Speaker Detail

ARIMA Model for Predicting Reservoir Pressure & Temperature in the Geothermal Power plant

With the increasing number of people, the amount of electricity demand will increase. Electricity does a commodity that cannot be separated from human lives. For example, the commuter line as the main transportation in Indonesia, especially in Jakarta, which connects the satellite cities like Bogor, Tangerang, and Bekasi, uses electricity to drive the motor inside the train. Every day more than 2 million people moving by this electric-powered vehicle. Even after the people arrive at the destination station, they will book online transportation like Gojek or Grab via the electric-powered mobile phone. Every time and everywhere we need electricity. The vacuum of electricity will affect human activities a lot, like the blackout accident that was happened on August 4th, 2019.

To provide the demand of electricity, people build power plants. There are many different types of power plants. In Indonesia, there are hydroelectric, geothermal, coal, and solar power plants. Each of them has its characteristics. But every type has its constraint, as for hydroelectric power plants is the drought season, coal availability for the coal power plants, and night period for the solar power plants. The case for geothermal is a bit different, as the source to rotate the turbine is from the heat inside the earth’s crust, which remains constant. But there is a challenge to maintain the temperature for longer life cycles by reinjecting fluid back.

Understanding whether the maintained process is done in the right way must be proven. The best way to check is by predicting the process variable, which is related to the heat condition. If the process variables have a downtrend, then the reinjection process must be adjusted again. If the opposite occurs or the trend is steady, then it means the process is done properly.

In this case, I will predict the reservoir pressure & temperature trend with the well-known method for predicting the stock market price— the autoregressive integrated moving average (ARIMA), which predicts the time series trend. With the same approach, I will test this method to the geothermal power plant. For the next phase, the machine learning method will create a model that describes the relationship between the input and output of the geothermal power plant. Power generation is the output, and the process variable is the input. The results will be shown in Esri's Geospatial Business Intelligence software.