Hydraulic and Morphological Consequences of Bank Protection Measures along the Jamuna River, Bangladesh

Transcription

1 京都大学防災研究所年報第 54 号 B 平成 3 年 月 Annuals of Disas. Prev. Res. Inst., Kyoto Univ., No.54B, 11 Hydraulic and Morphological Consequences of Bank Protection Measures along Jamuna River, Bangladesh Hao ZHANG, Hajime NAKAGAWA, Yasuyuki BABA, Kenji KAWAIKE, Md. Munsur RAHMAN* and Mohammad Nazim UDDIN** * Institute of Water and Flood Management, Bangladesh University of Engineering and Technology ** Department of Civil Engineering, Dhaka University of Engineering and Technology Synopsis Bank erosion is a severe problem in rivers of Bangladesh. Besides important scientific and engineering relevance, bank erosion also exerts significant social and economic impacts in this sourn Asian country. This paper describes typical bank protection measures as well as ir hydraulic and morphological consequences in lower part of Brahmaputra River (named Jamuna River in Bangladesh). Based on a series of field investigation results, attempts have been made to clarify mechanisms of bank erosion along this large alluvial river. Moreover, performances of existing bank protection measures are evaluated and possible solutions for furr enhancement are proposed. Special attention is paid to a historied and indigenous river training structure: Bandal. The recurrent use of Bandal-like structures is suggested for bank protection and channel stabilization of braided Jamuna River. Keywords: Brahmaputra/Jamuna River, bank protection, channel stabilization, field investigation, Bandal, spur dyke 1. Introduction The sourn Asian country: Bangladesh is one of most populated nations in world. The land of Bangladesh is covered by a complex river network system consisting of three major rivers: Ganges, Brahmaputra and Meghna, toger with ir numerous tributaries and distributaries. A brief overview of rivers in Bangladesh is referred to Oka (4). One important hydrological aspect of rivers in Bangladesh is that 9% of drainage basin of se rivers lies outside of country (FAP1, 1). Moreover, sediment within Bangladeshi rivers consists primarily of fine sands and silts with little clay matrix (Coleman, 199). Therefore, river banks are highly susceptible to erosion when flow conditions change. In past several decades, frequent flood and continuous erosion have consumed large areas of floodplains, made thousands of people homeless and destroyed a huge amount of infrastructures. The problems related to Brahmaputra River are particularly gigantic, and this river is selected as target for current study. The Brahmaputra River is named Jamuna River within border of Bangladesh. It originates from Kailas Range of Himalayas and flows across China, India and Bangladesh. The river meets with some major rivers in its lower part: Teesta, Ganges and Meghna and supplies sediment into one of world s largest deltas before finally entering Bay of Bengal. The location and catchment area of Brahmaputra/Jamuna River are shown in Fig

2 China Nepal Bhutan Brahmaputra R. India Ganges R. Jamuna R. India Catchment of Brahmaputra/Jamuna River River Bangladesh International border Bay of Bengal Fig. 1 Location and catchment of Brahmaputra/Jamuna River Myanmar As is known, Brahmaputra/Jamuna River ranks in top group of earth s large rivers in terms of both water and sediment discharges. In reach within Bangladesh, river wanders with a distance of approximately 4km and has a mean bankfull width of some 11km. The average annual flow discharge of Jamuna River is around,m 3 /s and sediment transported through river is about 59 million tons per year at Bahadurabad gauging station (Sarker et al., 3). The annual hydrograph of Jamuna River is characterized by low flows during winter dry season and high flows during summer time due to snowmelt in Himalayas and heavy local rainfall in monsoon. The river is braided with meta-stable islands and nodal reaches, mobile sandbars, shifting anabranches and severe bank erosion (Thorne et al, 1993). Systematic analysis of time-series of dry season satellite images and with supplement of available historical maps and aerial photographs, CEGIS reported that Jamuna River shows a persistent trend of westward migration (CEGIS, 4, 5, and 7). Analysis indicates that centerline of river has moved an average of 4.3km towards west since year 183 with a maximum westward movement of 13km at its norrn end. In recent years, Jamuna River is migrating westward at an average of 75m per year. Moreover, river is widening at an average rate of 145m/year, accompanied by annual creation of 19 hectares of char land from sediment trapped within channel. As one of most crowded countries in world and majority of population being still wholly dependent upon land holdings, bank erosion and migration of river have significant impacts on development of society and economy. According to combined analyses of population data with satellite images, it is found that during period of last 1 years, an average of almost 4, people were relocated every year due to bank erosion along Jamuna River (IWFM, 8). In order to protect lives and properties from frequent flooding, an earth embankment was built during later 195s and mid 19s along west bank of Jamuna River, extending for some km. The embankment is generally known as Brahmaputra Right Embankment (BRE). The on-going bank erosion by river, however, has led to breaches of BRE with attendant crop loss and damage to buildings and infrastructures as well as successive costly retirements of BRE over past several decades (Halcrow et al, 1994). Improvement of performance of BRE is refore an important part of flood protection and measures against bank erosion. In order to seek long-term strategy for protection of BRE, Government of Bangladesh has commissioned a series of studies since 199. A master plan was 478

3 formulated and construction of bank protection structures at different probable locations was suggested to save priority areas. From mid 199s, construction of major structures was started and re are now around 8 major bank protection structures along BRE. In general, se structures may be categorized into two kinds: one kind is intended to mainly strengn resistance of bank to be protected with insignificant interference on river flow, e.g. revetments. The or kind is aimed to decrease hydraulic impacts directly in front of protected area, e.g. spur dykes. It has to be mentioned that most of existing bank protection structures along BRE were designed using extrapolation methods based on research results and experiences of small rivers in or places such as some European countries. The resulted structures were generally very huge in both size and cost corresponding to scale of Jamuna River. Although positive roles of some of bank protection structures have been reported after an overall performance assessment based on available literatures, field investigations and discussions with engineers and stakeholders (BWDB, 1999), applicability and sustainability of se structures were still questionable for large Bangladeshi rivers. The fact is that most of m experienced frequent damages or failures after ir completion. On or hand, a locally developed structure: Bandal, deserves special attention. A Bandal structure may be simply described as a vertical screen mounted on a frame. The main construction materials are bamboo mats with a bundle of bamboo sticks. Bandal structures are quite cheap since bamboos are locally available and inexpensive labors are easily employable on site without any special training. When sediment-laden flow approaches Bandal, flow separation occurs: low sediment-concentrated flow at upper layer is diverted to main channel and is accelerated, resulting in bed degradation in main channel, while high sediment-concentrated flow at lower layer passes through Bandal and deposits behind it due to velocity reduction re. Bandal structures may be physically considered as combined structures of spur dykes and pile dykes (Zhang et al., 1). They have been successfully used in India subcontinent historically. Unfortunately, ir working principles are not well clarified and design guidelines are not available yet due to shortage of scientific evidences. In this paper, hydraulic and morphological consequences of typical bank protection measures in selected river reach are investigated. Based on investigation results, performances of se measures are evaluated. The mechanism of bank erosion in large alluvial rivers is discussed, improvement methods for existing bank protection structures are suggested and probable cost-effective solutions for training of Jamuna River are proposed.. Study sites and research methods Referring master plan report of Brahmaputra River training studies of Bangladesh government and based on first-hand knowledge and experiences from field visits, Sirajganj Hardpoint (revetment) and Betil/Enayetpur spur dykes were selected as target structures to investigate existing bank protection measures. These two sites ranked in top group of priority areas to be protected designated by government. Bandal structures near Jamuna Bridge at Randhunibari market are also included in study as a reference site. Field investigations to three sites were made in 8, 9 and 1. The locations of m are sketched in Fig.. No.8 No.7 No. No.5 Fig. Location of study sites (Left to Right: Study sites, Jamuna River and Bangladesh) 479

4 This study is conducted primarily on basis of literature review, field measurements, discussing with local people and collecting information from BWDB (Bangladesh Water Development Board), CEGIS (Center for Environmental and Geographic Information Services) and BUET (Bangladesh University of Engineering and Technology) with aid of our collaborators in Bangladesh. The authors have conducted a series of researches on spur dykes over past several years (e.g. Zhang et al.,, Zhang et al.., 7, Zhang and Nakagawa, 8 and Zhang et al., 9), knowledge acquired from those studies are partially extended to this research if necessary. Moreover, preliminary experimental and numerical resultss from authors research group on Bandal-like structures are also used as supplement information to understandd fundamentals underlying physical phenomena. As a field-based applied research, field trips and field measurement systems are of great importance. In this study, a measuring system consisting of an ADCP (Acoustic Doppler Current Profiler) and a GPS (Global Positioning System) is used to record 3D velocity vectors in river. The ADCP (1kHz, WH-ADCP Rio Grande by Teledynee RD Instruments) was mounted downward beneath water surface on a particularly designed plastic boat. And plastic boat was dragged by a local wooden boat during measurements. The ADCP and GPS are connected with a laptop installed with a real-timee data visualization and data analyzing/ post-processing program. A snapshot of field measurement is shown in Photo1, with specifications of measuring system. GPS Local boat Laptop Plastic boat Photo1 Field survey on Jamuna River ADCP Percentage finer (%) (μm) Fig.3 Sieve analysis result of sediment samples Before filed measurements, basic information on sediment properties and hydrological and morphological characteristicss related to study areaa is collected. The Jamunaa River is characterized by fine sedimentary environment with highly mobile beds and banks. The sediment is uniformly graded and has very low strength and transport resistance. Mean sediment sizes of river vary from um to 15um (FAP1, 1), showing insignificant changes from upstream to downstream. Sediment samples taken from floodplain near Sirajganj Town were analyzed during first field trip as shown in Fig.3. It is found obviously thatt sediment is much finer than that in typical Japanese or European rivers. The mean diameter is 181um, having a geometric deviation of On or hand, daily discharge at upstream Bahadurabad gauging station and monumented surveyed cross-sections (Section No. 5-No.8 as depicted in Fig..) near three study sites were obtained from BWDB are shown in Fig.4 and Fig.5. The hydrograph at Bahadurabad gauging station shown in Fig.4 ranges from year 1998 to year temporally. The low flow season and high flow period are evidently distinguished from year to year. The river generally peaks during period of July and August. Peak discharge is found to be generally around,cumecs, exhibiting insignificant changes over years. Extreme flood discharge occurred in 1998 and country suffered one of its worse ever floods between July and September in that year. 48

5 1 Discharge (m 3 /s) /1/1 1999/5/1 /9/7 //9 3//4 4/11/5 /3/ Date (Year/Month/Date) Fig.4 Daily discharge at upstream Bahadurabad gauging station Elevation (m) Elevation (m) Elevation (m) Elevation (m) Distance from left side milestone (m) Distance from left side milestone (m) Distance from left side milestone (m) Section No.8 Section No.7 Section No. Section No Distance from left side milestone (m) Fig.5 Temporal variation of typical cross-sectional profiles near study sites 481

6 The changes of bed elevation at four typical cross-sections in Fig.5 demonstrate highly active nature of mobile bed of Jamuna River in past several years. The locations of sections are depicted in Fig.. Section No.8 is in near upstream of Sirajganj Hardpoint. Section No.7 is very close to termination of Sirajganj Hardpoint. Section No. is in near downstream of Jamuna Bridge and Randhunibari Bandal sites and Section No.5 is very close to Betil and Enayetpur spur dykes. The river is obviously composed of numerous channels and se channels frequently shift year to year. In river, re are all kinds of sandbars and islands and some of which migrate frequently. With background information on sediment properties and flow discharges, morphological changes shown in Fig.5 might not be so much surprising. Neverless, difficulties lying in seeking suitable solutions to manage morphological processes really come to researchers and engineers as a surprise. Without a detailed understanding on mechanisms involved and more quantitative field data, decisions cannot be made confidentially and measures cannot be taken appropriately. 3. Revetments 3.1 Description of Sirajganj Hardpoint The old established town: Sirajganj is located in central Bangladesh, about 8km upstream of Jamuna Bridge and about 11km northwest of capital city: Dhaka. According to analyses results of satellite images, Sirajganj is observed to have highest erosion rate in catchment area (CEGIS, 7). The erosion of bankline since 195s has resulted in parts of historic town fronting directly onto river bank (Halcrow et al., 1994). In order to protect town, revetment works were launched by Bangladesh Government in 1998, known as Sirajganj Hardpoint. The hardpoint intends to control shape of BRE with lowest level of intervention to river flow. The hardpoint was constructed with cement concrete cubic blocks and consisted of a straight portion along BRE toger with a round upstream termination. The basics of Sirajganj Hardpoint are listed in Table1 and a photo is shown in Photo. Table 1 Basics of Sirajganj Hardpoint Length of revetment reach.55km Crest level over PWD 1.75m High flood level over PWD 15.75m Low water level over PWD.8m Apron setting level over PWD -4.m Design scour level over PWD -13.5m Thickness of apron 1.93m Side slope 1:3.5 Size of cement concrete block 55cm & 85cm Completion of hardpoint 1998 *PWD: Public Work Datum,.457m below seal level Photo Sirajganji Harpoint The Sirajganj Hardpoint is exposed to main anabranch channel of Jamuna River, receiving year-round hydraulic attacks. The hardpoint, especially area near upstream termination, has been almost damaged almost every year since 7. Velocity measurements were conducted during March -3, 8 and March 19, 9. In addition, bathymetry survey data around upstream termination was collected from branch office of BWDB. 3. Bed change around upstream termination The monthly change of bed elevation around upstream termination of Sirajganj Hardpoint from January 8 to February 9 was plotted in Fig.. The plot series provided valuable information on bed evolution characteristics in proximity of termination. Basically, bed morphology around termination was dominated by a scour hole and hole changed its shape as well as its bottom elevations almost all time even in dry season. The similarities in scour geometry and fluctuations in maximum scour depth from January to beginning of July indicated that transport of bed forms played an important role in bed evolution during low flow season. 48

7 N 77 January 1, 8 May 1, February 1, 8 June 19, March 9, 8 - July, April 11, 8 August 1, 8 Fig. Bed level change at termination (Unit: m) 483

8 N - 77 September, 8 January 11, October 8, 8 November 1, November, 8 Fig. Bed level change at termination (Ctd.) February 4, 9 Fig. Bed level change at termination (Ctd.) The scour hole exhibited significant changes during flood as shown in plots in August and in September. In particular, maximum scour depth was dramatically increased and location of deepest point also shifted a bit to downstream of termination. After flood, scour hole area was refilled with sediment and maximum scour hole depth returned to pre-flood level as obviously shown from plots after October. Although 8 flood enlarged scour hole, final scour hole in January of 9 was even much smaller and shallower than that in January of 8. It has to be mentioned that termination partially failed in September of 8 during flood. The severe scour at toe of termination resulted in stability problems of termination, i.e. sliding of apron and sand materials beneath it took place. The observations suggested that mitigation of scour due to flood and enhancement of termination against scour were probably keys to ensure effectiveness of Sirajganj Hardpoint. 484

9 3.3 Flow field at Sirajganj Hardpoint Although hardpoint did not exert direct influence on flow field, it altered flow field through morphological changes around it. The typical flow velocity patterns in year 8 and year 9 were plotted from Fig.7 to Fig.1 and compared with each or. The horizontal velocity vectors near water surface were shown in Fig.7 and Fig.8, toger with sketches of major sandbars based on field investigations and satellite images. In 8, approach flow channel was diverted into two parts by two sandbars. The flow attacked termination of hardpoint and resulted in two circulation flows in front of and behind termination, respectively. In 9, west part of big sandbar on east was washed away and width of approach flow channel was significantly enlarged. On or hand, a small sandbar appeared in approach flow channel just in front of termination. These changes affect a lot on flow structure near termination. The two circulations remained in 9, but became much weaker than those in 8. The flow in 9 was furr divided into two parts after passing hardpoint due to detachment of downstream sandbar from embankment. If one took a look at velocity field in vertical direction passing termination in Section P as shown in both Fig.8 and Fig.9, differences in flow structure could be evidently distinguished. A vortex occupied scour hole in front of termination in both years, but one in 8 was much stronger than that in 9. As this vortex was major engine for local scour around termination, it was n easily understandable that scour depth in 9 was smaller than that in 8 as have been confirmed by bathymetry data in Fig.. The observations indicated that movement of sandbars significantly influenced flow structure and bed morphology around hardpoint. Although problem related to hardpoint was a very local scale one, understanding migration characteristics of sandbars in a broader scale was of crucial importance. Moreover, horizontal and vertical velocity vectors indicated that local flow structure around termination was of three-dimensional nature. North (m) North (m) Vertical (m) East (m) Fig.7 Horizontal velocity.8m beneath water surface in 8 East (m) Fig.8 Horizontal velocity.8m beneath water surface in 9 North (m) Fig.9 Vertical velocity vectors at section P in 8 Vertical (m) Vortex Vortex.m/s Vortex.m/s North (m) Fig.1 Vertical velocity vectors at section P in 9 485

10 4 Spur dykes 4.1 Description of Betil/Enayetpur spur dykes The Betil and Enayetpur spur dykes are located in eastern part of Belkuchi and Chowhali Upazilla, about 5km south of Sirajganj town. The Betil and Enayetpur Bazar areas are historically well known for ir handloom textile production. There are lots of important public and private establishments such as schools, madrasahs, colleges, mosques, hospitals and industries. This area is one of high priority areas to be protected designated by government. Since 1914, bankline of this area has shifted towards west by around 5km (Halcrow, 1994). In order to protect bank, two spur dykes were constructed in. The distance between two spur dykes is around.5km. The Betil and Enayetpur spur dykes basically follow same design methods. Eir of m consists of an earn shank and a RCC (Reinforced cement concrete) head. The RCC head is not deep penetrated into riverbed but is supported by in-situ casted concretee piles with a diameter of.5m. Therefore, lower part will be permeable if bed level is degraded enough. The connection between RCC head and earn shrank is in form of a bell-mouth and has been found to be one of weakest points according to past failure experiences. Although RCC blocks and geo-bags additional protection to spur dykes, earn parts have are dumped into water to provide suffered from damages frequently in monsoon seasons since ir completion. The main parameters related to Betil/Enayetpur spur dykes are listed in Tab.. Table Basics of Betil/Enayetpur spur dykes Parameters Betil Length of earth shank 81m Length of RCC part 15m Crest level over PWD 15.5m Width of crest m Side slopes 1: (up) 1:3 (down) Design scour depth 18.5m Completion of spur Enayetpur 15m 15m 15.3m m 1: (up) 1:3 (down) 17.m Piles Photo3 The Betil spur dyke RCC head Photo4 The Enayetpur spur dyke Embankment Shank 4. Bed deformation around Betil/Enayetpur spur dykes during flood The Betil/Enayetpur spur dykes were constructed along secondary channel of river as depicted in Fig.. In dry season, almost no flow passed channel. But during flood period, channel might become active and made threats to embankment. The temporal variations of bed bathymetries around two spur dykes during 8 flood were plotted in Fig.11 and Fig.13. The information on flow discharge of river at upstream Bahadurabadd gauging station was plotted in Fig.1 for reference. Due to severee local scour at toe of Betil spur dyke, lower portion of RCC part became permeable. Therefore, hydraulic and morphological function of RCC part of Betil spur dyke would be similar to those of a Bandal. The downstream Enayetpur spur dyke, however, was still impermeable. It was evident from Fig.11 thatt local scour area at toe of Betil spur dyke exhibited insignificant changes although re was slight increase of maximum scour depth after flood peak. Taking into account previous scour records (e.g. Zhang et al., 8), it might be concluded that flood discharge would not always enlarge local scour. The continuous supply of sediment from upstream channel might prevent bed from continuous degradation. 48

11 N June 1, 8 Discharge (m 3 /s) June 1 July 19 August September 1 4/75/17 / /7/1 8/5 8/59/14 Date (Month/Date) Fig.1 Discharge in 8 monsoon at Bahadurabad N June 1, 8 July 19, July 19, August, August, September 1, 8 Fig.11 Bathymetry at Betil spur September 1, 8 Fig.13 Bathymetry at Enayetpuril spur 487

12 A large area of deposition was found behind Betil spur dyke, and Enayetpur spur dyke was situated immediately downstream of area. The bed level around Enayetpur spur dyke did not change so much in flood season (Fig.13), indicating flow and bed morphology were well controlled by upstream Betil spur dyke. Moreover, lowest bed located along earn shank of Enayetpur spur dyke instead of toe, which was completely different from that around a typical impermeable spur dyke. Similar to that of Betil spur dyke, land was created behind Enayetpur spur dyke due to sediment deposition re. 4.3 Flow structure around Betil/Enayetpur spur dykes during flood The velocity field was measured on July15, 8 and resulted velocity vectors in a horizontal plane near water surface around two spur dykes were plotted in Fig.14. Major sandbars were also sketched in figure according to field investigations and satellite images. Due to existence of various sandbars, approach flow did not attack head of eir spur dyke, which made scour phenomena around two spur dykes somewhat unique. Vertical (m) Vertical (m) Vertical (m) Vertical (m) m/s East (m) Fig.15 Velocity vectors at Section A m/s East (m) Fig.1 Velocity vectors at Section B North (m) Fig.17 Velocity vectors at Section C m/s North (m) Fig.18 Velocity vectors at Section D North (m) Vertical (m) m/s North (m) Fig.19 Velocity vectors at Section E East (m) Fig.14 Velocity vectors around Betil/Enayetpur spur dykes.8m beneath water surface When flow approached spur dykes, it was blocked, changed its direction paralleling to spur dyke and flowed towards spur dyke heads. The parallel flow had a potential to remove sediment on its way to head which gave 488

13 answer why severe scour took place along earn shank of Enayetpur spur dyke. At head of eir spur dyke, flow circulation was recognizable althoughh data was not fine enough to resolvee details. The velocity vectors in several vertical planes were also presented from Fig.15 to Fig.19. The locations of vertical planes were depictedd in Fig.14. The velocity vectors in Section A described typical parallel flow along earn shank of Betil spur dyke. And similar flow structure was expected around Enayetpur spur dyke. In Section B, strong downward flow detached from Betil spur dyke was observed. This downward flow was main cause for lateral enlargement of local scour hole. The downward flow attacked bed, turned upward and mixed with downward flow from sandbar in front of spur dyke. Section C showed longitudinal variation of flow velocity at head of Betil spur dyke. Two circulating cells were recognized: one was situated in upstream and or one was in downstream of spur dyke. They were major engines for development of scour hole in longitudinal direction. The circulating cells resulted from blockage of approach flow at earn shank and RCC head of Enayetpur spur dyke were shown in Section D and Section E, respectively. Althoughh bed profiles were quite different, circulating flows followed same direction. Both of m are capable of removing sediment from bed and result in scour at foot of spur dykes. water course with a length of about 1m and made an angle of 5 o - o to bankline. The Bandal was made of groups of vertical bamboo piles with a spacing of 4cm. The diameters of bamboo piles ranged from cm to 9cm. The piles were connected with each or by cross bamboos with a vertical spacing of around 75cm. Moreover, inclined bamboos were set at an angle of 45 o to bamboo piles to enhance stability of Bandal structure. The upper part of Bandal was closed with bamboo thatch while lower part is open. The Bandals were arranged in a group along secondary channel of river as show in Fig.. The distance between two consecutive Bandals is around 3m. The details of one of Bandals are shown in Photo1 which was taken during field investigation in July, 8. On or hand, Photo demonstrates sediment deposition behind Bandal structure along bank and photo was taken in July 9, one year after first field trip. It is found thatt a large area of land is created behind Bandal within only one year. The two photos give visible evidencee on deposition-promoting performance of Bandals. 5 Bandal and Bandal-like structures 5.1 Introduction The handloom enriched Randhunibari locates along west bank and immediately downstream of Jamuna Bridge as shown in Fig.. Oblique flow towards bankline during monsoon season put Randhunibari market nearby in a very dangerous situation. In order to protectt bank from erosion and to test performance of traditional river training structures, a group of Bandals was constructed by RRI ( River Research Institute) during period of year 7-8. Each Bandal structure protruded into Photo 5 Randhunibari Bandal (July, 8) Photo Randhunibari Bandal (July, 9) 489

14 5. Experiments A structure, having similar functions as a Bandal, is named a Bandal-like structure herein despite its construction materials, shape and layout. Hence, Betil spur dyke may be considered as a Bandal-like structure and flow velocity information of which may be of reference for Randhunibari Bandal site. In order to understand flow structure more precisely, laboratory experiments are conducted in Ujigawa Open Laboratory, Kyoto University. In experiment, powdered anthracite is used as model sediment, which has a mean size of.835mm and a specific gravity of A pair of Bandal-like structures is set perpendicular to left side of a flume. The flume is 1m-long and 8cm-wide and has a slope of 1/8. Each Bandal-like structure is made of 1 brass cylindrical piles, having a vertical steel plate mounted at upper part. The details of Bandal-like structure and experiment setup are shown in Fig.. The experiment is carried out under live-bed scour condition, i.e. approach flow velocity is larger than critical flow velocity for sediment entrainment. As a result, bed forms develop in whole movable bed area. It is case in Jamuna River. The hydraulic parameters in experiment are shown in Tab.3. The experiment starts from a flat bed and continuous sediment supply is ensured from upstream of flume. The total amount of sediment supplied is same as that collected from downstream, which is determined by some trial experiments. After hours, a dynamic equilibrium condition is reached. The pump is n stopped. When bed is drained out, a laser displacement meter is n used to measure bed deformation. The measured bed configuration is shown in Fig.1. In deformed bed, bed forms and large areas of sandbars occur due to sediment movement in whole movable bed domain. Local scour takes place at toes and along bodies of Bandal-like structures, particularly upstream one. Moreover, sediment deposition is observed in wake zones behind both structures. The local bed morphology around Bandal-like structure exhibits typical features of both an impermeable spur dyke and a permeable spur dyke (Zhang and Nakagawa, 9). y (cm) A Inlet tank Instrument carriage Fixed bed Flow Pump Flume side Bandal-like structure Sediment trap Movable bed 1. (c) Structure details: side view, top view & piles details Fig. Experiment setup (Unit: cm) Table 3 Hydraulic conditions of experiment u * : friction velocity, u *c : critical friction velocity Flow discharge 7.7 l/s Mean velocity 4.5 cm/s Approach flow depth 4. cm u * /u *c 1.91 (live-bed scour) Reynolds number 7,4 Froude number.387 Reservoir z (cm) x (cm) Fig.1 Final bed configuration in experiment Instant cement is sprayed to bed surface and deformed bed becomes fixed. Water is n pumped to flume again. The velocity on water surface is obtained with PIV (Particle Image Velocimetry) techniques and two electromagnetic velocimetries are used to measure 3D velocity y (a) Plan view z (b) Section A-A 3 x x A 49

15 in water column. Water level is recorded with a point gauge. 5.3 Numerical model Since flow around a Bandal-like structure is highly 3D, details of flow cannot be resolved without a 3D model. Moreover, a Bandal-like structure is generally sophisticated in shape and boundary of an actual river is commonly irregular. Hence, numerical model should be capable of resolving complex geometries as well. The authors have developed a numerical model satisfying above requirements (Zhang et al., ). In numerical model, complex flow field is obtained based on RANS (Reynolds-averaged Navier-Stokes) equations with k-εmodel for turbulencee closure. In near-walll domain, wall-function approach is applied and resistance of wall can be easily accounted for. The widely used FVM (Finite Volume Method) is adopted in formulation of numerical model and governing equations are discretized based on an unstructured mesh. The flow field on deformed bed in laboratory experiment is investigated with proposed numerical model. The measured water stage and bed level are used in mesh generation. In order to reproduce details of Bandal-like structures, a hybrid mesh consisting of different types of polyhedra is employed. The mesh system includes 5884 cells and 5787 nodes. The computational mesh in proximity of Bandal-like structuress is depicted in Fig.. At inlet boundary, all flow variables are known and are prescribed according to mean velocity. At outlet boundary, zero diffusion flux is assumed for all flow variables. The free surface is considered as a rigid lid and is unchangeable during computation. Near wall boundaries such as channel bed, Bandal-like structure and side of flume, flow velocity is assumed to be parallel to walls. 5.4 Results Flow velocities resulted from both experiments and simulations are plotted in Fig.3-7. The flow velocity on water surface, near bed and typical lateral and longitudinal sections is shown. y (cm) y (cm) Mesh number: 5884 Node number: Fig.3a Flow velocity (u, v) on water surface (PIV) 8 4 Fig. Computational mesh around Bandal-like structures Bandal-like structures x (cm) x (cm) Flow cm/s 4 cm/s 4 Fig..3b Flow velocity (u, v) on water surface (Sim.) According to Fig.3 and Fig.4, it is found that flow velocity on water surface is affected significantly by blockage of upper part of Bandal-like structures and 3D vortices in scour. On or hand, flow velocity at lower layer is closely related to bed geometries and existing of piles of Bandal-like structures. In general, numerical model reasonably reproduces experimental results. 491

16 8 cm/s 8 cm/s y (cm) 4 y (cm) x (cm) Fig.4a Flow velocity (u, v) at cm from initial bed (Exp.) x (cm) Fig.4b Flow velocity (u, v) at cm from initial bed (Sim.) z (cm) cm/s y (cm) Vortex Down flow Fig.5a Flow velocity (v, w) at x=-cm (Exp.) z (cm) cm/s y (cm) Fig.5b Flow velocity (v, w) at x=-cm (Sim.) z (cm) cm/s Vortex Up flow y (cm) Fig.a Flow velocity (v, w) at x=-1cm (Exp.) z (cm) cm/s y (cm) Fig.b Flow velocity (v, w) at x=-1cm (Sim.) z (cm) Vortex Vortex cm/s Vortex Down flow Down flow x (cm) Fig.7a Flow velocity (u, w) at y=7cm (Exp.) z (cm) cm/s x (cm) Fig.7b Flow velocity (u, w) at y=7cm (Sim.) If one takes a closer look at free surface velocity field, one may note that bed morphology influences a lot on flow structure. The flow separation angle at head of structure is much smaller compared with that of PIV measurement. As has been argued in previous research (Zhang et al., 9), over-estimation of current PIV method due to an over-accumulation of tracers is considered to be a probable reason. The inherent deficiency of eddy viscosity based turbulence model is also anor probable cause, but furr evidences are needed. 3D vortices are obviously confirmed in Fig.5-7, concentrated in extent of local scours and being similar to those observed around spur dykes (Zhang et al., 9). These vortices are engines for scour development. However, re are many properties unique to Bandal-like structures. Fig. and Fig.7 indicate that wake vortex behind upstream structure is not well developed. The vortex is very weak and is confined in shade of impermeable part of structure or than inside scour hole. It is attributed to strong flows which pass piles of structure and make great 49

17 disturbances to wake zone. It is also found in Fig.7 that re are vortices located in front of and behind downstream structure. Due to vortex in front of structure, flows passing piles lose ir energy and vortex behind structure develops more freely and even extends to inside of scour hole. The vortex in front of downstream structure is resulted from down flow and relatively small scour depth (hence, a small opening ratio). Similar vortex pair appeared in field measurements of Betil spur dyke as shown in Fig.17. Discussions The information on flow field and bed deformation of study sites provides a key to understand fundamental working principles and performances of typical structural measures against bank erosion along Jamuna River. Compared with conventional revetments and spur dykes, flow structure and bed morphology around Sirajganj Hardpoint (revetment type) and Betil/Enayetpur spur dykes (spur dyke type) are much more complex due to complex flow and bed conditions. Owing to its unique location, upstream termination of Sirajganj Hardpoint works like an impermeable spur dyke to some extent. Several vortex systems of obviously 3D nature form in proximity of termination and y are strongly dependent on local bed morphology as well as migration of sandbars nearby. The termination is a weak point of structure due to local scour at toe and possible sliding down of sand materials on which apron is set. The area a little downstream of termination is also very weak due to direct attack of return currents. Both parts experienced failures in past several years. The comparison of flow structures in 8 and 9 suggests that performance of Sirajganj Hardpoint strongly depends on sandbars in its neighborhood and that evolution of sandbars necessitates special attention. By introducing long earn shanks, area protected by Betil/Enayetpur spur dykes is enlarged and construction cost is reduced. On or hand, earn shanks become most frequently failed parts of two spur dykes due to direct hit by approach flow and parallel flow along shanks. Countermeasures against parallel flow are hence important for stability of two spur dykes. The RCC part of Betil spur dyke works like a Bandal structure which blocks flow in upper layer and allows water to flow at lower elevation. This kind of structure takes advantages of both impermeable spur dyke and pile dyke. As a result, excessive local scour is avoided but eir mainstream degradation or wake deposition is promoted. Moreover, return currents behind spur dyke are also weakened due to joining of flow passing through piles of RCC part. The downstream Enayetpur spur dyke furrmore prevents possible return currents caused by upstream Betil spur dyke from direct attacking embankment. Due to influence of Betil spur dyke, impermeable Enayetpur spur dyke does not suffer too much from local scour at RCC toe. The maximum scour depth around Enayetpur spur dyke is about 9.5m smaller than that of Betil spur dyke during field measurement. However, parallel flow is more dangerous along Enayetpur spur dyke compared with that along Betil spur dyke since former owns a longer earn shank and an almost impermeable RCC part. Making lower part of RCC part permeable to reduce effective length of spur dyke might be a solution to reduce erosion along foot of shank. Although detailed measurements are not conducted for Bandal structures in site, effectiveness of structures are confirmed from both Randhunibari test Bandal site and Bandal-like part of Betil spur dyke. In addition, laboratory experiment and numerical simulation preliminarily provides working mechanisms for this kind of structure. Recently, Teraguchi et al. (11) have made more systematic comparisons on hydraulic and morphological characteristics among conventional impermeable spur dykes, pile dykes and newly highlighted Bandals. It has to be mentioned that original Bandals are generally made from locally available bamboos and are constructed without reliable scientific design or 493

18 effective quality management. Therefore y are quite cheap (Table4) and easily implemented but generally suffer from stability problems especially during flood. Table 4 Cost of bank protection measures Measures Site Agency Cost (US$/m) Guide Jamuna Bank Bridge Foreign 33, Hardpoint Sirajganj Foreign 1, Solid spur Kalitola Foreign 1,5 Revetment Jamuna (Geobags) River Foreign -3 Revetment Jamuna River BWDB 38-4 RCC spur Jamuna River BWDB 95 Bandal structures Sirajganj RRI 9 *compiled based on Rahman et al., 7 Comparing Bandals with conventional impermeable spur dykes, it is noted that Bandals will generally exert less impact on flow and channel dynamism. Hence, it has a potential to provide a more nature friendly alternative for management of large alluvial rivers like Jamuna River. The Bandals are physically smaller and weaker than conventional impermeable spur dykes. Furrmore, Bandals may be quickly buried due to huge deposition around it and ir working life will be much shorter compared with conventional impermeable spur dykes. Therefore, Bandal structures should be better basically used as temporary structures in a recurrent way to cope with changing conditions adaptively. The total cost will include initial investments and timely maintenances. spur dykes play important roles in protecting BRE and that historied river training measures such as Bandals based on indigenous knowledge exhibit high potential for wider applications. The flow field and bed morphology around bank protection structures, eir intrusive or nonintrusive, vary spatially and temporally but maintain certain common features depending on type of structures. The strong vortex system in front of and return currents behind upstream termination of Sirajganj Hardpoint are major engines for morphological variations and are also main causes of structure failure. Moreover, intensity of flow around termination is closely related to evolution of sandbars nearby. The vortex system and parallel flow resulted from flow blockage at long shanks of Betil/Enayetpur spur dykes are great threats to stability of two spur dykes. The RCC part of Betil spur dyke works as a Bandal-like structure, providing a well control of incoming flow and sediment transport for downstream Enayetpur spur dyke. Compared with local scour at toe of RCC part of Enayetpur spur dyke, erosion along foot of long earn shank is a more serious concern. It might be a challenging problem unique to relatively huge structures constructed in large rivers. Field investigation suggests that Bandals developed from indigenous knowledge are capable of promoting sediment deposition efficiently. Considering migration nature of Jamuna River, cost-effectiveness and environmental harmony, wider application of Bandals or Bandal-like structures is recommended. However, more research is suggested to clarify associated mechanisms and to formulate guidelines on ir designs, constructions and maintenances. Acknowledgements 7 Conclusions This paper presented a study on hydraulic and morphological consequences of typical bank protection measures along Jamuna River of Bangladesh. The results indicate that conventional bank protection measures such as revetments and This research is supported by Grant-in-Aid for Scientific Research B, MEXT, Japan (PI: Dr. H. Nakagawa, Grant No ) and JST-JICA Program on Science and Technology Research Partnership for Sustainable Development (PI: Dr. H. Nakagawa). The authors would also like to express ir sincere gratitude to a lot of participating 494

19 members from BUET, CEGIS, BWDB, RRI and Kyoto University. References BWDB (1999): River Bank Protection Project. Evaluation of Performance of Hard Points, Review of Damage during 1999 Flood Season and Recommended Remedial Works. CEGIS (Center for Environmental and Geographic Information Services): Monitoring and Prediction of Bank Erosion Reports series, 4, 5, and 7. Coleman, J. M. (199): Brahmaputra River: channel processes and sedimentation, Sedimentary Geology, Elsevier Publishing Company, Amsterdam, Vol.3, pp FAP1. (1): Guidelines and Design Manual for Standardized Bank Protection Structures, Bank Protection Pilot Project, WARPO, Ministry of Water Resources, Bangladesh. Halcrow, Sir William and Partners, DHI, EPC and DIG (1994): River Training Studies of Brahmaputra River, Prepared for BWDB. IWFM (8): Field Based Applied Research for Stabilization of Major Rivers in Bangladesh, Technical Report 1, IWFM, BUET. Oka, T. (4): Floods in Bangladesh, Annuals of Disaster Prevention Research Institute, Kyoto University, No.47A. (in Japanese) Rahman, M.M., Hussain M.A., Hossain M.M., Sarker M.H. and Uddin, M.N. (7): Protective measures of flood embankment along Jamuna River in Bangladesh, ICUS Report 7, pp Sarker, M.H., Huque, I., Alam, M. and Koudstaal, R. (3): Rivers, chars and char dwellers of Bangladesh, International Journal of River Basin Management, IAHR&INBO, Vol.1, No. 1, pp.1-8. Teraguchi, H., Nakagawa, H., Kawaike, K., Baba, Y. and Zhang, H. (11): Alternative method for river training works: Bandal-like structures, Annual Journal of Hydraulic Engineering, JSCE, Vol.55, pp Thorne, C.R., Russell, A.P.G. and Alam, M.K. (1993): Planform pattern and channel evolution of Brahmaputra River, Bangladesh, Braided Rivers (eds. Best and Bristow), Geological Society Spec. Pub., No.75, pp Zhang, H., Nakagawa, H., Muto, Y., Baba, Y. and Ishigaki, T.(): Numerical simulation of flow and local scour around hydraulic structures, River Flow, Lisbon, Portugal, pp Zhang, H., Nakagawa, H., Muto, Y., Muramoto, Y., Touchi, D. and Nanbu, Y. (7): Morphodynamics of channels with groins and its application in river restoration, Annuals of Disaster Prevention Research Institute, Kyoto University, No.5B, pp Zhang, H. and Nakagawa, H. (8): Scour around spur dyke: recent advances and future researches, Annuals of Disaster Prevention Research Institute, Kyoto University, No.51B, pp Zhang, H., Nakagawa, H., Kawaike, K. and Baba, Y. (8): Local scour around bank protection spur dykes on Brahmaputra-Jamuna River, The 7th Annual Conference of Japan Society for Natural Disaster Science, Fukuoka, September 5-, pp.1-. Zhang, H. and Nakagawa, H. (9): Characteristics of local flow and bed deformation at impermeable and permeable spur dykes, Annual Journal of Hydraulic Engineering, JSCE, Vol.53, pp Zhang, H., Nakagawa, H., Kawaike, K. and Baba, Y. (9): Experiment and simulation of turbulent flow in local scour around spur dyke, International Journal of Sediment Research, Vol.4, No.1, pp Zhang, H., Nakagawa, H., Baba, Y., Kawaike, K. and Teraguchi, H. (1): Three-dimensional flow around Bandal-like structure, Annual Journal of Hydraulic Engineering, JSCE, Vol. 54, pp

20 バングラデシュ国ジャムナ川における河岸侵食防止対策が流れ及び地形に及ぼす影響 張浩 中川一 馬場康之 川池健司 ラマンムンスン * ウヂンナジム ** * バングラデシュ工科大学水資源 洪水管理研究所 ** ダッカ工科大学土木工学研究科 要旨バングラデシュ国の河川では, 河岸侵食が大きな問題となっている 河岸侵食は科学と工学的な問題だけでなく, この南アジアの国にとっては, 大きな社会や経済問題でもある 本稿は, ブラマプトラ川の下流 ( バングラデシュ国内ではジャムナ川と呼ばれる ) における河岸侵食防止対策及びそれらが流れ及び地形に及ぼす影響について報告する 数回の現地調査及び計測により, ジャムナ川における河岸侵食のメカニズムの解明を図った また, 既存の河岸侵食防止対策の機能を評価し, 可能な改善策を提案した 特に, 現地の歴史的な方法で土着工法の代表であるバンダル型水制について着目し, ジャムナ川の河道安定及び河岸侵食防止方策としては, バンダル型水制の設置の繰り返しが現地に適応した工法となり得る可能性を示唆した キーワード : ブラマプトラ川 / ジャムナ川, 河岸侵食防止, 河道安定化, 現地計測, バンダル, 水制 49