Thông tin tài liệu
POLITECNICO DI MILANO Dipartimento di Elettronica e Informazione RESEARCH DOCTORAL PROGRAM IN INFORMATION TECHNOLOGY ASSESSING AND OPTIMIZING THE OPERATION OF THE HOABINH RESERVOIR IN VIETNAM BY MULTI-OBJECTIVE OPTIMAL CONTROL TECHNIQUES Doctoral Dissertation of: Xuan Quach Advisor: Prof Rodolfo Soncini-Sessa Tutor: Prof Carlo Piccardi The Chair of the Doctoral Program: Prof Carlo Fiorini Co-supervisor: Dr Andrea Castelletti Dr Francesca Pianosi 2011 - XXIV Contents Abstract Introduction The Red River Basin and the Hoabinh reservoir 1.1 Physical and socio-economic system 1.2 Important issues of the system 1.2.1 Flood 1.2.2 Water shortages 11 1.2.3 Electricity 13 1.2.4 Navigation and water pollution 13 1.2.5 River bed erosion 14 1.2.6 Climate change 14 1.2.7 Transboundary issues 15 1.3 Institutional and normative framework for Hoabinh operation 16 1.4 Objectives of the study 19 1.5 Conceptualization: actions, criteria and design indicators 21 Modelling the Hoabinh water system 2.1 The Time Step 2.2 Upstream catchments 2.3 The Hoabinh reservoir 2.2.1 27 27 29 Data analysis 29 32 2.3.1 Characteristic curves of Hoabinh reservoir 34 2.3.2 Rating curve of Hoabinh reservoir 36 2.3.3 Mass balance equation of the reservoir 39 2.3.4 Minimum instantaneous discharge function 42 2.3.5 Maximum instantaneous discharge function 43 2.3.6 Validation of the reservoir model 44 2.3.7 Time constant of the reservoir 46 2.3.8 2.4 Model of the hydropower plant 48 The downstream river network 51 2.4.1 Background 52 2.4.2 The data 52 Contents 2.4.3 The model structure 53 2.4.4 Model identi cation 55 Improvement potential to Hoabinh reservoir operation 3.1 Historical operation of the Hoabinh reservoir 59 59 3.1.1 61 Hydropower production 61 The ideal operation of the Hoabinh reservoir 65 3.2.1 Solution of DDP problem 67 3.2.2 Discretization 67 3.2.3 3.3 61 Flood control 3.1.3 3.2 Water supply 3.1.2 Results and discussion 68 Summary of the main results Optimization by Stochastic Dynamic Programming 75 77 4.1 Review of application of SDP to reservoir operation 77 4.2 Formulation of the SDP problem 78 4.3 The catchment model 81 4.4 SDP results under `white in ow' assumption 83 4.5 SDP results under `not-white in ow' assumption 87 4.6 Summary of the main results 90 Optimization by Policy Search 93 5.1 Implicit Stochastic Optimization (ISO) 93 5.1.1 95 5.2 Evolutionary Multi-objective optimization (EMO) 5.2.1 Discussion of ISO results and its limits Non-dominated Sorting Genetic Algorithm (NSGAII) 5.2.2 5.2.3 103 The initialization procedure 104 Measure of Pareto frontier quality: the Hypervolume Indicator (HD) 5.2.4 101 104 Application results and discussion 105 5.3 EMO with hydrometeorological information 5.4 Summary of the main results 108 113 Comparison and discussion 115 6.1 Signi cance of the results for the Hoabinh management 118 6.3 Comparison of the reservoir optimization methods 6.2 115 Further research 120 List of Figures 1.1 The Red River Basin 1.2 Energy production of Hoabinh reservoir from 1989 to 2008 13 1.3 Percentage of monthly ow rates from China sub-catchment at Laichau (black), Tabu (gray) and Hoabinh (light gray) 16 1.4 Historical water level of Hoabinh reservoir (1995-2004) on the Da River 21 1.5 Yearly pattern of water demand at Sontay 24 2.1 The scheme of the Hoabinh water system and of its model 27 2.2 Minimum annual ows of the Da River at Hoabinh dam (blue), of the Thao River at Yenbai (green), and of the Lo River at Vuquang (red) 2.3 32 Mean annual ows of the Da River at Hoabinh dam (blue), of the Thao River at Yenbai (green), and of the Lo River at Vuquang (red) 2.4 33 Maximum annual ows of the Da River at Hoabinh dam (blue), of the Thao River at Yenbai (green), and of the Lo River at Vuquang (red) 2.5 The dam of the Hoabinh reservoir 2.6 Water level vs 2.7 Hoabinh reservoir: surface area vs storage 2.8 Downstream water level of Hoabinh reservoir: 33 34 storage (a); measured (solid) and esti- mated (cross) water level of Hoabinh reservoir (b) 35 36 observed (blue) and estimated by piecewise linear (green) and by polynomial of order (red) over the period 1995-2004 2.9 37 Release through penstocks vs hydraulic head (a); Release through penstocks vs storage (b) 2.10 Hoabinh reservoir: rating curve of bottom gates 37 38 2.11 Hoabinh reservoir: spillway rating curve 39 2.12 Hoabinh reservoir: daily unitary surface evaporation over one year 40 2.13 Hoabinh reservoir: instantaneous maximum and minimum releases 2.14 Validation scheme of the Hoabinh reservoir model 44 45 List of Figures 2.15 Observed (blue) and simulated (green) release; observed (purple), simulated (black) and dead (red) storage over the period 1995-2004 45 2.16 Linearizing the storage-discharge relation in correspondence with various storage values crease in the slope when operative storage s ¯ s ˇ s ˇ Note the sharp in- is greater than the maximum 47 2.17 Validation scheme of the Hoabinh hydropower plant model 49 2.18 Measured (solid line) and estimated (dotted) releases through the turbines at the beginning (a) and at the end (b) of the period 1995-2004 2.19 Observed vs 50 estimated downstream water level of the Hoabinh dam over the period 1995-2004 50 2.20 Historical annual mean of step energy production (thick line) and corresponding rate of release through the turbines (thin line) over the period 1995-2004 51 2.21 Cross-correlation at dierent lag values between observed time series at dierent stations in the RRB basin over the period (1989-2004) 54 2.22 Scatter plot of measured and modeled water level at Hanoi (a), and ow at Sontay (b) 3.1 Yearly pattern of the Hoa Binh level with historical oper- 3.2 Historical in ow (dotted line) and release (solid line) of 3.3 57 Hoabinh reservoir: Historical in ow minus release (blue ation over the horizon 1995-2004 Hoabinh reservoir over year 1996 60 61 line), water level (light blue line), normal water level (red line) over the period 1995-2004 3.4 62 Flow at Sontay over the period 1995-2004: historical value (vertical axis) and hypothetical value (horizontal axis) without regulation of Hoabinh reservoir 3.5 Total water de cit (a) and mean de cit step cost (b) in 3.6 63 Water level at Hanoi in 1996: historical value (solid line) each year over the period 1995-2004 63 and hypothetical value without the Hoabinh reservoir (dotted line) 3.7 Maximum water level in Hanoi (a), and mean ood step cost (b) in each year over the period 1995-2004 3.8 64 64 Total hydropower production (a) and mean of minus hydropower step cost (b) in each year over the period 19952004 65 List of Figures 3.9 Performances of some tion grids of control: ddp-policies with dierent discretiza- U14 (cyan), U20 (green) and U41 (blue) that dominate the historical one (red cross) over the evaluation horizon 1995-2004 (a); zoom the box in the left panel (b); The diameter of the circles is proportional to the hydropower production 3.10 The Hoabinh water level produced by ddp-25 over the evaluation horizon 1995-2004 3.11 ddp results: 68 71 release strategies of Hoabinh reservoir for reducing water de cit ddp-2 (a,b), for avoiding ood at Hanoi ddp-3 ddp-4 (e,f ) in 1998 and 2002 (release (blue), in ow (red), (c,d), and for compromising among the two water level (light blue), and normal water level (dotted red)) 72 3.12 Water level at Hanoi when the Hoabinh release is permanently equal to zero (horizontal) and produced by policy (vertical) 3.13 Result of ddp-4 ddp-4 73 Top panel: water level (blue) and dead water level (dotted red); middle panel: release of Hoabinh reservoir (blue), in ow to Hoabinh (red), ow of Thao (pink), and ow of Lo (yellow); bottom panel: water level at Hanoi (blue), water level at Hanoi without Hoabinh release (black), and ooding threshold (dotted red) in August, 2002 74 3.14 Hoabinh water level in 2004 under dierent policies (ddp1 (blue), ddp-2 (green), ddp-3 (red), and ddp-25 (light blue)) when the optimization horizon ends in 2005 (a) and in 2004 (b) 4.1 75 Autocorrelation of Da (a), Thao (c) and Lo (e) river ows; spatial correlation of river ows between Da-Thao (b), DaLo (d) and Lo-Thao (f ) 4.2 The periodic time patterns of 82 µ (a) and σ (b) of the model (4.7a,b) for Da catchment, together with the samples from which they were estimated (calibration period) 4.3 Flow distribution of Da (a), Thao (c) and Lo (e); autocorrelogram of 4.4 83 yt (b), Y ηt B (d) and V ηt Q (f ) Water levels at Hanoi with historical (red) and (blue) and sdp-ar0-4 (green) policies 84 ddp-4 86 List of Figures 4.5 ddp-4 (blue) and sdp-ar0-4 (green) The eects induced by during the ood event of 1996: water levels in Hoabinh reservoir (top panel); Hoabinh releases and Da (red), Thao (violet) and Lo (yellow) in ows (middle panel); water levels at Hanoi and the ood threshold at 9.5 m (red dotted)(bottom panel) 4.6 The in ow (red line); the release (continuous lines) and storage (dotted lines) produced by the sdp-ar0-4 4.7 87 ddp-4 (blue) and (green) 88 Event of January-February 1999: the eects induced by the ddp-26 (blue) and sdp-ar0-26 (green) policies: water levels in Hoabinh reservoir (top panel); Hoabinh releases and Da (red), Thao (violet) and Lo (yellow) in ows (middle panel); supply de cits (bottom panel) ξt+1 4.8 Auto-correlation of 4.9 The eects induced by and sdp-ar1-4 in equation (4.8) ddp-4 (blue), sdp-ar0-4 89 90 (green), (light blue) during the ood event of 1996: water level and the maximum level (red dotted) in Hoabinh reservoir (top panel); Hoabinh releases and Da (red), Thao (violet) and Lo (yellow) in ows (middle panel); water levels at Hanoi and the ood threshold (red dotted) at 9.5 m (bottom panel) 4.10 Performances of sdp-ar1 ddp sdp-ar0 policies (blue), 91 (brown), (cyan) policies that dominate the historical one (red cross) The diameter of the circles is proportional to the hydropower production 5.1 Procedure of Implicit Stochastic Optimization (iso) 5.2 92 Relationship: between time (day of the year), in ow and release decision (a) by ddp-15; 5.3 (green) (calibration dataset) Trajectories of ddp-15 (blue), 94 between time, in ow and interpolated release decision (b) by b iso-a (blue), and iso- iso-a (green), and 96 iso-b (red), in ow (light blue) over the evaluation horizon 19952004 5.4 Release decision of ddp (blue), water demand at Sontay (green), and interpolated release decision of over the evaluation horizon 1995-2004 97 iso-c (red) 98 List of Figures 5.5 Release decision by (red) and iso-c ddp-15 (blue), iso-c `interpolated' `simulated' (green) over the calibration horizon 1958-1977 (a) A blow up of the box in the left panel (b): the red and light blue dotted lines are the storage trajectories by 5.6 ddp-15 and iso-c iso-c with dierent values 99 99 Interpolated release decision of of storage while other inputs assume their historical values (t 5.7 = 6500) Release decision by (red) and iso-c ddp-15 (blue), iso-c `interpolated' `simulated' (green) over the calibration horizon 1958-1977 (a) A blow up of the box in the left panel (b): the red and light blue dotted lines are the storage trajectories by ddp-15 and iso-c+ 100 5.8 Procedure of Evolutionary Multi-Objective algorithm (emo)102 5.9 The hypervolume is the surface area of the grey region comprised between the reference point H and the points in the approximate Pareto frontier (black) divided by the surface area of the rectangle with vertices in H and U (dashed lines) 105 5.10 Performances of ddp policies (blue), emo policies of Ta- ble 5.3 (pink), and the historical one (red cross) The diameter of the circles is proportional to the hydropower production 107 5.11 Performance of historical (blue), ddp-25 (green), and emo- (brown) policy over the evaluation horizon 1995-2004 5.12 Water level at Hanoi produced by: historical (red), 25 (blue), and emo-6 109 ddp- (pink) policy in July and August of some years in the period 1995-2004 The red dotted line is the ood threshold of 9.5 m 110 5.13 Procedure of Input Variable Selection ivs 111 5.14 Location of hydrological and meteorological stations of the Red River Basin 6.1 112 Performances (in terms of design indicators) of several operating policies designed by dierent optimization methods The diameter of the circles is proportional to the hydropower production 6.2 116 Performances of several operating policies designed by different optimization methods in terms of physical indicators The diameter of the circles is proportional to the hydropower production (evaluation horizon 1995-2004) 119 Optimization by Policy Search Figure 5.14: Location of hydrological and meteorological stations of the Red River Basin Table 5.4 shows the results of several optimization experiments using emo and Arti cial Neural Network with input vector It , including dier- ent combinations of the above selected variables It can be seen that the quality of the Pareto frontier (measured by hd) is signi cantly improved with respect to the result in Table 5.2 Table 5.4: Quality of Pareto frontiers (hypervolume value HD) obtained by emo with exogenous information under dierent input vec- tor and number Input It size P ν of neurons in the decision rule, population and number of generations ν VQ Y T T |t, s, at , qt B , qt , T2 G , T3 G | VQ Y M M T T |t, at , qt B , qt , aHB , P2 T , P3 T , T2 G , T3 G | t−1 VQ Y M M T T T |t, at , qt B , qt , P2 T , P3 T , P2 U , T2 G , T3 G | VQ T M T T M Y |t, s, at , qt B , qt , P2 T , P3 T , P2 U , T2 G , T3 G | VQ Y M M T T |t, s, at , qt B , qt , aHB , P2 T , P3 T , T2 G , T3 G | t−1 VQ Y |t, s, at , qt B , qt T T M M , aHB , P2 T , P3 T , T2 G , T3 G | t−1 9 10 10 10 G P 250 250 250 250 250 250 G 100 100 100 100 100 150 HD (opt) 0.2589 0.3285 0.3372 0.3257 0.3611 0.4567 HD (val) 0.00182 0.04295 0.01769 0.11148 0.20793 0.26640 Table 5.5 shows the performances in terms of design indicators of the operating policies provided by con guration with V Y M M T T It = |t, s, at , qt B , qt Q , aHB , P2 T , P3 T , T2 G , T3 G | t−1 For convenience, selected policies of tached 112 It is seen that emo ddp, sdp, and and emo ν = 10 are also at- with exogenous variables is able to nd 5.4 Summary of the main results policies which nearly eliminate water de cit and have relative good values of hydropower production and ood control, for instance emo-exo-6 A more detailed analysis of the system trajectories under this policy is given in the next Chapter Table 5.5: Result age zon of emo values of with exogenous step costs of policies 1995-2004 over information: the found aver- validation by EMO V Y M M T T It = |t, s, at , qt B , qt Q , aHB , P2 T , P3 T , T2 G , T3 G | t−1 ν = 10 hyd Policy def GWh horiwith and o 2 (m /s) (cm) History -26.5 887 902 ddp-25 sdp-ar1-21 emo-6 emo-exo-1 emo-exo-2 emo-exo-3 emo-exo-4 emo-exo-5 emo-exo-6 -31.4 0.3 75 -28.5 73 422 -26.5 637 382 -27.4 283 351 -26.7 42 343 -27.9 509 359 -28.3 433 405 -28.4 510 438 -27.7 346 5.4 Summary of the main results This Chapter has presented two Policy Search approaches to design new operating policies for the Hoabinh reservoir The Implicit Stochastic Optimization (iso) identi es the decision rule by regression over simulated system trajectories from deterministic optimization Arti cial Neural Networks (ann) were chosen as regression function because of their ability in reproducing any nonlinear relationship Nonetheless, no operating policy Pareto-dominant over the historical one could be found by iso Subsequently, a novel method that combines Evolutionary Multi- Objective (emo) algorithm and ann was proposed and applied Many operating policies that dominate the historical operation and are equivalent to the policies found by sdp-ar1 (see Chapter 4) were found Fi- nally, exploiting the possibility given by the Policy Search approach of easily including exogenous (e.g meteorological) information in the decision rule, we showed that performances can be signi cantly improved when considering precipitation and temperature data 113 Comparison and discussion This Chapter compares and discusses the dierent methods presented in this thesis for designing the reservoir operating policies from the methodological point of view and for the signi cance of their results for the Red River Basin management Indications for future research are nally presented 6.1 Comparison of the reservoir optimization methods Table 6.1 shows the performances of dierent policies in terms of design indicators over the evaluation period 1995-2004 The rst row of the Table presents the system performances if the Hoabinh reservoir had not been built (`Nature') Under this assumption the hydropower production is zero and the irrigation supply de cit and the ooding costs are the ones that would have been produced by the natural hydrological regime The second row displays the performances of the historical Hoabinh operation (`History') The remaining rows show the performances of the best policies designed with the dierent optimization methods considered in the thesis The eects are measured in terms of the design indicators speci ed in Section 1.5 A 2D representation of the value in Table 6.1 is given in Figure 6.1 Policy ddp-25, designed by Deterministic Dynamic Programming (Chap- ter 3), cannot be applied in practice because it assumes a perfect knowledge of the future ows, but its performances set a goal to pursue, provide a benchmark for the other methods and help understanding the structural limits of the Hoabinh water system It shows that while the de cit in the water supply can be brought to zero, oods in Hanoi cannot be completely avoided even with perfect ow forecasting capacity Finally, ddp-25 performances are very close to the Utopia point (i.e minimum possible performances by ddp for all objectives simultaneously), mean- ing that if the future ows were known in advance, the con ict between the three objectives could be almost resolved Figure 6.1 also shows that there is a big gap in performances between the historical and the ideal (ddp-25) operation, i.e there exists a large room for improvement of 115 Comparison and discussion Table 6.1: Performances of dierent operating policies (and Utopia point of ddp) in terms of design indicators over the evaluation pe- riod 1995-2004 hyd GWh Nature def (m /s) o cm 7806 1379 History -26.3 887 902 Utopia-ddp -32.1 75 ddp-25 sdp-ar0-25 sdp-ar1-21 iso-c+ emo-6 emo-exo-6 -31.4 0.3 75 -29.7 102 869 -28.5 73 422 -31.2 596 1442 -26.7 637 382 -27.7 346 Figure 6.1: Performances (in terms of design indicators) of several operating policies designed by dierent optimization methods The diameter of the circles is proportional to the hydropower production 116 6.1 Comparison of the reservoir optimization methods Hoabinh operation This is a motivation for the application of other optimization methods to obtain policies that can be implemented in practice Policies sdp-ar0-25 and sdp-ar1-21, designed by Stochastic Dynamic Programming with an AR(0) and AR(1) model of the upstream catchments respectively (Chapter 4), are both Pareto dominant over the his- sdp-ar1-21 are signi cantly betsdp-ar0 in terms of water de cit and ood torical operation The performances of ter than the ones provided by cost This result reveals the importance of developing a su ciently complex model of the catchment, especially for ood control purpose Policy iso-c+, designed by Implicit Stochastic Optimization (Chap- ter 5), instead, is not Pareto dominant over the historical operation In general, no Pareto dominant policy was found by this method Nonetheless, its application provided a useful knowledge base for the subsequent application of Evolutionary Multi-Objective Optimization (emo) Policies emo-6 and emo-exo-6, designed by emo using dierent in- puts to the decision rule, namely hydrological variables only or both hydrological and meteorological variables, are both Pareto dominant However, emo-6 is dominated by emo-exo-6, con rming the additional value provided by exogenous information In general, the results of emo are comparable to those of sdp For both methods, using additional information or increasing the model complexity signi cantly improve the performances of reservoir optimization However, while in sdp the use of more and more sophisticated models for the upstream catchments is subject to the limits imposed by the so called curses of modelling and of complexity [Bertsekas, 1976] (all the input variables of the decision rule must be regarded as state variables, i.e modeled by state transition functions, and the computing time increases exponentially with the number of state variables), in the emo approach any exogenous information can be potentially included in the decision rule at low modeling and computing costs (no further model identi cation is needed because historical time series are directly used in the optimization; computing time stays almost the same) As for the computing eort required by the dierent methods, the number of evaluations of the system transition function (2.6) can be used (while the computing time is not a good indicator since the dierent methods were implemented in dierent numerical languages and run on dierent computers) Under sdp, the number of function evaluations is 117 Comparison and discussion proportional to the number of state, decision and random input values in the discretization grid, multiplied by the number the cyclostationary period, the number equation and, nally, the number i.e in the case of Nw NI T of time instants in of iterations of the Bellman of weight combinations examined, sdp-ar(0): SDP −AR(0) Nf = Ns × Nu × Ny × NηY B × NηV Q × T × NI × Nw = 68 × 41 × × × × 365 × × 28 = 4, 615, 924, 320 and in the case of SDP −AR(1) Nf sdp-ar(1): = Ns × Ny × Nu × Nξ × NηY B × NηV Q × T × NI × Nw = 68 × 21 × 41 × × × × 365 × × 28 = 96, 934, 410, 720 As for emo, length h the number of function evaluations is the product of the of the simulation horizon and the population size and number of generations, i.e EM Nf O = h × P op × Gen = 7305 × 250 × 150 = 273, 937, 500 to get the results in Table 5.5 6.2 Signi cance of the results for the Hoabinh management In order to better understand the signi cance of the optimization results presented in this thesis for the management of the Hoabinh water system, the policies considered in the previous section will now be evaluated in terms of other, more physically sound, indicators They are: the average annual hydropower production; the average increase in the annual hydropower production with respect to history; the maximum level in Hanoi; the average annual ooding (sum of the ooding threshold exceedences); the average number of ooding days; the average annual water de cit; and average annual number of de cit days These physical indicators are more direct and intuitive than the design indicators used as objective functions in the optimization, which include non-physical parameters to account for the relative importance of dierent time periods and Stakeholders aversion to risk (see Section 1.5) Table 6.2 and Figure 6.2 report the value of the above indicators under dierent operating policies over the evaluation horizon 1995-2004 From 118 6.2 Signi cance of the results for the Hoabinh management Figure 6.2: Performances of several operating policies designed by different optimization methods in terms of physical indicators The diameter of the circles is proportional to the hydropower production (evaluation horizon 1995-2004) the Table, it emerges that policies sdp-ar1-21, emo-6, and produce nearly the same annual energy production: emo-exo- about TWh, against a historical gure of 7.82 TWh, corresponding to a saving of about million usd per year, given that the import price of electricity from China is about 0.051 usd/kWh As for ood mitigation, all policies can reduce the average number emo-6 emo-6 on of ooding days and all but detail about the eect of As for water supply, emo-6 can reduce the peak level (for more ood in Hanoi see Section 5.2.4) is similar to better than historical performance sdp-ar1-21 emo-exo-6 and de nitely is even better in meeting the water demand, with a total water de cit of only Mm /y, and, on average, days of de cit per year The results con rm the bene ts that can be provided by the application of multi-objective optimization to the Hoabinh management Table 6.2: Performances of dierent operating policies (and Utopia point of ddp) in terms of physical indicators over the evaluation period 1995-2004 Nature History Utopia-ddp ddp-25 sdp-ar1-21 emo-6 emo-exo-6 Hydropower Annual Increase production wrt history (TWh/y) (%) 7.82 9.20 18 9.00 16 8.00 7.99 8.02 Peak level (m) 13.79 12.22 10.39 10.39 11.86 12.41 11.90 Flooding Annual exceedance cm/y 1708 1503 247 254 1004 907 771 Flood days day/y 15.2 16.0 5.3 8.1 14.4 13.7 12.1 Water supply Annual De cit de cit days Mm3 /y day/y 455 28.0 66 6.0 0.0 3.8 34 5.0 33 2.7 1.6 119 Comparison and discussion 6.3 Further research All the results presented in this thesis rely on several simplifying assumptions and subjective de nitions that would deserve further discussion First, the analysis focuses on three main objectives of hydropower production, water supply and ood control, while neglecting other important issues like ecosystem conservation (e.g de nition of minimum environmental ow) and river bed erosion For the selected objectives, the de nition of the objective functions requires some subjective setting like the choice of parameters re ecting the stakeholder values and risk aversion These choices should be validated with the direct involvement of the relevant stakeholders Another topic for further research is improvement of system model, especially the ow routing model used to estimate ow at Sontay, which is currently not very accurate Moreover, robustness of optimization results presented in this thesis to modelling assumptions should be checked by uncertainty analysis Finally, since in the Red River Basin there are several other reservoirs and some new more under construction, optimizing the Hoabinh operation may constitute a preliminary exercise for future research on the coordinate management of this multi-reservoir network If future research will enlarge the system domain to cover the whole inter-reservoir system and consider a larger number of objectives, the emo approach appears to be the most promising because of its relative simplicity of application to problems with many objectives and high dimensional system 120 Bibliography R G Allen, L S Pereira, D Raes, and M Smith Crop evapotran- spiration - guidelines for computing crop water requirements FAO Irrigation and drainage paper 56 1, FAO - Food and Agriculture Organization of the United Nations, Rome, Italy, 1998 A.R Barron Universal approximation bounds for superpositions of a sigmoidal function IEEE Transactions on Information Theory, 39 (33):930945, 1993 R.E Bellman Dynamic Programming Princeton University Press, Princeton, 1957 D.P Bertsekas Dynamic Programming and Stochastic Control Aca- demic Press, New York, 1976 A Castelletti, F Pianosi, and R Soncini-Sessa Water reservoir control under economic, social and environmental constraints Automatica, 44 (6):15951607, 2008 I.A.T de Kort and M.J Booij Decision making under uncertainty in a decision support system for the red river Software, 22(2):128136, 2007 Environmental Modelling & K Deb, A Pratap, S Agarwal, and T Meyarivan tist multiobjective genetic algorithm: Nsga-ii A fast and eli- IEEE Transactions on Evolutionary Computation, 6(2):182197, 2002 A.J Draper and J.R Lund value Optimal hedging and carryover storage Journal of Water Resources Planning and Management, 130 (1), 2004 Dynamic Programming for Optimal Water Resources Systems Analysis, chapter Dynamic programming and water resources: A.O Esogbue Origins and interconnections Prentice-Hall, 1989 FAOSTAT, 2003 URL http://apps.fao.org 121 Bibliography D.M Fults and L.F Hancock Optimal operations models for ShastaTrinity system Journal of the Hydraulic Division ASCE, 98:1497 1514, 1972 Dealing with complexity and dimensionality in water resources management PhD thesis, Politecnico di Milano, 2010 S Galelli B.A George, H.M Malano, and N.V Chien System wide water man- International Commission on Irrigation and Drainage (ICID) 2nd Asian Regional Conference, Melbourne, March 2003 agement in the dan hoai irrigation scheme, vietnam In K.C Gilbert and R.M Shane TVA hydroscheduling model: theoretical aspects Journal of Water Research Planning and Management - ASCE, 108(1):2136, 1982 W.A Hall and N Buras The dynamic programming approach to water resources development Journal of Geophysical Research, 66(2):510 520, 1961 W.A Hall, W.S Butcher, and A Esogbue Optimization of the operation of a multi-purpose reservoir by dynamic programming Resources Research, 4(3):471477, 1968 Water K Hansson and L Ekenberg Flood mitigation strategies for the red river In International Conference on Environmental Engineering, An International Perspective on Environmental Engineering, Canada, delta July 2002 D.N Harris Water management in public irrigation schemes in vietnam Impact Assessment Series Report 43, Australian Centre for International Agricultural Research, Canberra, Australia, 2006 J R Stedinger Hashimoto, T and D P Loucks Reliability, resilience, and vulnerability criteria for water resource system performance evaluation Water Resources Research, 18(1):1420, 1982 M Heidari, V.T Chow, P.V Kokotovic, and D Meredith Discrete differential dynamic programming approach to water resources systems optimisation Water Resources Research, 7(2):273282, 1971 V.H Hoang, R Shaw, and M Kobayashi Flood risk management for the rua of hanoi 2007 122 Disaster Prevention and Management, 16(2):245258, Bibliography V.H Hoang, R Shaw, and M Kobayashi Flood risk management for the riverside urban areas of hanoi: The need for synergy in urban development and risk management policies Management, 19(1):103118, 2010 Disaster Prevention and E.R Hooper, A.P Georgakakos, and D.P Lettenmaier Optimal stochas- Journal of Water Research Planning and Management - ASCE, 117(5):556587, 1991 tic operation of Salt River Project, Arizona T Kim, J.H Heo, D.H Bae, and J.H Kim Single-reservoir operating rules for a year using multiobjective genetic algorithm Hydroinformatics, 10(2):163179, 2008 Journal of E.C Kipkorir, D Raes, and J Labadie Optimal allocation of short-term irrigation supply Irrigation and Drainage Systems, 15:247267, 2001 10.1023/A:1012731718882 R Krzysztofowicz and E.V Jagannathan Stochastic reservoir control with multiattribute utility criterion In T.E Unny and E.A McBean, editors, Decision Making for hydrosystems: forecasting and operation Water Res Publications, Littleton, CO, 1981 S N Kulshreshtha and K K Klein Agricultural drought impact evaluation model: A systems approach Agricultural System, 30(1):896, 1989 V Kurkova and M Sanguineti Error estimates for approximate optimization by the Extended Ritz method tion, 15(2):461487, 2005 J.W Labadie SIAM Journal on Optimiza- Optimal operation of multireservoir systems: the-art review state-of- Water Resources Planning and Management, 130(2): 93111, 2004 L Le, T Tran, and H.Phan Project of integrated water resource management in the red-thaibinh river basin report on hydrology Hydrological report 1, Institute of Water Resources Planning, Hanoi, 2007 H.A Loaiciga and M.A Mari no Risk analysis for reservior operation Water Resources Research, 22(4):483488, 1986 L Ngo, H Madsen, and D Rosbjerg Simulation and optimization modelling approach for operation of hoabinh reservoir in vietnam of Hydrology, 2007(336):269281, 2007 Journal 123 Bibliography L Ngo, H Madsen, D Rosbjerg, and Claus B Pedersen Implementation and comparison of reservoir operation strategies for the hoabinh reservoir, vietnam using the mike 11 model Water Resour Manage, 22(1007):457472, 2008 D Nguyen, K Nguyen, S Nguyen, H.Nguyen, and P HungTien Integrated water resource management in the red river basin - problems and cooperation opportunity Technical report 1, International Conference on Adaptive and Integrated Water Management, Germany, 2007a H Nguyen Improving water security for the future through iwrm and better water governance in the red - thai binh river basin TAC background paper 1, Institute of Water Resources Planning, Hanoi, 2010a M Nguyen Tree-based input selection for hydrological modelling Master's thesis, Politecnico di Milano, 2010b T Nguyen, D Nguyen, H Nguyen, and H Le Project of integrated water resource management in the red-thaibinh river basin report on hydrology Irrigation report 1, Institute of Water Resources Planning, Hanoi, 2007b T.C Nguyen, N.H Do, T.H Nguyen, and K Egashira Agricultural development in the red river delta, vietnam - water management, land use, and rice production Journal of the Faculty of Agriculture Kyushu University, 46(2):445464, 2002 R Oliveira and D.P Loucks Operating rules for multireservoir systems Water Resources Resource, 33(4):839852, 1997 C Piccardi In nite-horizon minimax control with pointwise cost function Journal of Optimization Theory and Applications, 78:317336, 1993a C Piccardi In nite-horizon periodic minimax control problem of Optimization Theory and Applications, 79:397404, 1993b Journal C Piccardi and R Soncini-Sessa Stochastic dynamic programming for reservoir optimal control: dense discretization and in ow correlation assumption made possible by parallel computing Research, 27(5):729741, 1991 Water Resources Dynamic Programming for Optimal Water Resources Systems Analysis, chapter A dual approach to stochastic dynamic programming E.G Read for reservoir release scheduling Prentice-Hall, 1989 124 Bibliography P.P Rogers and M.B Fiering Use of systems analysis in water management Water Resources Research, 22(9):146S158S, 1986 Proceedings of C.E Shannon Communication in the presence of noise the Institute of Radio Engineers, 37:1021, 1949 R Soncini-Sessa, A Castelletti, and E Weber ipatory water resources management Theory Integrated and particElsevier, Amsterdam, 2007a Integrated and R Soncini-Sessa, F Cellina, F Pianosi, and E Weber participatory water resources management Practice Elsevier, Ams- terdam, 2007b N Srinivas and K Deb Multiobjective dominated sorting in genetic algorithms optimization using non- Evolutionary Computation, 2(3):221248, 1994 J.A Tejada-Guibert, S.A Johnson, and J.R Stedinger The value of hydrologic information in stochastic dynamic programming models of a multireservoir system Water Resources Research, 31(10):25712579, 1995 P.K Toan, N.M Bao, and N.H Dieu Energy supply, demand, and policy in vietnam, with future projections Energy Policy, In Press, 2010 doi: 10.1016/j.enpol.2010.03.021 W.J Trott and W Yeh Optimization of multiple reservoir systems VA Truong and N Vu Assessing in ow to hoabinh reservoir under Journal of the Hydraulic Division ASCE, 99:18651884, 1973 climate change Master's thesis, Politecnico di Milano, 2010 A Turgeon Optimal operation of multi-reservoir power systems with stochastic in ows Water Resources Research, 16(2):275283, 1980 H Turral and N.V Chien Development and speci cation of a service agreement and operational rules for la khe irrigation system, dong, vietnam Irrigation and Drainage, 51(2):112, 2002 H.V Vasiliadis and M Karamouz Demand-driven operation of reservoirs Journal of Water Research Planning and Management - ASCE, 120(1):101114, 1994 using uncertainty-based optimal operating policies S Vorogushyn, B Merz, K.E Lindenschmitdt, and H Apel A new methodology for ood hazard assessment considering dike breaches Water Resources Research, 46(10):117, 2010 125 Bibliography S Wu, B Bates, Zbigniew W Kundzewicz, and J Palutikof Climate change and water Technical report 1, Met O ce Hadley Centre, United Kingdom, 2008 S Yakowitz Dynamic programming applications in water resources Water Resources Research, 18(4):673696, 1982 W Yeh Reservoir management and operations models: a state of the art review Water Resources Research, 21(12):17971818, 1985 E Zitzler and L Thiele Multiobjective evolutionary algorithms: a comparative case study and the strength pareto approach IEEE Trans- actions on Evolutionary Computation, 3(4):257271, 1999 R Zoppoli, M Sanguineti, and T Parisini Approximating networks and extended ritz method for the solution of functional optimization problems 126 J Optim Theory Appl., 112(2):403440, 2002 ... e ciency The in ow to the Hoabinh reservoir The storage of the Hoabinh reservoir The surface of the Hoabinh reservoir The unitary evaporation of the Hoabinh reservoir The evaporation of the Hoabinh reservoir. .. through the spillways of the Hoabinh reservoir The water level in the Hoabinh reservoir The water level downstream of the Hoabinh reservoir The hydraulic head of the Hoabinh reservoir Turbine e ciency... Table 2.1: The main variables of the model Meaning The release from the Hoabinh reservoir The release through the turbines of the Hoabinh power plant The maximum turbine capacity of the Hoabinh power
Ngày đăng: 24/03/2014, 20:22
Xem thêm: ASSESSING AND OPTIMIZING THE OPERATION OF THE HOABINH RESERVOIR IN VIETNAM BY MULTI-OBJECTIVE OPTIMAL CONTROL TECHNIQUES doc, ASSESSING AND OPTIMIZING THE OPERATION OF THE HOABINH RESERVOIR IN VIETNAM BY MULTI-OBJECTIVE OPTIMAL CONTROL TECHNIQUES doc
