Dam Breaks

Dam-breaks and consequences. X M Carreira

25th August 2012 Page 1 of 12

DAM BREAKS AND CONSEQUENCES

Xosé Manuel Carreira Rodríguez ([email protected])

1. Introduction 2. Hydraulic flow models

2.1. 1D numerical modelling 2.2. 2D CFD techniques

3. Risk assesment 4. Examples

4.1. Historic dam failures 4.2. A recent case: the Aznalcóllar case

5. Concluding remarks 6. References

1. INTRODUCTION The possibility of a devastating flood resulting from dam failure is a concern wherever these structures exist. Earthquakes, floods, landslides, and volcanic activity have resulted in catastrophic dam failures in a variety of environments. From 1946 to 1955, a total of 12 major dam failures were recorded and during the same period of time more than 2,000 dams were constructed worldwide. From years 1956 to 1965, a record of 24 failures and more than 2,500 new dams were constructed during the same period of time. [JANSEN88]. [JOHNILLES02] summarized 300 dam failures throughout the world. Dam failure can be primarily attributed to a number of major key factors including earthquake, differential settlement, seepage, overtopping, dam structure deterioration, rockslide, poor construction and sabotage [RICO08].

2. HYDRALIC FLOW MODELS To know the effects of a dam break, we have to know how dams may break and how a flood will propagate. The damage parameter (flow velocity times water depth) deriving from a dam break flood proved to be a useful tool for estimating consequences of a dam failure (property damage and loss of life) as well as for emergency response planning. Prior to the preparation of any emergency response plan, the dam operator has to carry out a risk assessment study which provide information on the covered area: - the near safety zone, flooded in less than 15 minutes after the dam-break. - the remote area concerned by the submersion wave or the limit at which there is no significant danger for the populations.



Numerical and physical models are used to answer these questions but the development of a dam break is a complicated extreme problem that contains a lot of uncertainties.

2.1.1D NUMERICAL MODELLING Even though, the probability of dam failure can be extremely low, but its occurrences can imply catastrophic consequences downstream, including loss of human lives, properties, natural resources and so on. Therefore, significant predictive data on hypothetical flood events such as flood flows, flow velocities, depths and flood wave arrival times at specific locations downstream of the dam become some of the most important pieces of information for disaster preparedness such as for the formulation of Emergency Response Plan (ERP) guidelines [TURA02]. General international practices on dam safety would include procedures that suit practical management of the dam conditions such as sending early warning and notification messages of emergency situation to the authorities, as well as information on inundation of critical areas for action in case of emergency. Generally, dam break analysis aims at predicting downstream hazard potential systematically in equitable approaches. Numerical modelling process simulations can be carried out based on the topography of a catchment area using an appropriate grid size of approximately 200 m. Generally, a scenario discharge may be assumed in the simulation and flood affected areas may be predicted over a distance of 25 km downstream of the dam, and 1 to 2 km in width [BOSS99]. Currently, there are a number of dam break simulation models widely used by researchers and consultants such as the national weather service dam break forecasting, Mike- 21 (Danish Hydraulic Institute), HEC-HMS/HEC-RAS flood hydrograph (U.S. Army Corps of Engineers), BOSS DAMBRK hydrodynamic flood routing and soil conservation service (SCS) TR#66 uniform dam failure hydrograph. Downstream hazards may include potential loss of human lives, properties (such as residences, commercial buildings, industrial facilities, croplands and pasturelands), infrastructures and utilities located downstream of the dam [TURA02]. The 1D modelling for the dam break hazard analyses is based on an implicit finite difference scheme. The cross-sections used in the model can be taken either from a GIS terrain model or they can be on-site measured cross-sections. 2.2. CFD TECHNIQUES In the case of very complicated topography, the use of a 2-dimensional model seems to be more reasonable than the use of a 1D model. One-dimensional model needs a lot of experience since the cross-sections have to be put at the right locations. The use of a 2-dimensional models is more straightforward. The impact flow on a vertical wall resulting from a dam break problem can be simulated using a Navier-Stokes (NS) solver. The NS solver uses an Eulerian Finite Volume Method (FVM) along with a volume of fluid (VOF) scheme for phase interface capturing. One of the most common Computational fluid dynamics (CFD) packages for simulations of free surface problems is



FLUENT [FLUENT]. Results show favorable agreement with experiments before water impact on the wall. However, both impact pressure and free surface elevations after the impact depart from the experiments significantly. Hence the code is assessed to be good only for qualitative studies. In particular, we have examined the classical dam break problem and subsequent water impact on a plane vertical wall. The FLUENT results for the initial stages of the problem closely agreed with other numerical techniques and experimental results. However, there was some disagreement in water tip location between numerical results and experiments. This is perhaps due to the imperfect initial conditions and some physical effects not numerically modeled. The water impact pressure was numerically measured and compared with experiments. Although the first peak agrees with the experimental measurements of [ZHOU99], the second peak was largely underestimated. This suggests that FLUENT is acceptable for qualitative studies only. In general, the problem after the initial impact could not be modeled with the desired accuracy. Further research is needed to strengthen the features of the software which are not suited for these types of applications. Free-surface reconstruction (complex geometry) including fluid discontinuity and the treatment of entrained air are some of the areas that require further investigations. With careful modelling and accurate data the results of different modelling approaches may be relatively close each other. However, there is a lot of uncertainties in the modelling and specially in the one-dimensional flow modelling where the user of the model can have a significant effect on the results by selecting the locations of cross-sections carelessly.

3. RISK ASSESMENT Risk assessment is the process of deciding whether existing risks are tolerable and present risk control measures are adequate and if not, whether alternative risk control measures are required. Risk assessment incorporates, as inputs, the outputs of the risk analysis and risk evaluation phases. Risk assessment involves judgements on the taking of risk and all parties must recognize that the adverse consequences might materialize and owners will be required to deal effectively with the consequences of a dam failure. In 1988 the U.S. Department of the Interior [USDI98] classified downstream hazards in terms of two major potential adverse impacts on:

1) the number of human lives in jeopardy and 2) economic losses (such as properties, infrastructures, outstanding natural resources and other

developments) downstream of the dam. Based on Downstream Hazard Classification Guidelines published by USDI, downstream hazards may further be classified as “low” for zero live loss associated with minimal economic loss; as



“significant” for 1-6 lives in jeopardy associated with appreciable economic loss and as “high” or >6 lives in jeopardy associated with excessive economic loss. Downstream hazards can further be categorized into: 1) low danger zone, 2) high danger zone, and 3) judgement zones. The judgement zone could be determined from depth-velocity danger level relationship for 1) adults, 2) children, 3) houses and 4) passenger vehicles. For instance, a depth of flooding >1.00 m associated with flow velocity of > 3.0 m/s is considered as “high danger level” for adults, children, houses and passengers. Downstream risks (consequences of dam failure) have been estimated concerning property (infrastructure, buildings, agriculture) and loss of life. The most severe damage from the flood wave would be caused to buildings such as residences, industrial and business buildings, offices and stores. A dam breach flood would also affect bridges, the telecommunications network, the power grid, the water supply and sewage system, the street and road systems, the railway stations, traffic and agriculture. Depending on the severity of the dam break flood event only a certain part of the population will be confronted directly with the flood, i.e. population at risk near the flood inundation boundaries. The loss of life evaluation should be prepared with different impacts on population at risk, different dam break cases and different warning and emergency/rescue scenarios in mind (hours in a day, workday and weekend, and seasons). There are two ways to conduct an evaluation of the loss of life potential caused by a dam failure: One way is to use observations on life-loss associated with dam failures in the past and deal with the problem on statistical base. The other way is to model the expected flood event and its impact to the population at risk. There are also environmental consequences and consequences to the society deriving from potential dam accident. Although environmental and social consequences might be very large, estimating of those in monetary value is impossible. The consequences to society would be significant when taking into consideration the effects on living, infrastructure and working. In the past, dam-break risk analysis were kept secret. In modern open societies dam owners have to inform the populations about the risks they are exposed to. Geographical Information Systems (GIS) are recognised as a powerful mean to communicate with the public in a scientifically correct and yet simple manner. That is why simple applications of GIS have been developed in order to help the dam operator to list all the territorial divisions, roads and railways located in the flooded area, to determine the number of inhabitants and to prepare the public information. Progresses are still expected by a better coupling between water engineering and social and psychological approaches. [RESCDAM].



4. EXAMPLES

4.1.HISTORIC DAM FAILURES Dam failures are generally catastrophic if the structure is breached or significantly damaged. During an armed conflict, a dam is to be considered as an "installation containing dangerous forces" due to the massive impact of a possible destruction on the civilian population and the environment. As such, it is protected by the rules of International Humanitarian Law (IHL). The main causes of dam failure include spillway design error (South Fork Dam), geological instability caused by changes to water levels during filling or poor surveying (Vajont Dam, Malpasset), poor maintenance, especially of outlet pipes (Lawn Lake Dam, Val di Stava Dam collapse), extreme rainfall (Shakidor Dam), and human, computer or design error (Buffalo Creek Flood, Dale Dike Reservoir, Taum Sauk pumped storage plant).We can conclude that no two dam-breaks are the same. A notable case of deliberate dam failure (prior to the above ruling) was the Royal Air Force 'Dambusters' raid on Germany in World War II, in which three German dams were selected to be breached in order to have an impact on German infrastructure and power capabilities. [WIKILDF]

FIGURE 1. Puentes (Murcia, Spain) dam break in 1802. The 50 m heigth masonry dam was designed with empirical methods. Due to a foundation failure 608 people died under the water after 11 years of service. [JANSEN80].



FIGURE 2. Photograph of the Old Dale Dyke reservoir embankment in Sheffield (UK), shortly following its collapse in March 1864. After about thirty minutes the flood gradually subsided leaving a trail of destruction more than eight miles long. In addition to the massive loss of life of 238 people; total or partial destruction occurred to 415 dwelling houses, 106 factories, 64 other buildings, 20 bridges and 4478 cottage/market gardens. [BRIN93].

FIGURE 3. View looking north at “tombstone” remnant of Saint Francis dam following failure in 1928 (California, USA). Scene of landslide is visible in upper right-hand corner of image (Photo courtesy of Santa Clarita Valley Historical Society). 450 people died. [WIKIFRAN] and [BEGSAND07]. FIGURE 4. Ribadelago (Zamora, Spain) dam break. On 9th January 1959, torrential rain caused the dam to burst. A wall of water that swept down the head valley. The deluge lasted for 14 minutes and reached nine metres in height. 144 people were killed but only 28 bodies were ever found. [WIKIRIBA] and [SANA].



FIGURE 5.Vajont (Pordenone, Italy) dam break in 1963. This was an arch dam 267 m high. During the test filling of the dam, a land slide of volume 0.765 Mm3 occurred into the reservoir and was not taken note of. During 1963, the entire mountain slide into the reservoir (the volume of the slide being about 238 Mm3 , which was slightly more than the reservoir volume itself). This material occupied 2 km of reservoir up to a height of about 175 m above reservoir level. This resulted in a overtopping of 101 m high flood wave, which caused a loss of 3000 lives. [WIKIVAJO].



FIGURE 6. Malpasset (France) dam break in 1959. An arch dam of height 66 m, with 22 m long crest at its crown. When the collapse occurred, the dam was subjected to a record head of water, which was just about 0.3 m below the highest water level, resulting from 5 days of unprecedented rainfall. The failure occurred as the arch ruptured, as the left abutment gave away. The left abutment moved 2 m horizontally without any notable vertical movement. The water marks left by the wave revealed that the release of water was almost at once. The volume of water relieved was 4.94 Mm3 of water. 421 lives were lost. [WIKIMAL].

FIGURE 7. Tous (Valencia, Spain) dam break in 1982. During the incident most of the town near the river side was flooded and the water level reached more than 15m at the bridge and almost 8m at some areas of the town according to the estimated data. The flash flood affected a total population of approximately 200000 inhabitants, caused 8 deaths and an enormous material damage.[WIKITOUS].



4.2.A RECENT CASE: THE AZNALCOLLAR CASE The Aznalcollar tailings dam failed catastrophically in 1998. Nobody died but the breach caused one of the worst-ever environmental disasters in Spain. The dam was part of a large open mining complex that had been in operation for decades in the province of Seville, Spain. It became clear soon after the failure that a translational movement was the reason for the breach that opened in the embankment. There was no evidence of other possible mechanisms, such as overtopping of the embankment, instability of the embankment slopes or internal erosion. A deep translational slide displaced 600 m of embankment and its foundation towards the east. The failure surface was located inside blue clays (the colour is due to the iron). The displaced mass included the embankment, the alluvium terrace and about 10 m of the blue clay. [ALOGEN06].

FIGURE 8. Aerial view of the breached dam a few hours after the failure. The orientation of the initial outflow is indicated by the depositional fans observed outside the breached embankment. Because of the proximity of the Doñana National Park, which fortunately was protected from the direct flood, and the dam breaks in the past, this failure had considerable impact on the public opinion. There was a great interest in the case, and several possible causes of the failure were openly proposed in newspapers by people of widely different background and interests. [WIKIAZNA].



5. CONCLUDING REMARKS

The regulation of referable dams is based mainly on the population at risk in the event of a failure. However, dam designers may also wish to consider other potential consequences to determine design standards for a dam. These other consequences may include according to [ICOLD98]: • economic loss of the asset. • damage to property and infrastructure. • commercial losses and social impacts. • impacts due to loss of water supply. • environmental damage. Some of the common scenarios to be considered in consequence assessments include: • dam break event for the “sunny day” condition and a range of flood events. • remote floods: flood surges well downstream of dam which can coincide with storage release. • upstream floods: backwater effects of the dam during floods. • water supply loss: failure of pumps, outlet facilities, reservoir pollution etc. • operational problems: accidental opening of flood gates, equipment malfunction etc. Not only do dam owners face a responsibility towards society in order to avoid or diminish the probability a dam-break but the public opinion is becoming more concerned about dam safety as well.



6. REFERENCES [ALOGEN06] E. Alonso, A. Gens. Aznalcóllar dam failure. Part 1: Field observations and material properties. Géotechnique. 2006 [BEGSAND07] L. Begnudelli and B. Sanders. Simulation of the St. Francis Dam-Break Flood. ASCE. 2007. [BOSS99] BOSS DAMBRK hydrodynamics flooding routing user’s manual. 1999. [BRIN93] C. Brinfield, The History of the City of Sheffield. 1993. [FLUENT] On-line Fluent user's guide. http://jullio.pe.kr/fluent6.1/help/html/ug/main_pre.htm [ICOLD98] ICOLD. Dambreak Flood Analysis.Bulletin 111. 1998. [JANSEN80] R. Jansen. Dams from the beginning. U.S. Department of the Interior, Bureau of Reclamation.1980. http://ussdams.com/ussdeducation/Media/damsfrombegin.doc [JANSEN88] R. Jansen. Advanced dam engineering for design construction and rehabilitation, 1988. [JOHNILLES02] F. Johnson and P. Illes. A classification of dam failures. 1976. [RESCDAM] RESCDAM . Development of Rescue Actions Based on Dam-Break Flood Analysis. Final Report June 1999 – March 2001. [RICO08] M. Rico, G. Benito and A. Díez Herrero. Flood from tailing dam failures. 2008. [SANA] http://www.iberianature.com/material/lake_sanabria.html [TURA02] A. H. Turahim, N. A. Mohd. Analysis of dam break for disaster preparedness. 2nd. World engineering congress. 2002. [USDI98] U.S. Department of the Interior. Dam-Break Flood Analysis. Review and recommendations. 1988. [WIKILDF] List of major historic dam breaks: http://en.wikipedia.org/wiki/List_of_dam_failures. [WIKIAZNA]http://es.wikipedia.org/wiki/Desastre_de_Aznalc%C3%B3llar [WIKIFRAN] http://en.wikipedia.org/wiki/St._Francis_Dam



[WIKIMAL] http://fr.wikipedia.org/wiki/Malpasset [WIKIRIBA] http://es.wikipedia.org/wiki/Cat%C3%A1strofe_de_Ribadelago [WIKITOUS] http://es.wikipedia.org/wiki/Presa_de_Tous [WIKIVAJO] http://it.wikipedia.org/wiki/Disastro_del_Vajont [ZHOU99] Z.Q. Zhou, , J.O.D. Kat and B. Buchner. A nonlinear 3-D approach to simulate green water dynamics on deck. 1999.

Documents

Dam Breaks