Emergency Departments (ED) play a crucial role in the healthcare system, acting as the primary gateway for most hospital admissions. As primary entry points to hospital services, it is essential that EDs receive focused attention to ensure patients experience a smooth and uninterrupted healthcare journey. However, EDs face considerable challenges, with overcrowding being a major issue. Various solutions have been proposed to tackle this challenge, among which forecasting ED arrivals stands out as a foundational approach. By accurately predicting the number of patients arriving at the ED, healthcare providers can better prepare and manage resources, aiming to reduce the impact of crowding effectively.This study advances ED arrivals forecasting by predicting hourly patient arrivals for one-hour ahead, focusing on ESI level forecasts to improve resource allocation decisions. It introduces a dynamic, rolling base method for model training, a notable improvement over the traditional static approach. The research compares the performance of widely used forecasting models with more accurate yet straightforward proposed models. The proposed forecasting framework applies Multiple Linear Regression (MLR) and develops a Hierarchical forecasting approach, with MLR as the forecasting method for top-level and three different top-down reconciliations. Proposed models are compared with some state-of-the-art models. Model accuracy is assessed using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). Among all the models, the proposed model performs better for most of the ESI levels. Following this, the Diebold-Mariano test (DM test) is applied to determine if there is a significant difference in accuracy between forecasting models.