Floods in India to be Avoided Using Fuzzy Logic

IF A washing machine can figure out how dirty the laundry is, then it can calculate how long it should wash, how much detergent to add and how much water is needed. This is called the fuzzy logic technology and it works on a simple "if-then" principle. At present machines like air conditioners or microwave ovens have fuzzy logic technology operating them. Researchers from India have now suggested using fuzzy logic to improve dam operations.

Experts attribute unprecedented floods, like the one in Orissa last year, to a sudden release of water from reservoirs (see 'The right level', Down To Earth, October 16-31, 2008). They blame this on lack of appropriate operational policies for multi-purpose dams. During monsoons, water level of the rivers is high and release of excess water from dams exacerbates the problem.

A team of researchers suggested a system to operate dams that uses fuzzy logic technology with artificial intelligence which works like our nervous system. It converts various signals like water level, expected inflow and expected demand into a single input. The technology then takes the input as "if" and predicts how much water to release (output) as "then".

The team applied its model to the reservoir across Ramganga river, a tributary of the Ganga. With a storage capacity of 2,448 million cubic metre, the reservoir is used for power generation, drinking water supply to Delhi, irrigation and flood control downstream during the monsoon.

According to present operational procedures, water released from the reservoir depends upon its level in the reservoir and time. Reservoir level from June to October should not go beyond 361.95 m. The level in the month of October should not rise above 362.86 m and the level from November to May should not fall below 316.99 m.

The amount of water allocated for power generation is also supposed to cover irrigation needs. Therefore, at times of water crisis, the entire water released through the powerhouse goes into hydropower generation and irrigation demand is considered zero. The agricultural sector suffers.

"We need a procedure which can determine releases against different demands using data on current state of the reservoir," said Rama Devi Mehta, a scientist at the Roorki-based National Institute of Hydrology and the lead author of the paper. For the model the researchers used four sets of data from previous years: reservoir levels (during monsoon and dry period) water inflow and demands. Using the new technology the model converted these inputs into a single output in the form of amount of water to be released from reservoir. The results were consistent with the actual release value. The study was published in the February 2009 issue of Water Resource Management.

"Theoretically, models based on fuzzy logic will show improved results. But in reality the changes will be minor and not have much practical use," said Rajib Kumar Bhattacharjya, assistant professor at the Indian Institute of Technology, Guwahati. The team is hopeful that the new model will manage dam operations despite uncertainty in water inflow and demand, which may worsen with climate change, allowing operators to take quick decisions.

Source: Down To Earth, March 2009