OBSERVATION AND ANALYSIS ON CHARACTERISTIC OF NATURAL RIVERS

615Journal of Hydrodynamics, Ser. B, 2005,17(5):615- 620China Ocean Press，Beijing - Printed in ChinaOBSERVATION AND ANALYSIS ON CHARACTERISTIC OF NATURAL RIV-ERS"LU Jin-you, XU Hai-tao, YAO Shi- mingYangtze River Scientific Research Institute, Wuhan 430010，China，E- mail:lujy@ mail:crsri. cn(Received Aug. 12， 2004)ABSTRACT: Based on the previous study on turbulent Nanjinguan. Based on the field data, the large-scalestructure, field experiments on fluctuation velocity were car- turbulent structure was investigated. In this paper,ried out in section Huangingmiao of main stream of Y angtzeanalysis is carried out based on the field data of sec-River with a 3-D Doppler velocimeter. And a number of datation Huanglingmiao.such as velocity of different discharge and different verticalline were obtained. According to the field experiment data，period，frequency， period function and probability densityfunction of large scale coherent structure in natural river wereprescribed by means of quantitative analysis. And the problemabout distribution of time average velocity, turbulent intensity8069.44 3#， 14# 2#and Reynolds stress along depth were further analyzed andcalculated. The results in this paper will deepen the under-王40-stand of large- scale turbulent structure in natural river and somake the base for applying“ turbulence theory” to water con-5C200400600servancy，hydroelectric, environment engineering and so on.L./mKEY WORDS: natural river, large scale turbulence，turbu-Fig.Landform of huanglingmiao hydro-section and ar-lent mechanismrangement of observation vertical-lines2. GENERAL SITUATIONS OF OBSERVATIONS1. INTRODUCTIONThe landform of Huanglingmiao hydro-sectionMost of flows in natural rivers are turbulent,and arrangement of observation vertical-lines areind many fluid dynamic questions of engineeringshown in Fig. 1. By placing a 3-D Doppler velocimemust be solved by turbulence theory. But at pres-ter at the vertical line about lm deep under the wa-ent, turbulence theory has not reached the level toter，point velocity fluctuation processes of everyobtain a general solution for fluid dynamic questionsvertical line were observed. Observations were perof practical engineerings， many practical questionsformed at different vertical lines at thalweg or nearcouldn't be solved absolutely.In the past, abundant researches on turbulencethe bank，under different discharges of high andmiddle water level. Velocimeter can set observationfeatures of flows have been performed by manypoints at the vertical line at discretion. Flow genesscholars[1-8]，but most of those works aimed atof river reaches of each observation are shown influme experiments, few aimed at the turbulence ofTable 1. Each group of observation data includesnatural flow. In this item, with a 3-D Doppler velo-depth of observation point h，longitudinal velocitycimeter, turbulent structures in natural rivers were中国煤化工vertical velocity u。observed at different sites: estuary of Yangtze Riv-synth.MHCNMHG on. Sampling inter-er, section Huanglingmiao， the outlet of open di-vals are between U. bs -3s，and observations allversion channel of the Three Gorges Project andProject supported by the National Natural Science Foundation of China (Grant No: 50179001).Biography :石方熬据u (1963-). Male, Mater, Professor Senior Engineering616Table1 Flow genes in section Huanglingmiao when observation was carried outNumber of observation vertical line1#2#3#4#Distance (m)350310470Sampling intervals (s)2. 472.272.00.570. 57Observation last (h)1. 070.681.00.6Discharge (m3 /s)3000030000 4220042200 1120011200Water depth (m)60. 6317. 163.2 17.8 60.87 25. 9Width of water surface (m)504511495Hydraulic radius (m)34. 134.134. 8734.87 33. 1333. 13Slope of Water Surface (%)0. 1020. 1020. 1810.1810.019.0. 019Profile Average Velocity (m/s)1.674 1. 6742. 2772.277 0. 6550.655Friction Velocity (m/s)0. 1850. 1850. 2490.249 0. 0780. 078Reynolds Number (X10' )6.39.02.4Bed Materials doo (mm)0. 3550. 4010. 3130.876 0. 331). 251lasted more than 1h. Consequently, the observation 20s 30s，while the 1. 105m/s velocity of about 80s-data reflect the relatively strong fluctuations. Ultra- 100s. Through analysis， we draw a conclusion: thesonic launch frequency of velocimeter is 300Hz. As average point velocity value of long- term observa-a result, the data collected is not instantaneous val- tion comes forth at a minimum interval, the intervalue，but average value of samplings in the pre-estab- will both increase by adding to or subtracting fromlished time range Ol .the average velocity a same value， and the increas-ing extents are approximately equal. All of the fluc-3. ANALYSIS ON TURBULENT STRUCTUREtuation progress lines can be decomposed into sever-3.1 Quasi periodicity of water turbulenceal different periodic functions， so large scale turbu -Considering a certain point of every vertical line lence can be looked as a superposition of several vorespectively, velocity progress curves of three directexes of different scale and several syntonic waves oftions are drawn along the time. Despite of the different turbulence periodicity.difference of observation time， vertical line and 3. 2 The probability distribution of fluctuationpoint, the fluctuation progresses of velocity are ba-velocitysically similar. Fig. 2 shows fluctuation progressesDo probahilitv statistics of us, uy，and u。ofof three directions of the point at Line 3 # where each中国煤化工rin the emergence fre-the relative depth is 0. 96. From Fig. 2，we learn quencC N M H Gvelocity extension andthat large- scale turbulences have quasi- periodicities draw the frequency curve. The results in Fig. 3 andof different sizes. Statistics indicate that the average Fig. 4 show the experiential frequency curves of lon-velocity of longitudinal direction is 0. 709m/s， the gitudinal and transverse fluctuation velocities in ver-0.805m/方号数据Y comes forth at interval of about tical lines. Figure 5 shows both experiential and617Y=γ(1)hwhere Y is the relative water depth, y is the depthunder the water surface of the observation point inthe vertical line， and h is the depth of the verticalline. From Fig. 5 we learn that the probability dis-1020000tribution of instantaneous velocities in free turbu-15- - Longiudinal velocityTransverse veiocity一Vertical veloclylence region is approximately normal, namely:Fig.2 The variation of velocity with the time at the pointwhere the relative depth is0.96 at Line3# (Q= f(u) =2yv2(2)(2π)1/2σ。11200m* /s)where f(u) is the probability density function of in-stantaneous velocity u，u is the temporal average2.0 [1 Y=008 3Y=0.42 sY=0786Y= 097value of u，and σ。is the average variance of u .2Y=024 4 Y= 0.6042200m/sHowever, the probability distribution of instantane-是1.0ous velocity in the strong shear turbulence regionnear the wall is deflective. Contrast between the ex-0.0periential and theoretic frequency curves of the.1.0 .1.03.0measurement point where the relative depth is 0. 97u/ms'in Fig. 5 shows that two curves have distinct differ-Fig. 3 Probability distribution of longitudinal fluctuationence. Curves in Fig. 3 and Fig. 4 reflect that, withvelocity of different depth at Line 1 #the increase of the water depth， at every measure-ment points，the value of temporal average velocityapproaches to 0 and the average variance of instanta-. I Y=0.0842200m'/s2 Y= 0.24neous velocity increases gradually， except that the.0 t3Y= 0.42average variance decreases at the point where the54r=0.60.relative depth is 0. 97.5 Y= 0786Y= 0973.3 The tem poral average velocity distribution a-long the vertical line .The temporal average velocities of each point a--2.0)02.0long three directions were calculated, and the distri-u/ms"Fig. 4 Probability distribution of transverse fluctuationbution regulations of the temporal average velocityalong the vertical line were analyzed. Using bothvelocity of different depth at Line 1#power and logarithmic velocity formula, the results .were calculated and contrasted. The power velocity1 Y= 0.08. theoretic 5 Y= 097. thcoreticformula is:2.0 t 2Y= 0.08. experientiad 6 Y= 0.97. xpricntial3 y= 042. thcoretic4 r= 0.42. experientialu,= up(1+ m)4-二y)m(3)where_ u, is the temporal average velocity of the0.0.023.point中国煤化工re average velocity ofw/ms'the、:MYHC N M H Gwer coefficient, com-Fig.5 A contrast between experimential and theoreticalmonly, m= 1/7. The logarithmic velocity formulaprobability distribution of longitudinal fluctuationis:velocity along depth at Line 1#theoretical frequency curves of longitudinal fluctua-“= 5. 75log[-30.2(h- 2)x](4)tion veloct2数握ine 1#. In the figure:uK.618uo-log[-30.2(h- x](5)where u。is the friction velocity， K, is scale value oflog(30, 2hx )K,roughness，commonly, dgo of riverbed materials ’size distribution curve is chosen, x is emendationvalue which is the function of K,/δ， δ is the thick-where uo is the temporal average velocity of theness of laminar layer.point where the relative depth is the minimal amongall the measurement points. From Fig.6 and Fig. 7we learn that the variation tendency of velocity dis-Verical Transverse LongitudinalField valuetribution along the vertical line can be described by2 Formula (3)exponential or logarithmic curve. Only the formula2Formula (5)， 0.6used must be tested by field data. By comparison,1.0exponential velocity formula accord to field data bet--0.50.5.525ter.w/ms'3.4 The lurbulence intensity and the relalive lur-bulence intensityFig. 6 Contrast between field and counted value of tempo-Commonly, the average variance of the fluctua -ral average velocity along depth at Line 1#tion velocity is defined as the turbulence intensity ，namelyσn, = (u;?)1/2 . The absolute value of fluctua-gTransverse Longitudinal Verticaltion velocity u; also can be defined as the turbulence0.2Formula(3)intensity, namelyσn, =| u;|. The ratio of the tur-0.6Frmula(5)bulence intensity and the corresponding temporal.0average velocity is defined as the relative turbulence-0.1.2.5intensity, namely or0=|u;|/u;or0= (u7)12/u;u/ms'Fig.7 Contrast between field and counted value of tempo-ral average velocity along depth at Line 4 #0rUsing the logarithmic velocity formula， the。0.4-+-. Transverseslope value of water surface is needed in order t(-十- Verticalcalculate u.，therefore， the logarithmic velocity0.8 tformula only used to calculate the longitudinal ve-locity distribution. Figure 6 gives the temporal av-0.4erage velocity distributions of Line 1 # when thedischarge is 42200m*/s along the depth both theo/mfield value and the calculation value obtained withpower and logarithmic velocity formula, and Fig.'Fig. 8 Distribution of turbulence intensity along depth atLine 4#gives the distributions of Line 4 #。Something mustbe indicated is that the calculation value by the loga-rithmic velocity formula has obvious difference toIn this paper, the formulasσn, = (u?)1/2 and Othe field value, the reason is that choosing the val- = (u?)V/2/ u; were used. The turbulence intensityues of natural u. , δand K, has power function char- distribution of three directions at Line 4# along theacter. The values in Fig. 6 and Fig. 7 have beenre- vertical line is showed in Fig. 8, and the distribu-vised, the formula is:tion qA turhulonce intensity is showed in中国煤化工，Fig.that the turbulenceu, 5. 75log[inten:MYHCN M H Ge is quite evenness atmostly area (where the relative depth is 0-0. 6 tak-u,= uo=u. 5. 75log(( 30.2hx)ing the water surface as the baseline), and approxi-mates linear variation, it reaches the maximum nearthe riverbed， then decreases dramatically. Because619of the setting of the measurement apparatus itselfand the limits of the precision，the point showed in-0.2the Figure with a relative depth 1 is not a real river-bed point but the point having a less than 1m dis-0.2tance to the riverbed. Considering the temporal av->，erage values, we learn that the turbulence intensi-0.6, Eyyties of longitudinal and transverse direction are basi-cally equivalent， while the turbulence intensity of0.1vertical direction is the minimum, which is approxi-(t/p)Vm's2mately 1/3-1/4 of the two formers.Fig.10Distribution of normal Reynolds stress alongdepth at Line 1#0r3.5 The distribution of the Reynolds stress alongthe vertical line十Longitudinal0.4十- TransverseThe expression of the Reynolds stress τi is:+ Vertical0.8Tij =一ρu;u;(6)812In this expression， when i = j，τi is the normal0Reynolds stress，wheni ≠j, tτg is the shear Reyn-Fig.9 Distribution of relative turbulence intensity alongolds stress. Based on the fluctuation velocity of eachdepth at Line 4# .point，each component of the Reynolds stress can becounted directly. Since Reynolds stress τ。can beThe distribution of the relative turbulence in- looked as a symmetrical two -order tension, only 3tensity along the vertical line takes on different Fig- normal stresses and 3 shear stresses need to beure with the change of the situation of the investiga- counted. The distribution of normal Reynolds stresstion line. At Line 4# (which is near the bank)，the and shear Reynolds stress. at Line 1井when thetransverse and vertical relative turbulence intensity discharge is 42200m2/s along the vertical line, aredecreases gradually with the increasing of the depth showed in Fig. 10 and Fig. 11 respectively. Thewhen the relative depth is less than 0.9，and rea- stress values in the Figures are absolute values cal-ches the minimum at the situation where the relative culated by expression (6).depth is 0. 6-0. 8，then increases dramatically withthe increasing of the depth，and reaches the maxi-mum at some point near the bed. The relative tur-bulence intensity values of the transverse and verti--0.cal direction in this extent are all larger than 1.0.While the longitudinal relative turbulence intensityis almost invariable when the relative depth is lessthan 0.8, as showed in Fig. 9. At Line 1 # (which .is near the thalweg)，the relative turbulence intensi-0.05ties approximate linear distributions when the rela-tive depth is less than 0. 9，the values of the longi-(t/p)/m's2tudinal and vertical direction are almost invariable,中国煤化工Fig.nolds stress along depthand the values of three directions are all less than 1.YHCNMHGA common character of each observation line is thatthe relative intensity of longitudinal direction is theAs shown in the Fig. 10，at the thalweg，withleast one contrasting with that of transverse andthe increasing of the depth， the longitudinal andvertical direction.transverse normal Reynolds stress，firstly, increa-ses gradually， then decreases gradually after reac- mic velocity distribution formula has more deflec-hing the maximum value (the relative depth=0. 6- tions to the field value since the effect genes are not0.8)，finally, in the layer near the bed， the total easy to be confirmed accurately. (4) The verticaltendency is increase while the value varies greatly. turbulence intensities at each observation line， inThe vertical normal Reynolds stress is basically in- the extent with relative depth less than0. 6，is quite .variable along the depth. Near the bank, the distri- evenness, approximate to linear variation, and reachbution of longitudinal and transverse normal Reyn- the maximum near the river bed, then decrease rap-olds stress is similar to that at the thalweg，except idly. Based on the field data， the regulation of thethe position where the stress reaches the maximum Reynolds stress distribution along the vertical line isis more near the bed. The vertical normal Reynolds discussed, the field turbulence data in natural riversstress，similar to that at the thalweg，is basically are offered to study the characters about the Reyn-invariable along the depth. The tendency mentioned olds stress further and calculate the Reynoldsabove illuminates that, in some extent of the layer stress.near the bed，turbulence stress increases with thedistance to the wall， namely, viscous stress decrea-ses relatively. In upper flow region, the turbulence REFERENCESstress is prior. As shown in Fig. ll, the vary tend-ency of the shear stress τxy at the thalweg and near [1] ARNDT E. V. and IPPEN A. T. Turbulence measure-the bank is similar to that of longitudinal and trans-ments in liquids using an improved total pressure probe[J]. Journal of Hydraulice Research, 1970，8(2): 132-verse normal stress of itself， the vary tendency of158.tx andτy are all similar to that of the vertical nor-[2] HINZE J. , O. Turbulence[ M]. Second Edition, Newmal stress at the same line, namely, in the regionYork: McGraw- Hill 1975，229.where the relative depth is less than 0. 9，the value[3] RAICHLEN F. Some turbulence measurements in wa-of Tx and Tye has no evident change along the depth,ter[J]. Proc. ASCE,J. Engin. Mech. Div. ，1974， 93in another word, the value is almost invariable.(2): 73-97.[4WANG Xing- kui and Qian Ning. Turbulence character-4. CONCLUSIONistics of sediment-laden flow[J]. Journal of HydraulicIn this paper， based on the experimental inves-Engineering. ASCE, 115(6): 781-800.tigation, we have further understand on the turbu-[5] DOU Guo ren. Turbulent mechanics[ M]. Beijing: HighEducation Press, 1981， 11. (in Chinese)lent structure in river flow: (1) The fluctuation ve-[6] LIANG Zai-cao. Engineering turbulence[ M]. W uhan:locity progress can be decomposed into several dif-Huazhong Uniersity of Science and Technology Press,ferent periodic functions, and can be looked as a su-1999，91-121. (in Chinese)perposition of several vortexes of different scale and [7] WANG Xing kui. The turbulence features of open flow .several periodic syntonic waves of different turbu-[J]. Journal of Waterpower, 1993, (4): 12-21. (in Chi-lence periodicity. (2) The probability distribution ofnese)fluctuation velocity approximates to normal distri- [8] DENG Lian-mu, YE Min, WANG Dao -zeng. Oberser-bution in the free turbulent region，while it is par-vation and analysis on large- scale turbulence structure ofnatural river as well as Reynolds-stress calculation[J].tial distribution in strong shear turbulent regionJournal of Hydrodynamics, Ser. A, 2001，16(2): 187-near the wall. (3) Power velocity distribution for194. (in Chinese)mula accords to the field data better, while logarith-中国煤化工MHCNMHG