Against the backdrop of urban renewal, the transformation and functional enhancement of Existing Urban Industrial Areas (EUIAs) play a crucial role. Focusing on the rational development of underground space in EUIAs, this study explores forecasting methods for the development demand of such underground space, aiming to alleviate the contradiction between the protection of industrial heritage and intensive land use in EUIAs. This paper systematically sorts out the forecasting methods for the scale demand of underground space. Firstly, through a literature review, two major categories of factors influencing underground space demand—driving factors and conditional factors—are summarized, and an indicator system consisting of 23 indicators is constructed. On this basis, the modified Delphi method is employed to screen 7 dominant indicators, including the protection value of industrial heritage, the spatial distribution of industrial heritage, existing underground space, development functions, rail transit, spatial location, and surrounding supporting facilities. Based on the matrix of industrial heritage protection levels and land use nature, the development potential of underground space is evaluated, and a demand level correction model is introduced. Demand intensity is quantified through expert experience-based assignment with reference to typical domestic cases, thereby establishing a demand forecasting model for the underground space scale in EUIAs. Finally, the model is applied to the Saint-Gobain Industrial Area. Through the analysis of its industrial heritage value assessment, land use planning, and location characteristics, the areas with demand for underground space are delineated and their levels are corrected, forecasting a total underground space demand of 224,600–454,600 m². The research results provide a theoretical basis and methodological support for the underground space planning of EUIAs, and offer references for the development practice of similar regions.

Metro-connected underground commercial spaces present complex fire safety challenges due to enclosed layouts, dense pedestrian flows, shared evacuation routes, and diverse combustible materials. This study conducts a scenario-based fire evacuation assessment of a typical metro-connected underground commercial space in Nanjing, China, using coupled FDS–Pathfinder simulations. Three fuel scenarios were considered: polyurethane (PU), polyvinyl chloride (PVC), and a combined PU–PVC fuel scenario. Available safe egress time (ASET) was evaluated using visibility, CO concentration, and smoke temperature thresholds, while smoke-affected evacuation performance was assessed using a visibility-based walking speed attenuation model. The results show that visibility reached the critical threshold earlier than CO concentration and smoke temperature in all scenarios. The PU-only scenario produced the fastest smoke spread and the shortest ASET, with Stair 2 reaching the visibility threshold at 244 s. Under smoke-affected conditions, the total evacuation time increased from 364 s to 378 s. More importantly, a local ASET/RSET mismatch was identified at Stair 2, where 62 occupants remained near the node after the local ASET threshold was reached and local clearance was completed at approximately 250 s. By contrast, Stairs 4 and 7 retained safety margins of 133 s and 113 s, respectively. These findings indicate that evacuation safety in the simulated case was controlled not only by smoke-induced walking speed reduction, but also by route convergence, local stair capacity, and uneven exit utilization. The study provides case-based evidence for node-level fire evacuation assessment and evacuation management in metro-connected underground commercial spaces.