黄万里文集.pdf

I 序序 黄万里教授是蜚声中外的著名水利工程学专家，今年 8 月是他九十华诞。作 为晚辈和学生，与他相处也已经数十年了，总觉得应当做点甚么事，以表达我们 对老先生生日的美好祝愿。黄万里教授一生涉猎群书，知识渊博，视野广阔，著 述丰盛。他立论新颖，常常语出惊人，又能仗义执言，逆流顶风，坚持己见，独 战多数每每成为学术界和舆论界争论的焦点。世间对此褒贬不一，同时，对他也 朦罩着一缕缕神秘的色彩。然而，由于种种原因，他的诸多言论文章，只是散见 各处，未能集于一册，供人阅读、评判和研究。近十余年来，我们在帮助先生打 印整理他的讲义和文稿时，手边留下一些资料。于是几经酝酿，方才有了编辑出 版这本黄万里文集的动议。这得到黄先生的应允，也得到泥沙研究室和水利 系师友们的鼓励和支持。将我们现有的资料，经过适当筛选，编辑成册，再请先 生过目审校，以期能够赶在他生日之前印出。 黄万里教授早年在唐山交通大学学习，有深厚的数学和力学功底。1932 年毕 业以后，任浙赣铁路见习工程师。1931 年和 1933 年长江、黄河的大水灾，促使 他改行立志，学水利，治黄河，救国救民。在留学美国康奈尔大学、爱沃华大学 和伊利诺大学期间，他不仅学习水利工程的科目，更潜心研读有关的水文、气象 和地理等学科。1935 年和 1936 年先后获得硕士与博士学位。1937 年回国后，任 经济委员会水利处工程师，四川省水利局工程师，涪江航道工程处处长，从底层 的实际工作做起。1947 年，担任甘肃省水利局局长兼总工程师，又兼任水利部河 西勘测设计总队队长，主持陇西农田水利工程。1948 年应邀去东北解放区任东北 水利总局顾问。全国解放以后，到唐山铁道学院任教。1953 年全国高校院系调整， 方来清华大学水利工程系担任教授。1957 年，他力陈黄河泥沙问题的严重性，批 评苏联专家建议的三门峡水库规划是错误的。指出建库后泥沙淤积将使黄河北干 流与渭河两岸大量耕地淤没，居民将被迫迁移，三门峡水库不可以修建。同年， 因一篇花丛小语 ，被定为“右派” 。1964 年，三门峡水库因泥沙淤积严重而讨 论工程改建时，他不顾自己仍然戴着“右派”帽子，积极提出改建意见。 “文革” II 中他更遭厄运，作为“牛鬼蛇神”被扫地出门，从清华新林院的教授洋房被赶到 了地板下积着陈年脏水的北院小屋，每月领得 20 元生活费。后又被送到江西鲤鱼 洲农场“劳动改造” ，1973 年派到清华大学三门峡基地打扫厕所和接受批判。1978 年，这时他几乎是全国最后的一名“右派” ，终于也得到平反改正。以后他在清华 大学泥沙研究室工作，为教师和研究生开设统计与随机理论 、 治河方略和 治水原理等课程。同时继续研究连续介体动力学最大能量耗散率定律，分流 淤灌治理黄河策略，华北水资源利用，长江三峡工程，以及明渠不恒定流力学等 问题。九十年代以来，他极力反对长江三峡工程的开工，提出了许多十分尖锐的 问题，引起世人瞩目。 黄万里教授的一些学术观点和意见， 常常不为人所赞同和理解， 被斥之为 “异 端邪说” ，遭到反对和批判，得不到公开发表和申辩的机会。有的在被历史证明确 实是正确意见之后，仍然受到许多不公正的待遇。当然，他的见解有的不无道理， 有的也确有值得商榷之处。鲁迅说过， “倘要完全的书，天下可读的书怕要绝无； 倘要完全的人，天下配活的人也就有限。 ”但是，一个完善的社会应当有充分的大 度和包容。何况，在影响到国计民生、影响到子孙后代的重大工程技术问题中， 多一些对立面，多一些思考和论证，对于正确的决策和更加完善的规划设计，总 是一件好事。而且有的问题认识正确与否，还有待历史的检验。如果学术上没有 百家争鸣，只有长官意志和“一言堂” ，必将堵塞认识真理的道路，阻碍科学技术 的进步与繁荣，最后受害的将是国家和人民。 数十年来，黄万里先生所经历的坎坷磨难，所遭遇的升降沉浮，在我国知识 界中是十分少见的。但是，不论在甚么情况下，他对学术的严谨和认真态度，对 民众父老、对国家民族的一片赤诚之心，始终没有改变。在他还戴着“右派”帽 子的时候，毅然勇敢地站出来，坚持自己认为正确的意见。文化革命中，他一边 接受批判和劳动改造， 一边却在研究和草拟他的 “治理黄河方略” 。 改革开放以后， 他怀着极大的喜悦和高昂的热情，培养研究生，为青年教师讲课，指导他们进行 科学研究。他常常感激国家给予他这么高的工资，而自责未能对国家做出多少贡 献。为了水利系的课程设置，他多次找有关同志，提出应当开设“治河工程学” 的建议。1998 年长江大洪水以后，他更倍感焦急，责备自己过去教学方面的缺陷， 主动要求重上讲台，为研究生和教师讲授治河原理课程。他对生活充满希望，坚 信真理必将为人们所理解和接受，总能够保持乐观向上的精神风貌。近些年来， 在他身上相继发现多处癌症。他一面积极治疗，与病魔做斗争，一面仍然醉心于 长江、黄河等问题的研究，积极向有关方面提出自己的意见。他对事业执著，勇 III 于坚持真理；为人胸怀坦荡，处事光明磊落；对晚辈关怀爱护，真诚平等相待。 他在我们泥沙研究室和水利系的师生中，赢得了普遍的赞誉和钦佩。 黄万里先生生活的这九十年，是多么珍贵、多么难得的九十年啊。在他九十 华诞之际，我们愿以这本黄万里文集 ，表达对他的尊敬和祝福。 由于时间仓促，除我们现有的资料以外，未能专门去收集其他的资料。连黄 先生自己手边的资料也未能帮他进行整理。所以文集中所列文稿，远非先生 著作的全部。但是，他对水利工程学的一些基本理论问题的研究，对黄河治理与 长江三峡工程等重大问题的基本观点， 文集尽量予以反映。另外， 文集还 收录有黄先生的部分诗词,散文和几篇记者访谈录。我们希望通过这本文集 ， 可以大体了解到黄万里教授主要的学术成就和对一些重大科技问题的见解，可以 观察到像他那样一代学人为追求事业、追求真理的执着、艰难和曲折的历程，也 可以多少能够从中感受到他那鲜明的个性和高尚人品，欣赏到他那优雅的情趣和 秀美的文采。 编辑出版小组 2001 年 8 月 于清华园 IV 目目 录录 序序 水利工程学理论 水利工程学理论 钢筋混凝土拱桥二次应力设计法钢筋混凝土拱桥二次应力设计法唐山交通大学论文，1932 年 存目 铆钉接头中各铆钉应力推算法铆钉接头中各铆钉应力推算法 唐山交通大学论文，1932 年 存目 混凝土沙石配合最大容重决定强度论混凝土沙石配合最大容重决定强度论唐山交通大学论文,1932 年存目 暴雨洪水统计分析暴雨洪水统计分析 （康乃尔大学工程硕士论文，1935 年）存目 瞬时流率时程线学说 瞬时流率时程线学说 （伊利诺大学博士论文，1937 年）存目 洪流估算洪流估算 （水利电力出版社，1956 年）存目 工程水文学工程水文学 （水利电力出版社，1957 年） 存目 沙流连续方程意义的简释沙流连续方程意义的简释 1 连续介体动力学最大能量消散率定律 连续介体动力学最大能量消散率定律 11 连续介体动力学最大能量消散率定律的解释连续介体动力学最大能量消散率定律的解释 25 The Extremity Laws of Hydro-Thermodynamics Applied Mathematics x, t ] []tAstxTVPss,,;,,,, 11 tt xxLLL VhPtxVhP Q P ⋅ 自变量参变量消散率 最大热能量增率ΛΛΛΛ presents a ula for determining the quantity of water resources and argues that China is endowed with the most abundance of water resources in the world which meets the comprehensive requirements because of the appropriate distribution of time and space. This is against the state published data of 2.71012m3 for the average yearly quantity of water resources considered to be in short and unappropriately distributed in time and space. The problem is worth while for public discussions, as it affects so much the tactics of planning in hydraulic engineering. 82 The Velocity Profile ula along Section of Open Channel Flow Determined by the Law of Maximum Rate of Energy Dissipation William W.L. Huang, M.C.E, Ph.D. Professor Emeritus, Tsinghua University Synopsis This paper presents a ula of velocity profile along section of open channel flow determined by the law of maximum rate of energy dissipation proposed by the author in 1975. The analysis partly follows the mixing length theory for turbulent flow in that the length 1 is a dimensionless multiple of the depth y of the level of flow line, but replaces the Karman constant k0.4 by a variable ηηy which varies from 1 at the bottom to 0 at the surface of flow. The total rate of energy reserved in the section shall invariable be a minimum. The calculated velocity profile by the proposed ula has been checked precisely by the experimental data measured by the U.S. Geological Survey in Denver, Colorado. The discrepancy of results in using the Prandtl-Karman ula with the measured data is manifested in comparison. Recapturation of Historical Development Early in the 17th century numerous Italian and German engineers curiously believed that the velocity in a vertical counting downward from the surface increased with the depth even comfirmed by fallacious experiments. In 1848, Dupuit developed from theoretical consideration the equation 83 ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −− h y uuuu minmaxmax Until 1858, Bazin, an assistant to D’ Arcy, developed the parabolic curve of velocity profile from results of experiments in the middle of a natural river. The equation proposed was 2 max 20⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − h y hJ uu in which u is the velocity of flow at the depth y, h- the maximum depth, J – the slope. Later, Pressey, in America, Jasmund and Bolte, in Germany, improved the Bazin’s result of the constancy of the value 20 by introducing the effect of the roughness of channel on the increase of curvature of the profile. R.Jasmund 1893 – 97 examined 445 velocity profiles based on his observations on the Elbe. He proposed four types of curves, i.e., parabolas with horizontal and vertical axes, hyperbola and logarithmic curves for trials in fitting the data, and concluded that the latter was the best fit u a b lg y c where a, b, and c are constants for a particular stream. Not until 1883, when the essence of turbulent vs.laminar flows was fully understood through the works of O. Reynolds, different ulas were developed for the two regimes. The Prandtl-Karman semi-rational approach to the logarithmic ula for turbulent flow has been popularly accepted. Nevertheless, the distribution of velocity along a vertical of flow still remains void of reason. The subject, however, is of wide interest to hydraulics in practice, so as to answer the requirement of verifying the Prandtl-Karman ula; as well as to the mechanics of sediment transport which is closely related to the shape of the vertical velocity curve. On the Inconsistencies in the Prandtl-Karman Analysis L. Prandtl and Th. von Karman have successively developed the mixing length theory and velocity deficiency Law of turbulent pipe flow by coordinating theoretical analysis partially with experimental research. Nevertheless, these fruitful results 84 remain with inconsistencies in both theory and practice. Firstly, for a definite shear stress o τ γhJ, the effect of wall roughness is assumed to be only limited to a shallow region of viscous sublayer adjacent to the wall, while the velocity distribution in the turbulent core is assumed to be identical for all conditions of flow. Secondly, assumption has been made that the turbulent core velocity curve uy joins abruptly the viscous film straight line at the point of depth y δ11.6ν/u*, where u* is the shear velocity and ν, the kinematic viscosity. This is far from truth as shown by many recent velocity measurements close to the boundary 1. The slopes of velocity gradients dy du for the two regimes are radically different. Thirdly, in computing the average velocity by summation, the effect of flow within the range yδ is entirely neglected, which is reasonable only for the part of thin laminar film, but is not valid for the part of turbulent flow actually existing in the hypothetical layer yδ. Fourthly, the expression of mixing length 1 ky 0.4 y is not true even near the flow boundary. From the Nikuradze diagram of ub/u*lg ν * u D, ub – velocity near bed, D – effective diameter of bed particals it can be seen that * * / lg udu D v u d k b varies with different D. Other Nikuradze curves of 1/roy/ro also show that the slope k dy dl varies right from the outset and throughout the depth of flow y o h. There is no such universal constant k 0.38 ≈ 0.4, exists in nature. k0.4 exists only when 4 . 0lg * D v u 92 . 0 ln * D v u , 51 . 2 * D v u and ub/u*7.8 where the straight line tangent to the curve. As pointed out by M.S.Yalin2, at present state of knowledge, the exact of the variation of 1 with y throughout the thickness of the flow is not known. Hence it is not reasonable to leave k outside of the integration h y kky dy u u h o ln 1 * ∫ 85 The Law of Maximum Rate of Energy Dissipation The law of maximum rate of energy dissipation 3 4 was proposed by the author in 1975 as the second law of contimuous dynamics as an addition to the St. Venant conservation laws of matter and energy as the first category of laws. Only by simultaneous application of both laws, may a dynamic problem be solved. The corollary of the law in applied hydraulics is the Belanger-Boss theorem of minimum specific energy. The law has been proved by means of variational principle. The law provides that the mean velocity um and depth h of given constant flow shall be so distributed along the flow in x-direction that the total loss of head detween the two control section x0 and xxc is a maximum, i.e., ∫ c x o m dx RC u .max 2 2 R-hydraulic radius≈h 1 Then, among different combinations of h and g um 2 2 at the control section xxc, there is only one such that the total head representing the sum of kinetic and potential energy per unit weight of the fluid above the channel bottom must be a minimum, i.e., min 22 22 y p g u h g u mm γ 2 For any section x other than the control, with given values of um and h determined by backwater or drop-down curve the values of u and p, shall also be so distributed along a vertical that the rate of energy conserved is a minimum. This is the law proposed by the author to underlie the means of determination of the shape of the velocity profile along a vertical. The Tangential Conjunction of the Turbulent Core and Laminar Film According to the contemporary mechanics of turbulent flow the shear stress τ at a level y is composed of the laminar part τl is due to molecular viscosity μ and the part due to turbulent fluctuations t τ 2 2 ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ dy du dy du tl ριμτττ 3 86 the laminar partτl is significant only in the laminar film and the flow in the turbulent core can be considered to be independent of viscosity although the macroturbulent eddies do dissipate its energy into heat through overcoming of viscosity. Introducing the expression ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − h y u h y ghJ h y o 111 2 * ρ τ ρ τ 4 We may solve Eq. 3 and 4 simultaneously ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −≈−⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −ΛΛ 2 4 * 4 2 * 2 * 2 2 * 1 2 8 1 1 2 11 22 1 h y u l v h y u l v h y u l v l v h y u dy du ι 5 after expansion into series, which is convergent. It can be shown through numerical examples, that even at the conjunction of the two flow regimes, v/2l is not significant when compared with u*, so that we may write h y u dy du l−1 * 6 Let yηη be an unknown function of y to replace k as a constant in Karman’s hypothesis, such that h y yl−1η 7 Combining 6 with 7, we have in turbulent core y u dy du η * 8 On the other hand, the velocity profile in the viscous film is a straight line with a gradient v u dy du o 2 * μ τ 9 Naturally, the curve must tangent to the straight line with a common gradient in order to be continuous. Thus, 87 * u v y η 10 At the depth y01,/ * ηuv; * 2 * ,//uuvuyu ooo 11 This point of conjuction of the two flow regimes may be taken as the initial of the velocity profile at the bed. It should be at a distance * /uvyo above the sand bed, or approximately at the top of sand bed, since yo is too small to be measurable. Derivation of the ula of Velocity Profile The velocity u at a level y from the sand bed along a vertical may be integrated from Eq. 8 ∫ ≈ y yo u y dy uu 0 ** η 12 in which η is an unknown function of y to be found. The problem confronting us is that besides all physical and boundary conditions, the constant discharge Q and the surface curve including total depth of flow h and slope J at the given section of flow are all given, it is required to find the velocity profile uy along the vertical of the section. Since ghJu * and BhQum/ are known, from gCuum/ * , the Chezy constant reflecting roughness of the boundary of flow is also given. Evidently, the uy relation must satisfy the requirement of total discharge Q after integration, i.e., BQhudyu m h o / ∫ 13 Among an infinite number of such uy relations, there exists only one relation true that obeys the law of minimum rate of energy reserved r E in the section, which is equivalent to the law of maximum rate of energy dissipation ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ∫y p g u BudyE h o r γ γ 2 2 88 .min 2 3 − ∫ hQdyu g B h o γ γ 14 since hy p γ , and QudyB h o ∫ . Thus the problem reduces to one of finding the function of velocity distribution along the vertical such that .min 3 ∫ dyu h o 15 whith η,yuyuu, where yηη. Substituting Eq. 12 into 15, we have .min1 3 3 * 3 ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ∫∫∫ dy y dy udyu y o h o h o η 16 The equation shows that the required condition of minimum holds for any multiple of u, i.e., ukua, as ghJu * is given, This offers a to determine the absolute values of u so as satisfy Eq. 13. Let us change u to ua in Eq. 12 and 16 for the time being .min 33 ∫∫ dyukdyu h o a h o 17 According to the variational principle, the Euler’s ula provides, the extremum occurs at 0 2 ∂ ∂ η a a u u since 0≠ a u, the minimum condition is 0 ∂ ∂ ∫ y o y dy ηη 18 The integral equation is solved through the following processes 0 1ln 2 −∫ ηηηd dy y yd y o 89 2/ 2/ 2/ ln 2/ lnln 2 2 222 η η ηηηd dd d yd d yddyd ⋅− 2/ 2/ ln ln 2/ 2/ 2 1 2 2 2 2 η η η η d dd yd yddd − 1 2 2 ln 2 lnlnln2/ln 2 1 Cdyd ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − η η ydCdln 22 1 2 2