure 5.24 Graphical summary of the 2D numerical simulations with Iber. ............. 88 Figure 5.25 Joint Blockage probability (totB) for the most influential independent variables for blockage at the bridge pier, in congested wood transport regime. .......................................................................................................................... 89 Figure 5.26 Predicted (black line) and observed (red dots) flow velocity in the cross sections upstream of the pier for Fr=0.3 (upper panel) and Fr=0.5 (lower panel). ............................................................................................................... 91 Figure 5.27 Local flow velocity values measured (observed) and simulated (predicted). ....................................................................................................... 92 Figure 5.28 Top (upper panel) and side (lower panel) view of the trajectory of one single log (LW1) and of the same log in presence of more logs (3, 6, 9, 15). . 93 Figure 5.29 Wood-pier interaction reproduced with Flow-3D, in the case of single log and submerged conditions (A), and more logs in semi-submerged conditions (B). .................................................................................................. 93 Figure 5.30 Top (upper panel) and side (lower panel) view of the trajectory of logs in “congested” transport simulation with 6 logs. ............................................. 94 Figure 5.31 Log centre orientation (upper panel) and correspondent drag coefficient (lower panel) plotted for one single log (LW1) and for the same log in presence of more logs (3, 6, 9, 15 logs). .......................................................... 94 Figure 5.32 Orientation of logs, with respect to the flow direction, before encountering the pier (A) and at the moment of the collision with the pier (B). .......................................................................................................................... 95 Figure 5.33 A) Normalized drag coefficient from simulation with Flow-3D compared with data from Gippel et al. (1992); B) elevation of the log centre versus the orientation respect to the flow direction; C) log orientation along the x coordinate of the flume. ................................................................................... 96 Figure 5.37 Graphical summary of the 3D numerical simulations with Flow-3D. ..... 97 Figure 6.1 Experimental observation on orientation and travelled path logs approaching the flat pier (upper panel) and the rounded pier (lower panel) at 4 different time intervals (Flow direction from right to left). .......................... 99 Figure 6.2 Streamlines at a flat pier shape (left) and semi-circular pier shape (right). Flow from right to left. ................................................................................... 100 Figure 6.3 Example of “Effective Blockage Probability” (MinBP) in flume experiments: the log stopped at the pier at the end of the tests. ..................... 101 Figure 6.4 Congested wood transport simulated with Flow-3D positioning the logs in two overlapped layers. ............................................................................... 103