Assessment of the Possible Extent of Loss of Life (LOL) Downstream of Large Dams due to Dam Failure Flooding

Transcription

1 Assessment of the Possible Extent of Loss of Life (LOL) Downstream of Large Dams due to Dam Failure Flooding Saqib Ehsan 1,* and Walter Marx 2 1 Head Civil Engineering Department, NFC- Institute of Engineering & Fertilizer Research (IEFR), Faisalabad, Pakistan 2 Ex-Deputy Director, Department of Hydraulic Engineering and Water Resources Management, Institute for Modeling Hydraulic and Environmental Systems, Universitaet Stuttgart, Stuttgart, Germany. ABSTRACT High flooding downstream of a dam could cause severe damages to the people and property. The most important damage is the loss of life which does not have any substitute. In this study, the extent of loss of life (LOL) downstream of a large dam due to dam failure has been focused. The LOL extent has been considered in terms of downstream distance. For systematic hydraulic analysis, the Jhelum river valley, about 329 km long downstream of Mangla dam in Pakistan has been taken into consideration. Using the flood routing results of different dam failure cases modeled in MIKE 11 and available data of population at risk (PAR), the possible loss of life (LOL) has been estimated by an improved LOL estimation method, which has been developed by the main author. The extent of life loss has also been determined in terms of downstream distance for different dam failure cases. In different dam failure cases, about 8% of the total loss of life occurs up to 5 km downstream and 2% occurs after 5 km up to 286 km downstream of the dam. This study verifies and also contradicts some past predictions about the possible extent of LOL downstream of very large dams due to dam failure. The results of this research would be useful for long-term disaster mitigation planning of large dams in Pakistan as well as in other parts of the world. KEYWORD Loss of life (LOL), Extent of life loss, Mangla dam, dam failure, MIKE 11 I INTRODUCTION n this study, the possible extent of loss of life (LOL) downstream of a dam due to dam failure has been discussed. The extent of life loss downstream of a dam has been considered in terms of downstream distance. For *Corresponding Author: r: Telephone Number r: Fax. Number r: hydraulic analysis, the Jhelum river valley downstream of Mangla dam in Pakistan has been taken into account. Mangla dam completed in 1967, is one of the largest earth and rock-fill dams in the world. The original height of the main dam is about m above river bed [8]. The raising of the dam by about 9.15 m started in 24 and it was planned to be completed by 29 [9]. For this study, Mangla dam has been considered with raised conditions. The crest length of the main dam is about 2561m [1]. The original catchment area of the reservoir is about 3336 km 2 & water surface area (at conservation level) is about 253 km 2 [8]. Fig. 1 shows the top view of Mangla dam and its reservoir. Based on official and GIS data, the Jhelum reach downstream of Mangla dam has been modeled in a one dimensional program, MIKE 11. The flood routing of the Jhelum river valley (Fig. 2) has been done for unsteady flow conditions by using different failure outflow hydrographs which were determined by dam break simulations in MIKE 11. The LOL has been estimated by an improved LOL estimation method using the available data of PAR and flood routing results by MIKE 11. EXTENT OF LOSS OF LIFE In the literature, there is no exact guideline available about the extent of life loss downstream of the dam due to dam failure. There are some statements available about small and very large dams which are based on the empirical data of past events [5]. The extent of loss of life has been discussed in terms of the downstream distance mainly for smaller dams. Based on the available database of small dams, different predictions have been made for the possible extent of life loss downstream of very large dams. Some of these past predictions cannot be considered to be correct due to the unavailability of data for very large dams. This study provides realistic and precise estimates of possible loss of life downstream of a very large dam i.e. Mangla dam in Pakistan for different dam failure cases. The results of this

2 study verify some predictions and also contradict some ideas about the possible extent of life loss downstream of very big dams due to dam failure. MODELING IN MIKE 11 One dimensional flood routing in MIKE 11 is based on an implicit finite difference scheme developed by Abbott [1]. MIKE 11 is capable of using kinematic, diffusive or dynamic and vertically integrated equations of conservation of continuity and momentum (the de Saint Venant equations), as required by the user [6, 7]. The basic equations are derived considering the conservation of mass and conservation of momentum [6]. The model has been calibrated for the 1997 flood with a peak discharge of 12,798 m 3 /s downstream of Mangla dam. For validation of the model, the past highest flood of 1992 (peak discharge downstream of Mangla dam: 26,293 m 3 /s) has been considered. Different cases of dam failure have been modeled in MIKE 11 and the resulting outflow hydrographs have been further utilized for flood routing of the Jhelum river valley downstream of Mangla dam. LOSS OF LIFE (LOL) ESTIMATION A) Population at Risk (PAR) For LOL estimation, the number of people at risk (PAR) was required. According to the available census data (1998) and maps, PAR was estimated at the locations downstream of Mangla dam. Depending on the extent of flooding and available location of houses on GIS map, the estimated PAR has been approximately related to 92- Flood (Past highest flood: Peak Q: m 3 /s). No extra information about the location of houses in the flood plains was available for higher scenarios. Obviously the population at risk (PAR) will be more for higher flooding scenarios. For scenarios of higher flooding, the PAR has been increased quantitatively at each downstream location with respect to the ratio of flooded area in a higher flooding scenario to the flooded area in 92-flood. The PAR downstream of Mangla dam is rural and urban. But rural PAR is more than urban PAR. Fig. 3 shows the number of people at risk downstream of Mangla dam according to 98- census data [2]. PAR downstream of Mangla dam (98-Census data) 2 Fig. 2: Jhelum river valley downstream of Mangla dam PAR (No. of People at risk) Total PAR : Downstream chainage (m) Fig. 3: PAR downstream of Mangla dam (98-census data) and the 92-flood event Rural PAR : 63% Urban PAR : 37% PAR B) LOL Estimation Method In this study the loss of life estimation has been done by using an improved and elaborate method developed by the main author [2, 3, 4 ]. The suggested LOL estimation method has been developed by considering thoroughly the weaknesses and useful ideas in different available methods. Depending on the available data of the Jhelum river valley downstream of Mangla dam, different factors have been considered. The generalized form of the method is given in Equation (1) [2]. Fig. 1: Top view of Mangla dam (by Google earth) LOL i = PAR i FAT BASE F sv F age F mt F st F h F war F ev (1)

3 Where, LOL i = Loss of life at a particular location i downstream of the dam PAR i = Population at risk at a particular location i downstream of the dam FAT BASE = Base fatality rate of.15 from [5] (recommended fatality rates for LOL estimation) for the worst case of medium flood severity. An average value of 1. for all other factors with average conditions is assumed. The other factors could differ from 1. depending on the specific available information. These factors have been defined after a qualitative plausibility analysis according the related information available in the literature. F sv : Flood severity factor based on flood severity indication F age : Age risk factor depending on different age groups in PAR F mt : Material risk factor F st : Storey risk factor F h : Health risk factor F war : Warning factor depending on the initiation of warning and flood travel time F ev : Ease of evacuation factor Depending on the available data, all factors for LOL calculation due to dam failure have been estimated by following the procedure given by the new LOL estimation method. More details of LOL estimation by this method can be found in the related reference [2]. RESULTS OF LOSS OF LIFE Two cases of bridges have been considered for the unsteady flow simulations. In the first case, model has been run for unsteady flow conditions with the existing bridges at different locations downstream of the dam. Whereas, in the second case model runs were made without bridges assuming that the bridges might be washed off due to extreme flooding.loss of life estimation has been done by using the flood routing results of dam break simulations. For LOL estimation the worst case of warning initiation of 3 minutes after the failure has been considered [5]. Fig. 4 shows the percentages of total loss of life for different dam failure cases (with and without bridges). The percentage of total loss of life is more in case of no bridges for different failure cases. The percentage of total loss of life has been calculated with respect to the corresponding PAR for different failure cases. The increase in %LOL is mainly due to increase in flooding which subsequently results in decrease of warning time for higher failure cases. The estimated percentage of total LOL is close to 2.5% for the maximum failure discharge. Fig. 5 shows the cumulative loss of life (LOL) with respect to the downstream distance for different failure cases (with bridges). About 8% of the total LOL occurs up to 5 km downstream of the dam for all failure cases. The LOL results for different failure cases clearly verify some past predictions and also contradict some predictions about the possible extent of LOL downstream of very large dams % Total Loss of Life for Different Failure Cases 3 % Total LOL (% dead people) Worst Case for Warning Initiation: 3 minutes after Failure %LOL (with bridges) %LOL (without bridges) Max. Discharge (m 3 /s) Fig.4 :% Total loss of life for different failure cases

4 Cumulative Loss of Life due to Dam Failure 1 9 Cumulative LOL % Cum. LOL up to 25Km: about 68% of Total LOL % Cum. LOL up to 5Km: about 8% of Total LOL % Cum. LOL up to 1Km: about 9% of Total LOL Total LOL Worst Case for Warning Initiation: 3 minutes after Failure Failure Case1 Failure Case Downstream chainage (m) Fig. 5: Cumulative loss of life due to dam failure (with bridges) VERIFICATION OF SOME PREDICTIONS Based on the dam failure data, it was stated by Graham that a high percentage of life loss due to dam failure occurs in the first 25 km downstream of the dam. Further it was said that for smaller dams this downstream distance is considerably less than 25 km. It was also predicted that there may be some very high dams or those storing very large quantities of water where severe flooding could extend to 161 km or more downstream of the dam. In these cases, the loss of life studies may be extended for more than 5 km downstream of the dam [5]. The above statements have been verified by this study as explained in the following. Firstly, this study verifies the general statement of Graham about the occurrence of high percentage of life loss due to dam failure in the first 25 km downstream of the dam. In Fig. 5, it is clearly shown that about 68% of the total loss of life occurs up to 25 km downstream of Mangla dam for different failure cases. Secondly, the results of this research clearly verify that for very high dams with very large storage capacity severe flooding could extend to 161 km or more downstream of the dam. In this study, Mangla dam in Pakistan has been considered as one of the largest earth and rock-fill dams in the world. The extension of flood severity is up to 286 km in the Jhelum river valley downstream of Mangla dam for different failure cases. Thirdly, this study also verifies the prediction about the extension of loss of life studies for more than 5 km downstream of the dam. In this research, life loss estimation has been done for 286 km of the Jhelum river valley downstream of Mangla dam as shown in Fig. 5. CONTRADICTION OF SOME PREDICTIONS This study also contradicts some past predictions about the extent of life loss downstream of the dam due to dam failure. Graham stated that generally the life loss for more than 5 km downstream of a dam should be very small as compared to the life loss estimated for the areas closer to the dam. Further it was stated that the life loss from 5 km to onwards downstream of the dam would not change the results of the dam safety recommendation [5]. The results of this study clearly contradict the Graham s statements about the extent of life loss for more than 5 km downstream of a dam. In this study the estimated loss of life for more than 5 km downstream of Mangla dam is not very small. The loss of life up to 5 km downstream of Mangla dam is about 8% of the total loss of life for different failure cases as illustrated in Fig. 5. The 2% life loss occurs after 5 km up to 286 km downstream of the dam. The life loss increases to 9% of the total loss of life up to 1 km downstream and then the remaining 1% occurs up to 286 km downstream. The loss of human beings is the most severe damage due to extreme flooding downstream of a dam with or without dam failure. In Fig. 5, the life loss of 2% after 5 km downstream shows

5 considerable number of fatalities for different dam failure cases. Finally, this 2% loss of life after 5 km downstream to onwards 286 km cannot be ignored for dam safety studies of very large dams like Mangla dam. It will certainly influence the results of a dam safety recommendation. This study emphasizes the consideration of possible loss of life due to dam failure for more than 5 km downstream of very large dams for any dam safety recommendation. CONCLUSIONS The estimation of the possible extent of LOL downstream of a dam due to dam failure is very important for a dam safety analysis. The available predictions about the possible extent of LOL in terms of downstream distance have been discussed. This study provides realistic and precise estimates of possible LOL in terms of downstream distance for different dam failure cases of Mangla dam in Pakistan. The LOL has been estimated by using the new LOL estimation method for about 286 km downstream of Mangla dam with respect to the available data of PAR and flood routing results. In different dam failure cases, about 8% of the total loss of life occurs up to 5 km downstream and 2% occurs after 5 km up to 286 km downstream of the dam. The results of this study verify some past predictions and also contradict some predictions about the possible extent of LOL downstream of very large dams like Mangla dam. This research is intended to provide new useful guidelines for realistic estimation of possible loss of life downstream of a dam in case of dam failure. Further, this study strongly recommends the consideration of possible LOL extent over downstream reach length due to dam failure flooding in risk assessment and mitigation studies of large dams. [3] Ehsan, S., Marx, W. and Wieprecht, S.Estimation of possible residual risks for the long-term safety of dams, Proceedings of the 2nd International Conference on Longterm behavior of Dams, Graz, Austria, 29, ISBN: [4] Ehsan, S., Marx, W. and Wieprecht, S.Estimation of possible damages due to catastrophic flooding for longterm disaster mitigation Planning, Proceedings of the Risk and Planet Earth International Conference (Vulnerability, Natural Hazards, Integrated Adaptation Strategies), Universität Leipzig, Leipzig, Germany, 29, ISBN: [5] Graham WJ. A procedure for estimating loss of life caused by dam failure (DSO-99-6); U.S. Department of the Interior, Bureau of Reclamation, Denver Colorado: [6] Reference Manual MIKE 11. A modeling system for rivers and channels; 24. [7] User Guide MIKE 11. A modeling system for rivers and channels; 24.\ [8] WAPDA Mangla Dam Raising Project, Feasibility Study Report Volume I: Main Report; Pakistan: 21. [9] WAPDA Mangla Dam Raising Project; Pakistan: 27. ( [1] WAPDA Mangla Dam Raising Project, Tender design Report, Volume I: Section 3, Embankment Dams; Pakistan: 24 ACKNOWLEDGEMENTS Authors would like to acknowledge the support of respective authorities in Pakistan, mainly Water and Power Development Authority (WAPDA) and National Engineering Services Pakistan (Pvt.) Ltd. (NESPAK) for providing the necessary data of Mangla dam and Jhelum river valley. Further, the continuous technical support provided by the Institute for Modeling Hydraulic and Environmental Systems, Universitaet Stuttgart, Germany is also highly appreciated. REFERENCES [1] Abbott MB, Ionescu, F. On the numerical computation of nearly- horizontal flows, Journal of Hydraulic Research; 5: [2] Ehsan S. Evaluation of Life Safety Risks Related to Severe Flooding; Institute of Hydraulic Engineering, Universität Stuttgart, Germany: 29, Vol. 18, ISBN: