Global transportation systems rely on rail networks as an important infrastructure that drives economic connectivity, reduces carbon emissions, and provides an efficient, cost-effective method of moving goods and passengers across distances with minimal environmental impact. However, railway accidents pose a significant risk to human lives and assets. Therefore, this research focuses on prioritizing zones to allocate budget for safety project investments using a mathematical model. We consider two main factors: economic risk or gross provincial product (GPP) of each zone and accident risk to identify the severity of accidents based on historical accident data, with three levels of budget. Two effective metaheuristics, Simulated Annealing and Multi-Start Simulated Annealing, are proposed to solve the problem. The results identify selected cities as areas with significant economic importance, corresponding to the model's objective. In terms of algorithmic performance, the quality of solutions from both methods is similar. Nevertheless, Simulated Annealing outperformed Multi-Start Simulated Annealing in computational efficiency. This research provides important guidance for improving safety in railway transportation.