Current static evacuation path planning methods are unable to meet the demands of real-time changes in fire scenarios. Therefore, a bi-objective dynamic planning (BODP) method for fire evacuation paths is proposed not only to minimize evacuation time but also reduce the thermal radiation dose received by personnel during the evacuation process. The BODP method includes both an equivalent rule proposed for the first time based on the linear weighting method, and an improved Dijkstra algorithm designed in this paper. Because of them, the BODP method can normalize evacuation time and thermal radiation dose to the same degree and perform multi-sink dynamic planning on a hexagonal grid map, in order to find the optimal evacuation path with the lowest equivalent cost in the event of a tank farm fire. This method accounts for changing fire scenarios and avoids potentially high-risk paths. Finally, a case study of a chemical plant tank farm is conducted to demonstrate the effectiveness of the method. The results indicate that the BODP method is better suited for evacuation path planning in complex and dynamic fire scenarios. The BODP method resolves the issue found in static path planning: the first-degree burn probabilities derived from estimated thermal radiation doses are 34% lower than those derived from actual thermal radiation doses, potentially resulting in unnecessary casualties. Additionally, the BODP method is also applicable to scenarios involving domino effects triggered by fire incidents. This method provides scientific and reliable technical support for emergency evacuation in tank farm fires, greatly contributing to the safety and protection of personnel.