Research And Application On Key Issues Of Controlling Reservoir Flood Limited Water Level

With the social economy rapid development and the population growth in the 21st century, China's flood control and water supply task are more and more hard. Adhering to flood control and drought relief, scientific control of reservoir flood limited water level is an effective non-engineering measure to improve flood resource utilization under flood control security during flood season. The reservoir flood routing is the foundation of design and control of flood limited water level, and the latter is divided into static and dynamic control. Aimming at key issues of the above-mentioned three aspects, this article launches the research in the following five parts:(1) Reservoir flood routing includes two cases of plain and mountain reservoirs. Most numerical solutions of plain reservoir flood routing ignore the local truncation error and the local stability. Runge-Kutta-Fehlberg (RKF) method appraising the local truncation error with relative error between four steps and five step precision numerical solution, a Runge-Kutta-Fehlberg-Reverse (RKFR) method is advanced to rout plain reservoir floods, in which the step length will further be rectified following the RKF method using a local stability parameter, which is based on the fact: as the local truncation error is little enough, the local stability might approximately be characterized by the relative error between the initial dependent variable associated with every step and the feedback which may be calculated through a Runge-Kutta method. The presented RKFR method is verified with two numerical examples and a reservoir flood routing program, and the results demonstrate that it is more accurate and stable than the RKF.(2) The dynamic capacity data collection difficult associated with floods, especially catastrophic floods, through spillways of a mountain reservoir, most existing methods is unfit for routing large floods through a mountain reservoir. A two-stage model is presented to rout a flood through spillways of a mountain reservoir. Based on an introduction of a physical concept: transmitting rate T [m~3/s] of dynamic volume into static volume, the diffusion from the inflow at upstream end to T can be estimated through Muskingum formulation since the impact from the water level before the dam on dynamic volume is far less than from the inflow rate, and the process from T to reservoir outflow may be approximated into a surge wave through a plain reservoir. Subsequently, the trial algorithm is provided to solve the two parameters of Muskingum formulation of the first stage, and the recurrence of halving algorithm is offered to dispose the case of a discontinuous outflow hydrograph derived from multi-prevention standards and multi-discharge equipments of a reservoir. While the two parameters are finalized, the model can be applied to rout floods, involving great floods, through a mountain reservoir. The instance results demonstrate the proposed model is more accurate.(3) The static control of flood control level includes two types of based on flood season staging and flood forecast and regulation. The focus of the former is to determine the staging of flood season and flood control standard of the staging. Now, in most regions of China, climate data is still relatively scanty. Therefore, this paper presents a new method of flood season staging, which carried out a cluster analysis using the rainfall data of catchment areas controlled by reservoir and performs the revision of staging results by flood data. The cluster analysis to rainfall data can usually be simplified a multi-dimensional time series clustering problem, which requires the clustering method capable of reasonably classifying for a multi-dimensional and time series sample, and of finalizing the appropriate number of cluster. However, the existing methods don't meet the above mentioned requirements simultaneously, so a new clustering principle that is a new category should be formed by two contiguous categories with minimum relative distance and a new clustering validity criteria that is the sum or average of distance between similar samples should be smaller, at the same time the average distance between two contiguous categories on the timeline should be bigger when the classification is appropriate, have been advanced in this paper. Coupling the new clustering principle and validity criteria, a new time series relative hierarchical clustering method is established to deal with the multi-dimensional time series clustering problem of flood season staging. The case studies show that the result of flood season staging is reasonable. In addition, a new method of determining the flood control standard of staging is also proposed based on the statistical basic principles, which is keeping the same design value with no-staging according to the standard requirement in main flood season, and in other flood season, the flood control standard of staging adopts the value from the flood control standard of year divided by the numbers of staging. According to the new method, a conclusion can be gotten that is not the more staging numbers, the better, which is consistent with what the standard requires.(4) The key of static control of reservoir flood limited water level, based on flood-forecasted control, is designing flood limited water level. Coupling specific flood forecasting foreseeable periods of different watersheds with the output of traditional design method, an object-oriented constrained forward shift method is advanced to design flood limited water level, which overcomes many existing methods' puzzles, from introducing flood forecasting characteristic elements into the indicators of differentiating flood order, such as choice of forecast elements, simulation of elements processes, risk analysis of processes errors. The instance results demonstrate economic benefits are marked. The proposed method is practical to the reservoir in the planning or design stage, and especially to the one in operational phase, which need to check flood limited water level due to the change of control mode.(5) Rainfall forecast information, especially medium-term rainfall forecast information, is crucial for static control of reservoir flood limited water level to ensure safety of flood control and to increase benefits of water supply. However, 24h short-term rainfall forecast is mainly concerned by most static control methods. With introducing three parameters, the medium-term numerical precipitation predication information is pretreated, and then for evaluating the mid-range forecast precision, the precision indicators is defined and the evaluating means is given. Around defining the real-time relatively-safe period of dynamic control, the real-time relatively-safe period method, with the above-mentioned three parameters, is proposed, on which the process of risk analysis is just the one of parametric optimization. Therefore, for optimizing the parameters, a model is constructed, adopting a strategy of controlling risk within the decision-makers affordability followed by pursuing risk-benefit biggest, with defining the indicators of dynamic control risk and risk-benefit and the indicators of capability of decision-makers enduring risk. The instance results demonstrate the real-time relatively-safe period method improves the economic efficiency with increasing the rainfall foreseeable period, under flood control security.At last a summary for the paper and problems to be further studied are discussed.