Flood Risk for Power Plant using the Hydraulic Model and Adaptation Strategy

2.1 Model Development

The model was developed in Step 1, and the calibrated and validated in Step 2. The first step includes: (1) obtaining and analyzing the 90 m resolution DEM as map format; (2) con-verting to geometric data by using HEC-GeoRAS tool; and, (3) importing the boundary conditions to HEC-RAS. The model was then calibrated and validated to make sure it was applica-ble (Step 2). In this step, there is 4 main sub-steps including: (1) calculation of Nash-Sutcliffe (NS) values after calibration; (2) validation in a different year; (3) calculation of NS values after validation; and, (4) finishing the model if NS values meet the highest value (0.75≪NS≤1). If NS values are lower than 0.75, there will be 4 extra steps, including: (3’), (4’), (5’) and (6’). Then it is calibrated again until NS meet the highest value (Fig. 2).

Fig. 2.

Steps of model development.

This model is a tool for assessing the impact of climate change on the power plant, especially flood events. The applicable model can be used to predict the impacts of climate chan-ge under the future scenarios.

2.2 Model Description

DEM of the study region: The 90 m resolution digital ele-vation model (DEM) was collected from the Consultative Group on International Agriculture Research-Consortium for Spatial Information (available at www.cgiar-csi.org). The DEM of South Korea was obtained to creating geometric data and focused on Yeongwol district, Gangwon Province, which is located near the power plant (Fig. 3). The DEM was used for cross section interpolation, and the cross-sections are then imported to HEC-RAS program to develop the model.

Fig. 3.

Digital elevation model of study area.

Geometric data: The main geometric data included stream centerlines (the line between two river banks); river cross-sections, created from DEM by using HEC-GeoRAS in combination with GIS program; and, Manning’s n roughness coefficients. The cross-sections were created based on the conditions: (a) the change of channel slopes; (b) the change of river widths; and, (c) the change of curves of the channel. The total length of the Namhan River reach in this study area was approxima-tely 8,000 m (Fig. 4).

Fig. 4.

Geometric data in the model.

Boundary conditions: There are two upstream inputs, inclu-ding flow rates in Yeongwol Bridge, located on Dong River, and Palgoe Bridge, located on Namhan River (Fig. 4). The flow rates (m³·s^-1) were measured every 10 minutes and published on HRFCO (availableat www.hrfco.go.kr).

The downstream boundary, which is friction slope, is esti-mated by using the energy equation (Eq. (1)) (Chow et al., 1988) as

where and : elevations of the main channel inverts (m); and : water depths of downstream and upstream (m), respectively; and : average velocities of downstream and upstream (m·s^-1),respectively;: gravitational acceleration (m· s^-2); and, : energy head loss (m). The representation of terms of energy equation for open channel flow is shown in Fig. 5.

Fig. 5.

Representation of terms in the equation for open channels.

The friction slope (Canestrelli et al., 2014; Cheikh, 2015; Wu et al., 2015) is calculated by Eq. (2).

where : fiction slope; and, : channel length (m).

The friction slope and channel slope have the same value ( = ) in case of unsteady uniform flow (Chow et al., 1988). Applied for the model, the friction slope can be estimated as Eq. (3). Then the value was used as downstream boundary input data.

where : channel slope.

In this model, the estimated total length from upstream to downstream reaches is L = 8,000 m. The upstream elevation is = 188 m, and the downstream elevation is = 176 m. Therefore, the friction slope is calculated as = 0.001 and used as the downstream boundary condition.

Manning’s n roughness coefficients is the value indicating the flow obstacle in the rivers. There are many n values depen-ding on different types of channel. The n values are limited between 0.012 and 0.100 for the main channel of natural streams (Alegre et al., 2017; Bilskie et al., 2016; Brunner, 2010; Chow et al., 1988; Savage et al., 2016). Manning’s n equation is important for hydraulic study, it indicates the relation between flow rate, water level, n coefficient and friction slope (Eq. (4) and Eq. (5)).

where : flow rate (m³·s^-1); n: Manning’s n roughness coefficient; A: flow area (m²); R: hydraulic radius (m); and, P: wetted perimeter (m).

Model running: The year 2013 was chosen for running the model because there was a peak water depth of the period bet-ween 2009 and 2014, and the data of two upstream boundary conditions are sufficient. At Yeongwol Bridge Station, the maxi-mum water level is 7.14 m on July 15^th, and the maximum flow rate is approximately 7,000 m³·s^-1. At Palgoe Bridge Station, the maximum water level is 9.40 m, and the maxi-mum flow rate was nearly 5,700 m³·s^-1.

2.3 Calibration and Validation

J. E. Nash and J.V. Sutcliffe created one index called Nash-Sutcliffe index (NS) that represents the accuracy of hydrolo-gical and hydraulic models (Habert et al., 2016; Nash and Sut-cliffe, 1970). Harmel et al. (2010) developed a watershed model was calibrated by using NS index, and it shows that the NS values are greater than 0.90 resulting in a good applied model (0.75≪NS≤1) (Moriasi and Arnold, 2007). The NS index was also used to calibrate the hydraulic model. For example, Panda et al. (2010) and Paiva et al. (2011) used NS to calibrate the hydraulic model for river networks. Besides, there were vari-ous studies using NS for comparing the simulated and observed water levels and discharge in 1D hydraulic model recently (de Paes and Brandão, 2013; Lotsari et al., 2014; Meert et al., 2016; Saleh et al., 2013).

The NS value is calculated in Eq. (6). If the NS closes to one (100%), the model will be acceptable and can be applied.

where NS: Nash-Sutcliffe Index; $$ X_{sim,i}$$ : simulated value at the time i; $$ X_{obs,i}$$: observed value at the time i; $$ {\overline{X}}_{obs}$$: average observed value in the period of calibration; m: number of hours.

In this case, X value represents the water level of the channel. Therefore, $$ X_{sim,i}$$, $$ X_{obs,i}$$, and $$ {\overline{X}}_{obs}$$ were simulated, observed, and average observed water levels, respectively. In addition, the output data was simulated hourly and decided as 2,208 hours, from July 1^st to September 30^th, and that means m = 2,208.

The Manning’s n roughness coefficients of the main channel were adjusted to make the simulated water levels closed to observed values. The model was tested with different n coeffi-cients in a range from 0.012 to 0.100, and it would be stopped if the NS index met the highest values.

After calibration, the model was validated by running in the different periods, i.e. 2011, 2012 and 2014, to make sure that it is reasonable for other years. If the NS values were much lower than one, the n values would be adjusted again until it was applicable. The model was then computed in the first period, i.e. the year 2013, to make it applicable (Fig. 2).

Table 1 shows the Nash-Sutcliffe index of two upstream stations under calibration and validation. The results indicated that the NS values were very high, and it can be apply for simulation and prediction. Moreover, it was found that the Manning’s n coefficient in the main channel of Namhan River and Dong River was 0.036, and it was used for all river cross sections.

Table 1.

NS index after calibration and validation

The comparisons of water levels in the two upstream stations are shown in Fig. 6(a) and Fig. 6(b), and explained for the accu-racy of the model (NS index). The periods of the highest water level was chosen to compare. The peak water level usually occurred in the period from 2^nd to 16^th of July in 2011 and 2013. Moreover, the simulated water levels almost equal to the observed values. The maximum water levels in 2012 and 2014 were quite low compared to the other years. The maximum observed and simulated water levels were 3.57 m and 3.54 m in 2012; and 2.47 m and 2.34 m in 2014, respectively.

Fig. 6.

Simulated and observed water levels at Yeongwol and Palgoe Bridge Stations in 2011 and 2013.

2.4 Future Scenarios

The comparison of the discharges shows that the flood events in 2006 and the highest water level in 2013 occurred in the same period, i.e. the period from July 15^th to July 17^th, at Yeongwol Bridge Station. The scenarios were determined based on the extreme flood event in 2006 and strategies to resist these problems. The maximum discharge in 2006 (about 19,000 m³·s^-1) was approximately three times greater than that in 2013, which was nearly 6,000 m³·s^-1.

The data of discharges and water levels at Yeongwol Bridge Station in 2006 is sufficient; however, the hydrodynamic data at Palgoe Bridge station in 2006 is missing. It was assumed that the future discharges at Yeongwol Bridge Station are three times greater than the flow in 2013 (Extremely High Flow). In addition, it was assumed that there will be three conditions for Palgoe Bridge Station in the future including: (1) the future flow equals to the flow in 2013 (Low Flow); (2) the future flow is double the flow in 2013 (High Flow); and, (3) the future flow is triple the flow in 2013 (Extremely High Flow). According to the data collected from HRFCO, a riverbed expansion strategy was decided in order to prevent the impact of flood event. The cross sections of the rivers were expanded to the maximum width and depth but the stable flow was still ensured. In this case, the average expanded depth is approximately 4 m, such as cross sections 48.7 and 31 [Fig. 4, Fig. 7(a) and 7(b)].

Fig. 7.

Cross section data before and after widening at the upstream and downstream of the lower reach of Namhan River (AMSL: above mean sea level).

There are six scenarios based on the combination between future climate change conditions and the adaptation strategy (Table 2).

Table 2.

Future scenario description

3.1 Water Surface Elevations

According to the survey, the minimum ground surface ele-vation of the power plant is approximately 197 m, two-meter higher than the elevation of the road elevation (195 m). The cross section 42.5, located near the power plant, was chosen to analyze. Fig. 8 shows the simulated maximum water sur-face elevations from 2011 to 2014. The lowest elevation is in 2014 (black line) and the highest is in 2013 (yellow line). The historical maximum water elevations are below the bank sta-tion (road). The highest elevation in 2013 is approximately 193.3 m, and the lowest elevation in 2014 is 186.9 m. The depths in 2011 and 2012 are 191.7 m and 189.3 m, respectively. The maximum velocity is ranged from 1.8 m·s^-1 (2014) to 3.4 m·s^-1 (2013).

Fig. 8.

Maximum surface water elevation from 2011 to 2014.

The results shows that the maximum water surface eleva-tion (Max WSE) of Sce. 3 is extremely higher than the power plant ground elevation (PPGE). After widening the riverbed, Max WSE Sce. 6 is slightly greater than the PPGE. The flood depths under most of scenarios are extremely high compared to the road elevation (RE) except Sce. 4 (Fig. 9). Although the water flow of Dong River is extremely high, the risk will be low if the flow of the upper Namhan River is low (Sce. 4). Other scenarios shows that the extremely high water flows in both upper reaches (i.e. upper Namhan and Dong River) can cause the flood. Therefore, the higher flow of the upper Nam-han River affects the downstream reach more significantly than Dong River.

Fig. 9.

Simulated water surface elevations following the future climate change scenarios.

The flood depths under Sce. 1, Sce. 4 and Sce. 5 do not affect the power plant. Sce. 2 and Sce. 3 affect the power plant with the maximum depth are 0.43 m and 1.52 m above the PPGE respectively. Despite widening the riverbed, Sce. 6 still impacts on the PPGE with the maximum depth is 0.34m (Fig. 9 and Table 3).

Table 3.

Inundation depths under six scenarios (m)

In the comparison between Sce. 1 and Sce. 4, only flood depths under Sce. 1 affects the RE. Therefore, the riverbed wi-dening strategy is highly effective for this case (low flow of the upper Namhan River). For the second case (high flow of the upper Namhan River), the widening strategy is useful for preventing the flood impact on the PPGE, with the maximum flood depth decreases from 0.43 m in Sce. 2 to 0.00 m in Sce. 5; however, it is not effective for the RE. The widening stra-tegy may reduce the flood depths but may not be helpful for both PPGE and RE in the third case (extremely high flow of the upper Namhan River).

3.2 Velocities

A small part of lower Namhan River, from cross-section 43 to cross-section 40, was chosen to compare the velocities among the scenarios. It can be seen that the maximum velocities of the river will decrease if the riverbed is widened. The velocity under Sce. 4 is lower than Sce. 1; the velocity under Sce. 5 is lower than Sce. 2; and, velocity under Sce. 6 is lower than Sce. 3 (Fig. 10). The very high velocities cause river bank ero-sion, the riverbed widening strategy therefore help to reduce the risk of bank erosion.

Fig. 10.

Maximum velocities in the part of river near power plant.