Dam break analysis of the Nagmati and Dhap dams using HEC-RAS Open Access

Climate change impact assessment has predicted the possibility of droughts (Naumann et al. 2018; Balting et al. 2021) and floods (Tabari 2020; Janizadeh et al. 2024), making water resources management more difficult and decreasing future water availability. One of the most effective strategies for tackling these challenges is the construction of dams, which serve multiple functions such as flood control, fulfilling water demands, power generation, facilitating irrigation, increasing flow in the river, supporting recreational activities, and boosting tourism (Boulange et al. 2021; Schmitt & Rosa 2024; Pandey et al. 2024), playing a vibrant role in the country's economy. Although dams provide numerous advantages, maintaining their sustainability is challenging as they also have long-term environmental and social impacts, such as habitat destruction, displacement of communities, sedimentation problems, and potential failure (Wang et al. 2005; Tilt et al. 2009; Fung et al. 2019; Miranda et al. 2021; Mohammed et al. 2022; Rana et al. 2022; Hwang & Lall 2024). The catastrophic consequences of their failures can lead to rapid flood inundation downstream, resulting in substantial property damage, and disastrous impacts on human safety, infrastructure, and ecology (Olson et al. 2000; Spero et al. 2022; Gee et al. 2024; Zhu et al. 2024). Simulating dam break events and successive floods is essential for understanding and mitigating the risks associated with potential failures of dams (Xiong 2011; Aureli et al. 2024; Jibhakate et al. 2024). It is essential for developing emergency action plans, including identifying the flood risk zones and determining the rescue times (Ansori et al. 2021). The failure mode classifications include flooding due to overtopping, malfunction or breakdown of gate spillways, piping, slope instability, and seismic activity (Foster et al. 1998). Overtopping failures result from the erosive action of water due to the uncontrolled flow of water over, around, and adjacent to the dam (Mattas et al. 2023). About 80% of dam accidents are due to the failure of embankment dams, while 20% are due to the failure of concrete dams (Kostecki & Banasiak 2021). Overtopping is the most common failure mode representing 48% of the reported dam failure cases (Foster et al. 1998).

Previous studies have performed several dam break analyses in recent years. The destruction of the Kakhovka Dam on the Dnipro River during the Russia–Ukraine war, which resulted in the deaths of 50 people, underscores the potential use of dam failures as a strategic weapon in conflict and highlights the devastating downstream consequences of such breaches, severely impacting both populations and infrastructure (Vyshnevskyi et al. 2023). Shrestha et al. (2024) performed a dam break analysis on the Dudhkoshi storage hydroelectric project in Nepal using the shallow water equation (SWE). Xiong (2011) performed a dam break analysis based on three scenarios (dam break, no dam break, and without dam) on the Foster Joseph Sayers Dam, United States, for unsteady flow conditions considering probable maximum flood (PMF) to predict the stage and flow hydrographs at different locations downstream. Abhijith et al. (2017) examined the dam break situation at the Idukki Dam in India employing the Hydrologic Engineering Center-River Analysis System (HEC-RAS) to assess the highest water level, the greatest flow rate, and the timings of their occurrence at various points. Furthermore, the dam break at the Hidkal Dam in Karnataka, India, has been investigated by Bharath et al. (2021) by simulating a dam break model for unsteady flow conditions using PMF as flow data for overtopping and piping failure modes to understand the dynamics of a dam breach.

The Bagmati River Basin Improvement Project (BRBIP) has undertaken the construction of two artificial dams, the Dhap and Nagmati dams, in the upper region of the Bagmati River in Kathmandu Valley, Nepal, to improve water security and enhance resilience against potential climate change impacts (Rana 2013). These dams are 24 m high and 94.5 m high with water storage capacities of 0.85 and 8 million m³, respectively. Their primary objective is to sustain perennial flow, particularly during the dry season (Singh & Subedi 2020). Breaching of these dams will potentially cause significant life loss and property damage in Kathmandu Valley, the densely populated capital city of Nepal; however, no such studies have been performed before. In addition, it makes two United Nations Educational Scientific and Cultural Organization (UNESCO) World Heritage Sites (Pashupatinath Temple and Bouddha Stupa) vulnerable. Therefore, it is important to assess the potential risk and plan to mitigate the effect of the dam break.

Although dam break problems have been modeled extensively, this study offers new insights by concentrating on the Dhap and Nagmati dams, two relatively unknown dams in Nepal. This study especially looks at the effects of dam breaches in an area of great cultural and ecological significance, including the susceptibility of UNESCO World Heritage Sites, in contrast to earlier research that mostly focused on generic dam break evaluations. In the dam break analysis, numerous tools are employed such as Flow 3D, HEC-RAS 1D, HEC-RAS 2D, MIKE 11, FLDWAV, and DAMBRK (Nugusa Duressa 2018). HEC-RAS is one of the common hydrodynamic models used for dam-break flood modeling; moreover, this model can also perform 1D steady flow simulations, 1D and 2D unsteady flow simulations, sediment transport calculation, mobile bed computation, and water temperature/water quality modeling (Brunner & CEIWR-HEC 2021; Marasini & Pandit 2021; Vashist & Singh 2023). The simulation of dam breaking and routing the flood using HEC-RAS 1D requires several assumptions when setting up the system. The 1D model cannot give an accurate result when the slope is steeper than 1:10 and they also lack the velocity distribution on a plane, especially in urban and complex areas (Newmiller et al. 2017). On the other hand, the HEC-RAS 2D model can provide velocity and directional variation and is more accurate, stable, and easier to use (Urzică et al. 2020); so, in this study, we decided to employ HEC-RAS 2D. The other reasons for selecting this model are its ability to simulate the hydraulic behavior of flood waves, produce dynamic flood maps, free availability, and user-friendly manual (Ehsan et al. 2016). HEC-RAS is utilized to compute dam breach parameters, generate inundation maps, and simulate the hydraulic behavior of flood waves. The development of the breach was modeled through the simultaneous iteration of input parameters, which were integrated into the hydraulic simulation model. The other required breach parameters are calculated using the empirical equations that are derived from the characteristics of the dam, such as dam height and storage volume (Wahl 1997; Bharath et al. 2021).

The main goal of the study is to assess the impact of the Nagmati and Dhap dam breaching under the overtopping mode of failure. To this end, several specific objectives have been drawn up. First, the study aims to predict the flood intensity and travel time at various locations and subsequently prepare the resulting hydrographs. Then, it portrays the flood inundation, assesses flood velocity, determines flood duration, and maps flood hazard vulnerability within the study area. Last, the research aims to conduct a detailed sensitivity analysis of Nagmati Dam breach parameters by focusing on their impact on peak discharge, arrival time, and water surface elevation. The research aimed to address two key questions: (1) What would be the extent of inundation from the Nagmati and Dhap dam breaches under various scenarios? and (2) How could this data aid in calculating the time required for rescue operations and locating safe zones within the impacted area? The dam break analysis was performed under three different scenarios: (1) failure of the Dhap Dam at the full supply level (FSL), (2) failure of the Dhap Dam with the Nagmati Dam at the FSL, and (3) failure of the Nagmati Dam at the FSL with sensitivity analysis.

The accuracy and dependability of data analysis rely on various factors, including terrain, the quality of satellite imageries, rainfall record data, and crucial details about the dam, which are essential for providing necessary physical information for the analysis (Sumira et al. 2023). It includes the structural dimensions parameters of the dam and the capacity curve. Supplementary Table S1 shows the data sources used for the dam break analysis. Table 1 shows the salient features of the Nagmati and Dhap dams and Supplementary Figure S1 shows the volume–elevation curve of the Nagmati Dam and the Dhap Dam.

A 2D mesh with dimensions of 50 m × 50 m was generated to depict the downstream land surface, utilizing combined terrain data. The chosen mesh size strikes a balance between representing the flat floodplains adequately and ensuring a stable solution within the required computation time interval. The storage area and the downstream area were connected using an inline structure (the Nagmati and Dhap dams) on the HEC-RAS geometric data editor. The storage area corresponds to the upstream portion of the dam axis, while the 2D flow area represents the inundation study area downstream. Manning roughness values for the 2D flow area in the mesh are given by Chow (1959). A mesh size of 50 m is employed to ensure stability, given its ability to reduce both computational time and memory usage (Vashist & Singh 2023).

In this study, we preferred to breach the dam using an overtopping mode of failure rather than piping because of its severity. Overtopping failure occurs when the water level rises above the dam crest elevation because of extreme rainfall and not enough spillway capacity (Khosravi et al. 2019; Sumira et al. 2023).

The choice of appropriate boundary conditions is essential as they heavily influence the floodwater downstream flow and must accurately represent the actual site conditions (Bharath et al. 2021). The probable maximum precipitation (PMP) represents the maximum possible rainfall that could occur within a specific duration under existing meteorological conditions (Awal et al. 2024). The PMP was thus superimposed with a synthetic unit hydrograph (SUH) to generate a PMF. The upstream boundary condition is explained using the PMF. PMF is the largest flood that could happen in a specific area (Panday et al. 2023; Awal et al. 2024), like near a dam. It is calculated by considering the worst possible weather conditions, including the most extreme rainfall and, if applicable, snow melting, along with the conditions of the surrounding land that can contribute to flooding. The condition at the downstream boundary is defined as a normal depth, determined through the calculation using a slope value of 0.01 for the initial iteration. As the iteration progresses, the slope value will adjust itself to find the correct slope and normal depth.

For the determination of PMF, two main methodologies are usually adopted: the traditional SUH method and the utilization of river basin or hydrological process models (Devkota & Shrestha 2021). While the latter method encompasses abundant data and the inclusion of multiple factors, it may be challenging in areas where observed runoff data, necessary for calibrating hydrological models, are unavailable (Mattas et al. 2023). In such scenarios, the SUH method becomes the suitable method. Supplementary Figure S2 represents the upstream boundary condition, i.e., PMF for both dams. In this study, the flood hazard vulnerability map is prepared according to the Australian rainfall and runoff (ARR) guidelines given by Smith et al. (2014). It is a national guideline document and software suite that can be used for the estimation of design flood characteristics. This guideline developed the quantitative hazard classification threshold by integrating people's stability (Cox et al. 2010), vehicle stability (Shand et al. 2011), and building stability (Mason et al. 2012). So, for any area to fall into a particular hazard zone, it must satisfy one of the three criteria: limiting depth, limiting velocity, or their combination, as depicted in Table 2.

Based on the presence of the dam, three scenarios were created. In the first scenario, the Dhap Dam was considered without the presence of the Nagmati Dam downstream. In the second scenario, the capacity of the Nagmati Dam's freeboard to mitigate the flood from the Dhap Dam is evaluated. Finally, in the third scenario, the Nagmati Dam was breached at its FSL, and a sensitivity analysis was performed. The scenarios are described in detail under the following headings.

The investigation included an unsteady flow simulation to examine the dam break scenario at the Dhap Dam, specifically focusing on an overtopping failure mode. We simulated the PMF from 12:00 pm to 10:00 pm, with the results corresponding to this time period. The simulation simulated the PMF between 12:00 pm and 10:00 pm, with the results aligning with this time frame. The breach was initiated at 7:00 pm from the onset of the simulation, revealing a maximum breach discharge of 164.69 m³/s, as depicted in Supplementary Figure S3.

In this study, we investigated the potential for managing floodwater resulting from the breach of the Dhap Dam by assessing whether the Nagmati Dam, situated immediately downstream, could effectively mitigate the incoming flood. To evaluate this, we simulated a flood hydrograph corresponding to the breach of the Dhap Dam at the location of the Nagmati Dam, as shown in Supplementary Figure S4. The area under the flood hydrograph represents the volume of water that would accumulate in the Nagmati Dam. Subsequently, we employed a capacity curve of the Nagmati Dam to ascertain whether the Nagmati Dam's storage capacity, particularly its freeboard capacity, would be sufficient to contain the incoming flood volume. This analysis aimed to assess whether the Nagmati Dam could effectively mitigate downstream inundation risks by recessing the floodwater resulting from the breach of the Dhap Dam.

From the flood hydrograph at the Nagmati Dam, as shown in Supplementary Figure S4, the volume of the flood from the Dhap Dam was calculated by taking the area under the hydrograph that was determined to be 1.02 Mm³. The dam crest elevation is 1,911.5 m. From the capacity curve, it can store a volume of 39 Mm³ up to the crest. Similarly, the freeboard of the dam is 3 m. The water surface elevation of the dam is 1,908.5 m and can store 35.5 Mm³ up to this level. The water that can be stored in the freeboard can be obtained by deducting the volume at crest level from that of water surface elevation and it was determined to be 3.5 Mm³. Our analysis shows that the volume of floodwater resulting from the Dhap Dam breach (1.02 Mm³) is smaller than the capacity of the Nagmati Dam's freeboard (3.5 Mm³). Consequently, the floodwaters are effectively recessed within the available freeboard capacity of the Nagmati Dam, suggesting its capability to mitigate downstream flooding risks under these circumstances.

The breach width also exhibits notable influences on the downstream flood dynamics, with wider breaches resulting in increased peak discharges and faster flood propagation. For example, at Gokarna, widening the breach from 54 to 72 m increases the peak discharge from 7,243.73 to 7,393.36 m³/s and marginally reduces the arrival time. This highlights that the breach width plays a crucial role in determining the volume and speed of water flow, thereby exacerbating flood impacts in downstream areas. Likewise, changes in the breach side slope and weir coefficient result in minor yet evident variations in peak discharge and arrival times, highlighting their role in shaping flood propagation dynamics. The time it takes for a breach to form has a considerable impact on the arrival time, with quicker breach formation (e.g., 0.8 h) resulting in earlier flood peaks and greater discharges than slower formation times. These findings emphasize the crucial influence of breach parameters on the intensity and timing of downstream flooding. We can see that the parameter ‘breach elevation’ is relatively more sensitive followed by the breach width than other parameters. The sensitivity analysis of breach parameters is illustrated in Table 3.

The Nagmati and Dhap dams are situated in the same region of Kathmandu Valley, Nepal, with the Nagmati Dam being larger than the Dhap Dam in terms of surface area and discharge. So, the failure of the Nagmati Dam at the FSL has a larger impact compared with that of the Dhap Dam. The discharge and flood travel time vary when moving downstream from the dam. From the result, we have found that the maximum breach discharge of the Dhap Dam is 164.69 m³/s and that of the Nagmati Dam is 2,106.38 m³/s. The potential hazard is higher during the breaching of the Nagmati Dam than the Dhap Dam due to larger flood discharge in its breaching. The travel time of the flood from the breach of the Dhap Dam at Gokarna is about 4 h whereas, in the case of the Nagmati Dam, it is found to be 0.81, 1.08, 1.38, and 2.175 h, respectively, at Gokarna, Jorpati, Bouddha, and Pashupati. The flood from the Dhap Dam has no significant changes downstream from Gokarna, as the results were seen only up to Gokarna. The flood from the breaching of the Nagmati Dam is found to have effects up to the lower portion of Kathmandu Valley in Balkhu and Chobhar.

The total area inundated in the failure of the Dhap Dam is 2.667 km² and that due to the failure of the Nagmati Dam at the FSL is 29.401 km². In the case of the Dhap Dam, the study reveals that the cropland is the most heavily affected by inundation, covering 1.866 km². Similarly, the forest experiences inundation over an area of 0.54 km². Conversely, grassland and built-up areas show minimal inundation, with areas of 0.017 and 0.254 km², respectively. Moreover, around 430 buildings and 22.137 km of road are affected by this flood. Similarly, in the case of the Nagmati Dam, the study shows that the breach of the Nagmati Dam inundated 29.401 km². The land use pattern in inundated areas shows that the built-up area is heavily inundated by flood, with a total inundation of 21.169 km². The cropland of 5.425 km² is inundated. Similarly, the forest, grassland, water body, and bare rock show minimal inundation of 1.149, 1.631, 0.02, and 0.007 km², respectively. Moreover, around 66,220 buildings and 626.315 km of road will be affected by this flood. This shows that the breach of the Nagmati Dam at the FSL is significantly more dangerous than that of the Dhap Dam.

This study focuses on four key regions: Gokarna, Jorpati, Bouddha, and Pashupati. In these areas, the flood caused by the Nagmati Dam breach flows through the Bagmati River without any significant lateral inflows or tributary contributions to these sections. However, the Manohara River and its tributaries join this catchment after the Pashupati area. Due to the absence of a suitable hydrological station, we could not consider these tributaries in our study. Similarly, in the upstream of the dam, there is only one meteorological station, requiring us to rely solely on its data. However, using data from a single station may not adequately represent the spatial variation in precipitation in the calculation of the PMP.

Since the elevation data of the study area are not surveyed, we have to use satellite data such as the DEM, which is not 100% accurate, so uncertainty might occur in fewer margins. The accuracy of this study may be affected by different factors such as model assumption and uncertainties in hydraulic modeling.

The arrival time of failure of the Nagmati Dam at the FSL in Pashupati is 2.175 h, which signifies that the population residing in these areas and downstream can be safely relocated to safer places by studying the potential hazard map if the effective flood warning system is established. This signifies the importance of the study regarding the health and safety protocol of the affected population. Building upon our findings, future research could explore the various factors like impacts in Kathmandu Valley due to simultaneous breaches of both dams and structural and non-structural measures that can be applied to reduce the impact of floods.

The failure of a dam is complicated and uncertain. Creating maps of disasters and assessing risks is crucial for planning successful strategies to prevent and mitigate such events. Evaluating and simulating potential dam failures are vital for recognizing and reducing risks. Furthermore, accurately gauging flood depths and predicting when waves will reach key areas downstream are essential for making effective emergency plans. In this study, HEC-RAS was utilized to analyze the hydraulic model of the Dhap and Nagmati dam breaches under the overtopping failure mode, where the PMF was employed as a crucial parameter. Based on the findings of this, it is possible to conclude that:

• Breaching of the Dhap Dam at the FSL has no effect below the Nagmati Dam as 1.02 Mm³ of calculated floodwater from the Dhap Dam is recessed in the 3 m freeboard of the Nagmati Dam having a capacity of 3.5 Mm³ volume of water.

• The flood after the breach formation in the Nagmati Dam takes about 0.8 h to reach the beginning of the settlement area, Gokarna, making it difficult for the initiation of rescue operations, whereas, at Pasupathi, the flood reaches after 2.2 h of breach, allowing safe evacuation of people from the flood risk area. Ultimately, the flood reaches the end of the valley at Chobar after 11 h of breach.

• Because of the breaching of the Dhap Dam, around 430 buildings, 23 km of road, and 2,150 human lives will be affected.

• A total of 66,220 buildings are estimated to be inundated (at least 0.5 m or more), putting 331,100 human lives at risk. Around 637 km of road (including highways, district roads, and feeder roads) will be affected by the breaching of the Nagmati Dam.

• The hazard vulnerability map, created using ARR guidelines, shows that areas like Gokarna, Jorpati, and Boudha fall under the H6 (very high hazard) zone, while Pashupati is in the H5 (high hazard) zone, with the UNESCO World Heritage Sites of Pashupatinath Temple located in the H5 zone and Boudhanath Stupa in the H6 zone, highlighting the flood risk in these areas.

• From the sensitivity analysis of the Nagmati Dam break, it can be concluded that the flood stage and peak discharge are more sensitive to breach height and moderately sensitive to breach width and coefficient, whereas the side slope and breach time are less sensitive.

We acknowledge the Department of Hydrology and Meteorology for providing the meteorological data. Moreover, we extend our gratitude to the Civil Engineering Department and Administration of Himalaya College of Engineering for supporting us during the study report.

U.M. conceptualized and supervised the study. S.P., R.N., and P.K. collected the data. S.A. and P.S. are responsible for the analysis of data. P.K. developed the model. P.K. and R.N. prepared the first manuscript. U.M. is responsible for overall guidance and feedback on writing the overall manuscript. S.D. reviewing and editing. R.K. reviewing and editing. H.C. reviewing. All authors contributed to the interpretation of the result and manuscript preparation.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.