Fires in the engine compartments of surface and underground non-rail heavy duty vehicles are still highly frequent. Statistics show that most of the reported fires commenced in the engine compartment and that these were not promptly detected by the drivers. Fires which were not detected rapidly, spread oftentimes beyond the firewall of the engine compartment having notorious economical and environmental repercussions; furthermore, endangering the safety of the occupants. Detecting fires in the engine compartments of heavy duty (HD) vehicles with inexpensive and simple automatic detection systems is in general challenging. High air flows and large amounts of suspended pollutants, together with the complicated geometry and wide range of surface temperatures typically occurring during the normal operation of the vehicle, complicate the reliable operation of almost all types of detectors. This work presents a theoretical study assessing the effectiveness of different detection systems in a simulated fire scenario. Results from calculations and Computational Fluid Dynamics (CFD) simulations of a well defined fire in a standardised engine compartment are used to determine the gas temperature and smoke concentration across the compartment and how these are affected by different engine configurations and driving conditions. The effectiveness of different detection systems is studied by means of simulating a fire and analysing predictions measurements of temperatures and smoke concentrations resolved in time and space in the virtual compartment.