Understanding and predicting human behaviour in complex situations, particularly when influenced by environmental factors such as airflow, is a central challenge in safety research. Pedestrian simulation frameworks address this challenge by incorporating varying levels of behavioural and environmental detail. Vadere is an open-source pedestrian simulation software that integrates pathogen exposure models developed by Rahn et al. [1] to simulate the spread of infectious diseases. In this exposure model, infectious agents emit aerosol clouds that susceptible agents inhale. Airflow was neglected in this first model. [2] have extended the exposure model by shifting the aerosol clouds along a precomputed airflow. This work describes a major restructuring of Vadere's exposure model to allow for a grid-based approach, thus enabling a more detailed and realistic representation of the distribution of airborne substances. The airflow can now be imported from external high resolution simulations. Based on the new structure, we extend the model to simulate the inhalation of fire-related toxins instead of pathogens. Pathogen concentrations are replaced by fire toxin concentrations. Their spread is driven by data computed with the FDS fire simulator. From a computer science point of view, it is very easy to add agents' reactions to exposure, such as changing direction or slowing down. However, behaviour effects of toxic smoke exposure vary greatly and depend strongly on the toxicity [3]. Therefore, our model currently represents toxin exposure visually. Restructuring the Vadere exposure model to a grid-based approach lays the foundation for more accurate modelling of any kind of foreign substance, such as pathogens or fire toxins, and enables future extensions towards exposure-dependent health and behaviour models.