As organic porous soil, peat is prone to self-heating ignition, a type of spontaneous initiation of fire that can take place at ambient temperatures without an external source. Despite the urgency to tackle peat fires, the understanding of the self-heating ignition of peat is insufficient. In this study, a computational model that integrates the mechanisms of heat transfer, mass transfer and chemistry is incorporated with a three-step reaction scheme that includes drying, biological reaction and oxidative oxidation to simulate the self-heating ignition of smouldering peat. The model is first validated against 13 laboratory-scale experiments from literature. For critical ignition temperature (Tig), the model gives accurate predictions for all experiments with a maximum error of 5°C. The validated model is then upscaled to predict Tig for field-size peat soil layers and compared with the predictions using a one-step scheme. The three-step scheme is shown to give more reliable predictions of Tig than the one-step scheme. According to the simulation results, for a 1.5-m-deep peat layer, self-heating ignition can occur at an average ambient temperature above 40°C. This is the first time that a multi-step scheme is used to simulate the self-heating ignition of peat, aiming to help in the prevention and mitigation of these wildfires.