PREDICTIONS OF LiFePO4 BATTERY DISCHARGING TEMPERATURE USING AUTOREGRESSIVE INTEGRATED MOVING AVERAGE AND HOLT'S DOUBLE EXPONENTIAL SMOOTHING

• Main Author: Revano Mege, Christio
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43197
• Institution: Institut Teknologi Bandung

Alternative energies become the solution in the middle of global climate change due to CO2 emissions. However, alternative energies such as wind power and solar cell are intermittent that require energy storage such as batteries. Lithium ion batteries are in the frontline as energy storage for electricity produced by alternative energy. One of the benefits from this kind of energy storage is the high energy density. However, performances of lithium cells in a module highly depend on operating temperature between 00C – 400C. Temperature increasing that exceeds the limit can significantly reduce battery performances Therefore, temperature predictions are required in order to know when battery could exceed the 400C limit. In this research the effect of temperature rising on battery performances such as depth of discharge and electricity generation efficiency had been conducted. After that temperature data acquired from data acquisition process is used as training data and test data to predict temperature using Autoregressive Integrated Moving Average (ARIMA) dan Holt’s Double Exponential Smoothing (DES Holt). Then the models evaluated using Root Mean Square Error (RMSE) dan Mean Absolute Deviation (MAD). Models are said to have good predictions if RMSE and MAD values are smaller than half of the standard deviation of measurement data. The results show that at 0.7C, cells temperatures inside module reached 35.40C, rising about 4.90C. The temperature rising is greater than single cell that rose 30C to 29.70C. Then at 1.4C the module temperature reached 38.60C rising about 8.30C. Single cell temperature at 1.4C reached 35.70C, rising 9.40C. At 2.1C, single cell reached 45.10C with temperature increasing of 18.50C. Module temperature at 2.1C reached 480C with 190C increasing. Efficiency of electricity generation of single cell at 0.7C is 92.58%. The efficiency reduced to 84.48% at 1.4C rate. Then at 2.1C rate, single cell only capable of generated energy about 23.3Wh with 76.82% efficiency. Module at 0.7C has electricity generation efficiency of 91.58%. At 1.4C, the efficiency reduced to 83.38%. At 2.1C rate, the efficiency was getting smaller to 72.9%. Predictions conducted show that both ARIMA and DES Holt can predict the temperature rising in single cell. In module temperature predictions, training data was taken from one cell only to predict the rest of the cells. At 0.7C, both methods can predict six out of eight cells well. Five out of eight cells could also be predicted well using ARIMA and DES Holt at 1.4C. However at 2.1C, just four cells could be predicted well. The predictions accuracy of ARIMA and DES Holt decreased when the temperature uniformity in module decreased as the C-rate increased.