| Title | Time Series Analysis for Long-term Monitoring and Forecasting of Subsurface Temperatures for a Campus-scale Geothermal Exchange Field |
|---|---|
| Authors | Shubham Dutt ATTRI, Evan HEEG, James TINJUM, Dante FRATTA, David HART, and Orhun AYDIN |
| Year | 2024 |
| Conference | Stanford Geothermal Workshop |
| Keywords | district-scale geothermal, ground heat exchange, machine learning, exponential smoothing, regression analysis, subsurface temperature predictions |
| Abstract | We investigate the application of autoregressive time-series models in predicting the subsurface temperature of a low-temperature, geothermal heat exchange (GHX) system. We use subsurface temperature data from 2596 152-m-deep boreholes in a 280 m by 360 m, cooling-dominated, district-scale GHX field in the Midwest region of the United States. We monitored the temperature for over seven years via the deployment of fiber-optic distributed temperature sensing (FO-DTS). This study aims to impute a two-year gap in temperature measurements using the first three years of data and test the out-of-sample performance of three forecasting models. We use autoregressive time series forecasting models (including ARIMA and Holt-Winters Triple Exponential Smoothing) to forecast subsurface temperature using previously observed time series patterns. To predict subsurface temperature, we define the forecast model with three exogenous variables—air temperature, humidity, and the energy exchanged for heating and cooling the campus buildings. We observe that the best MSE value among all the models is 0.0100 deg C^2 for a prediction horizon of one month and 0.0665 deg C^2 for a horizon of six months using the Holt-Winters smoothing. Findings showcase a gradual, seasonal rise in subsurface temperature and offer valuable insights for designing more efficient GHX systems, conducting improved energy balance assessments, and creating long-term ground-temperature change models. We demonstrate the potential of autoregressive time-lag models in forecasting geothermal heat exchange. |