1 DAMS AND FLOOD CONTROL Carlos E. M. Tucci Institute of Hydraulic Research University Federal of Rio Grande do Sul Porto Alegre RS Brazil Abstract: Floods are a natural process in which the river spills over its banks. Hazards occur when the population occupy these areas during a sequence of small floods. In South America flood impacts have occurred mainly due to the lack of flood plain management. Flood hazards in some communities on the banks of major rivers have had major social and economic impacts which can be seen many years after the event. There is an explicit conflict between dams for energy production and their use to for flood control. In the Brazilian electricity sector, these dams have been operated since the end of the 1970s, with a waiting volume that aims not only at improving their safety but also at reducing the impact of floods. In this article, the potential impacts of dams are stressed, with operational features for flood control and the description of a conflict generated between a hydropower dam and the community of two cities. FLOODS Floods may occur due to the natural conditions or those generated by land use, such as urbanization or hydraulic structures. In this article we will discuss the natural floods and those related to dams for flood control or other Water Resources uses. The main flood in the regions occur on the banks of the major rivers of South America such as the Prata/Paraná, Amazon, São Francisco and Tocantins. In Figure 1 the basin of the Prata river can be seen with the main reaches where flooding occurs and there are dams (hydropower dams). In the Paraguay river basin and in the lower reach of the Paraná river, floods are the result of prolonged rainy periods which affect large areas and produce flood levels for several months. In the reach involving Argentina, the areas are flat and usually flooded for several months. In the upper part of the Paraná river and in the Uruguay river floods are of short duration lasting only a few days. Flooding is a natural process in which the river floods its first bottom depending on the intensity and duration of floods and initial humidity conditions. When changes occur in the river, such as dam construction, flooding levels may change as a result of operational conditions in the dam. The main impacts on the population occur due to lack of: (i) knowledge regarding the occurrence of flood levels, and (ii) planning for space occupancy according to the risks of flood events. Levels can be forecasted for the short term (real time forecasting) or for the long term that is the statistical forecast or risk of occurrence of a given flood. The first type allows flood warning a few hours or even days in advance, reducing the resulting damages. The second type allows planning the occupancy of spaces at risk, or building protection works.
2 Control measures are non-structural and structural. The non-structural control measures involve the population living near the river by means of warnings and the zoning of areas where floods could occur and the structural measures involve the modification of the river by constructing protection works such as dykes, dams, canals and others. The costs of the former type are generally much lower than those of the structural measures. No systematic flood control management program is observed in the region, only single actions to build protection works, while very few non-structural measures are taken. In some river reaches flood alerts are given by simple information regarding levels or short-term forecasts, as in the lower reach of the Paraná river, the Pantanal reach of the Paraguay river and in Iguaçu river. In the city of São Paulo there is a warning forecast system based on meteorological radar.
3 The common scenario of impact is due to the fact that the population occupies the flood plain during a sequence of years with low maximum annual levels, since the flat areas are favourable to settlement. When years with higher floods return, damages are significant and the population demands that governments take action to build control structures such as dams and others. Below are examples of this scenario: In figure 2 the floods levels of Iguaçu river at União da Vitória can be observed. For a long time, floods remained below the 5-year return period. The floods after 1982 produced significant damages to the community (table 1). Table 1 Flood losses at União da Vitória and Porto União(JICA, 1995) Year Losses US$ millions Figure 2 Maximum flood levels in Iguaçu river at União da Vitória (a basin of approximately 25,000 km2), Tucci and Villanueva, 1997) For the Itajaí river there are records which began in 1852, with all the levels above the low-flow bed of the river. In figure 3 the variability of these levels can be observed, and also that between 1911 and 1983 no levels were higher than m (15-year risk, approximately), whereas before and after this period there were several levels of up to 17.19m. Both in this case and in the previous one, the continuous series of records that began in 1930 presented a sample bias for risk assessment. In this case the losses were also significant; in 1983 (approximately 30-year risk) they represented 8% of the GDP of the State of Santa Catarina at the time. One of the largest wetlands of the world, called Pantanal, is located in the Upper Paraguay river. In this region the environment and the population have always lived together in peace. In figure 4 the maximum flood levels can
4 be observed at Ladário since the beginning of the century. Table 2 shows the values of the maximum mean level of flooding and of flooded áreas of the Pantanal, at three different periods. The great difference between the 1960s and the other periods can be observed. During this period the flood valleys were occupied. This population was dislodged during the following decades and, due to the changes in the bed caused by the variability of the Pantanal flows, they had to abandon their properties and are living in poverty on the outskirts of towns and cities in the region. flood level, m flood levels lower flood level year Figure 3 Flood Levels in Blumenau, S. Catarina, Brazil Table 2 Estimated values for levels and flooded áreas in the Pantanal (approximate values) Period Mean Maximum Level m Mean flooded area in the Pantanal * Km , , ,49 50 * approximate values obtained from Hamilton ( 199x )
5 7 Annuall Flood levels, m Years Figure 4 Annual flood water levels at Ladario in the Paraguay River and the mean of the periods: (i) ;(ii) ; n(iii) DAM CHARACTERISTICS Dams in South América were usually designed for one of the following purposes: power production, water supply, irrigation and navigation. Very few enterprises aim exclusively at flood control in the region. The Itajaí-Açu river basin in Santa Catarina, in Brazil, is an example of mediumsized basin ( approximately 12,000 km2) with three dams built solely in order to control floods. Medium-sized and large dam construction in South América has always been associated with hydroelectric power generation. Hydropower developments usually have a significant volume and are located downstream from large basins (>2,000 km2). Since the energy of a development depends on the discharge and head, the larger the streamflow regulated by the reservoir and the head generated by the dam, the more power will be generated. Thus, the volume and basin should possess characteristics that will render the enterprise economically feasible. The small dams have generally been built for irrigation or water supply, but they are for small volumes and in basins with an area of less than 1,000 km2. These dams will not interfere very effectively in flood control. Navigation dams only maintain the water levels and also have an insignificant volume for flood control. Thus, this article will analyze flood characteristics, flood control by means of dams, hydropower development characteristics, their potential impacts, the multiple use of a dam for flood control and power production and several examples. DAMS AND FLOODS Conditions of dams as to floods The dams designed should take into account the impacts they may produce downstream and upstream from the development. Downstream: Generally, downstream from a dam are areas subject to flooding. When the dam is built, the tendency is for the reservoir to dampen the floods, in these
6 riparian areas, if no operational problems occur in the dam. However, if the downstream area is not occupied, people will move in because of the proximity to the development, and it then becomes subject to floods. If the development does not dampen the floods, the tendency is that they be required by society to reduce the downstream impacts. Thus, the downstream constraint becomes the maximum streamflow Q1, from which point onwards the river floods its banks. During the flooding periods there will be events in which the dam will not be able to dampen the streamflow and floods will occur The public perception of this situation is generally to blame the dam for what has happened. Therefore it is necessary for the development to have an effective operational system and reliable observation of the hydrological data needed to show the operational conditions and justify their actions Upstream: The construction of a reservoir could produce the following impacts upstream: (i) (ii) Depending on the inflow, the operational rule and runoff capacity, the backwater curve could flood or provoke damming upstream; the conditions of the previous items could change over time due to sedimentation in the reservoir, which initially occurs in the reach lying farthest upstream. For this reason, the previously designed flood levels may rise, reaching areas outside the boundaries of expropriated properties. Dam failure: the possibility of dam failure should be considered with the area that may be affected and runoff velocity as a base for a civil defense plan downstream from the development. Considering the occurrences, the main causes are presented in table 3. In the same table the proportion can be seen according to type of dam, in which the earth dams are outstanding because there are more dams of this type (Serafim, 1981). The risk of dam failure is generally on the order of 10-4 (based on the probability of spillway sizing). Thus, during its service life of 100, a dam would have a 1% chance of failing. During the last few decades, with increased hydrological information, improvement of building methods, forecasting and flood warning, this risk is expected to diminish (Serafim 1981). Marengo(1996) shows that depending on the period when it was constructed and on age the risk of failure diminishes, converging to values of According to Lave et al. (1990) and Serafim (1981) the use of a warning system with an advance alert of at least two hours may reduce the number of deaths to zero.paté-cornell and Tagaras (1986) mention that a monitoring system could reduce the chances of dam failure by 12%, the number of deaths by 44% and the losses by 12%. Brown and Graham (1988) identified two types of floods: (i) time longer than evacuation, estimated as 90 minutes; (ii) time less than 90 minutes. In the former case, the loss of lives is estimated as 0.04% of the population threatened and in the latter 13%. It should be considered that the highest risks occur at the flood passage crosssection, where velocities are higher. Few countries have a legal system that obliges developments to include prevention for this type of event. In Argentina this process began with the privatisation of the electricity system. In Brazil, as yet, there are no regulations to prevent the impacts resulting from this situation. In France a 1968 decree requires an emergency plan for all dams over 20 meters high, or reservoirs with a capacity of more than 15 hm3 (Benoist, 1989). In the United States there are dams under federal legislation and others under state legislation. Only dams managed by federal agencies or private dams
7 that participate in hydropower projects are under federal law. During the nineties, the federal government developed a dam safety standard to unify the laws of the different states. Table 3 Dam Break characteristics Causes Underdesigned spillway 35 Foundations, infiltration, landslide and earthquake 25 Others: construction and design problems; quality of the 40 material, war sabotage, etc Type of dam Earth 63 Concrete arch 1 Concrete gravity 6 Rockfill 4 Others 26 Flood control dams Events % Traditional flood control developments are designed with constraints only downstream from the development. In this case the project is based on the Q l limit of outflow discharge at the outlet from the dam and on the useful volume Vu necessary to dampen the flood. The project should have overflow outlets that will allow the discharge downstream of all the inflow up to streamflow Q1, using the useful volume above this limit. The operation of this system may become even more efficient if there is a warning system that enables decisions regarding the use of the useful volume throughout the flood. As mentioned above, dams are rarely built for flood control. When they exist they are built on medium-sized basins where the dam volume may minimize flood impacts. An example of a system of dams built for flood control is that of the Itajaí-Açu river basin in Santa Catarina, Brazil (figure 5). The West dam, located on Itajaí-Oeste river, upstream from the town of Taió (finished in 1973), the South dam on Itajaí do Sul (finished in 1975) upstream from the town of Ituporanga, Ibirama dam on river Hercílio (finished at the end of the eighties). The latter did not exist during the 1983 and 1984 floods. The design of these dams uses bottom spillways with a capacity that tends to retain a large volume in the reservoirs, requiring a very long time to empty. The contribution of the first two dams to flood control in 1983 was insignificant due to the large volume of rainfall that occurred during 7 days. In the case of the 1984 flood,which lasted only 2 days, the contribution was greater. Examining the maximum discharge series before and after the dams were built, an unexpected result was observed, the increased mean and standard deviation of the dams to one of the downstream sections of one of the dams. However, the result of this increase was due to increased rainfall in the basin, precisely between the two periods. In table 4 are presented some statistics of this comparison. The West dam that did not show an increase proved effective at containing the floods, whereas the South dam apparently does not have the design volumes appropriate to provide a significant reduction in floods.
8 West Dam Hercílio Rio do Sul South Dam Itajaí Blumenau Figure 5 Itajaí River basin and Flood Control Dams Table 4 Statistics before and after dam construction in the Itajaí Basin Statistics West Dam m 3 /s South Dam m 3 /s Annual Rainfall mm Rainfall (1) mm Mean Before the Dam 292,2 488, ,1 After the Dam 274,5 513, ,7 Standard deviation Before the Dam 73,2 267,1 After the Dam 56,2 356,6 Period Before After (1) Rainfall of the month in which the annual maximum flood occurs Multiple Use: flood control and power production Considering that most of the reservoirs built in the region are for hydropower production, it is initially found that there is a potential conflict, since the hydropower developments aim to keep the water level as high as possible during the rainy period (greater hydropower generation), with a consequently smaller damping volume. Under these circumstances, the development may worsen flooding conditions, both upstream and downstream from the development. Most hydropower developments were not designed to contain floods. Since building reservoirs tends to regulate downstream discharge, the low risk floods no longer occur due to damping caused by the reservoir. Thus, a larger amount of risk areas is occupied, because of the structures and the development of flat áreas. When the higher risk floods occurred, that the reservoirs were not prepared to dampen, large floods occurred, with conflicts between the population and the developments. During the 1970s, several relevant facts occurred in the Brazilian electricity sectors which produced a change of attitude as regards hydropower dam operation. First two dams broke on Pardo river in the state of São Paulo. The rupture of the first
9 due to operational and design failure led to the break in the second one due to the rupture wave. The second fact were the floods in São Francisco river, in 1979, which produced impacts downstream and strong public pressure. For this reason, criteria were established for these systems, so that, besides producing power, they could contain part of the floods by creating a waiting volume. Figure 6 shows the stages and the definition of the waiting volume of a reservoir. This volume is kept free to receive the flood volumes and reduce streamflow downstream, trying to satisfy the upstream and downstream constraints. There are several methodologies to estimate this volume, based on the statistics of the historical series of dam flows. The methods used in the Brazilian electricity sector have been the Volume x Duration Curve Method (adaptations of the methodology presented by Beard, 1963), or the method of critical pathways (Kelman et al. 1983). The former uses the historical series observed and the latter, the series of streamflows generated by a stochastic model. Both methods determine, statistically, the waiting volume that should be maintained by the reservoir during each day of the rainy period for a given risk of analysis. Normal maximum operation level Maximum operation Level for flood control Maximum design level Net volume Dam height Waiting volume Minimum operation level Dam bottom Figure 6 Operation levels in a Hydropower These procedures do not take into account the information existing in the basin during the flood period. For basins where seasonality is not clearly defined, the model may underestimate or overestimate the waiting volume with significant losses. On the one hand losses due to flooding and, on the other, due to loss of power generated. The use of information that exists in the basin involves the real-time forecasting of inflows to the reservoirs. For a safe forecasting system the waiting volume could be reduced if the inflow can be predicted in advance, and therefore, the volume could be increased according to need, taking downstream constraints into account. Silveira (1996) showed the usefulness of this approach to predict inflows to Sobradinho on São Francisco river, based on a simple flow forecasting model from upstream flows. Moro (1998) showed the use of a group of models to predict the inflows to the Foz do Areia reservoir and, based on this forecast, to optimize system operation (example presented in the last item).
10 CASE STUDY: União da Vitória x Foz de Areia The conflict In recent years (1983 and 1992), União da Vitoria and Porto União in Brazil suffered severe floods, after a long period (50 years) of normal to low floods. The economic losses to some industries, businesses and homes created a depression and psychological impact on most of the population. The population believed that the main cause of the floods was the impact of Foz do Areia Dam, a Hydropower Plant located downstream from the city. It created a major conflict lasting more than ten years between the population and the State Power Company. Cities and flood conditions The cities of União da Vitória and Porto União (figure 7) are located on the banks of the Iguaçu River, State of Parana, Brazil, where, for about 50 years (1930 to 1982, see figure 2), only low or medium floods occurred. During the early 80 s the Foz do Areia Dam was built, about 100 km downstream. The backwater curve for the dam design flood may reach the cities, depending on its operation water level. While designing the dam, two alternatives were studied to cope with this influence, protection levees or operational rules to prevent the rise of flood levels in the cities. The first operational level chosen was 744 m. Studies carried out by Parana State Power Company (owner and operator of the dam) in 1982 and 1985, showed that 744 m was a high level, and recommended 742 m at first, and later m. In 1983 an extreme flood occurred, which caused severe economic losses. The flood level was the highest in 107 years (estimated return period of 170 years, and a duration of 62 days), and the estimated losses were U$S 78,1 millions. At that time the population blamed the Power Co. for the Dam operation and high flood levels. The flood recurrence was estimated as being about 1000 years, using continuous records ( ). This calculation, however, did not take historical marks (figure 2) into account, and this led to an overestimate of the return period. When historical marks (found after some research done by local personal) were included in the statistical analysis, the return period decreased to 170 years. In 1992 there was another flood, smaller than the last, but of similar magnitude and impact (return period of 30 years, duration of 65 days and U$S 54,6 millions in losses). As the people had been told that the risk was very small, and, in less than ten years, the cities were flooded again, their reaction was very strong, and distrust towards previous studies and official statements was widespread.
11 Figure 7 Iguaçu River at Foz de Areia and União da Vitória A NGO (Non Governmental Organization) called CORPRERI (Regional Permanent Commission for Flood Prevention of Iguaçu River) was created by the population to deal with the flood problem. This organization has been the representative of the cities in the discussion with the Parana State Eletric Power Company (COPEL). Tucci and Vilanueva (1997 ) presented the alternative study described here. Simulation of Iguaçu River with Dam operation In order to answer most of the questions asked by the population, a hydrodynamic model was applied (Tucci, 1978), taking into account the characteristics of the flood valley and main channel in the reach between the cities and the dam. Usually the flood plain is represented only by a storage function assuming an infinite roughness. Since depths in the flood plains can reach more than 5 m, there are flow dynamic effects in this part of the river system. The model used to describe the dynamic effects was the Lateral Distribution Method described by Wark et al (1991) and Villanueva, (1997) to compute the conveyance and velocity distribution coefficient for each section. The boundary conditions used for the simulations (fitting and scenarios analysis) were the discharges in the cities (upstream boundary) and the level at the dam (downstream). Model fitting was performed comparing simulated and recorded levels at União da Vitória (section M8, figure 7) and discharge and level at Porto Vitória (figure 6) for the 1983 and 1992 events. The discharges in those floods varied between 240 m 3 s -1 to 5,000 m 3.s -1 (figure 8).
12 Flood impacts Figure 8 Simulation of the floods at União da Vitória The first question to be answer was the influence of the dam operation in the cities floods and the causes of the high levels. Some other questions were raised by the population related to other solutions for flood alleviation such as improvement of channel conveyance and bridges impacts. Using the two major floods, 1983 and 1992, two different downstream boundary conditions were used in the model: (i) the actual operational levels at the time of the flood; (ii) safe operational level at the dam, established at m by previous studies. The results showed that the operational procedure in 1983 increased the flood levels at the cities by 14 cm during the first part of the flood but did not show any influence during the flood peak. During the 1992 flood the operational levels at the Dam did not influence the flood levels at the cities. Flood Control Measures Non-structural measures: The non-structural measures developed were: (i) Flood zoning and (ii) flood forecasting. Flood zoning: In order to define the flood levels for the land use map, levels were calculated for each of the available cross sections, for several return periods, using the mathematical model. A 1:2,000 map was used. Flood level lines were established for 5, 10, 50 and 100 year return periods (considering historical marks). Figure 9 shows a view of the flood map. Three zones were established for the flood control areas: (i) Preservation area: below , reserved for environmental protection and parks. This area was purchased by the Power Company that owned the dam. A park was constructed in the
13 urban part of the land after an agreement between the communities and the Company; (ii) Water resources protection area: between the former area and the 10-year levels. Remove public building such as school and hospital from this area; progressive taxes to be imposed for any use but the recommended ones: parks, sport fields, agriculture, and others. A tax deduction was recommended for these uses; (iii) Low density area: less restrictive than the former, but needing special care and protection against flood damage in the buildings. Figure 9 União da Vitória Flood Map Flood forecasting: The flood zoning must also have a real time flood forecasting system, working together with the Civil Defense Authority. The proposed actions were the following: (i) definition of an alert system: mathematical model, forecasting range and alert steps; (ii) County Civil Defense Authority must be created, so as to be prepared to act, with well pre-established plans, during the floods; (iii) emergency plans must be prepared for the different parts of the city. Three forecasting conditions have been recommended: (i) watch condition: from that level on, the behavior of the river must be accompanied carefully. Real time
14 forecasting begins at this condition; (ii) alert condition: when the m level is to be reached within 12 hours; (iii) emergency condition: when the m level is to be reached within 12 hours; Mine (1998) developed a model to forecast the flow and operate the dam taking into account the upstream and downstream restriction. The flow forecasting for União da Vitória ( km 2 ) was done by an empirical model, the flow from the basin between União da Vitória and Foz do Areia (5.000 km 2 ) was simulated by the rainfall runoff model IPH II (Tucci et. al. 1981) and the flow in the river reach was simulated by the hydrodynamic model used in the other simulations. In the real time flood forecasting there were the following options: (i) zero rainfall for the lead time; (ii) rainfall forecasted by empirical model; (iii) known rainfall. This flood forecasting model was integrated to a operation model based on the upstream and downstream flood restrictions. The forecasts were done with a lead time of 24 hours and updated every 4 hours. The Dam level has to be below 742,0 m due to upstream restrictions. In figure 10 are shown the results for the 1983 flood. It can be seen that the actual operation used during this event was above the limit for a few hours due to the amount of water from the tributaries and operation brought the level down very fast. Using the operational system with well-known rainfall, the operation is more efficient because it stays below the restriction and increases the dam level after the risk, improving energy production. Figure 10 Levels at Foz do Areia actual operation and with flood forecasting and operation management model (Mine, 1998) Structural measures: The structural measures studied were: (i) changing river characteristics and; (ii) levee protection along the city. Some of the potential alternatives to decrease the flood levels were to modify the
15 characteristics of the river at some critical reaches downstream from União da Vitória. These river modifications are: straightening and enlarging some bends, duplication of the channel and by-pass of the curve immediately downstream from the cities, and even deepening the channel stretch between the Porto Vitoria rapids and the cities (about 50 km long). As a general conclusion of the analysis, it can be said that, even though the critical points contribute to increasing the levels, none of them is alone responsible for the high levels. Not even their joint effect increases the floods critically. The main problem is the lack of river flow conveyance for the floods in the sections along the cities and some contractions. Other factors such, as river bends and contractions far downstream have little influence. In fact, discharges for both floods considered (1983 and 1992) were very high, and the 1983 flood was the historical maximum. The historical records, however, show that flood levels of a similar magnitude had already occurred. In a previous study, JICA (1995) considered the levee alternative as a control measure and recommended a feasibility study for this solution. Most of the structural solutions studied, relating to the river conveyance, were not feasible due to the high intervention cost. A feasibility study was recommended for this system and concluded that this was the acceptable combination of protection and cost. The levee system would be constructed protecting the areas above a 10-year return period. This engineering system will change the cities characteristics and protecting it from low risk floods. The analysis done in this study led to some questions that the population had to answer before it decided in favor of this system: a system of that kind must be under the jurisdiction of the city authorities. This implies a cost that needs to be supported by taxes. The investment maintenance costs are high which may create a major impact on the city economy. the protection will benefit mainly the areas between the 10-year flood level and the 1983 flood level. Will the whole population be willing to pay for the benefit to only part of them?; the impact of the levees on the flow conditions upstream and downstream of them must be carefully studied; not only the technical and economical aspects of the levees must be considered, landscape and urban environment issues also need to be included. Questions such as this have to be answered: Would you like to live with a 6 m wall, all along the city, which will protect levels above 10 year flood or create a new urban development toward more safe areas and use zoning measures? Until now the decision has been to use non-structural measures in order to cope with flood conditions. But for political reasons there are pressures to show some structural measure that in this case represents high costs. The analysis of the problem leads to some interesting conclusions regarding its origin and development. A long period without severe floods induced a false feeling of security in people, who began settling in the flood valley. Also as a result of this false security, no flood protection measures or planning were adopted. When the 1983 flood showed that there was an actual risk, it was disregarded, based on seemingly dependable (50 years data) statistical analysis. This analysis, however, did not take into account existing and very valuable information (the historical flood marks). When
16 another severe flood came along in 1992, the population was upset, and distrusted the technical studies. This problem was aggravated because of communications problems with the responsible institutions. It must be stated that, except for the statistical analysis, the existing technical reports were basically right in their diagnosis and conclusions. The lack of flood protection planning and preventive measures has caused losses evaluated at about U$S 150 millions, and the solution will have a similar cost, not to mention the indirect impacts in both cases. Several decades without severe floods are not an unusual situation, it is logical that floods with high return periods seldom occur. The consequences of long lags between this kind of floods are also common: occupation of the river valley and disregard for protection measures. CONCLUSIONS A long period without severe floods induced a feeling of false security in people, who began settling in the flood valley. Also as a result of this false security, no flood protection measures or planning used to be adopted. The lack of flood protection planning and preventive measures has caused losses evaluated as able to destroy the economy of the communities. Several decades without severe floods are not an unusual situation, it is logical that floods with high return periods seldom occur. The consequences of long lags between this kind of floods are also common: occupation of the river valley and disregard for protection measures. Flood control should not be performed with single solutions, but by means of a preventive program for the occupancy of higher risk spaces, the development of efficient flood warning systems with advance alerts that will allow the minimization of the space and effective performance by Civil Defense. The non-structural measures are usually most difficult to implement due to their constraint characteristics and the population is always expecting that some structure will provide Salvation! Dam planning and operation involves responsibility for the effects which may be produced upstream and downstream of the valley. The simple expropriation of the estimated flooding areas and traditional operation of power production levels do not exempt dam owners of responsibility. A preventive system for flood warning will be necessary, an organized prevention system to prevent the impact of dam breaks and to perform independent monitoring of levels upstream and downstream from the development in order to control operations and their social impacts. REFERENCES BEARD, L.R Flood Frequency Techniques. Austin: Center of Resources University of Texas, Austin BENOIST, G Les études dóndes de subermsion des grands barrages d EDF. La Houille Blanche. No.1 p BROWN, C. A; GRAHAM, W.J Assessimg the threat to life from dam failure. Watere Resources Bulletin Vol.24 N. 6 p December. JICA, The master study on utilisation of water resources in Parana State in the Federative Republic of Brazil. Sectoral Report vol H- Flood Control. LAVE, L.B. RESENDIZ-CARRILLO, D. McMICHAEL, f.c., Safety goals for highhazard dams: are dams too safe? Water Resources, V. 26 n. 7., p , July. MARENGO, H.M Análisis de riesgo de falla en presas, estatísticas y parámetros de referencia. Ingeniaria Hidráulica en México. Vol XI, N.2 p.65-77
17 MINE, M., Método determinístico para minimizar o conflito entre gerar energia e controlar cheias. Tese de Doutorado. Instituto de Pesquisas Hidráulicas. UFRGS. PATÉ-CORNELL, M.E.; TAGARAS, G Risk costs for new dams.: Economic analysis and effects of monitoring. Water Resources Research, V22 N.1 p 5-14 January. SERAFIM, J.L Safety of Dams judged from failures. Water Power and Dam Construction Sutton V.33 n.12 p December TUCCI, C.E.M Hydraulic and Water quality simulation in a river network. PhD dissertation Civil Engineer Department Colorado Statr University. Fort Collins Co. TUCCI, C.E.M; VILLANUEVA, A, Controle de enchentes das cidades de união da Vitória e Porto União. CORPRERI, 117 p. VILLANUEVA, A O N, Dynamic floodplains simulations: compound channels and wetlands. PhD thesis IPH-UFRGS (in portuguese) WARK, J. B. ; SLADE, J.E. RAMSBOTTOM, D. M., Flood Discharge Assessment by the Lateral Distribution Method. Report SR 277, Dez Hydraulics Research Wallingford