Document Type : Research Paper
Authors
1
Department of Geography and Urban Planning, University of Tabriz
2
Department of urban and regional planning, Faculty of planning and environmental sciences, University of Tabriz, Tabriz, Iran.
10.22034/gp.2025.68673.3443
Abstract
This study investigates temporal trends and projects future electricity consumption patterns across diverse urban sectors (residential, industrial, agricultural, public, street lighting, and miscellaneous) in Tabriz, a city characterized by a cold semi-arid climate.
Biannual electricity consumption data, spanning 2008–2023, were sourced from the Tabriz Electricity Distribution Company and disaggregated by sector. Trend analysis was performed using the Mann-Kendall test, with inter-sectoral relationships evaluated via Spearman’s rank correlation coefficient. Forecasting until 2031 was conducted using a Neural Network Autoregressive (NNAR) model. Time-series analyses incorporated Augmented Dickey-Fuller (ADF) and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) stationarity tests, Seasonal-Trend decomposition using Loess (STL), Box-Cox transformation, and model accuracy assessment through Root Mean Square Error (RMSE), Mean Squared Error (MSE), and Mean Absolute Error (MAE).
All sectors exhibited statistically significant upward trends in electricity consumption. The most pronounced trends were observed in the miscellaneous (τ=0.572) and aggregate (τ=0.570) categories, with substantial increases in residential (τ=0.496) and industrial (τ=0.500) sectors. Spearman’s correlation revealed strong associations between residential, industrial, and miscellaneous sectors with total consumption (ρ≥0.91). Residential consumption displayed marked seasonal variability, with peak demand during July–August and January–February, indicative of climatic influences. The NNAR model demonstrated robust predictive accuracy across all sectors (RMSE<6%).
Despite consistent consumption growth across sectors from 2008 to 2023, projections indicate relative declines by 2031 in residential (2.3%), agricultural (2.14%), and street lighting (5.61%) sectors, potentially attributable to recurrent power outages impacting model inputs. These findings advocate for energy policies in Tabriz prioritizing suppressed demand management, grid resilience enhancement, and tailored seasonal and sectoral strategies. The integration of the NNAR model, Mann-Kendall test, and a data-driven framework for energy policy formulation in semi-arid climates constitutes a novel contribution, offering actionable insights for sustainable urban energy management.
Keywords
Main Subjects