Resolving the conflict between the need for electricity generation upstream and the water supply for economic development downstream in the Ca River Basin, Vietnam

The Ca River Basin is located in the North Central region, with geographical coordinates from 18° 15′ to 20° 10′30″ north latitude; 103° 45′20″ to 105° 15′20″ east longitude. The starting point of the basin is located at coordinates 20° 10′30″ north latitude and 103° 45′20″ east longitude. The bay outlet is located at 18° 45′27″ north latitude and 105° 46′40″ east longitude. The Ca River Basin is located in two countries; the upstream part is located on the land of Phong Sa Van and Sam Nua provinces of the Lao People's Democratic Republic. In Vietnam, the river basin is in the territory of three areas: Thanh Hoa, Nghe An, and Ha Tinh provinces.

The Ca River is crucial in supplying water for the socio-economic development activities of Nghe An and Ha Tinh provinces. Annually, the total water demand of all sectors in the basin is about 2.5 billion m³, of which the most significant water demand is concentrated in the dry season months from February to July; the most critical water user is still agricultural production, accounting for about 70% of total water demand.

To meet the current water demand in the Ca River Basin, 3,472 works have been built, including 1,588 reservoirs, 600 dams, 1,266 pumping stations, 18 main sluices, and 13,343 km of canals to irrigate a 170,030 ha area. Significant works are taking advantage of synthesis on the mainstream; the tributaries are irrigation lakes, hydroelectric steps, and irrigation works systems to bring water along the river. These works regulate water sources, generate electricity, fight floods, and supply water to vital economic regions downstream. These are significant works closely related to water supply and flood control activities. Therefore, the topic will focus on researching and proposing a plan to coordinate the operation of irrigation reservoirs, hydropower plants, and irrigation systems to exploit water sources on the Ca River mainstream to meet water needs and effectively and sustainably combat floods downstream.

Areas benefiting from the Ca River system are located mainly in the downstream areas of Nghe An, and Ha Tinh provinces. These include a series of large water intake systems such as the 45,700 ha Do Luong irrigation system, the Nam Hung Nghi system, 30,200 ha and the Nghen River, 27,000 ha.

There is a high demand for agricultural water, especially in January, May, and June, which significantly affects the water collection efficiency of irrigation works in the downstream area.

The flow regime of the Ca River system has changed significantly. Indeed, the reservoir system can help push the water flow downstream during the dry season while limiting the flow during the flood season. The lowering of the Ca River water level has made collecting downstream water in the dry season difficult. As analyzed above, the influence of upstream reservoirs on the downstream flow regime (at the Dua hydrological station) is relatively significant compared to that during the period before the construction of upstream irrigation works. However, the flow in the dry season was sometimes improved thanks to additional discharge reservoirs built to help meet the downstream demand. Sometimes, this is due to lower flow to the Dua hydrological station than the average of many years before the upstream works.

According to the Dua hydrology station statistics, during the period from 2009 to 2015, there was a massive fluctuation in dry seasons, especially from 2010 onwards. In particular, in 2009, the water volume was at its lowest in April and May with a flow rate of 88.4 m³/s, which was 16.9 m³/s higher than the flow with a frequency of 85%. During a 6-year period starting from 2010 to 2015, only in 2012 was the water flow relatively good, with the lowest flow rate of 114 m³/s in late April and early May. In 2010, 2011, 2013, and 2015, the flow at Dua Station in the driest periods was extremely low. In 2010, during the driest period starting from mid-February to mid-May, at the driest time, the flow was only 51 m³/s, which was 20.5 m³/s lower than the flow with the frequency of 85%. In 2011, the lowest flow was only 50.4 m³/s during the driest February–April period. In 2013, the lowest flow was only 48.1 m³/s in the driest month – April.

In the downstream area of the Ca River Basin, there are two large irrigation systems, namely Do Luong and Nam Nghi. These two systems operate widely across various districts and cities, potentially leading to water shortage at key irrigation works and significantly affecting regional production and daily activities. The drought area in recent years of the Ca River downstream is as follows: 16,101 ha in 2014, 18,027 ha in 2015, 18,262 ha in 2016, and 17,040 ha in 2017. Of which, the irrigated area from Do Luong dam is 5,000-6,000 ha, the irrigated area from Nam Dan drain is about 5,000–7,000 ha. The total estimated damage cost caused by drought and saltwater intrusion along the Ca River in 4 years from 2014 to 2017 is 35,317 million VND. The average annual loss caused by drought is 88,293 million VND.

Currently, the total discharge from the Ban Ve reservoir to the downstream areas has not experienced a significant change over the years. However, from 2014 until now, the water level at Nam Dam drain, under the regulation of the Ban Mong reservoir, has always been lower than the designed water level. During the summer–autumn crop, the water level at the drain is about 0.29 to 0.4 m (the design water level is 1.15 m). In this regard, the water level of the Lam River remains too low, reducing the water supply to the lower end of the systems, such as Nghi Loc, Hung Nguyen, and Vinh City of Nghe An province. However, Nam Dan drain has always opened its two gates to obtain the maximum water volume.

Furthermore, the discharge of hydropower reservoirs is on an hourly basis and in accordance with the market's electricity price requirements, making it difficult to collect water from the downstream. In fact, hydropower plants often generate electricity only during peak hours to obtain strong economic benefits. On average, electricity is generated for 14–15 h/day. Therefore, although the discharge to the downstream is guaranteed by the requirements of the hydropower plant operation procedure, the limitation of power generation as mentioned above would cause difficulties for the water collection of the agricultural irrigation systems in the downstream areas.

Ensuring sufficient water supply in areas significantly affected by tides, such as the Nam Dan sluice, is even more difficult. The river water level remains very low most of the time, and a high level, if any, often lasts only several hours. Therefore, Nam Dan drain has been facing many difficulties in collecting water.

This study uses a model simulating the optimal operation of four hydroelectric reservoirs of the Ca River upstream to optimize power generation and water regulation in an attempt to ensure sufficient water supply for the downstream area and for four hydropower reservoirs of the Ca River upstream, including Ban Ve, Khe Bo, Chi Khe, and Ban Mong reservoirs.

This study develops different water supply scenarios for the Ca River downstream area based on the calculation of water demand for irrigation activities, domestic and industrial use, in consideration of the water collection conditions of the irrigation works at the downstream area as well as saltwater intrusion prevention at the Nam Dan drain. Five scenarios have been developed with the elevation of water determined at the water elevation control points of Nam Dan drain.

In recent years, reservoir operations processing has become increasingly important (Wada et al. 2017), providing water for domestic demand, economic sectors, and reducing flood risk (Billington & Jackson 2017). In addition, changing climate extremes and social demands amplify and reshape uncertain stressors, ultimately altering decision-makers' preferences and risk perception (Hall et al. 2014; AghaKouchak et al. 2015; Mallakpour et al. 2019).

An optimal approach in multi-reservoir operation in river basins to ensure optimal benefits for water users is one of the leading approaches in planning and managing water resources (Molle & Hoanh 2011; 2030 Water Resources Group 2017).

Extensive literature exists on applying optimization techniques to operate hydropower reservoir systems. Studies vary in several ways, including the objective optimized, time horizon for optimization (long- vs. short-term), system size and configuration, and the representation of uncertainty. These factors determine the optimization techniques most suitable for each case. This review focuses on studies where hydropower generation is the primary objective to be optimized. This focus still includes multi-purpose reservoirs whose primary purpose is water supply and hydropower, where hard constraints are imposed to satisfy non-hydropower requirements.

Optimal approaches include multi-objective optimization and single-objective optimization. Performing the set of optimization problems usually includes linear programming (LP) (Zessler & Shamir 1989; Ribeiro et al. 2012; van der Vat 2016), nonlinear programming (NLP) (Yeh 1985), and dynamic programming (DP) (Opricović et al. 1991; Bertsekas 1994; Jaafar 2014). These approaches apply to single-reservoir or multi-purpose reservoirs where the pools ensure power generation and water supply for downstream economic development and flood prevention during the rainy season (Wang et al. 2011; Chou & Wu 2015; Yang et al. 2015).

In Vietnam, there are several studies on optimal reservoir operation in river basins (Rockwood 1968; Promwungkwa et al. 2019), such as the Red River Basin (Ngo 2007; Nguyen 2023). In the Ca River Basin study, the optimization problem for multi-reservoir operation was for reservoirs at Ban Ve, Ban Mong, Khe Bo, and Chi Khe. The problem's objective function will be the reservoir's power generation benefit function. The water supply requirements for economic development in the lower Ca River are calculated according to the water demand and use scenarios and included in the problem as a minimum flow constraint at downstream control points. In addition to the rules on water supply requirements, there are many constraints related to the operation mechanism of each reservoir, the balance of confluence and distributive points, flow balance on the system, requirements on flow (or water level elevation) at control points, and flow to ensure the environment. The objective function is to optimize the power generation of reservoirs upstream of the Ca River. The problem's constraints are constraints on water balance in reservoirs, flow balance, and water supply requirements at downstream control points over time.

This solution turns from the multi-objective optimization problem of electricity generation and water supply for economic development in the lowland area to a single-objective problem of power generation and water supply for economic growth in the downstream area constraints in the optimization problem. Then the solution to the problem is the discharge at the reservoirs corresponding to each calculation period, the amount of electricity generated for each reservoir for each calculation period, the water flow (or the water level if there is a relationship between the flow rate Q m³/s with H water level) at the construction sites (ex. pumping station) or the control points where the flow needs to be controlled.

Building the problem of coordinated operation and discharge of water between four reservoirs so that both power generation benefits are optimized and water supply regulation is ensured for downstream economic development is a top priority. This research presented the establishment of a multi-objective multi-reservoir coordinated operation model according to the model of an NLP problem and used the general algebraic modeling system (GAMS) (Rosenthal) to solve the problem.

The water use scenarios are analyzed considering the current state of socio-economic development and the economic development orientation of the downstream regions, including Nghe An and Ha Tinh provinces.

Based on the hydraulic simulation calculation model, based on the requirement of taking water for agricultural irrigation at the sluices operated by the irrigation works operators Nam Nghe An and Nam Hung Nghi irrigation system five scenarios (see Table 1) for calculating water demand in the downstream area are as follows:

• Scenario 1: The water level at Nam Dan drain (H_{Nam Dan} = 1.15 m) will meet the water demand according to the design water level at irrigation works from Nam Dan drain and below.

• Scenario 2: The water level at Nam Dan drain (H_{Nam Dan} = 0.83 m) corresponds to the water demand calculation according to the regulations operation process of the irrigation system of Nghe An province.

• Scenario 3: The water level at Nam Dan drain (H_{Nam Dan} = 0.7 m) corresponds to the alternate agricultural irrigation solution.

• Scenario 4: Water level at Nam Dan drain (H_{Nam Dan} = 0.6 m) corresponds to alternate irrigation water demand.

• Scenario 5: The water level at Nam Dan drain (H_{Nam Dan} = 0.4 m) is calculated based on the alternate irrigation option and requires the intervention of other additional sources from ponds and reservoirs.

Based on the water level scenarios at the Nam Dan drain, the study used hydraulic data and the correlation between flow and water level to calculate and determine the flows to be maintained at the Dua hydrological station (see Table 2).

From April to July 19 in the Nghe An region, large amounts of water are needed to serve the winter–spring crop. Next is land preparation, sowing the summer–autumn crop. With the summer characteristics of high temperature and strong southwesterly winds, evaporation causes significant water losses. However, in the operating process of this period, the discharge from Ban Ve reservoir is only from 50 to 125 m³/s, about 50 m³/s lower than the requirement for water intake.

Constraints:

S_Max,t,i is the maximum capacity of the lake i at time t, usually in the volumetric model corresponding to storage of operation level or storage of flood control level (m³); S_t,i is the recovery volume i in period t (m³)

Q_min is the minimum supply value for water user j in time t.

Remark in model

The goal of the optimal model

• (i) Establishing a model to determine discharge at Ban Ve, Khe Bo, Chi Khe, and Ban Mong reservoirs at different times to achieve optimal benefits in terms of power generation and, at the same time, meet the discharge volume minimum at the Dua hydrology station, or water level at Nam Dan drain. In the study, the minimum discharge at the Dua station corresponded to the minimum water level at Nam Dan drain according to a mathematical function relationship;

• (ii) Determining the feasible scenarios for the combined operation of four hydropower reservoirs corresponding to five scenarios of minimum flow at Dua station, in which the hydrological boundary of the discharge to the upstream reservoirs corresponds to the frequency of the water flow rate P = 85%;

• (iii) Based on possible scenarios in the operation of four reservoirs, Ban Ve, Khe Bo, Chi Khe, and Ban Mong choose the scenario that brings the highest efficiency in power generation. The diagram of the hydrological boundary and calculation network in the GAMS model of the four reservoirs, Ban Ve, Khe Bo, Chi Khe, and Ban Mong.

In this study, the calculated hydrological margin selected corresponding to the frequency P = 85% and the chosen year of calculation was as follows:

The initial assumption for reservoirs at the beginning of operation was that they were full of water. Because it is calculated for a dry season, an assumption is that the water of all reservoirs is full or that reservoirs are responsible for storing enough water to serve the dry season.

With the calculation scheme, the coordination of water discharge at two reservoirs upstream of the flow, Ban Ve and Ban Mong reservoir, is essential. The computational network has parallel regulation coordination and serial regulation between reservoirs.

When calculating the corresponding five scenarios of water use downstream, it will be in order from high water demand to lower demand (Q_min at the Dua control point will be from high to lower).

The results in Figures 4 and 5 correspond to scenario 1, in which both upstream reservoirs cannot regulate the entire dry season, that is, lack of water, due to water discharge to the dead water level. Ban Ve reservoir will be discharged to the finished water level on 3 June, and the rest of the dry season from 3 June to 19 July will no longer be able to operate and regulate. Ban Mong reservoir also discharged to the dead water level until 5 July, the period from 6 July to 19 July, and was no longer able to hold water and generate electricity.

With scenario 2, when both main regulating reservoirs were discharged by 5 and 6 July, both reservoirs ran out of water, while the additional flow from the central areas and Khe Bo reservoir were not enough to meet the minimum flow amount at Dua control point.

Therefore, scenario 3 meets the requirements of inter-reservoir coordination upstream of the river and the water requirements downstream (flow control point at Dua station). To Scenarios 4 and 5 are viable and easy to operate synergistically between reservoirs. Calculation results of coordinated discharge between reservoirs in each period are shown in Table 3. The results of coordinated discharge between reservoirs help build a basis for making an optimal coordination mechanism between reservoirs. The result has been checked with Q_min at Dua station according to scenarios 1–3 as shown in Table 4.

In the early stages of the dry season, due to the low discharge required downstream, the discharge regulation reservoirs with low discharge capacity of Ban Ve reservoir discharge from 60 to 70 m³/s, and Ban Mong reservoir releases with a discharge of 23–70 m³/s, it is explained that although they want to remove more to generate electricity, the reservoirs need to store water for the later stage. In the periods after 1 April every year, the water demand downstream is significant, so the discharge volume in the corresponding period also increases. However, the more substantial discharge is still at Ban Ve reservoir, so the role of Ban Ve reservoir in regulating water is still high. Ban Ve reservoir was essential, determining the coordination mechanism between the four reservoirs and those with the largest water reserve, generating the most significant amount of electricity. For the other two reservoirs, Khe Bo and Chi Khe, the roles of water regulations at Dua are a bit different. Only the Khe Bo reservoir is operated daily. The capacity is also tiny so it will depend entirely on the water flow discharge from Ban Ve reservoir and increased clearance in the middle area. Chi Khe reservoir cannot regulate the reservoir; it only creates a water column to generate electricity so that inflow will equal water discharge during the day. The value of the objective function of power generation of the four reservoirs was calculated based on the amount of electricity generated and the electricity price set by Viet Nam Electricity. In this study, it was taken according to the average price. And the benefit value of the four reservoirs will be 1,009.8 billion VND in which Ban Ve, Khe Bo, Chi Khe, and Ban Mong, respectively, were 363.48 billion VND; 213.95 billion VND; 315.23 billion VND; 117.14 billion VND. In fact, the benefit of generation can be higher because electricity can be used at a higher price, especially during rush hours.

This research focused on researching and building the multi-objective optimization problem, power generation, water supply for irrigation, and water for socio-economic development in downstream areas, into a single-objective optimization problem of electricity generation for easy solving with the GAMS tool. The water supply targets in downstream areas for economic development and irrigation were calculated as water use scenarios and entered into the problem model as minimum discharge constraints. In this study, the minimum discharge includes the discharge to maintain the ecological flow of the environment. The study also presented a straightforward approach to building five scenarios for water uses downstream of the Ca River Basin, taking the Dua hydrology point as the flow control point. The hydropower generation optimization calculation showed that feasible scenario 3 was suitable for the operation of upstream reservoirs. The corresponding water level at Nam Dan drain was 0.7 m, based on solving the hydropower generation optimization problem to determine the coordinated discharge between reservoirs as a basis for proposing inter-reservoir operating mechanisms upstream of the Ca River. The optimal approach to finding the inter-reservoir coordination mechanism corresponding to typical hydrological years was suggested. The results of the pseudo-optimal study can be used to propose joint operating mechanisms for the reservoirs upstream of the Ca River to bring about the highest economic benefits in the integrated use of water resources. The results of the research could be helpful for irrigation system management in Nghe An and Ha Tinh provinces to answer how and when to take and pump water into the irrigation system. This result also demonstrates that a new inter-reservoir operating process is needed upstream of the Ca River.

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

The authors declare there is no conflict.