Solar and flexibility꞉ key to Türkiye’s rising cooling challenge

Downloads

Report - Solar and flexibility: key to Türkiye’s rising cooling challenge

Methodology

In this study, Türkiye’s hourly cooling consumption was calculated using a top-down approach. First, NASA population data (with a resolution of approximately 25 km × 25 km) was obtained and then hourly temperature data at 2 meters above ground level was extracted using ERA5 datasets. The temperature values were weighted hourly using population data to reduce the influence of high temperatures in low-population areas on the estimated cooling consumption. Then, cooling degree hours were calculated by subtracting a 22 °C cooling threshold from the hourly temperature values.

Hourly consumption data was obtained through the real-time electricity consumption API provided by EPİAŞ’s Transparency Platform.

To estimate the impact of cooling on total electricity consumption, an autoregressive model was developed using the Python programming language. To prepare the data for modelling, the time series was first transformed into a stationary series using the commonly applied first-differencing method. Stationarity was confirmed via the Augmented Dickey-Fuller (ADF) test. A multiple linear regression model was then constructed using the stationary data series. Electricity consumption follows certain recurring patterns due to the regularity of daily life and often repeats periodically. To capture these patterns, a second differencing was applied to the already differenced consumption data to identify the effects of the previous day. To account for annual consumption growth – while excluding its influence from cooling-related consumption – differences from matching days in the previous year were added to the model as independent variables. For instance, consumption and CDD (cooling degree day) values for Monday, 3 June 2024, were compared to those for Monday, 5 June 2023. All days from all years in the study period were matched to the same weekday of the same week in the previous year and their differences were included in the model.

Moreover, holidays are a significant factor in electricity consumption in Türkiye. To account for drops in consumption on national holidays, each holiday in the study period was identified and included in the model as an independent variable. A similar approach was applied for weekends, and holidays that fell on weekends were incorporated into the model as weekend days.

Finally, to measure the impact of cooling on total electricity consumption, cooling degree days were included in the model as an independent variable.
The final form of the model equation is as follows:

Cooling consumption = β₁ × CDD_differences + β₂ × first_difference_of_consumption + β₃ × non-working_day + β₄ × annual_consumption_difference + ε

In this equation, the β coefficients represent the slope of each corresponding independent variable in the regression model, while ε denotes the sum of the error terms not explained by the model. After running the model, actual cooling-related consumption was calculated on an hourly basis using the coefficient for cooling degree days.

For estimating hourly peak demand, the annual average of hourly consumption – obtained via the real-time consumption API from EPİAŞ’s Transparency Platform – was compared to the actual hourly peak demand for each respective year. Then, using the National Energy Plan’s 2030 and 2035 demand projections, along with the projected share of cooling consumption in total demand, estimates for peak demand in 2030 and 2035 were calculated. The slope coefficient for cooling degree days (CDD) derived from the model was used to estimate the impact of a 1 °C increase in air temperature on peak demand.