Concerns over mitigating methane leakage from the natural gas system have become ever more prominent in recent years. Recently, the U.S. Environmental Protection Agency proposed regulations requiring use of optical gas imaging (OGI) technologies to identify and repair leaks. In this work, we develop an open-source predictive model to accurately simulate the most common OGI technology, passive infrared (IR) imaging. The model accurately reproduces IR images of controlled methane release field experiments as well as reported minimum detection limits. We show that imaging distance is the most important parameter affecting IR detection effectiveness. In a simulated well-site, over 80% of emissions can be detected from an imaging distance of 10 m. Also, the presence of "superemitters" greatly enhance the effectiveness of IR leak detection. The minimum detectable limits of this technology can be used to selectively target "superemitters", thereby providing a method for approximate leak-rate quantification. In addition, model results show that imaging backdrop controls IR imaging effectiveness: land-based detection against sky or low-emissivity backgrounds have higher detection efficiency compared to aerial measurements. Finally, we show that minimum IR detection thresholds can be significantly lower for gas compositions that include a significant fraction nonmethane hydrocarbons.