The Chi River basin in Thailand experiences regular flooding downstream of two major dams, which causes damage to infrastructure and impacts human lives. To help the Hydro Informatics Institute (HII) address this challenge, DHI updated and improved HII’s real-time flood forecasting system for north-eastern Thailand. By applying Model Predictive Control (MPC), DHI was able to decrease downstream discharge peaks significantly.

Download the case story flyer

Challenge

The Chi River is one of the major river basins flowing through Thailand. Because of today’s changing climatic conditions, flooding happens frequently, which has a great effect on economy, human life and properties in the area. Therefore, finding efficient ways of managing floods to reduce impacts is an important priority at national as well as regional levels. As such, HII needed insights of real-time control to optimise discharges from EGAT’s (Electricity Generating Authority of Thailand) and RID’s (Royal Irrigation Department) two major dams, Ubonrat and Lampao, in order to provide decision support to the National Water Resource Committee of Thailand.

Solution

To manage real-time control of releases from the two dams located upstream of the Chi River system, DHI applied a MPC framework. The method combines a fast-solvable optimisation model with evaluation and realignment by a hydrodynamic flood model created in MIKE HYDRO River. Using MIKE OPERATIONS to run the hydrodynamic flood model automatically, dam releases and downstream discharges can be controlled in real-time.

The MPC framework optimises the release of the combined water from the two dams and thereby improves the flow of water through the river systems. The optimisation is based on a linear surrogate model of the Chi River network that is calibrated against a full-blown hydrodynamic model, solving the one-dimensional flow equations. In real-time mode, the full-blown hydrodynamic model is used to initialise each optimisation run and simulate the MPC derived releases.

The optimisations take the expected future river system state into account when deciding on the best operation. Control actions for the two dams are optimised over a 7-day forecast horizon. Every day a new forecast is available and optimal controls are updated based on the most recent system information. This principle is called ‘rolling horizon’. By using the MPC framework, the downstream discharge peaks were reduced with more than 300 m3/s during a medium inflow year when releases from the Ubonrat and Lampao dams were controlled in combination.

This research project—made possible with support from EGAT and RID—tests the MPC approach and does not include factors such as side flow from other small rivers and other impact.

Results

Real-time optimisation of multi-reservoir systems

Decreasing flood peaks significantly

Calculation of optimal releases from reservoirs in 20-30 seconds

‘Collaborating since 2012, DHI has supported HII with cutting-edge technology and capacity building of the real-time flood forecasting system. As a result of this partnership, HII has become a national expert within flood forecasting. Thanks to DHI for the support on this excellent outcome.’

Dr Piyamarn Sisomphon
Head of Hydroinformatics Modeling System Section
Hydro Informatics Institute

About the client

The Hydro Informatics Institute (HII) is a public organisation that supports effective water management by focusing on research and development of advanced tools and technologies for managing water resources in Thailand.

Software used

Case story - Highlights flyer

Let us help

Want to take the next step in solving your challenge?