FLUSHING SEDIMENT THROUGH RESERVOIRS *

Transcription

1 Iranian Journal of Science & Technology, Tranaction B, Vol. 8, No. B1 Printed in Ilamic Republic of Iran, 004 Shiraz Univerity FLUSHING SEDIMENT THROUGH RESERVOIRS * N. TALEBBEYDOKHTI ** AND A. NAGHSHINEH Dept. of Civil Engineering, Shiraz Univerity, Shiraz, I. R. of Iran taleb@hirazu.ac.ir Abtract About 1% of the total torage capacity in the world reervoir i lot annually due to edimentation. Sediment can alo block intake in reervoir and damage tunnel or turbine. One of the mot effective technique to remove thee ediment i fluhing, whereby water level i lowered ufficiently to re-erode depoit and fluh them through the intake. Outflow ediment dicharge may well be related to the parameter uch a the ediment characteritic in the reervoir, during fluhing and geometry of fluhing channel. In thi tudy, laboratory experiment were performed on a 1-D reervoir model in a flume in the hydraulic laboratory of Shiraz Univerity to invetigate the fluhing operation procee by uing polymer particle. The polymer particle were lightweight and non-coheive with an average grain ize of about.40 mm and denity of (kg/m ). The model wa intalled in a flume; 0 m long, 1 m wide and 0.75 m height. The length of the tet ection wa 11.5 m, and ediment were placed at a length of 4.8 m long uptream from the dam poition. Experimental run have been performed for two flow condition; m / and m /. The very low inflow dicharge helped for better monitoring and meauring of the effective parameter. A luice gate wa placed at the central bottom of the dam (a the bottom outlet) and wa opened at a contant rate to make the complete drawdown. Reult howed that the rate of ediment fluhing i trongly aociated with outflow rate, water urface gradient with the dam ection and the width of the fluhing channel. The reult from thi tudy were in agreement with that in the literature. It i conidered that the low denity of the particle caue them to behave a very fine and non-coheive ediment particle, like loe ediment. Keyword Reervoir edimentation, fluhing, retrogreive eroion, cone formation, fluhing channel 1. INTRODUCTION Dam contruction ha increaed during the lat decade, particularly in the tropic and emi-arid area where high ediment yield are prominent, a well a reervoir edimentation. In 1900 there were 4 large dam (i.e. higher than 15 m) while in 1950 and 1986 there were 5,68 and about 9,000, repectively [1]. Sutainable ue of exiting reervoir ha become an important iue, ince building new reervoir i rather difficult due to the new environmental regulation, high cot of contruction, and lack of uitable dam ite. Therefore, technique for reervoir deiltation have received increaing attention recently []. Approximately 1% of the torage volume of the world reervoir i lot annually due to ediment depoition []. In ome developing countrie where population preure on fragile upland ecoytem have led to accelerated rate of oil eroion and reervoir torage i being lot at much larger rate, the following approache have been adopted in order to control meaure to reduce ediment inflow into the reervoir [, ]: 1) Increaing ediment tranport through the reervoir during high flow with heavy ediment concentration, ) Fluhing reervoir-ediment accumulation through the reervoir, ) Bypaing high flow with heavy ediment concentration from entering the reervoir, 4) Fluhing ediment from reervoir by denity current, 5) Removing the reervoir ediment by mechanical mean uch a dredging and iphoning. In certain part of the world, a combination of three approache: 1, and 5 i an attractive practice. The bypa of flood flow and ediment from entering the reervoir require certain topographical and flow condition, and thi approach i not commonly ued. We do know the neceary condition, but not the Received by the editor October 9, 00 and in final revied form March 18, 00 Correponding author

2 10 N. Talebbeydokhti / A. Naghhineh ufficient condition for the exitence of the denity condition, and thu it i difficult to rely on the proce of denity current to fluh ediment [4]. Fluhing i the only mean of recovering lot torage without incurring the expenditure of dredging or other mechanical mean of removing ediment. Where feaible, fluhing can offer an attractive mean of recovering and maintaining a ueful torage capacity when compared with the cot of alternative method (e.g. dredging and iphoning) []. a) Sediment fluhing Fluhing i the couring out of depoited ediment from reervoir through the ue of low-level outlet in dam to lower water level, thereby increaing the flow velocitie in the reervoir. The technique i not widely ued becaue: (a) It i uually effective in narrow reervoir, (b) It involve large volume of water being paed through the dam, and (c) It require the reervoir to be emptied. Drawdown i the lowering of the water level in a reervoir. Drawing down a reervoir (for a few week or month during a flooding eaon) i alo a form of fluhing, although the principal purpoe i to pa the high ediment load carried by flood flow through the reervoir. In the literature thi practice i commonly termed a Sluicing []. Brandt [1] mentioned that fluhing i ued to erode previouly depoited ediment, and luicing i ued to dicharge incoming ediment through the reervoir without necearily drawing down the water level. In thi paper, however, luicing i conidered to be a particular kind of fluhing []. b) Reervoir-ediment depoitional pattern Sediment particle are carried by flow into a reervoir and depoited in the reervoir due to the increae of flow area, and thereby the reduction of flow velocity. A hown in Fig. 1, the deltaic depoition conit of four part [, 4, 5]: (a) Front reach (Bottomet), (b) Frontet (Foreet), (c) Topet, and (d)- Tail reach. In high ediment laden flow, ediment may reach the downtream end of the reervoir quickly, reulting in a wedge-haped depoition with it apex very cloe to the dam. A wedge-haped depoition can alo be formed if the water level i allowed to be drawn down regularly. In addition, deltaic depoit may migrate toward the dam and the entire reervoir depoitional pattern will become wedge-haped. Thu, the wedge-haped depoitional pattern could be the equilibrium tate of certain reervoir in a long run []. For a wide reervoir (with lateral width much greater than width of flow through outlet), the lateral ditribution of ediment depoit may not be uniform [4]. c) Fluhing procee A dicued by Lai and Shen [], fluhing procee may include the following two type: The firt type i to ue flow to remove previou reervoir ediment depoit. The econd type i to pa heavy ediment concentrated flow through the reervoir during high flow. When the water tage in the reervoir i high, a indicated in Fig. 1, only a local fluhing cone i formed and the fluhing proce i not very effective. However, when the water tage i low (the topet of ediment depoit i cloed to the water tage), the fluhing of ediment can be very effective in removing previou reervoir ediment depoit and alo to pa ediment in the flow, provided the frontet of the ediment depoit reache a location very cloe to the dam [4]. Fig. 1. Schematic ketch of depoitional pattern in the longitudinal direction [4] Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

3 Fluhing ediment through 11 After forming the deltaic depoition in a reervoir, Fig. illutrate the fluhing procee correponding to the drawdown operation in the longitudinal direction with variou water level. Auming that the luicing capacity i large enough for lowering the reervoir water level and there i no ediment clogging at the luicing outlet, the luice gate i opened to the lower water level for drawdown fluhing. During the drawdown period, the backwater may diminih from the uptream of the reervoir and progreive eroion in the tail reach can occur to carry previou depoit or incoming ediment toward the dam. If the reervoir water level i low enough to cour the apex of the deltaic depoition, the apex of the delta can move retrogreively and a fluhing channel i developed by cutting through the depoit. Thi eroion pattern i called Retrogreive Eroion which propagate from downtream toward uptream []. Fig.. Schematic diagram of fluhing procee in the ediment delta [] The developing channel deepen and widen a a reult of large-cale eroion in the reervoir. Thu, an effective fluhing proce occur. However, ediment coured from the delta reach may move toward the dead-torage zone near the dam and ettle down before they can be fluhed out. Under uch a condition, the fluhing may not be effective. If drawdown fluhing i operated with a wedge-haped depoition, ediment depoit can be coured in the vicinity of the luice-gate opening within a very hort period of time under preurized flow condition, and a funnel-haped crater called fluhing cone will be formed by the fluhing flow. Once the fluhing cone ha been formed and there i no ediment moving into the cone, the flowing water i clear through the opening, meaning that the formation of the fluhing cone i fairly table and no ediment will be removed from the fluhing cone afterward. If the water urface i to be drawndown ignificantly to generate high flow velocity near the luicing outlet, the flowing water will tart to erode the rim of the fluhing cone and retrogreive eroion may occur []. d) Formation of the fluhing channel During depoition procee in a reervoir, ediment depoit in the main channel and then extend laterally acro the full width of the reervoir to create a nearly horizontal depoition level. Neverthele, by operating drawdown near the luicing outlet, flow condition in the impounded reache will be imilar to the original riverine pattern. In other word, through fluhing procee, channelization by retrogreive eroion create a channel flow cutting in the depoit to reetablih a new cro-ectional pattern with the main channel and floodplain. e) Effect of local fluhing A mentioned previouly, fluhing in a hort period could create a table fluhing cone. Field obervation by Wan [] at Sanmenxia reervoir in China indicated that a fluhing cone wa formed within ten to twenty minute. Compared with the eroion cale of drawdown fluhing, the cale of the fluhing cone i relatively mall. In general, the main function of the fluhing cone i to reduce the ediment concentration around the entrance of the intake tower, and thereby prevent hydraulic tructure from abraion by ediment. f) Objective and cope of thi paper The objective of the preent tudy are a follow: Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

4 1 N. Talebbeydokhti / A. Naghhineh 1- To invetigate the mechanim of the fluhing operation in an experimental, one-dimenional model. - To decribe the behavior of the cone formation and fluhing channel evolution. - To variate the outflow water and ediment dicharge, water depth in the uptream and downtream end of the reervoir with time in regard to the fluhing operation in the experimental model. 4- To compute the coefficient of erodibility (E) baed on the Tinghua Univerity method with outflow water and ediment dicharge, water urface lope and width of the fluhing channel in the reervoir. 5- To etimate the fluhing efficiency (F e ) and it relationhip with outflow ediment dicharge, cumulative volume of fluhed ediment and experimental duration for each run. 6- To compare of thee reult with previou work. 7- To preent recommendation for ediment fluhing operation in exiting reervoir.. LITERATURE REVIEW A large number of tudie including theoretical and experimental work have been carried out to explain reervoir behavior during fluhing ediment through the bottom outlet. Fluhing i not a new technique. The firt known method of fluhing (which wa in Spain in the 16 th century) wa reported by D Rohan (reported in Brandt [1]). However, Bouvard (reported in Brandt [1]) tated that the reervoir hould be reaonably large o that it would not be neceary to performe fluhing too often, diturbing the normal operation of the intake. Ramirez and Rodriguez (reported in Brandt [1]) divided the fluhing proce of the Cachi reervoir in Cota Rica into three phae. The firt conit of 5 day of low water releae, lowering the water table one-meter per day down to a few meter above minimum level for power generation. The econd phae conit of the rapid releae of the remaining water, a lowering of 45 m during a period of approximately five hour. The third phae conit of free flow water out of the reervoir for two or three day. Lai and Shen [] pointed out that fluhing hould be done regularly before depoit conolidate, epecially for coheive clay depoit. Sluicing operation hould be timed to meet the higher ediment concentration brought in by flood flow [7]. Du and Zhang (reported in Brandt [1]) tated that the reitance of coheive ediment in front of the dam i much greater than the andy ediment in the middle part of the reervoir. Therefore, the eroion of coheive ediment near the dam region govern the progre of retrogreive eroion in the reervoir. When eroion of the reervoir bed ha commenced, the flow oon become hyper-concentrated. When concentration increae, the ettling velocity decreae, thu the flow can tranport with very high concentration of ediment under a relatively mall flow (Wu, reported in Lai et al.) [6]. Nagle (reported in Brandt [1]) made obervation on the cro-ectional eroion pattern during fluhing (Villar reervoir, Rio Lazoya, Spain). There, ediment wa removed only from the old tream channel and the high bank of depoit were left on either ide. The main channel underwent aggradation and degradation alternately, and the flood plain alone kept riing lowly (Zhang et al., reported in Brandt [1]). In reache far from the dam in the Henghan reervoir, China, the eroion at the foot of the bank broadened the channel, leading to the formation of a rectangular cro-ectional form and cauing the coare floodplain depoit to collape vertically into the channel [8]. Initially a carved-out channel wa narrower than the original river width, but with periodic fluhing the coured channel will approach the pre-dam width of the river (Mahmood, reported in Brandt [1]). Baed on data from four different reervoir, Atkinon [] found that the channel width formed during fluhing correlated well with the fluhing dicharge alone, with no apparent enitivity to lope or ediment propertie. The ide lope, which will develop during fluhing, depend on the ediment propertie, the degree of conolidation, the depth of the depoit, and perhap the extent of water-level fluctuation during fluhing []. The fluhing channel may deepen under either mall or large flow until it encounter the armored bed of the pre-impoundment river. Afterward, further eroion can occur only by widening of the channel by bank failure, which i the primary mechanim involved in the widening of fluhing channel [8]. Di Silvio (reported in Lai and Shen []) found that under the preurized flow condition, a fluhing cone in the vicinity of the luice gate could be formed with a wedge-haped depoition. He added that if the Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

5 Fluhing ediment through 1 water level i allowed to be drawndown, two type of eroion pattern can occur: progreive eroion and retrogreive eroion. In the latter cae, retrogreive eroion back-cutting in the depoit reult in trong and large-cale eroion, which retore ueful torage capacity. Fan and Morri [8] preented everal model to imulate the fluhing procee under draw down operation. Baed on the field data they obtained during retrogreive eroion, everal empirical formula have been propoed to etimate ediment outflow dicharge which may be related to flow dicharge, energy lope or channel width. Shen, et al. [6] ued a dimenional analyi for non-coheive ediment to etablih a formula to etimate the eroion depth for local fluhing in front of the luice gate. Then Lai and Shen preented a one-dimenional experimental and diffuion model to imulate the general trend of the bed profile evolution, and the amount of reervoir ediment removal during fluhing, in order to evaluate the applicabilitie and limitation of the diffuion model []. They found that the imulated reult of the diffuion model agree well with the laboratory data in a narrow flume (with eentially 1-D flow) with a nearly uniform flow condition after rapid draw down operation. At lat, Shen [4] reviewed the current tatu on the fluhing ediment through reervoir tatitically and alo treed the need of an analytical olution for incorporating the rik analyi for the planning ediment operation through dam. There have alo been many numerical and theoretical tudie. For example, Ju (reported in Lai et al. [6]) preented one-dimenional (1-D) diffuion model which were olved analytically or numerically to imulate the removed ediment volume and bed profile change with contant dicharge and channel width during fluhing. White (reported in Lai et al. [9]) developed numerical model dealing with edimentation and fluhing to tudy the feaibility and effectivene of fluhing operation for mall reervoir in Zimbabwe. Olen and Melaaen (reported in Olen [10]) ued three dimenional model to calculate the local cour and ediment depoition in a reervoir, and in a and trap they gave a limited decription of the twodimenional numerical imulation of the fluhing proce.. DIMENSIONAL ANALYSIS During drawdown fluhing, the following variable hould be conidered: C = volumetric outflow ediment concentration; u = mean flow velocity; h = water depth; S = energy lope; g = gravitational acceleration; γ = pecific weight of water; γ = pecific weight of ediment; and w = fall velocity of particle. Uing the Buckingham theory [11] for dimenional analyi, function of variable i then preented a The above function i changed to f ( C, u, h, S, g,, γ γ, w) = 0 γ (1) C g us γ u f 1,,, = γ w γ γ gh 0 () If g and γ are conidered contant, we can bring one of the dimenionle parameter to the left ide and ay C g γ γ u S γ γ ghw m () Or one can write it impler C γ g γ u = k ghw γ γ m (4) and a g and γ are contant o Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

6 14 N. Talebbeydokhti / A. Naghhineh C = k o γ u γ γ ghw m (5) In which k o = k γ g and we can ubtitute ome of above parameter with new parameter G γ = G γ γ γ 1 γ 1 = = (6) γ G 1 γ γ In which k o and m are coefficient to be determined; G i the pecific gravity of ediment. The value of m wa determined from field data in China and are about unity []. For practical purpoe, let m equal unity. Auming the approach flow to be uniform during fluhing, the Manning equation can be applied a below (in metric ytem) Having a wide rectangular flume lead the relation change to h 1 w 1 u = R S (7) n 1 4 q Q o 1 k k o u o u = R h q = m u = S R C = = 1, h,, 10, w h S h (8) h B n ( ) ( ) w G 1 gqw gw G 1 n Let take k k = gw G o ( 1) Then we will have C w k Q o 1 = S n (9) n B S h 1 w From thi point an iterating method will begin. We hould ubtitute h with (q/u) and put (Q o /B) intead of q and ue the Manning equation for the parameter u, until their power i fixed within 10% accuracy, and aume the power which are too mall to conider, uch a zero. We will write the reult of ome iterating run below Iteration#1 Iteration#7 Following the iteration, the end reult would be 1 4 Q o S h 8 w B C k = (10) n k Q 187 o S 187 h 500 w B 187 C = (11) n C = n k Q B o S w 1 h C = gw ( G 1) k o n.4 Q o B 0.6 S 1. w (1) Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

7 Fluhing ediment through 15 So the outflow ediment dicharge by weight will be a follow: Q o = γ k o Q o 1. γ k o Q o S w γ C Q o = S Q Q.4 w o o = (1) gw ( G 1) n B gw ( G 1) n B The main formula for calculation of outflow ediment dicharge i a follow: Q o S w 0. 6 Q o = E (14) B where Q o =outflow ediment dicharge (ton/ec); Q o =outlet water dicharge (m /ec); n=manning roughne; B=width of fluhing channel (m); S w = water urface lope = S o = bed lope; and E = coefficient of erodibility which i a below [, ] E = gw γ k. 4 ( G 1 ) n o (15) 4. METHODS AND EXPERIMENTS a) Setup and procedure From both previou laboratory and field data by Lai and Shen [], it ha been found that outflow ediment dicharge can well be related to ome hydraulic parameter, which are function of outlet dicharge, waterurface gradient and fluhing channel width. A one-dimenional experimental model preented herein i employed to imulate the general trend of bed profile evolution and the amount of reervoir ediment removal during fluhing in order to evaluate the applicabilitie and limitation of thi model. A rectangular concrete flume 0 m long, 1.0 m wide and 0.75 m in height, located in the hydraulic laboratory at the Civil Engineering Department of Shiraz Univerity in Shiraz wa modified to model a reervoir. Part of the flume wa elevated up to 0.16 m to prevent the outlet from being ubmerged at the downtream of the dam (thi value ha been found baed on the dimenion of the Sefid Rud reervoir, which i the only reervoir in Iran that ha been aved by fluhing operation). Water wa re-circulated by a pump at two deired teady flow rate of and (m /ec) by controlling a liding valve at the end of the pipeline (ee Fig. ). Fig.. Schematic ketch of the experimental model (a) Top-view, (b) Sidei A one dimenional model with a central luice gate, dimenion of about 0.15 m in height and 0.09 m wide wa made and ued (ee Fig. 4a and 4b). The ediment material ued in the experiment wa a Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

8 16 N. Talebbeydokhti / A. Naghhineh lightweight and non-coheive polymer with the traditional name of Polytyrol 14E, which wa one of the BASF product. The choice of the ediment type and ize wa baed on the dimenion of the flume and the capacity of water upply. Polytyrol 14E had a bulk denity of about 650 kg/m, pecific gravity of about 1.065, and median diameter of about.4 mm. Furthermore, uing the Polytyrol 14E repeatedly in the experiment did not change the ediment propertie. The denity of thi material wa very cloe to the denity of water and thi made problem in ditinguihing the threhold time for the retrogreive eroion in each experimental run. At the beginning of each experimental run, ediment wa wetted and paved without any compaction and conolidation at a 4.8-m length by 0.06 m height uptream of the dam. The zero elevation wa et at the bottom of the luice gate (which wa 0.16 m above the original flume bed). Fig. 4a. Downtream face of the dam and luice gate in the experimental model Fig. 4b. Uptream face of the dam and luice gate in the experimental model The ame procedure of wetting and placing wa performed for all the ediment particle (110 Kg of the polymer), and then the ditributing and moothing operation began. A T-haped device wa ued for ditributing and moothing. The moothne of the polymer particle (ediment) wa checked on the idewall of the flume. The level wa exactly 0.06 m above the raied wooden bed. The middle part of the ediment bed wa alo checked by a point gage located at the centerline of the flume to make ure of the moothne of the whole ediment in the tet ection. The central point gage wa moved from the dam to the uptream end of the tet ection to reach to the ame height of the ediment (0.06 m) by compenating or removing polymer particle. Thi operation uually took about 1.5~ hour. The pattern of the gate wa modified by adding a mall teel channel at the back of the gate to repreent the original river channel. The dimenion of thi pattern were a follow: 0.09 m width (equal to the width of the luice gate) and 0. m long. The polymer particle within thee wall were removed prior the experimental run to repreent the natural condition (bottomet and foret of the long-term ediment depoit in reervoir) at or near the bottom outlet (luice gate). The teady tate condition wa attained by regulating the luice gate opening. It hould be mentioned that one of our main aumption in thee experimental run wa the teady tate flow condition. To obtain thi condition, the inflow dicharge mut be equal to outflow dicharge. The inflow dicharge wa contant in each experimental run. The outflow dicharge wa controlled only by the opening of the luice gate. The water depth wa kept contant of 0.10 m height in all-experimental run at the tet ection. Due to the length and ize of the experimental model, the change in gate opening wa gradual to prevent unneceary water fluctuation and wave in the reervoir. Becaue the udden change were caued the lightweight polymer particle floatation, the gate wa opened lowly and gradually in everal tep to obtain a teady tate condition without any ediment diturbance. The main experimental run began right after reaching a teady tate condition. It i worthwhile to mention that the height of ediment, the length of ediment, water Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

9 Fluhing ediment through 17 depth and the inflow dicharge for each experimental run were kept contant in order to evaluate the fluhing ediment behavior through the reervoir. To tart the experiment, the luice gate wa manually opened at a contant rate and the outflow waterediment mixture were ampled every two minute. The total time of each experiment wa about 50 minute. Thi i equivalent to around 5 ample. Two point gage; one at 0. m from the dam and the other at 4.8 m from the dam (at the uptream end of ediment depoit) were intalled to meaure water urface elevation. The former one wa ued to meaure the head of water above the weir to calculate the outflow water dicharge. Baed on previou work [9] and cited obervation in the literature, the width of the fluhing channel in each time tep wa meaured in the middle reach of the reervoir (at.4 m from the dam) where the effect of uptream and downtream end are minimum. At each experimental run, the bed ediment profile wa meaured by a point gage at the centerline of the flume. Water temperature ranged from 5 to 8 degree Celiu in the experimental run. b) Sediment fluhing obervation A detailed decription of a repreentative experimental run provide a general overview of the erie of experiment. For thi purpoe, Flow1 wa elected a the repreentative run to be decribed in thi ection. Following experimental procedure, Flow1 wa initially et with zero bed lope and with an initial channel. To reach a teady tate flow for the initial etup with a contant inflow dicharge rate of m /ec and a water depth of 0.10 m above datum, the central luice gate wa opened by 1 cm; meanwhile, a fluhing cone wa haped by the fluhing water, which removed ediment depoit only in the vicinity of the gate opening. Two to three minute after the gate wa opened, the outlet dicharge wa almot clear and a table cone wa formed. The ide lope of the fluhing cone wa cloe to the angle of repoe of the ubmerged polymer particle. At thi phae, the table condition wa defined a the initial etup of Flow1. Opening the luice gate at the incremental rate of cm per minute wa performed at the tart of the experimental Flow1 until the opening reached 6 cm in height. The water urface then dropped becaue the outflow capacity wa much greater than the inflow dicharge. When the water urface elevation near the gate wa fully drawn down and became cloe to the rim of the fluhing cone (or the apex of the depoition), the flow condition changed from preurized flow into the free urface flow condition. At thi tage, water urface gradient near the luice gate increaed ignificantly. The flow firt eroded the rim of the fluhing cone and began cutting through the ediment depoit. Conequently, retrogreive eroion took place along the initial channel and propagated toward the uptream creating omehow a traight fluhing channel. At thi phae, a ignificant amount of ediment depoit wa fluhed through the reervoir, and the initial channel deepened and widened a a reult of largecale eroion in the reervoir. Lateral eroion on the floodplain wa alo oberved before the fluhing flow wa drained and confined into the fluhing channel, epecially near the dam. The entire experiment lated for about 50 minute. The bed form oberved in thi traight fluhing channel wa in the form of a plane bed. The experimental data are ummarized in Table 1. Flow No. Inflow dicharge (m /) Ex. No. Table 1. Summary of experimental data Initial water tage at dam (m) Initial depoit depth (m) Length of ediment bed (m) Initial bed lope(%) Running time (min) ,,, 4, , 7, 8, RESULTS AND DISCUSSION a) Behavior of cone formation procee and fluhing channel evolution After conducting two of the initial experimental run, it wa een that the hape of the luice gate entrance wa the dominating factor in dictating the fluhing channel to follow a traight path. In experimental run Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

10 18 N. Talebbeydokhti / A. Naghhineh no.1 (Flow1), the lack in the luice gate entrance caued the fluhing channel to be formed at the left bank of the flume, right behind the gate. Therefore, it modified the entrance of the luice gate by intalling two teel plate with a length of 0 cm and height of 10 cm at each ide of the gate (ee Fig. 4b). But the fluhing channel till followed a meandering pattern. Modifying the entrance of the gate alo made the fluhing cone bigger and more evident. Figure 5-7 how the cone formation, extenion and detruction near the luice gate during an experiment with modified luice gate entrance. At firt, the ediment depoited in the modified ection wa fluhed out and then the fluhing phenomena locally progreed retrogreively. In thi ituation, the retrogreive eroion did not extend into the reervoir and the ediment were eroded in the vicinity of the luice gate entrance. When the uniform flow condition wa atified, the gate wa opened manually at a contant rate. At thi tage, the upper layer of the polymer particle were all moved through the whole of the reervoir to the gate, and for the firt five minute, there wa no clear local cone or fluhing channel formation. But after thi, by further lowering the water level, the reervoir reached to a temporary table ituation and the fluhing cone formed. Figure 5 how the heart-like fluhing cone formation. The hape of thi fluhed formation indicated the ymmetry of the fluhing operation in the reervoir. Figure 6 how the fluhing cone five minute later when the heart hape wa getting bigger and extending laterally. A few minute later, when the water level in the reervoir wa lowered completely and reached the open channel flow condition, the cone ymmetry vanihed and extenive change took place. At the end of the experiment, when the inflow and outflow dicharge were almot identical to each other for the econd time, the heart like hape of the fluhing cone wa een again, but it wa no longer ymmetrical (ee Fig. 7). Fig. 5. Local fluhing at the beginning of an experiment, near the luice gate, fully ymmetric Fig. 6. Local fluhing after a few minute of the ame experiment, near the luice gate, till ymmetric Fig. 7. Fluhing formation near the luice gate at the end of the experiment, unymmetrical A fluhing channel wa formed about 15 minute after opening the gate, when the water level wa le than 6 cm (the optimum ize of gate opening). The fluhing channel could be recognized when the left and/or right bank in the reervoir were evident. Gradually the fluhing channel deepened and it width Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

11 Fluhing ediment through 19 decreaed. Figure 8 and 9 how the ketche and photo of the fluhing channel at the end of experimental run no.1 (Ex.1) of Flow1 and experimental run no.6 (Ex.6) of Flow, repectively. The reulted fluhing channel wa not a traight a expected baed on reult of previou work [4]. There were ome reaon that explain the meandering pattern of the fluhing channel. The major caue of thi behavior tem from the non-uniform ditribution of ediment particle or mall change in polymer particle hape. (a) (b) Fig. 8. Flow1, Qin = (Lit/ec), Ex.1: a) Plan-view of the fluhed reervoir, b) The ame experiment (looking downtream) (a) (b) Fig. 9. Flow, Qin = 0.68 (Lit/ec), Ex.6: a) Plan-view ketch of the fluhed reervoir, b) The ame experiment (picture i taken from the uptream end) It ha been een that fluhing characteritic of thee type of ediment particle are (e.g., polymer particle with denity cloe to water and without coheion) imilar to loe ediment. A large amount of the ediment wa fluhed in the firt 0 minute of fluhing duration. Of coure in thi duration, the fluhing channel might have had ome change in it bed, but the general bed trend remained table. After thi time, the ediment outflow dicharge wa almot contant and decreaed at a gradual rate, but the fluhing channel wa tabilized and it bed no longer changed. Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

12 10 N. Talebbeydokhti / A. Naghhineh The hape of the fluhed ediment at the uptream end of the reervoir where the fluhing channel finihed, looked like a funnel. Thi i becaue of the preence of the fluhing channel, which led the water to pa through the reervoir. Sometime, it wa oberved that the lateral bank of the fluhing channel lid into the fluhing channel and fluhed through the reervoir which wa related to the angle of repoe of the ediment particle and pore water preure. Of coure thi wa a long-term behavior and couldn t be oberved right after lowering the outflow ediment dicharge. b) Variation of meaured parameter with time due to fluhing For better undertanding of the fluhing procee, the data have been preented graphically. Figure 10 and 11 how the variation of outflow water dicharge, outflow ediment dicharge and water depth at the uptream and downtream end of the reervoir with fluhing time. Figure 10 i related to Flow1 and Fig. 11 i for Flow Outflow ediment dicharge (kg/), Water Level (m) Outflow water dicharge (Lit/) W ater Depth in 0.m From Dam (m ) W ater Depth in 4.8m From Dam (m ) Sediment O utflow Dicharge, Qo (kg/ec) O utflow W ater D icharge, Q o (L it/ ec) Fluhing Duration (m in) Fig. 10. Relationhip between ediment outflow, water outflow and water depth at two end of the reervoir (Flow1, Qin = (Lit/), Ex.1) Outflow ediment dicharge (kg/), Water Level (m) Outflow water dicharge (Lit/) W ater Depth in 0.m From Dam(m) W ater Depth in 4.8m From Dam(m) Sediment Outflow Dicharge, Qo (kg/ec) O u tflow W ate r D ich arg e, Q o (L it/ec) Fluhing Duration (min) Fig. 11. Relationhip between ediment outflow, water outflow and water depth at two end of the reervoir (Flow, Qin = 0.68 (Lit/), Ex.6) All the figure how the ame trend for water depth variation with time. The water depth variation with time at the uptream end of the reervoir i higher than the downtream end for all the experimental run. The water urface profile in the reervoir after opening the luice gate repreent M profile. After analyzing the data of the experimental reult, it wa found that the meaurement hould be performed right after opening the luice gate. Unfortunately, we mied recording and documenting the peak of the water and ediment dicharge. By calculating the approximate falling rate of the outflow water and ediment dicharge for each experimental run, it wa found that the falling rate of the outflow ediment Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

13 Fluhing ediment through 11 dicharge i about 0% higher than outflow water dicharge. Thi mean that the fluhing wa performed uccefully. c) Tinghua univerity method (TUM) The tranporting capacity of fluhing flow can be etimated uing an empirical method reported in IRTCES []. Thi method i baed on obervation of fluhing at reervoir in China, where the predominant practice i annual fluhing and o relatively little conolidation occur between fluhing operation. The method i baed on the following equation [,, 4, 5]: f S 0. 6 Q Q = E (16) B where Q i ediment tranporting capacity (ton/), Q f i fluhing dicharge (m /), S i bed lope, B i channel width (m), and E i a contant et from the ediment type: 1600 for loe ediment, 650 for other ediment with a median ize finer than 0.1 mm, 00 for ediment with a median ize larger than 0.1 mm, 180 for fluhing with a low dicharge. Equation (16) wa attributed to Tinghua Univerity by IRTCES (1985) and i referred to here a the Tinghua Univerity method (which wa derived in ection- a Eq. (14)). The validity of the Tinghua Univerity method (TUM) wa examined for calculation of tranporting capacity of fluhing flow for Polytyrol 14E particle. All the parameter in TUM (i.e. outflow ediment dicharge and the parameter contain the effect of outflow water dicharge, water urface lope and width of the fluhing channel in each time interval) were meaured in thee experiment. Figure 1 how the relationhip between Q o and Q 1.6 o S 1. w /B 0.6. The coefficient of erodibility (E) i the lope of the bet-fitted line (which hould follow the format of Y = ax). 1E- 1E-4 Qo (ton/ec) 1E-5 1E-6 1E-7 1E-9 1E-8 1E-7 1E-6 1E-5 [(Qo^1.6)*(Sw^1.)]/(B^0.6) Fig. 1. Relationhip between ediment outflow and Q o 1.6 S w 1. /B 0.6 from the experimental data (TUM) All the data of all experimental run (both Flow1 and Flow) were ued for drawing Fig. 1. The formula for the bet-fitted line wa a follow: Y = X (17) with the correlation factor of: R = more than / of the data point follow the bet-fitted line trend. From the reult of thi tudy, the coefficient of erodibility for the polymer particle i around 900, which i within the range of above the value. The polymer particle denity i cloe to water, the particle are noncoheive, and they behave the ame a loe ediment and fine-grained particle. Thi reult validate the reult which were mentioned by Chinee field data []. Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

14 1 N. Talebbeydokhti / A. Naghhineh d) Efficiency of fluhing Efficiency of fluhing ediment through the reervoir i important in determining the feaibility of fluhing operation when the value of tored water i high. In the preent tudy, fluhing effectivene wa analyzed in one of the fluhing event through the flume. With repect to the amount of water uage and the value of fluhed ediment, fluhing efficiency, F e, i defined a [1, ] F V V o i e = (18) Vw in which V o i volume of fluhed ediment in the time interval t; V i i newly-added volume of depoit from the uptream ediment upply in t, and V w i water volume ued in t. Without ediment upply from the uptream of the reervoir, fluhing efficiency, F e, outflow ediment dicharge, Q o, and cumulative volume of fluhed ediment, V t, are plotted againt time in Fig. 1 and 14 for Ex.1 of Flow1 and Ex.6 of Flow, repectively. A een in the mentioned figure, the general trend of the curve i imilar. Table clarifie the behavior of the parameter V t and cumulative volume of fluhed ediment for both inflow. Thi table indicate that the average of the peak of V t in Flow1 (with lower inflow dicharge) i about 10% le than that in Flow (with higher inflow dicharge), which i in agreement with what wa expected. Within the firt ~4 minute of the fluhing duration, the cumulative volume of fluhed ediment reached to fifty percent of it ultimate amount, and then increaed moothly. Thi i the general trend for all V t curve. Comparing the average percentage of the time to reach the 50% of the maximum amount, a difference of about 8~10% between Flow1 and Flow i evident. Due to higher inflow dicharge in Flow, more ediment fluhed in a horter period of time in comparion with Flow1, which had le inflow dicharge. Ex. number Table. Statitical reult of the cumulative volume of fluhed ediment in Flow1 and Flow Max. cumulative volume (m ) Flow1 (Q in = Liter per econd) Time of reaching 50% of the Max. over the fluhing duration* Time of reaching 50% of the Max. over the fluhing duration (%) Time of reaching 50% of the peak over the total time Time of reaching 50% of the peak over the total time (%) / / /50 8 0/ / / / / /56 8 6/66 9 Ave Flow (Q in = 0.68 Liter per econd) /4 19 / /50 8 0/ / / / /66 4. Ave *The fluhing duration i about 10~15 minute le than the total experimental duration becaue of the delay in formation of fluhing channel. Table how the value of maximum fluhing efficiency value for all the experimental run of Flow1 and Flow. Comparing the average of the maximum fluhing efficiencie (F e ) of each inflow dicharge how that F e for Flow i 7% higher than Flow1 (with lower inflow dicharge) and thi i in agreement with what we had expected due to the literature. Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

15 Fluhing ediment through 1 Table. Reult of the maximum fluhing efficiency in all experimental run of Flow1 and Flow Flow1 (Q in = Liter per econd) Ex. number The Max. of fluhing efficiency (F e ) Flow (Q in = 0.68 Liter per econd) Ex. number The Max. of fluhing efficiency (F e ) Ave Ave The trend of outflow ediment dicharge (Q o ) rie rapidly at the beginning (tart of the experimental run) till it reache a peak, and then gradually fall till the end of the experimental run. The comparion between two curve of Flow1 and Flow (ee Fig. 1 and 14) how that the falling rate of Flow (with higher inflow dicharge) i % le than Flow1 (with lower inflow dicharge) E xperiment Duration (min) Fluhing Efficiency, Fe Sediment Outflow Dicharge, Qo (kg/ec) Cumulative V Fig. 1. Fluhing efficiency and cumulative volume of fluhed ediment v. time (Flow1, Q in = (Lit/), Ex.1) Experiment Duration (min) Fluhing Efficiency, Fe Sediment Outflow Dicharge, Qo (kg/ec) Cumulative Fig. 14. Fluhing efficiency and cumulative volume of fluhed ediment v. time (Flow, Q in = 0.68 (Lit/), Ex.6) 6. COMPARISON WITH PREVIOUS WORKS In 1985, IRTCES (International Reearch and Training Center of Eroion and Sedimentation) preented Equation (5-1), which wa attributed to Tinghua Univerity and i referred to here a the Tinghua Univerity Method (TUM) [5]. The data ued to develop the method were collected from reervoir in China, where fluhing practice and predominant ediment characteritic may not be repreentative of other region. Therefore it eemed neceary to invetigate and control the Tinghua Univerity method by collecting more data about fluhing with different type of ediment particle. Lai and Shen [] performed an experimental one-dimenional model of a reervoir to examine the TUM method. They modified a rectangular flume, 50 m long,.44 m wide, and 1.5 m high by elevating a part of Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

16 14 N. Talebbeydokhti / A. Naghhineh it bottom up to 0.60 m to prevent the outlet from being ubmerged at the downtream of the dam []. Water wa recirculated by a pump at the deired rate ranging from to (m /ec) by controlling the pipe valve []. Three luice gate were intalled on the dam and manually operated by rotor; each gate could be et eparately. But only the central luice gate (0.5 m high and 0.15 m wide) wa ued []. The ediment material ued in the experiment wa the walnut hell grit, a non-coheive lightweight material. The choice of the ediment ize wa baed on the incipient motion criterion, the dimenion of the flume and capacity of water upply. The denity of walnut hell grit wa about 190 kg/m, a median diameter of 1.5 mm, and poroity of 0.55; the gradation coefficient i 1.18 and the angle of repoe for ubmerged walnut hell grit i about 5 degree []. Lai and Shen, [] mentioned that at the beginning of each run, ediment within 9 m from the dam wa placed in the depoitional depth of 0.1 m above the datum (i.e. the zero elevation et at the bottom of the luice gate) with zero bed lope. A total of ix inflow dicharge have been performed in the Lai and Shen tudy and each experimental run lated for 0 minute. The initial water tage at the reervoir for all the experimental run were about 0.17 m []. Table 4-6 how the comparion between the preent tudy and that of the Lai and Shen []. By referring to thee table, it can be een that Lai and Shen [] have had a larger flume, which behaved cloer to the prototype, and alo with particle denity higher than that in the preent tudy. Thi made their work fater in the fluhing operation and the fluhing channel wa een at the beginning of it formation through the ediment particle. Table 4. Sediment particle characteritic in preent tudy and Lai and Shen, [] Characteritic of the Sediment particle Preent tudy (001) Lai and Shen (1996) Denity (Kg/m ), γ Median diameter (mm), D Gradation coefficient, C c Angle of repoe (degree), θ 0 5 Poroity, n Table 5. Dimenion of the experimental model in preent tudy and Lai and Shen, 1996 Dimenion of the experimental model Preent tudy (001) Lai and Shen (1996) Length of the flume (m) 0 50 Width of the flume (m) Height of the flume (m) Height of the elevated flume bottom (m) 0.16 * 0.60 Length of the tet-ection in the flume (m) 11.5 Cloe to 19 Width of the luice gate (m) Height of the lice gate (m) It follow the relationhip between the bottom outlet poition in Sefid-Rud Reervoir with total height of the dam. Table 6. Experimental data in preent tudy and Lai and Shen [] Experimental data Preent tudy Lai and Shen [] Number of inflow 6 Range of the inflow dicharge (m /) and to Length of the ditributed ediment particle in the reervoir (m) Thickne of the ditributed ediment particle in the reervoir (m) Initial water tage in the reervoir (m) About 0.10 About 0.17 Initial bed lope and topet lope in the depoit (%) Opening ize of the luice gate (m) Running time (min) About 65 * About 0 ** * Thi i the total time of each experimental run. But the fluhing tarted with 1-15 minute delay (after complete drawdown in water depth and tarting the retrogreive eroion and the fluhing channel appearing in the reervoir). ** The delay in Lai and Shen [] wa about 5~7 minute. Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004

17 Fluhing ediment through 15 A higher amount of denity for the ediment particle in Lai and Shen work made the data point to follow the trend of Curve III with a coefficient of erodibility (E) of about 180 [, 4]. Comparing the calculated coefficient of erodibility (E) from the preent tudy and that of Lai and Shen [] how very high value for thi parameter in the preent tudy. The E value i etimated to be about 900. Thi i becaue of the very low amount of denity for the ediment particle in the preent tudy (cloe to water). The high value for E prove that the fluhing operation i much more ucceful in thi tudy. Comparing the maximum amount of the fluhing efficiency and cumulative volume of fluhed ediment how that thee value are about twice of thoe in the preent tudy in comparion with the tudy of Lai and Shen []. Thi i due to the type of ued ediment particle (polymer particle with very low denity). Lai and Shen [] alo have mentioned that fifty percent of the total volume of removed ediment wa fluhed out in one third of the experiment duration, which i cloe to the preent tudy. The general trend of Lai and Shen [] curve and the preent tudy are the ame. Thi mean that the fluhing operation ha had a unique trend for thee two different type of ediment particle. Figure 15 how all the reported data point in one graph. It contain the range of the field data point reported by IRTCES [] and that of the tudy of [], beide the data point of the preent tudy. Fig. 15. Comparion between reported data point of fluhing operation, uing TUM method, contain field data point of Chinee reervoir (IRTCES, 1985), Lai and Shen [] data point of an experimental model (with the coefficient of erodibility, E, about 180) and experimental data of the preent tudy (E i about 900) [, 4, 5] A een in Fig. 15, the field data point (which wa collected from the Chinee reervoir) tand higher than the data point of the laboratory model and thi i becaue of the dimenion of the reervoir, water depth, water and ediment dicharge and fluhing channel width in the prototype. Lai and Shen [] data point are lower than the preent data, and that i becaue of more reitant ediment particle againt fluhing. The data point in the preent tudy follow the ame trend a other data point, but are located in the lower part of the graph and that i due to the characteritic of polymer particle. It eem that the very low denity of particle in thi model made them behave imilar to very fine ediment particle in the prototype. So the data point are located cloe to the area of loe ediment particle. Winter 004 Iranian Journal of Science & Technology, Volume 8, Number B1

18 16 N. Talebbeydokhti / A. Naghhineh 7. CONCLUSIONS AND RECOMMENDATIONS A one-dimenional experimental model wa ued to invetigate the fluhing procee uing very lightweight, uni-ized polymer particle a the depoited material in a reervoir. Two different inflow rate and m / were ued (i.e. Flow1 and Flow, repectively), and a total of nine experimental run were performed for both inflow. The reult are preented in two part, qualitative and quantitative reult. In the qualitative part, the procee of cone and fluhing channel formation were oberved and documented and are in agreement with the literature. The heart-haped and ymmetric cone formation at the beginning of the fluhing period and the meandering effect in the fluhing channel were alo oberved. In the quantitative part, the Tinghua Univerity Method (TUM) wa ued by meauring the outflow ediment and water dicharge, water urface lope and fluhing channel width in every time interval of two minute to calculate the coefficient of erodibility (E) and to examine the TUM method. It wa found that the value of E are in the expected area (and wa about 900). The reult i in agreement with the phyical characteritic of the ued polymer particle, and thi validate the TUM equation. Value of the fluhing efficiency, F e, for each experimental run how that F e for Flow i 7% higher than Flow1 (with lower inflow dicharge), which i in agreement with what we had expected. Within the firt ~4 minute of the fluhing duration, the cumulative volume of fluhed ediment, V t, reache to fifty percent of it ultimate value and then increae moothly. Thi i the general trend for all V t curve. Comparing the average percentage of the time to reach 50% of the maximum value, how a difference of about 8~10% between Flow1 and Flow. Higher inflow dicharge in Flow reulted in more ediment fluhed in a horter period of time than in Flow1 with le inflow dicharge. Comparing the preent reult with the previou reult of TUM how a imilar trend for all data. REFERENCES 1. Brandt, S. A. (000). A review of reervoir deiltation. International Journal of Sediment Reearch, 15().. Lai, J. S. & Shen, H. W. (1996). Fluhing ediment through reervoir. Journal of Hydraulic Reearch, 4().. Atkinon, E. (1996). The feaibility of fluhing ediment from reervoir. Report OD17, HR Wallingford. 4. Shen, H. W. (1999). Fluhing ediment through reervoir. Journal of Hydraulic Reearch, 7(6). 5. Atkinon, E. (1998). Reervoir operation to control edimentation: technique for aement. In: The Propect for Reervoir in the 1 t Century, Proceeding Of the Tenth Conference of the BDS held at the Univerity of Wale, Britih Dam Society. 6. Shen, H. W., Lai, J. D. & Zhao, D. (199). Hydraulic deiltation for non-coheive ediment. Proceeding of the 199 Annual ASCE Hydraulic Engineering Conference, San Franico, H. W. Shen, S. T. Su and F. Wen, ed. 7. Brune, G. M. (195). Trap efficiency of reervoir. Tranaction, America Geophyical Union, 4(). 8. Fan, J. & Morri, G. L. (199). Reervoir edimentation. II: Deiltation and long-term torage capacity, Journal of Hydraulic Engineering, Lai, J. S. & Shen, H. W. (1995). Degradation fluhing procee in reervoir. HYDRO 000, 4, Thoma Telford, London. 10. Olen, N. R. B. (1999). Two-dimenional numerical modeling of fluhing procee in water reervoir. Journal of Hydraulic Reearch, (1). 11. Streeter, V. L. & Wiley, E. B. (1985). Fluid mechanic. 8 th edition, New York: McGraw-Hill Book Company. Iranian Journal of Science & Technology, Volume 8, Number B1 Winter 004