IO, Surabaya Industrial waste generally contains hazardous substances that can pollute the environment and have an impact on human health. In view of this, three students of the Sepuluh Nopember Institute of Technology (ITS) who are members of a team called UCiFi designed an artificial intelligence modeling to facilitate more efficient and environmentally-friendly waste processing in the oil and gas industry.
Fitria Kusumaningrum, Citra Annisaa Nurul Ain, Nuzulul Syaqawati Azzahra are undergraduate students of the Department of Engineering Physics, Faculty of Industrial Technology and Systems Engineering at ITS. They innovated the use of artificial intelligence in the form of an Artificial Neural Network (ANN) to predict the accuracy of Chemical Oxygen Demand (COD) levels contained in oil and gas industrial waste.
Ulul, the nickname of Nuzulul Syaqawati Azzahra, the team leader, said that the waste produced by the oil and gas industry is in dissolved in water. Oil exploration results in pollutants, including H2S (Hydrogen Sulfide), oils and fats, NH3 (Ammonium), and COD. “The levels of these pollutants exceed quality standards, so the levels of pollutants must be reduced first before they are returned to nature,” she said in a release received by the Independent Observer, Friday (21/5/2021).
One way to reduce pollutant levels is to use a polishing unit. The input data on the polishing unit is in the form of high COD levels and several other parameters, such as pH, temperature, NO3, PO3, MLSS, TSS, and SVM. “After the data are processed in the black box polishing unit, the output of COD will be lower. To predict the COD levels, an accurate predictor is needed, namely, ANN,” said Fitria.
She added that apart from the ANN model as a predictor, it is necessary to apply optimization techniques to support minimal COD results. One of the most widely used optimization techniques is Genetic Algorithm (GA) optimization. The UCiFi team believes that the use of ANN + GA has better prediction accuracy results than other artificial intelligence models, such as ANFIS. “Although both ANN and ANFIS have very many data input capabilities, ANN has a hidden layer so that data predictions will be much more accurate.”
When asked about the constraints, Fitria revealed that initially, her team did not understand information processing techniques using ANN+GA, so they needed more time to research this matter. “However, with persistence and cooperation between team members, we were able to complete the papers that we needed to submit,” she said.
The hard work of the UCiFi team has resulted in a proud achievement. Guided by Totok Ruki Biyanto ST MT Ph.D., the UCiFi paper entitled “Artificial Intelligence in Oil and Gas Wastewater Treatment” won second place in the 2021 Petrolida Paper Competition, at the end of last month. In the competition held by the ITS Student Chapter of Petroleum Engineer Society, the UCiFi team outperformed papers of the other nine teams.
In the future, Fitria hopes that the UCiFi team can continue to work together to participate in other competitions. “By utilizing artificial intelligence for waste treatment, we want this innovation to be a breakthrough to improve the performance of waste treatment in the oil and gas industry,” she said hopefully. (est)