Abstract: To address the engineering pain points of 'large-model parsing timeout and poor heterogeneous format adaptation' in the lightweight scheduling of BIM models for pumped storage power stations and hydropower stations on the Luantu Collaborative Platform, this study proposes a dynamic scheduling architecture driven by a 3D quantification table. By constructing a three-dimensional service carrying capacity quantification system of 'format - size – component', combined with the Current carrying Capacity (RC) algorithm, the precise matching of scheduling services and multi-lightweight engines is achieved. Simulation experiments based on 100 pumped storage model samples from the Luantu Collaborative Platform show that when processing hydropower station models, the total parsing time is reduced by 51% compared with fixed 3-task scheduling, the average resource utilization rate is increased by 119%, and the maximum RC value is ≤ 1.0. In the next step, an automatic component count extraction algorithm for BIM models will be developed. Engineering verification will be carried out in the Jingshuitan Pumped Storage Project and the Philippines Pakil Pumped Storage Project to provide a implementable scheduling solution for BIM collaboration in water conservancy and hydropower engineering.