PROCEEDINGS OF THE 3RD INTERNATIONAL CONFERENCE ON HYDROPOWER TRONDHEIM/NORWAY130 JUNE - 2 JULY 1997 Hydropower '97 Edited by E.BRdcH & D.K.LYSNE The Norwegian University ofScience and Technology, Trondheim, Norway N.FLATAB0 Norwegian Electric Power Research Institute, Trondheim, Norway E.HELLAND-HANSEN Norconsult International AS, Sandvika, Norway A.A. BALKEMA/ROTTERDAMI BROOKFIELDI 1997 Hydropower'97, Broch, Lysne, FlatabfJ & Helland-Hansen (eds) 1997Balkema, Rotterdam, ISBN9054108886 Table of contents Preface Xli Acknowledgements XN Organization XV 1 Hydropower in the environmental context Application and comparison of computermodelsfor quantifying impactsof riverregulation 3 on fish habitat KAlfredsen, WMarchand, T.H.Bakken & A Harby Dulyn/Eigiauwater transfer project- Integration of environmental issues 11 I V. Baxendale Planning anddesignof desilting basins inHimalayas - Acasestudy 17 1.Chandrashekhar &A Rengaswamy Environmental aspectsof theLowerKihansiHydropower Project, Tanzania 23 IH.Gerstle, S.L.Mhaville &I Lindemark . ..: TheEIAprocess: Amulti-dimensional perspective 33 E.Helland-Hansen Economic aspectsof removalof sediment fromreservoirs 39 T.Jacobsen Hydropower and environment: Decision makingin Norway 47 HiKaasa Weirconstruction as environmental mitigation in Norwegian hydropower schemes 51 1.H.L'Abee-Lund & IE.Brittain Aframework for a 3Dnumerical modelfor hydropower reservoir waterquality 55 N R B Olsen .. Fishbypass charmels: Desi .arameters, evaluation, discharge, costsof construction 61 , . ';'it'\andoperation B.Pelikan v Hydropower projects and environmental impact analysis 5. Rajani . 67 The proposed controversial Upper Kotmale Hydro Power Project in Sri Lanka and its environmental and technical aspects NRupasinghe 73 Environmental issues block hydroelectric project: A case study of Baspa-I Hydroelectric Project R.CSharma, S.P.Bansal & Y.Attri 79 Austria's hydropower and its importance to the environment WSteininger 85 Split & settle - A new concept for underground desanders RStt/Jle First world development in a third world environment: The challenges and solutions to environmental impact mitigation during construction phases of hydropower projects in Tanzania, East Africa P.A.McCauley Terhell 95 105 . Hydropower development in harmony with environment NVisvanathan & UBhat 111 Hydro power and environment problems in Lithuania IVycius 119 . Neelum-Jhelum hydroelectric project - Environmentally sound hydropower D.AWright & MAMalhi 125 Resettlement methods at Lianhua Hydropower Station of China YongZhao 133 The Shi Sanling pumped storage power plant and its environment ZhaoZheng, Liang Hai-Bo & WuXiao-Feng 137 Environmental issues of Three Gorges Project DexiangZhu 141 2 Hydropower in mixedsystems Coordinated operation of a hydrothermal power system: The case of Nepal D.B.Basnyat & AD.Gupta 153 Game model for optimizing a river regulation plan Qiguang Chen & Changming Wang 161 Temperature dependency of demand in mixed hydro-thermal systems G.LDoorman &B.Mo 167 Norway: Europe's pumped-storage system -Necessary modifications of power plants A Elstrom 173 VI The value of hydropower import into a thermal system 181 P.B.Eriksen &1 Pedersen Long to short termoperationplanningand modelling of hydrothermal power systems 187 B.A Flechner & H Wolter Portfolio management in a deregulated hydropower based electricitymarket 197 S.-E.Fleten, WT.Ziemba & S.WWaliace Model experiment of steel lining and reinforced concrete back penstock in the Three-Gorge 205 Hydropower Station Xiong De-yan, Fu Yi-shu, Ma Shan-ding, Gong Guo-zhi, WuHan-ming, Yang Xue-tang &YangYao A case of hydro scheduling with a stochastic price model 211 A Gjelsvik, MB.Belsnes & MHaland Neural network based simulationtool for improving the control of hydro cascade system 219 R.Golob, D'Grgic & T.Stokelj Stochasticoptimization of weeklygeneration schedules: Solutionof the hydraulic 227 subproblemswith interior point methods I-P.Goux, A Renaud, S.Brignol &l-CCulioli Analysison selection of pumpedstorageplant in North China 235 Liang Hai-bo, Gu Zhao-qi, Ma Ji-ming &Zhang Ming Benefitof capacity expansionin hydropower stations in viewof power exchange contracts 239 and transmissiongrid utilization KS.Hornnes, O.S.Grande & T.G.Borg The challengeof hydro-thermal schedulingin a deregulatedpower market regime 245 A Johannesen Incorporationof thermal stochasticelements into a hydro-thermal model 251 CJfjJrgensen &HF.Ravn An optimizationmodel for regional generation schedulingat Hydro-Quebec 259 Ll.afond Tuningthe planning chain of hydroelectric systems 267 CL Correa de So, Jr & CLyra Filho Evaluatinghydro expansionor refurbishment in a deregulatedelectricitymarket 271 A Haugstad, HMo & MBelsnes Integratinglong- and short-termmodels for hydro scheduling 279 B.Mo, A Haugstad & Q.B.Fosso Estimation of the economical and operational impacts of a HVDC-link between Norway 287 and a continental power system T.1Larsen, S.Niefien & HIHaubrich Dailygenerationplanning of a hydrodominated hydrothermal system 293 O'Nilsson & L.S6der VII Stochastic hydrothermal scheduling in a competitive environment M V. R Pereira &N.M Campod6nico V. 301 Effects on thermal power systems in coordination with hydropower systems SiSather 309 Advantages and disadvantagesin exchange of power between a hydro- and a wind-dominated electricity supply system P.Meibom, T.Svendsen & B.Srensen 317 Maximizing the profit of hydro generation when taking into account the extra costs of co-generation with thermal power LVinnogg 323 Experiences from the Norwegian deregulation experiment LWangensteen & 0. Rismark 327 Integration of river water temperature constraints within real time operation of hydroplants RWelt, S.Hachem & R.Kahawita 333 Dynamic models for real time management of hydroplants S.Hachem, AG.Hammadia, RWelt & MBreton 341 3 Dam safety and risk analysis Performance as an indicator of the safety of arch dams with special reference to the wide spanned arch dam Sta. Maria P.Beyeler, WHauenstein, P.Lier & B.Otto 349 A role for risk assessment in dam safety management D.S.Bowles, LR.Anderson & T.RGlover 359 Experience of failure mode, effect and criticality analysis on UK hydropower schemes CBeak, lW.Findlay &D.LAikman 369 Some problems discussed in design of the Three Gorges Project Gu Zhao-Qi, Peng Shou-Zhuo, Cai Jun-Mei, Ma Ji-Ming, Zhang Ming, Liang Hai-Bo & Guo Jian-Jun 377 Stability analysis of rock foundation of a plant dam section . Zhang Ming, Peng Shou-Zhuo & GuZhao-Qi 383 Credibility and defensibilityof dam safety risk analyses D.N.D.Hartford & G.MSalmon 387 Evaluating the probability of failure of an earth dam by seismically induced liquefaction D.N.D.Hartford, K Y.Lum, M K Lee & R.A Stewart 395 Rehabilitationof the intake structures at the VerseDam, Germany CHeitefuss &H-J.Kny 405 Incorporating risk analysis in dam emergency planning LJenssen 413 VII' Mohale CFRD, design considerations P.Johannesson, CGratwick & S.Nthalw 419 Risk analyses of three Norwegianrockfill dams P.MJohansen, s.GVick & CRikartsen 431 Dam safety legislation and guidelines - A UK. perspective lP.Millmore & T.AJohnston 443 Seismic risk analysis of concretegravity dams - Problems and solutions D.S. Kisliakov 451 Numerical modelling of2-dimensionaldam-break flow 1 P. Laasonen 459 Ice in spillways in connection with dam safety LLia 467 Hydrodynamic forces from steep waves in rivers A Levoll 473 Large scale model test on the hydraulic properties of differenttypes of inlets to the penstock Ma Jiming & Liu Dechao 481 Hydraulic prototype observationof Geheyan Hydropower Station Ning Tingjun, Cheng Yuanqing & Wang Shipeng 487 Dam safety and risk analysis- Experience of E. S.B. Ireland' lD.O'Keeffe 493 A 3-dimensional numerical model for determination of spillwaycapacity N.R.B.Olsen & HMKjellesvig 501 Visco-plastic analysis for the lock slope of the Three-Gorge Project Peng Shou-Zhuo & Guo Jian-Jun 507 Meadowbank Dam spillway review- A case study in the use of risk analysis and non-structural solutions LPolglase . 513 A new approach to probable maximumflood studies lD.Cattanach, lv.Q..Chin & GMSalmon 521 Estimating the magnitude and probabilityof extreme floods GM.Salmon, lv.Q.Chin & V.Plesa . 531 The dam safety business N.P.Robins & GAWeller 539 The hydraulicproblems existing in Xiaolangdi hydraulic project T.Xiang, B.Wu, lMCai, lMFen & lv.HYarig 545 Parameter uncertaintyin modellingdam breach and its flood lX.'LYang 551 IX 4 Tunnelling and underground works Development of tunnellingtechnology in Nepal by use of local resources P. P.Adhikari 563 A new method for in situ determinationof the roughnesscoefficient of the hydropowerplant tunnels P.Boeriu & V.Doandes 575 Hydraulic jacking tests for unlined high pressure tunnels E.Broch, T.S.Dahl & S.E.Hansen 581 Ertan hydroelectric project: Experiences during construction QianYang, P.K.Edvardsen &K.ICarstens 589 Rehabilitationin the unlinedrock tunnels of NedreRessaga after 40 years of operation T.Carstens, S. E. Hansen & B.Undrum 597 Shotcrete-linedhydropower tunnels S.Elfman 605 Optimal design of hydropower plants 1 Eliasson, P.Jensson & G. Ludvigsson 611 Monitoringsurvey and feedback analysisof underground powerhouse ofMing Tombs pumped storage plant LiangHai-Bo, Gu Zhao-Qi, Zhang Ming & MaJi-Ming 619 TBM-tunnellingat Sauda Power Project RMoe, RHolen, E.D.Johansen &B.Aspen 623 Rebuildingof the 70 years old Nore 1 Power Plant 1 Hope, APalmstrm &KFinnerud 631 Rock mechanical engineeringto the design of the underground tunnelling works of Bakun River diversionproject in Sarawak, Malaysia W.R.Jee & llChoi 639 Stabilitystudy of an underground power cavern in sandstone WeichengJin, MLu & E.Broch 647 Modeling and back analysisfor a large scale undergroundpowerhouse complex Zhong-Kui Li, Ai-MinWang & Xing-HuaMuo 653 Head losses due to airpockets in hydropower tunnels E. Tesaker & S. Lunde 659 Prediction of rock support in Melamchi Tunnel, Nepal P.Pradhan 667 New method for estimationof head loss in unlined water tunnels P.-E.Rnn & MSlwg 675 Economic design of hydropower tunnels P.-E.Rnn & MSkog 683 x Floor paving in unlined hydropower tunnels (Z).Solvik & E.Tesaker 691 Unlined invert impact on the free-flow tunnels drainage capacity V.D.Tashev & K.T.Daskalov 697 Method of calculating pressure transferred by soft layer surrounding penstock Yao Shuang-Xi, Gu Zhao-Qi &Liang Hai-Bo 703 Author index 707 XI Hydropower'97, Broch, Lysne, Flatabe & Helland-Hansen (eds) 1997 Balkema, Rotterdam, ISBN 90 54108886 Preface These Proceedings contain the papers presented at the International Conference on Hydropower Development in Trondheim, Norway, June 30 to July 2, 1997. The first conference in this series, Hydropower'87 in Oslo, concentrated mainly on underground hydropower plants, the second conference in Lillehammer in 1992 had a wider scope, covering technical, enviromnental and economic issues. Hydropower'97 broadens the scope even further to also cover safety and training. In a world where the demand for electric energy is steadily increasing, hydropower holds a unique position as the renewable energy source with the highest potential in a medium-range perspective. The regions of the world where the need for energy is most pressing, all possess huge hydropower resources. With a century of experience to draw on, the hydropower community has an extensive basis of knowledge and skills in support of every aspect of hydropower development. This includes how to deal with potentially negative effects of inadequate planning and design. Hydropower'97 does focus on issues that are vital in this respect, such as hydropower in an environmental context, dam safety and risk analysis. Planning of hydropower developments is still a great challenge covering a wide range of technical, economic and enviromnental issues. The objective of the conference has been to address these issues and highlight the ways in which hydropower can be developed in a flexible manner to meet varying demands and changing conditions. These are important issues both for professionals and for the general public. The papers were selected on the basis.of a general invitation, except for a few specially invited lectures. The editors thank all contributors, who have made it possible to collect documentations on many recent scientific and technicaladvancesinhydropower engineering. We hope that the proceedings will form a valuable basis for further progress of hydropower development. . The proceedings have been produced by the offset printing method. All papers are typed by the authors in accordance with given instructions. The Editors are therefore not responsiblefor misprints or errors in the text. The opinions expressed are those of the authors and not necessarily those of the Editors. ' Trondheim, 2nd April 1997 E.Broch E. Helland-Hansen N.Flatab0 D.K.Lysne Editors XIII Hydropower'91, Brach, Lysne, Hetsbe & Helland-Hansen (eds) 1991 Belkeme, Rotterdam, ISBN 90 54108886 Planning and design of desilting basins in Himalayas - A case study 1. Chandrashekhar & A. Rengaswamy Central Water Commission, New Delhi, India ABSTRACT: The paper highlights the planning and design aspects of desilting basins in Himalayan rivers with specific reference to Nathpa Jhakri underground hydroelectric project under execution. The project is a run-of-the-river project on river Satluj which experiences severe sediment transport. The project comprises of an underground desilting basin complex with four parallel basins, with hoppers and flushing conduits at the bottom to control undesirable sediments from entering the turbines. The design is aimed at eliminating sediments of particle size O.2mm and above in the basins. The hydraulic performance of the proposed basin as studied on a physical model as well as simulated in a numerical model is highlighted in this paper. 1. INTRODUCTION Himalayan rivers carry huge quantum of sediments both during snowmelt season as well as during monsoon. The planning of run-of-the-river hydropower projects in Himalayan rivers call for careful handling of the sediments. Undesirable sediments in the water diverted for power generation will cause significant wear and tear of the electromechanical parts of the power station. Diverting absolutely sediment free water for power generation from such rivers is practically impossible keeping in view the economic viability. However the sediment content in the diverted water can be controlled to permissible limits at the headworks using settlingidesiltingjbasins.T'roper planning and design is of utmost importance. Physical model studies and numerical modelling of large settling basins help in better understanding of the water and sediment flow behaviour. This paper discusses in brief the planning and design studies carried out for the desilting basins of Nathpa Jhakri Hydroelectric project: 2. THEPROJECT The Nathpa Jhakri Hydroelectric project is an ongoing run-of-the-river project on river Satluj . The proposed installed capacity of the project is 1500MW, through six Francis units of250MW each, utilising a discharge of 405cum/s. under a design head of 425m. The project is an underground power project and comprises of mainly: - concrete gravity diversion dam 65.5m high. - four desilting basins, each 525m long, 16.31m wide at centre,27.5m high - lO.15m dia. 27300m long headrace tunnel - 21.6m dia. 301m deep open surge shaft - 4.9m dia., three steel lined pressure shafts - powerhouse cavern 20m wide, 49m high, 216m long - transformer cavern 18m wide, 27.5m high, 196m long. - 10.l5m dia., l012m long tail race tunnel. The layout plan of the project is shown in Fig.l. 3. SEDIMENT CHARACTERISTICS The river carries significant amount of bed load as well as suspended sediment load. The particle size distribution of the measured suspended sediments indicates the presence of very high percentage of fines. The average distribution of coarse (0.2 to 2mm), medium (0.075 to O.2mm)and fine (less than 17 TAIL RACE TUNNEL 10.15 M.rp, 1012 M.LONG JHAKRI POWER HOUSE COMPLEX NAT HPA DAM,INTAKE &DESILTING COMPLEX HEAD RACE TUNNEL 10.15Mct>,27.3 KM. LONG Fig. 1. Nathpa Jhakri H.E. Project - Layout Plan 0.075mm) sediments observed in this river and averaged over 15 years is 17%, 25% and 58% respectively. The bed material gradation measured indicates that the bed load ranges from O.lmm to 200mm. The analysis of suspended sediments for mineral composition indicates a very high percentage of angular quartz i.e. about 40%. Besides these, other minerals with Mho's hardness more than 7 viz. Zircon, Garnet etc. constitute about 8%. These sediments could be very much damaging for the turbines if not controlled. 4. DESILTING ARRANGEMENT The desilting basins for this project has been planned fully underground as topography doesn't permit a surface one. Four independent intakes and inlet tunnels feed four parallel desilting basins flowing full under pressure. Various alternatives constituting three to six basins were considered and four basins were finally found to be technically and economically sound. The layout plan of the desilting basins is shown in Fig.2. The desilting basins are Dufour type( hopper type). The design criteria laid down by the turbine manufacturer calls for excluding sediments above 0.2mm in size in the desilting basins with the stipulation that the units will trip when the sediment concentration exceeds 5000ppm. Accordingly the basins have been designed and dimensioned keeping the flow velocity around 0.3m/s. The proposed basins are each 525m long, l6.31m wide at the centre and 27.5m high. The inlet discharge into each basin is 121.5cumls which includes 20.25cumls for continuous flushing which is 20% of the outlet discharge. The incoming flow into the basin expands uniformly in all directions through a 50m long transition (diffusor). The outlet tunnel at the end of the basin is provided with a controlling gate to isolate and empty any of the basins for maintenance. The sediment controlled water is thereby led into the 27300m long headrace tunnel. The efficient performance of settling basin calls for ejection of the sediments as it settles down. The continuous hopper 5m deep at the bottom of the basin has a 3m wide settling trench with inlet holes of varying sizes for the settling particles to gradually travel into the flushing conduits. The basin is divided into three sections of 175m each.. Three flushing conduits are provided to sectionalise coarse, medium and fine fractions of the sediments. One conduit runs in full length of the basin from the upstream end carrying coarse sediments while the second and third conduit starts at 175m and 350m 18 from the upstream end carrying medium and fine sediments respectively. Six gates for each basin i.e. two for each flushing conduit at the end shall control the flushing operation and lead the sediment laden water into a free flowing flushing tunnel and finally back into the river downstream of diversion dam. The velocity in the flushing conduit gradually increases along its length from 3m/s to 3.75m/s which is of the same order of magnitude as in the inlet tunnel. Thus all sediments entering the flushing conduit shall be flushed out. The longitudinal section and cross-sections of the desilting basins is shown at Fig. 3 . 5. MODEL STUDIES 5.1 Physical modelling Physical model studies of the desilting basin were carried out (CWPRS, 1990) on a 1:30 scale model covering the entire basin from inlet to outlet. The particle size distribution of the suspended sediment inflow had do = O.OOlmm, dss = 0.075mm, dS3 = 0.2mm and dlOo= 1.Omm . The physical model study indicated that the 525m length of the basin is adequate for 90% settlement of the sediments coarser than O.2mm for an overall inflow concentration of 5000ppm by volume. The overall settling efficiency of the basin worked out to be about 36% for the gradation of the particles considered. The size of the flushing tunnel is adequate for continuous flushing of the settled sediments with 20% of the design discharge for flushing. The length and slopes of bed and roof of the diffusor were finalised based on the results of the model studies so as to permit full expansion within the transition without flow separation. 5.2 Numerical modelling Numerical modelling simulating the entire desilting basin from inlet tunnel to the outlet tunnel was carried out for studying the hydraulics and functional efficacy (Chandrashekhar, 1994). A three dimensional model called "Sediment Simulation in Intakes with Multiblock option" (SSIIM) (Olsen, 1994) was used to study the water and sediment flow in the ehtire domain of the basin. The velocity field in the basin was found to be predominantly favourable except for the inlet zone where recirculation was observed. All particles above OAmm were seen to settle within the first half of the basin. The concentration of sediments of 0.2mm size is very near to zero at the end of the basin, but nevertheless some particles find their way into the tunnel . All particles of size 0.05mm and below enter the power tunnel . The particles that are under suspension in the upper half of the basin towards the outlet are getting carried away into the tunnel by the high velocities and turbulence . As a result, significant percentage of fines in the range 0.05mm to 0.2mm enter the headrace tunnel. The study indicates that the desilting basin is found to be adequate for 90-95% removal of particles upto 0.2mm size for an overall inflow concentration of 5000ppm. The overall settling efficiency of the basin is predicted as 37%. 6. ROCK SUPPORT SYSTEM AND LINING The basic rock support system designed after characterising the rock mass comprises of pattern rock bolting with steel fibre reinforced shotcrete . The internal lining for the basins was originally conceived as 30cm thick concrete lining with welded mesh reinforcement on the inner side. The welded mesh was proposed to be anchored to the rock by 25 dia. anchors . Various other alternatives have been studied for lining the basin . The lining should be able to withstand the internal water pressure during operation and the external pressure head when one of the basins is emptied for maintenance purposes. It is now proposed to provide concrete lining for hopper portion and the inlet transition reach where the velocities are higher. The side walls and roof of the basin is to be finished with 150mm thick steel fibre reinforced shotcrete lining. The final surface shall be smoothened to the extent possible . Though the rugosity coefficient of shotcrete lining is .higher than that of concrete, due to very low velocity in the basin the increase in head loss is negligible. 7. SUMMARY Desilting basins form a major cost item in the headworks of any hydroelectric project. Hence the planning and design should aim at optimizing the 19 I 175 M I 17 5 M .. , 175 M I 150M ", 525M ., 'I " "GATES \ j FLUSHING TUN NEL. :I:'" ,.; SETTLING TRENCH r- -_J__: CHAMBER- 3 CHAMBER-2 CHAMBER- 4 r - CHAMBER-l :!: '" ui.. :... -"\ 'I. I -, \ \ \ ......, "\ ,' ............, .....\ , '" Ii .... \ \'"3, '7 'd. \ Vl \ . . \ I '\ I Fig.2 NathpaJhakri headworks - Layout Plan I- t- .525 H ., I:"LI LONGITUDINAL SECTION c (FI u5hing tunnel not shown) 400 1600 I' c e 'E 50 " 40 iI: 30 20 10 o 20 15 10 Time (minutes) ._----- ---_.__.....-- _. _..... _-._- 1', ... 1', 1', "' .loI -, Ps Ps .. \......., orsteinsson 1985) called vas. Included in order to estimate expected values of finn power 1:' in this is a set of formulas, which are designed for capacity of the total system and expected secondary calculating the approximate cost of hydropower power production. ; projects in the feasibility stage. These formulas are In this paper the design of hydropower schemes is .t.' exclusively used as technical basis in this work. For divided into three system stages, in order to have a complete reference see Eliasson & Ludvigsson relationship with conventional design levels. These (1996) and I>orsteinsson (1993). stages are defined as: Today, planning of extensions to the power system Plant stage (feasibility level) is usually divided into two main stages, feasibility Allocation stage (feasibility - project planning level) 611 Global system stage (project planning level) For complete description of these different levels see Eliasson & Ludvigsson (1996). This paper will concentrate on the plant stage, comparable with the conventional feasibility level. The other stages are subject for future improvements and therefore only introduced here. In the plant stage, calculations have the following variables (Eliasson & Ludvigsson 1996): E, The annual energy demand (finn power), or the area under the duration curve: t=lyear E = fN(t) dt [kWh/a] (1) o Ti, Annual load duration time, and the plant's load factor Ap: E Tk =--[hours]; (2)Nrnax Now, a reservoir with a storage capacity, V, and an inflow series, Qi, can deliver a maximum average discharge Qa < Average(Qi), to the plant for energy production. V and Qa are calculated from the inflow series and reservoir site data as functions of water level in the reservoir. The plant is, however, not designed for a constant average discharge, Qa, over the whole year. To satisfy the market it is instead designed for the discharge: (3) where the discharge load factor Ap is introduced because the power N is not a totally linear function of Qd due to headlosses in the waterways (thirddegree function of Qd). p, the cubic average of flow, is used to estimate average head losses: 13=!.J(Q(t))3dt (4)T 0 Qd Junge's load duration curve J(t) is used (Eliasson & Ludvigsson 1996): J (t ) =1- (1- A/ ) .!'". = N(t ) I Nmax os 1 (5) The flow duration curve, Bit), may be derived from the load duration curve, using: 1 - B(t)2!!.L] J(t) B(t) 1 _ (6) [ where Bit) = Q(t)IQd; hi = head loss at designs discharge, Qd; Hb = brutto head. B(t) is now obtained from (6): -2 B(t) = I cos ,,3hf I n, [ (l h, I I H.) + ; ] (7) Now we have: (8) = ! r(Q(t))dt = f,ear B(t)dt (9)T 0 Q 0d The time average of (6) gives: (10) d{1 i'h: ] After taking the time average, may be found from (10). 2 OBJECfIVE FUNCTION The logic of optimisation is the samefor all kind of problems. The goal is to find the minimum or maximum of a defined objective function, where the variables can be subject to some constraints. Mathematically put (maximisation problem): maxftxl, X2, ... xn) a, Xi b, for j = 1 to n (11) gixl, X2, ... xn) Cj for 'i/ j In this chapter the principle of optimal profit is introduced as our objective. This leads to a method that in fact includes many of the conventional local optimisation methods used so far. and can yield the same results. To be able to calculate the revenue of a hydropower plant with a energy capacity E the 612 energy demand of the market is needed (Eliasson & Ludvigsson 1996). Firm power is sold according to many different tariffs, and secondary power prices vary even more. Besides the market, sales of secondary power also depend on the river flow. Nevertheless, the average unit-price of energy is not very variable. To take an example, the average price of energy from NPCI varied between 1,7 and 1,83 IKR/kwh in the last 6 years. It is therefore reasonable to assume power sales to be at a fixed price, at least while no market limitations beside the load factor are taken into account (Eliasson & Ludvigsson 1996). By assuming an infinite energy demand and a fixed energy price, ke, the present value of the revenue of energy sale becomes: l - (1+ r )- NJ NPV =keE. r( (12) l - (1+ r )- NJ-Cv -C( r where: r is the interest rate for discounting N is the lifetime of the investment C is the project investment v is proportional operation and maintenance cost ke is the unit price of energy E is the annual energy capacity of the scheme As all costs, as well as revenues, are included in the objective function, the optimisation can be . considered global. The optimisation is performed by finding the maximum profit, NPV, with respect to all design variables. Consider the case where there are n mutually independent design variables (i.e. each one ; can be changed without affecting the others), that are to be optimised. Mathematically put, this means: n aNPV d NPV =(;; a Xi d Xi =0 (13) condition for global maximum of ,.a NPV =0 for all i = 1 to n (14), aX Differentiating (12) with respect to E results in: N d ) dC [1 -(l + rr:1 dE (k . E - C) r = 0 => k, = dE (15) Eq. (15), shows that for the optimum of the objective function, k is equal to the marginal cost of energye dC/dE. Due to this fact, k can be used as a pree selected marginal cost control parameter in the optimisation procedure. It still has to be kept in mind that the optimal solution found will not optimise the profit of the power production unless k is the actual sales price. e Other k values will give solutions optimal in some e other respect. The optimisation procedures of Mosonyi (1991) and later publications may be deduced from (14). Take a design variable such as a tunnel diameter D and put Xi equal to D in (14). The result is: aNPV aE sc (16)-a-D =0 => k, -aD- =-aD-Often this cost, i.e. variable costs of other project items than the conduit itself, are not taken into account, which results in a larger tunnel diameter than necessary. 3 CALCULATION OF THE NPV The basic method for the calculation of the system's NPV is quite simple. The first characteristic of this method is the division of the parameters into three main groups; main parameters, local parameters, and coordinates. These main parameters are mutual for all types of structures included in the scheme. From now on the vector MP indicates these parameters, and they are listed in Table 1. The user can define local parameters and coordinates of the nodes, Pi-I and Pi' (i = 0 to N), that locate the connection points of the structures and are used to calculate sizes such as length of Table 1. Main parameters Description Symbol Unit Design discharge Average discharge Head at design discharge m Averagehead m Brutto head m . Number I Length L m Scheduled cost C currency Power N MW Energy E GWh/a 613 i conduits. The nodes can be both fixed or free for optimisation. The scheme is calculated by starting at the highest point upstream (i = 0) and 'send' the main parameters through each structure U=I to M) on the way downstream to the lowest point (i = N). This is shown schematically in Figure 1. The values of the main parameters after structure M include: Scheduled construction cost Power gained Energy gained These parameters, along with the basic assumptions for the project, give the project investment cost, and the revenue of the power sale, and so, the net profit of the investment NPV (value of the objective function). For further explanation, the simple example in Figure 2 is introduced. _ The dam U= I) is furthest upstream, so MP 0 = 0 is the array entering it. Associated with the dam is some inflow data and a reservoir curve that according to a maximum reservoir level, H, gives the discharge, QI. The height of the dam is a function of the reservoir level, H, and the cost of the dam, CI, calculated according to it. The coordinate of the outlet, P" is known, so the head produced by the dam is HI = H - ZI. This gives: MP1.Q=QI; MP1.H=HI; MP.C=CI The next structure downstream from the dam is the pressure conduit U= 2). It has a diameter, d, and roughness ks. For the discharge MP I.Q there is a headloss, hi, in the conduit so the total head produced by the shaft is: H2 = MP .H + ~ - ZI - hi The discharge in and out is naturally unchanged, and the cost C2 is a function of the diameter, lining, tunnelling method, and H2 Thus: I Structure. ---t-Z-II-Z ---. P2 P3 ~ Direction of flow Figure 1. Schematic picture of a model Tailrace Tunnel Dam Pr. Shaft Powerhouse Tailrace Tunnel Figure 2. Example of calculation method MP 2.Q = MP I.Q; MP 2.H = H2 MP 2.C = MP I.C + C2 Next is the powerhouse U = 3). It's main properties are the type and power output of the turbines. Based on MP 2.Q and MP 2.H the power N3 and the energy E3 is calculated. The cost C3, is a function of the number of turbines, power capacity, and design head. Discharge is unchanged and the pressure head is reduced to zero by the turbines. This gives: MP 3.Q = MP2.Q; MP 3.HEEO MP 3.e = MP 2.e + C3; MP 3.N =N3; MP 3.E = 3 A similar procedure is applied for the tailrace. No more structures are in this scheme so MP 4.e, MP 4.N and MP 4.E are used to calculate the net profit of the investment. 4 OPTIMISAnON PROCEDURE The method used to find the global optimum is called Genetic Algorithm, (see, e.g., Goldberg 1989; Pirlot 1994), but the following main parameters control the process: Population size Mutation probability Number of generations Each variable in the objective function is subject. to the following constraints: Each variable p must have an upper and lower bound, Pmin::; P ::; Pmax. If these bounds are not specified by the user, the program must do so. Coordinates that are to be optimised are only allowed to move along a line, not in a three dimensional region. 614 __~ d I - The resolution of each variable must be defined, i.e, the smallest allowed increment of a variable op. For instance a diameter of a tunnel only needs to be optimised with an accuracy of say 10 em, a coordinate with accuracy 5 m, and so on. For further explanation, the reader is referred to Eliasson & Ludvigsson (1996). 5 THE PROGRAM The program 'Hydra' is a 32 bit Windows 95 application, written in Visual Basic 4.0. It includes a .. SQL connection to the database Access 7.0, where all project data, unit prices, and results, are stored. Also included is a OLE connection to other . Windows applications, such as Excel and Word, so results and other data can easily be moved between programs. When using the computational model, the input ;'process consists of the following steps: , - Define Project - Assign main assumptions - Define Structures - Assign Structure Properties - Define Points - Connect Structures - Edit Unit Prices . When the input phase is finished, the model is [eady for calculation. The user can get one detailed solution by using the default variables, see how varying one variable affects the result, or perform a global optimisation. The optimisation procedure the following steps: . - Select the variables that should be optimised - Select the points that should be optimised . - Decide the size of population and number of enerations for the Genetic Algorithm . ,' - Run Optimisation , When the program has finished the optimisation . ccess, it presents the graphical evolution of best ult in each generation. If the development of best lution is satisfying, the user can perforin different eries on the [size of population] . [number of nerations] size solution set. Following queries are eluded: :, - Get all the results listed, sorted by profit, vestment, power capacity, energy capacity, breakyen price or the ratio; profit divided by investment. Select any listed solution and get a detailed scription of it, including cost summary and , ificant data. - Get a graphical presentation of the distribution of solutions less than N percent below best solution, as function of project investment, energy, load or break-even price. This is used to check if the objective 'hill' is flat or steep around it's maximum (used in sensitivity analysis) and if there are other slightly smaller 'hills' in the region. Besides performing direct optimisation the application can, for example, be used for sensitivity and what-if analysis, i.e. check how adjustment of one or more input parameters (including adding or removing structures) affects the optimal design and as a training tool to get a feeling for the importance of different structures and parameters. 6 CALIBRATION A 'simple' example, shown in Figure 3, for which the results can be mathematically derived, is used to test the validity of the model, and calibrate the parameters of the Genetic Algorithm. The mathematics are too long to show here, the results are shown in Table 2. When the results of the optimisation are compared with the mathematical solution, it is obvious that the runs where the GA parameters are optimally tuned, reach results very close to the true optimum, see Figure 4. Experience shows that running times lie in the vicinity of 2-4 minutes, depending on the size of population and number of generations. The result of the conventional local optimisation method is also calculated and it gives a optimum diameter, D, of 4.5 m, which is a 0.5 m difference in the diameter between methods. This shows clearly how inaccurate the conventional method can be, ignoring the fact that a change in the diameter not only affects the cost of the conduit but also the cost of, for instance, E & M Equipment and powerhouse. o 3. Powerboose 4. ThDDet Figure 3. Calibration example H, 615 --Table 2. Mathematical solution (bold) compared to optimisation results, NPV, for different number of individuals P, generations G and mutation probability IJ,. P G 11 D 4,0 HI 543,0 Hz 48,2 H] 44,9 100 100 100 100 0.001 0.005 0.Ql 0.025 4.0 4.0 4.0 3.9 543 543 543 543 44 49 49 48 39 46 46 45 W W 100 200 200 0.05 0.025 0.05 4.0 4.0 3.9 543 543 543 42 50 44 37 46 39 NPV 28594 28580 28594 28594 28590 28569 28593 28576 dNPV -14 0 0 -4 -25 -1 -18 29.00l "U''''' ".eoo j V .. ., , : ", " '.. .."" t.iJ:.. ... :.,....,.,."" ",,"= 128.00l l Z ZlJJCIJ 21.eoo 21.""" 21.2fXJ27.00J -:: \!! ;;; til :; :;; ill 10 III lloeIopment of BestSolution ...... ........' _0, SQ'O,0i5 .......... _"'"_. _. &rQ,OOl -100,01 l!! 10 III ;;; III FIgure 4. Development of solution for different number of individuals / mutation probability A more detailed description of the difference between the global optimisation and the conventional method is given in Eliasson & Ludvigsson (1996). 7 CASE STUDY: FUOTSDALUR PROJECT In cooperation with the National Power Company of Iceland, NPCI, and their engineering consultants, the model was used to perform a case study on the Flj6tsdalur Hydroelectric Project. The resulting design variables of the GA optimisation are then compared with the project planning report, PPR, from April 1991 (Eliasson & Ludvigsson 1996). Two runs are made, firstly, a Plant Stage run, and secondly, an Allocation Stage run. This is done in order to find out, by comparing the results, if the Allocation Stage could be simulated with the model. The project is located in eastern Iceland .on the river Jokulsa i Flj6tsdal, which runs from Vatnajokull glacier to .the north-east. The powerhouse location, in the Flj6tsdalur valley, is about 40 Ian south-west from the town Egilsstaoir.. i F1,j6tsdaI"" Hydroelectric Figure 5. Schematic layout of Flj6tsdalur Hydro. electric Project The project area extends from the powerhouse site some 35 Ian to the south-west on the Flj6tsdalshei6i plateau and onwards to the Eyjabakkar and Hraun area, and includes diversions of smaller rivers on both sides of the Jokulsa, The total drainage area is estimated 478 Ian2 including diversions (Flj6tsdalur Engineering Joint Venture 1991). A schematic layout of the project is shown in Figure 5. In a plant stage optimisation, the cost of all optimised structures are calculated according to 'Virkjanalikan Orkustofnunar', VOS, (lJorsteinsson 1993) by using unit-prices from December 1992. Four runs are performed on the Flj6tsdalur Hydroelectric Project for the Plant Stage. The results are presented in Table 3, where we have: PPR j : The model is calculated for all dimensions fixed according to the Project Planning Report to get a comparison to the optimisations. PPR2: The model is calculated for all dimensions fixed according to the Project Planning Report except the maximum reservoir level of the Eyjabakkar reservoir, which is 4 m higher. This run is made because this raising of the dam is mentioned as a possibility in the Project Planning Report (porsteinsson 1993). 01150: In order to simulate the effect of a global optimisation on the actual 210 MW scheme, a special run 0 1150 is introduced, where the maximum energy demand is kept fixed at 1150 GWhla for the lifetime of the plant. The optimisations is performed with the GA parameters: 51 individuals and 100 generations. 0 00: One feasibility level optimisation is performed with infinite energy demand. Following GA parameters are used: 51 individuals and 100 generations. 616 Table 3. Tabulation of significant data and net profit of the investment (optimised dimensions are bold faced). 60 BIKR == 1 billion $ Description PPR1 O ~ Reservoir level m.a.s.l, 664.5 668.5 665.1 667.6 Headrace tunnel dia. In 5.0 5.0 4.3 4.8 Pressure shaft dia. In 2.9 2.9 2.6 2.7 Power MW 213 239 211 233 Energy GWhla 1159 1300 1150 1278 Investment BIKR 21.16 22.91 19.92 21.96 Profit BIKR 10.90 13.44 12.28 13.86 lWrofitlMnvestment % / % 0/0 +23/+8 +13/-6 +27/+4 The 0 1150 optimisation seeks a slightly higher dam (increased discharge to the plant) to compensate for increased power losses in narrower conduits. The 0 00 optimisation results in a significantly higher dam compared to the PPR I. The explanation is that in the project planning report, the size of the .power plant and the size of reservoir is selected on basis of a. power market scenario at the expected construction time of the plant, but the optimisation , assumes infinite demand. The solution is however not far from the PPR2 arrangement. The global optimisation 0 1150, leads to a 0,7 m narrower headrace tunnel compared to the PPR. Local optimisation, considering only variable cost of the headrace, leads to the same result as in the PPR (5 m). The power capacity reduction due to increased headlosses, is compensated with a slightly larger reservoir (increased discharge). The 0 00 optimisation results in a slightly smaller headrace diameter compared to the PPR. It is quite ., natural when compared to 0 1150, that this :, optimisation seeks a larger tunnel, because there are "no market restrictions. The same logic can be used to explain the difference in the pressure shaft diameter. There is, however, a problem with the maximum velocity in the shaft. In both optimisations the diameter breaks the design criteria that the maximum velocity Should ,be below 8 mls. For both 0 1150 and 0 00, the minimum diameter that satisfies this constraint should be selected by the user, in both cases close to , d = 2,8 m, depending on design discharge. This has a minor economical significance in this case, but is however a good example of how dependent , constraints g(x, y) ~ 0 have to be considered in the future development. The way to handle this is to develop and add a penalty function, 4l(x), to the construction cost of the pressure tunnel type object , (and other objects where necessary), that 'penalises' the tunnel if it's water velocity exceeds the allowed value but is otherwise zero. This prevents the Genetic Algorithm from breaking this constraint. The 0 00 optimisation results in a larger energy output than in the PPR I. This is natural, as this optimisation assumes plant stage, which means no market restrictions and no extra benefit for the system. The extra benefit is that interactions between Flj6tsdalur Power Plant and the existing power system produces substantial extra energy (estimated 250 GWhla firm energy in the PPR) through better utilisation of the water resources. The project investment is 6% lower in optimisation 0 1150 compared to the PPR I, resulting in a 13% higher profit, which is a significant improvement. The optimisation 0"" on the other hand leads to a 4% higher investment and a 27% higher profit. When it is kept in mind that the PPR I plans a future raising of the dam to reservoir level 668,5 m.a.s.l (Fljotsdalur Engineering Joint Venture 1991), the result of 0 00 is very close to the PPR2 version. In order to ensure the best possible result in the global optimisation the cost estimation of the whole scheme is completely revised. The VOS construction cost functions are removed and replaced with new cost functions, specially prepared by the engineering consultants (Helgason, pers. comm.). Now similar runs as for the Plant Stage are performed. The results are presented in Table 4. The 0 1150 optimisation leads to a similar arrangements as the plant stage optimisation. The 0 00 however shows significant changes. This is because the new cost formulas do not represent the true variation of the costs except in a narrow region around the PPR I values. Therefore the results of the 0 00 optimisation are hardly applicable. However a comparison of the columns 0"" in Tables 3 and 4, shows how important it is that the cost formulas in the optimisation are accurate. It may therefore be concluded that it is worth the effort for the consultants, to take the time and trouble to have the Table 4. Tabulation of significant data and net profit of the investment (optimised dimensions are bold faced). 60 BIKR == I billion $ Description PPR1 PPR2 OllSO O ~ Reservoir level m.a.s.l. 664.5 668.5 665.1 669.6 Headrace tunnel dia.. In 5.0 5.0 4.3 5.3 Pressure shaft dia. m 2.9 2.9 2.6 2.8 Power MW 212 239 210 242 Energy GWh/a 1159 1300 1149 1325 Investment BIKR 22.78 24.40 22.18 24.91 Profit BII(R 9.36 11.78 9.72 11.97 lWrofitlMnvestment %/ % % +26/+7 +41-3 +28/+9 617 cost formulas in Hydra improved with formulas specially designed by themselves, in order to improve the accuracy of optimisations performed. In such an optimisation, care must be taken that all the constraints are correct. Othervise the optimisation can find an unpractical solution, typically tunnel diameters below the minimum required by safe operation. The new dimensions in Table 4 are being checked for this by the owner and the consultants. 8 CONCLUSION The main result of this work is the development of the program 'Hydra'. The algorithm reaches the true global optimum for the calibration example. The Flj6tsdalur Hydroelectric Project case study leads to a improved design, with similar energy and load capacity compared to the project planning report, but approximately 6% better economical result, due to different dimensions of dam and conduits in the plant stage, and 3 % in the allocation stage. It is our belief that this new approach finally makes it possible to globally optimise the design of hydropower projects, and in fact, many other engineering design problems. This gives better results than the conventional local optimisation methods. The Genetic Algorithm reaches an optimal solution in less than five minutes and could undoubtedly be used on a much more complex and detailed objective function with more variables. It should be noted that it is not possible, in all cases, to prove that the true global optimum has been reached. However, our experience with the Genetic Algorithm shows that very good solutions are always reached, if not the optimal one. Another benefit of.this method, that has not been much emphasised in this paper, is the possibility to optimise the locations of the structures as well as their design. This could lead to solutions that were not considered before, or would take a long time to reach. On the other hand it is also clear that much work has to be done to make this a really powerful tool. The skeleton is formed, but the flesh is missing. The most demanding improvements are: The cost functions should be improved so that each structure is designed according to its geometry, conditions on site, and forces acting on it. Then the quantities of material, and manpower needed, should be calculated, and multiplied with unit-prices from a detailed up-to-date price database. The model should be expanded to the project planning leve! (allocation ~ n d global system stage) as presented 10 chapter 7, i.e., include the eXisting power system, optimise the load factor of each plant against reserve thermal power and optimise the construction order of new plants (timing). Connect the application to a GIS database, to be able to calculate reservoir volumes, dam volumes (varying dam sites), overburden for conduits etc. Make the project capable of optimising a series of projects in the same river (Master Plan StUdies), instead of optimising single projects. - The conclusion: Promising beginning, long way to go.9 ACKNOWLEDGEMENTS This study was made possible by research grants from Landsvirkjun (National Power Company of Iceland), Orkustofuun(National Energy Authority), the Consulting Engineering firms Honnun and Verkfneoistofa Siguroar Thoroddsen and the Icelandic Research Council. REFERENCES Eliasson, J. & G. Ludvigsson 1996. Optimal Design ofHydropower Plants. Univ, Iceland, Engineering Research Institute, Water Resource Department. (Research Report). Fljotsdalur Engineering Joint Venture. 1991. Fljotsdalur Hydroelectric Project - Project Planning Report. Landsvirkjun. Goldberg, D. E. 1989. Genetic Algorithms in Search Optimization & Machine Learning. AddisonWesley. Mosonyi, E. 1991. High-Head Power Plants Volume Two/A & Two/B. Akademiai Kiado, Budapest, Pirlot, M. 1994. General local search heuristics in Combinatorial Optimization: a tutorial. Belgian Journal of Operations Research, Statistics and Computer Science 32(1-2). l>orsteinsson L. (editor) 1985. Virkjanalfkan Orkustofnunar I-V, OS-85121/VOD-07. as VOD, VST. I>orsteinsson L. (editor) 1993. Endurskooun a virkjanalikan Orkustofnunar I-V, OS-85121/VOD07. OS - VOD, VST. Helgason, S. Civil Engineer, VST, Iceland. 618 ________________J Hydropower'97, Brach, Lysne, Flatabe& Helland-Hansen (eds) 1997 Balkema, Rotterdam, ISBN90 54108886 Monitoring survey andfeedback analysis of underground powerhouse of MingTornbs pumped storage plant LiangHai-Bo, GuZhao-Qi, ZhangMing&Ma Ji-Ming DepartmentofHydraulic Engineering, Tsinghua University, Beijing, People'sRepublicof China ABSTRACT: In this paper monitoring survey is presented together with feedback analysis together with numerical calculation of the underground powerhouse ofMing Tombs pumped storage plant. Based on the monitoring materials and calculation results, it is drawn that the cavern can keep its monolithic stability. , I PREFACE ,The Ming Tombs pumped storage plant which is the third largest one in China is located in Changping . county 40 km far away from Beijing city. The underground powerhouse is built inside the Mang "mountain on the left bank of Ming Tombs reservoir. ,There 4 reversible turbines with total installed capacity of 800MW. The overburden above the underground powerhouse is more than 300 m of conglomerate, with medium strength and relatively high Yang's modulus. This cavern is 145 m long, 23 m wide and 46.60 m high. The orientation of the long axial is NW2800, with a small angle to the maximum 'earth stress. See Figure 1. The maximum earth stress is 7.54 - 11.60 MPa, of which orientation is NW300 - 310. There are 3 main sets of structural planes in this region. That is set ofNE, set ofEW and set of NW. During the operation of this plant, the ownstream water level is more than 30 m higher han the arch crown of the underground powerhouse. The arch crown is supported by 20 ern thick shotcrete, 140 thick concrete lining and bolts 28 of 300 em and 500 cm length. For the walls, 15 ern shotcrete and bolts 22 of 550 em and 800 ern length. 6 prestress cables(600kN) are also adopted for supporting the walls. 2 DESIGN OF MONITORING SURVEY Three sections are selected for monitoring survey of the underground powerhouse: section 0+33 m, section 0+83 m and section 0+ 119 m. Multiple deflectometers and bolt stress gauges are embedded at the arch crown, upper walls(EI. 56.0), middle walls(EI. 45.0) and down walls(EI. 32.0). The length of multiple deflectometers is 14 - 24 m. Some of them are pre-embedded and others are post-embedded. The bolt stress gauges embedded at arch crown are 6 m long and others are 8 m long. See Figure 2 - 4. igure 1. General layout plan of the MingTombs pumped storage plant 619 ,,-"-01.'Hydropower'97, Brach, Lysne, F/atabo & Helland-Hansen (eds) 1997 Ba/kema, Rotterdam, /SBN 9054108886 TBM-tunnelling at SaudaPowerProject H.Moe, H.Holen& E.D.Johansen Statkraft Anlegg a.s., Hevik, Norway ~ B.Aspen .' Statkraft Engineering a.s., Hevik, Norway . ABSTRACT: The Sauda river-system in Norway is from nature well suited for hydropower developments. . Today four power-stations with a total installed capacity of 160 MW are running, with an average annual production of 1020 GWh. The plants are constructed in different periods since 1913 when the river was first developed for electric powerproduction. Each construction was the optimum solution at the time, however, modern techniques like tunnelling by a .Tunnel Boring Machine - TBM - has created new alternatives for hydro-power design and greatly increased the economically exploitable potential of the river-system. The owner of the plants is therefore planning a new hydropower scheme which will increase the installed capacity to 520 MW and the average annually .production to 2120 GWh. . 1. INTRODUCTION orway is from nature blessed with a topography . ell suited for hydropower development. High ountains and deep valleys with natural lakes ombined with abundant precipitation has formed e potential of hydro-electric power development. The Sauda river-system, in the western part of ',prway is typical of this potential. Different needs . r power and the possible technical solutions at the ctual tirrie, have continuously involved engineers v.: dbuilders to develop or plan to develop the otential of the river. The harnessing of the river has ways been ruled by an optimalisation of benefits 'gainst cost. However; in the earlier construction riods in Sauda (1913-30) was the design mainly estrained by technical limits such as; Transmission and distribution of electricity were limited. There was no national grid and the 9onsumption 'of energy had to be closely linked to the power-house. In Sauda this was solved by the ;construction of a factory to consume the energy. The factory's need for energy was constant through all 8,760 hours of the year. Since hydropower was the only source of energy, the regulated flow in the river had to be steady over the year, i.e. .large reservoirs had to be secured. - The maximum size of the units with regard to both transport weight, flow rate and the head was limited. High heads had therefore to be developed stepwise and several smaller aggregates were installed in the power-house. - Tunnelling techniques with high cost limited the economical tunnel length and thereby the plant layout. - Few roads existed, and a very difficult terrain with adverse weather conditions in the winter made it both expensive and time consuming to construct and operate roads to remote sites. Today, new techniques make new solutions possible from a technical point of view. hnprovements in the last decades, within both .mechanical/electrical and civil engineering have made new concepts possible such as; - The existence of a high-voltage grid makes it possible to coordinate production and transport energy over long distances with limited energy losses. In this system the power may be generated from many different sources, however, the regulating flexibility of the hydropower aggregates makes them capable to be dimensioned for peak performance in this system. . 6.23 - Technological development of mechanical and electrical equipment has made larger units possible. - New tunnelling techniques have been developed. Unlined cost saving pressure tunnels have been used with success. - Increased flexibility in design layout, makes it possible to locate the sites at places with adequate access by road or helicopter. - Environmental aspects are, however, a new and limiting criteria when hydro-power projects are designed in 1997. The technical-economical aspects are not anymore the only design-criteria for the modem hydro-power engineer. The new tunnelling techniques have been a main factor for the increased flexibility in hydropower design, and the latest innovation in tunnelling by Tunnel Boring Machine - TBM - is opening up new visions for the engineer.. - In this paper, new possibilities created by the use of TBM at the Sauda Power Project are presented. 2. ACHIEVEMENTS IN TUNNELLING TECHNOLOGY Although the hard rock tunnelling techniques have a long 'history, a number of cost saving improvements through the latest decades have given the tunnelling process a higher degree of performance. These improvements have greatly economised the whole tunnelling process and in factmore than balanced the general price increase. The relative cost of hardrock tunnelling has thereby steadily been reduced. Without detailing the achievements some major developments in the tunnelling technique can be stated as; TBM's in hard rock. TBM machines have also entered the arena in hard crystalline rock. It is thereby possible to bore in all classes of rock-masses. The minimum practical diameter for a TBM in hard rock is 3 - 3.5 m . depending' on the length of the tunnel. Rock support techniques. Bolting and steel fibre reinforced shotcrete have improved the support methods. In a TBM tunnel the blast vibrations are eliminated and the circular profile normally needs less rock support than the blasted alternative However, the TBM is less flexible in tackling heav; stability support at the face compared to drill and blast tunnelling. The speed of tunnelling is increased. For smaller TBMs with diameter of 3-4 m, a weekly advance rate of 200 m is common with peaks exceeding 400 m, even in hard rock. Yearly advances of 6-10 krn for one tunneldrive is realistic. Technical possibilities of longer tunnels. TBMs with its reduced need for ventilation and increased progress compared to the drill and blast method, make it possible to construct longer tunnels from one adit. Headings of 20 km or more are possible within reasonable cost-and time frames. The distance between opposite adits may be 40 krn or more. Environmental considerations. Longer headings reduce the number of job-sites with less need of access roads and thereby the environmental impact. In addition, the reduced number of sites often means great savings in mobilisation and operation costs. 3. NEW DESIGN CRITERIA FOR THE ENGINEER. The new achievements in the tunnelling techniques and geological engineering promote a remarkable flexibility in designing underground hydro-power schemes. The lay-out of the hydropower scheme may be studied more freely as an alternative to the earlier designs, when the principle was to link the intake and the power house together with a waterway as short as possible. 624 Flexible lay-out of the scheme. ; Today, the engineer should not necessarily search only for the concentrated lay-out as this is probably jinot the most economically and environmentally '::favourable solution. A design with transfer tunnels, ;:pressure tunnels and underground located power ;houses make new arrangements possible, where theipower-house is not necessarily linked to the actual 'I'.river-course or waterfall. Intakes, reservoirs, tunnels and power-houses can located where the conditions are most favourable regard to' technical, economical and 'I'environmental parameters. 'f I(i t,:f'Additional catchment areas. " The possible pick-up of run-offs from additional jcatchment areas are possible with transfer tunnels. .iEven remote areas maybe intercepted within cost- and time-frames. These may be rivers not economically exploitable as single objects in their own water-courses. , 'fhorterconstruction period. 'it: 1,1';mgher tunnelling advance rates shorten the "onstruction time for the project. and tunnels, even with high water pressure !:an be constructed without lining, as the rock is used :;' a part of the final construction if the geological \\onditions are favourable.l, .' eadloss in tunnels. e headloss in a tunnel is reduced if the tunnel is led, however, an increase of the cross-sectional ea in an unlined tunnel is generally a cost-effective . pensation because of the low cost of tunnelvation compared to lining costs. A fully lined nel requires in addition an extension of the nstruction period. The TBM-tunnel implies significantly less adloss than the drill and blast tunnel and the ntour is in fact only marginally rougher than in concrete lined tunnel. Environmental aspect. The flexibility in lay-out gives the engineer the possibility of reducing the environmental impacts from reservoirs, access roads, spoil tips etc. to a more acceptable level. 4. THE SAUDA POWER PROJECT 4.1 The existing power-plants. The Sauda river-system is in fact not one single river. The basin that forms the river-system, consists of many smaller valleys with their own creeks that together join forces in the lower sections of the river. In addition there are several other rivers in additional catchment areas that can be connected by transfer-tunnels to the Sauda hydropower project. In the existing hydro power scheme, the technical achievements through this century are easily studied. Four different power stations constructed in different periods represents the optimum technicalJ economical solutions at the time period in question. The first powerstations. The section of the river with the most concentrated falls, between Lake Dalvatn and Lake Storlivatn (260 m head / 1200 m water-course), was utilised in the first powerstation, Sauda I, which was planned in 1913 and has been in operation since 1919. This powerplant consists of a 900 m horizontal tunnel, penstocks in steel and an open air powerhouse. The electricity is generated in five units in the powerhouse (each 5.4 MW / 2.6 m3/s). The energy was transformed to 13.5 kV and distributed to the factory through a 5 km powerline. In 1980 the voltage was increased to 66 kV. Sauda II powerplant came into operation in 1922 utilising the head between Lake Holmavatn and Dalvatn upstream of Sauda I. The lay-out were similar to the Sauda I with a 900 m horizontal tunnel, penstocks in steel and an open air powerhouse with three generating units. The powerstation was closed down and replaced by a modem 22 MW unit in 1978. Step by step, the technical development has , reduced the costs and made new solutions possible. Sauda Ill powerstation was planned in 1926 and has been in operation since 1931. This powerstation is located at sea-level in Sauda utilising the head from 625 Two photographsjrom the constructions of the oldest plants, 1922. Lake Storlivatn. The lay-out was similar to Sauda I and II with a horizontal headrace tunnel, penstocks in steel and a powerhouse on the ground. The tunnel was at the time impressive with its length of 7386 m and 21 m' constructed from 7 adits. In the powerhouse three units (each 20 MW / l lm'zs) supply the energy directly to the neighbouring factory at 12 kV voltage. The latest powerstation - Sauda IV. Sauda IV which was planned in 1965 and has been in operation from 1968, comprises an underground power-house and more tunnels than the three oldest all together. This includes transfer tunnels to catch additional run-off from neighbouring rivers. The electricity is generated in two units in the powerhouse (each 25 MW / 13.5 m3/s). The energy is transformed to 66 kV and distributed to the National grid. 4.2 The 1984-design. This design without the use of TBM consisted of several smaller power-plants. Seven new plants located all over the Sauda area should in addition to three of the existing plants increase the production capacity in the Sauda river by about 500 GWh. The oldest plant, Sauda I, should be closed down. The investments for this project was estimated to 230 mill. USD or 0.46 USDIkWh. This project was, however, not implemented. The investment and operating cost of the ten plants were high, and the owner of the plants started to spot a silver lining in the horizon. 4.3 The 1997-design. There are, in fact, many alternatives for further hydro power developments in the Sauda area. . The solution presented here includes boring of 51600 m of tunnels from only 3 sites. The achievements in tunnel boring technique has made this concept economically and environmenmjjv viable, since long tunnels can be constructed at a very competitive price from just a few work adits. The environmental impacts, like the construction of new roads will be very limited and considerably reduced as compared to more conventional methods. e.g. the only new access road that has to be constructed is in fact the 7.5 km to the site at Botnavatn. Site Botnavatn. An important element of the new design is the site at Lake Botnavatn. The run-off from the upper part of the Sauda rivers together with additional catchment areas in the north, will be transferred into the lake via a tunnel system. A TBM with diameter 3.5 m will from one adi .. west of the southern end flf the lake, bore a 17500 m long tunnel with 7 brook inlets between theVaulo river and the Lake Botnavatn. A : 2500 m long tunnel from north east with 3 brook inlets will also be connected to the TBM-tunnel. In average 214 mill. m' of water per year will be transferred into the lake via this tunnel system. Botnavatn power station. The average head of 314 m between Lake Nedre ( 'j 626 I ~ .__..-J '00 '00 800 700 900 '00 200 100 1100 1000 ma.sr NOT TO SCALE The 1984-design 627 ST.EINAVATNET $AUDAFJORDEN I. Senna 130MW 2. Sauda III 64 MW 3. Sauda IV 50 MW 4. Dalvaln 40 MW 5. Svartkulp 15 MW 6. Svartavatn 15 MW 7. Slettedalen 12 MW 8. Helgedalen II MW 8. Fetavatn 0,9 MW 10. Ferstavam 2,1 MW II. Sauda I CLOSED he existing plants The 1997-design EXISTING POWERSTATION EXISTING TUNNEL C NEW POWERSTATION NEW TUNNEL CATCHMENT BOUNDARIES ) 628 1l ,j, ,Ifi 'I Sandvatn and Lake Botnavatn will be utilised in a 28 MW powerstation transferring another 150 mill. l: 3 LakB;i' m into e otnavatn. 1'" Lake Botnavatn reservoir. 50 m high rockfill dam with a central core of moraine will be constructed at the outlet of Lake Botnavatn and together with a lake tap intake for a new Sauda 2N powerstation at Dalvatn, the reservoir capacity is calculated to be 88 mill. rn' in Lake Botnavatn. Site Dalvatn. The average head of 134 m between Lake Botnavatn and Lake Dalvatn will be utilised in the new 34 MW powerstation, Sauda 2N. The 10850 m long headrace pressure tunnel can be bored by a TBM with diameter 4.5 m directly from the powerstation area. Alternatively about 7.5 km can be bored by TBM from a work adit at Lake Berdalsvatn while i the remaining distance is excavated by the drill and blast method. A surge shaft will also be included. A new headrace tunnel will be constructed for the ;: old Sauda II power station using the head between f. the two lakes Holmavatn and Dalvatn. ! The new Sauda III powerstation, Sauda 3N. , J 2 x 2 twin power generating units in the same power :; station will be utilising the heads from the lakes l!Dalvatn (H) and Storlivatn (L) respectively. Total !:'installed capacity will be 385 MW.t To utilise the head of 550 m between LakeI;i;Dalvatn and the Saudafjord a 10580 m long low ';'pressure headrace tunnel and a 695 m long, 20 m', ;' 45 0 unlined pressure shaft needs to be excavated. A conventional 165 m long surge shaft at 45 0 in clination and a cross sectional area of 45 m' will be constructed at the end of the headrace tunnel. The headrace- and transfer tunnels for the new 'fSauda 3N power station are a big challenge for the bTBMs. The tunnel system is planned bored with a diameter 5 m, in two directions from one owork adit. Two transfer tunnels branching off from headrace tunnel towards the south, 8190 m long 4500 m long respectively, will be drilled by a 3.5 m in diameter. The tunnel system will a total of 7 brook inlets. t For utilising the lower head another unlined pressure shaft will be constructed between the 2 new power generating units in the Sauda 3N power station and the existing headrace tunnel, to replace of the old Sauda III power station with it's steel penstocks which will be abandoned. 5. CONCLUSIONS. The introduction of hard-rock TBMs at the Sauda Power Project has opened up for new design options. The concept of utilising the hydro power potential in the river is changed in the latest design scheme based on extensive use of TBMs. In this design five tunnels with the total length of 51 km are planned bored from only 3 adits. The new hydro power scheme for the Sauda area will increase the total annual production from 1020 to 2120 GWh at an estimated cost of 385 mill. USD. The oldest operating plants, Sauda I and III will together with the penstocks of Sauda II be closed down, while three new power-stations will be constructed. The total capacity of the stations will be 520MW. The oldest plants are in such a condition that they in any case have to be rehabilitated. The alternative to only rehabilitate them with no increasement in the generating capacity will cost about 65 mill. USD. Taking into account the alternative cost for. rehabilitation of the oldest plant, the cost of the new energy will beabout 0.29 USDIkWh. Conclusions - TBM's have changed the design concept. - The energy output from the river is higher than by the earlier designs, at a lower cost. - The environmental consequences are reduced. 629 Hydrapower'97, Brach, Lysne, FlatabB & Helland-Hansen (eds) 1997Balkema, Rotterdam, ISBN90 54108886 Rebuilding of the 70 years old Nore 1 Power Plant Johannes Hope Statkraft SF, Hovik, Norway ArildPalmstrom erdal Stromme AS, Sandvika, Norway jell Finnerud tatkraft Engineering AS, Hevik, Norway STRACf: At Nore 1 Power Plant the water was conducted from a low pressure head race tunnel to the bove ground power house by 8 exposed penstocks of 1000 m length. For safety reasons, Norwegian uthorities have ordered all so-called water gas welded penstocks to be taken out of service within certain 'me limits. A number of alternative schemes for the necessary rebuilding of Nore 1 were developed, The 1ected scheme was to replace the penstocks between the existing head race tunnel and the power house by a ew underground waterway system in rock. Required rock overburden of water bearing tunnels and length of steel lined tunnel were determined by means of rock stress measurements as well as numerical analyses. Most f the tunnels were excavated by drill & blast while the power plant was in operation with the old penstocks. the vicinity of the power house vibrations induced by blasting had to be kept within certain precalculated , its, To connect the low pressure head race tunnel to a new high pressure tunnel it was decided to excavate vertical unlined pressure shaft of 213m length. However, the main challenge to the project was to limit utage time for the turbine units during t ~ e construction and erection period for the new system, 6 of the 8 penstocks were of so-called water gas welded type. ,1 Historical development ore 1 was the first hydroelectric power project in 1.2 Watergas welding orway constructed and owned by the government. riginal data: Watergas welding is an old method of making ross head, m: Approx.360 longitudinal joints on steel pipes. This method was utput,MW: 200 commonly used up to about 1930, when it was ischarge, m3/s: 72 gradually replaced by electric welding. ean production: 1000 GWh/year Watergas welding is a form of forge welding, urbines: 8 horizontal Pelton of with heating applied by means of a water gas flame 25MWeach. instead of a coke furnace. Watergas is generated by .ear of construction: 1919 - 1954. injecting steam into a hot coke furnace and mixing The first stage of the plant, consisting of 4 with air. its, was constructed between 1919 and 1928. Later The prepared overlapping edges of the plate the plant was extended gradually from 4 to 8 were heated by the gas flame until they reached a its, This extension was completed in 1954. The mushy state, after which they were forge-welded, ter was conducted from the reservoir to the plant first by hand then later mechanically. Maintaining ough a' 5200 m long low pressure head race the correct heat was critically important. Over-rapid .nnel, cross section 40 m2, followed by 8 exposed heating led to irregular temperature distribution and . nstocks of approx. 1000 m length, diameter hence internal bonding defects, and also to scaling ing from 1600 to 1400 mm. As each turbine has and surface slag, Too low temperatures resulted in o runners, there was a bifurcation on each inadequate bonding, excessive hammering and nstock, just in front of the power house wall. embrittlement of the material. 631 The nature of the process and the requirements entailed by it precluded the achievement of a continuous, homogenous seam. Noticeable improvements were secured with the switch from manual .to mechanical forging, bur nevertheless it was still possible to keep coming across the typical, inherent faults of the process. Typical faults which can be observed in these welds consist of overheated zones, inadequate wall thickness, segregations, slag inclusions, inadequate bonding and cracks. 1.3Backgroundfor rebuilding Over the years, there have been a few ruptures in water gas welded penstocks in Norway. As a consequence of this the authorities, i.e. the Norwegian Water Resources and Energy Board, has ordered all owners of water gas welded penstocks in Norway to take these out of service within certain time limits. Following these instructions Statkraft SF decided to study a number of different alternatives for rebuilding of the plant. Fig. 1 Layout of rebuilding area. 2 SELECTION OF REBUILDING ALTERNATIVE. 2.1 Considered alternatives Among considered alternatives were: To continue production with the two electrically welded penstocks only. To replace the 6 water gas welded penstocks with 2 new steel pipes of larger diameter. To replace existing penstocks and power house by new underground pressure shaft, power house and tail raise tunnel in rock. To replace both Nore 1 and Nore 2 Power Plant by a new underground waterway system between the existing head race tunnel and Lake Norefjorden including new power house all in rock. Nore 2 is utilizing a head of 100m between Nore 1 and Lake Norefjorden which is located 4 kID downstream of Nore 1. The selceted alternative. 2.2 Selected rebuilding scheme The selected scheme, which due to calculations provides the best Net Present Value, was to replace the penstocks by an underground waterway system between existing head race tunnel and the power house. This system consists of the following main parts: An approx. 100 m long low pressure tunnel, branched off near the downstream end of the .existing headrace tunnel. In this tunnel a new fixed wheel gate is installed. Adit tunnel, upper level, with bulk head An approx. 220 m high vertical pressure shaft An approx. 400 m long unlined pressure tunnel An approx. 465 m long steel lined pressure tunnel, pipe diameter 3.7 m. Adit tunnel, lower level, with bulk head A complete branch pipe arrangement, including in total 15 bifurcations, 8 emergency closure valves, interconnecting pipes, flange pipes etc., for distribution of water to the in total 16 runners. The new emergency closure valves (butterfly type) have also replaced. the function of the 16 old turbine valves, which have been removed in order to achieve simplified operation and maintenance, and reduced head loss. The water way system upstream of the steel lined tunnel may also form part of a possible new Norel Power Scheme in rock in the future. 632 Fig. 2 Longitudinal section of rebuilding area. 3 GEOLOGY AND TUNNEL DESIGN 3.1 Geology and ground quality. A simplified geological setting is shown in Figure 3. The rocks are Precambrian supracrustal rocks, i.e. rocks which were formed on the surface (such as basalts and sediments) some 1800 mill. years ago. Later, during several metamorphic orogenesis they have been formed into green schist,' greenstone, quartzite and metarhyolite. The rocks are mainly steep dipping with strike almost along the tunnels. Joints or other weakness planes occur along the schistocity (foliation) of the rocks. Another set of joints occurs approximately at right angle to the foliation. The quality of the ground according to the Q-systemis Q = RQD/Jn x Jr/Ja x Jw/SRF =(80 to 90)/4 x 1.5/(1 to 2) x (0.66 to 1)/1 = 10 to 25 (good) 3.2 Unlined pressure shaft and pressure tunnel Occasionally, swarms of joints reduce the quality of the rock mass to Q =0.5 to 5 (poor to fair) Already during the preinvestigations the rock mass conditions were found suitable to ,have "the pressure shaft and part 'of the pressure tunnel unlined, i.e, without steel lining. For this solution the following conditions are essential for a successful result: 1. No leakage and no deformations of importance, should take place during the lifetime of the power plant. This is achieved by ensuring that the minimum rock stresses are higher than the internal water pressure in the tunnel or shaft. A verification of the stresses by measurements in the tunnel is a prerequisite, especially for safety factor lower than F =1.5. 2. Large, open or filled joints which can cause deformations (for example increasing the width of the joint), or lead large quantities of water into the rock masses surrounding the tunnel or shaft, must be sealed. This is especially important for joints normal to the direction of the minimum rock stress. 3. When designing unlined pressure waterways a main condition is that the minimum rock stress is higher than the water pressure. Large savings (approximately USD 10,000 per metre tunnel) could be achieved if the inlet cone could be moved outwards, thus reducing the length of the embedded steel penstock. A safe location of the cone with a limited length of the steel lining is therefore important. The following calculations and measurements were performed during planning and construction to meet this requirement: In the early phase of the planning preliminary evaluations of the location of tunnels and shaft were made based on assumed magnitude of the rock stresses. The most important part of the unlined waterway is the location of the inlet cone, i.e. where the embedded steel lining towards the powerhouse starts. A preliminary location of the cone was chosen 620 m from the power house, applying a safety factor F = 1.5. Here, the overburden is 200 m. The access tunnel to the inlet cone was designed so that the location of the cone could be changed after stress measurements performed during construction of the tunnne!' 633 500 400 400 green schist meta-basalt, phyllite 1. hydraulic splitting test 2. hydraulic splitting test adjusted rock surface I HWL =736 ~ pressure shaft ,,,, I,, I I I I, I '" ...... . 500 m 1--------11 o I 900 700 ' .. quartzite N 400 @ hea race tunnel-800 80 400 Figure 3 Upper: Plan showing simplified geology and section used in the FLAC calculations. Lower: Section with pressure waterway, location of inlet cone and hydraulic splitting tests. 634 r , When the construction works started numerical calculations were performed using FLAC version 3.22. This 'finite difference code' model is suitable for continuous ground. It simulates the behaviour of constructions in rock by the use of relevant input parameters of the rocks, stresses and topography. The topography is represented in a 2-dimensional model by elements or zones comprising a network adapted to the local conditions. For Nore 1 a simplified model of the topography had to be used because, as shown in Figure 3, the actual area consists of a ridge between two valleys. This ridge gives a reduced contribution to the rock stresses in tunnels and shaft. The adjusted topography used in the FLAC model is shown by dotted lines in Figure 3. The following input parameters were used in the FLAC calculations: The density of the rock mass 'Y = 0.027 MN/m3 The E-modulus of the rock mass E = 2510gQ = 18 GPa The bulk modulus of the rock mass M = FJ[3(l-2v)] =10 GPa The shear modulus of the rock mass G = E/[2(l+v)] :: 7.5 GPa The ratio horizontal/vertical stress k :: 1.5 The results from the FLAC calculations are shown in Figure 4. (The irregular course of the stress lines at the right end of the figure is caused by the small bend in the rock surface here.) To verify the stresses found in the FLAC model, hydraulic splitting tests were performed. The first measurements, showing the minimum stress cr3= . 2.3 MPa, were performed 242 m into the access tunnel. Here, the minimum rock stresses calculated in the FLAC model were close to the measured ones. From this an interim location of the cone was chosen 595 m from the power house (at chainage 540 in the access tunnel in Figure 4), based on a safety factor of F = 104. The second hydraulic splitting tests were performed when the access tunnel had been excavated 480 m. This time the results were approx. 9% lower than those calculated by FLAC (cr3:: 3.2 MPa compared to 3.5 MPa). Based on this the cone was finally located 580 m from the power house (at chainage 525 in the access tunnel, see Figure 4), applying a safety factor of F = 1.25. The measurements and calculations made (at a cost of usn 40,000) showed that the steel lining could be 40 m shorter (by moving the location of the cone from 620 m to 580 m). This gave savings in the order of usn 400,000. -----j..2.>O 1.100 Figure 4 Results from FLAC calculations showing the magnitude of the minor stress. (The chainage given along the access tunnel is 55 m shorter than the distance along the steel lined tunnel from the power house) The first water filling of the un