The latest changes to the UL-217 & UL-268 standards challenge current and existing smoke detectors and necessitate the development of better performing devices that are more sensitive towards polyurethane fires and simultaneously more resistant to nuisance alarms. Possible design strategies that include multi-color or multi-angle scattering techniques further increase the complexity and requirements for new detectors. Optical simulation is a key instrument for the design and development phase to improve the performance of photoelectric smoke detectors to meet the more stringent requirements.Going hand in hand with vertical manufacturing that allows detailed knowledge and control of each component the goal was to build an optical model that strives for achieving qualitative and quantitative performance predictions. This model starts from the optoelectronic component level and goes up to the fully assembled smoke module. We tested the optical model against experimental data of an existing photoelectric smoke detector and improved it to show convergence between simulated and experimentally measured data.On this basis, we investigated and optimized key optical design parameters to achieve better signal-to-background levels that pave the way for enhanced smoke sensitivity and an easier discrimination between smoke sources and non-hazardous nuisance sources. The optimization of interaction volumes, wavelength dependent scattering angles, and surface features that directly control the path of light with the consideration of manufacturability and robustness against assembly tolerances was of particular interest. The outcome of this process is an optimized dual color smoke detector whose performance is consistent with the predictions of our optical model.