The implementation of forced convective boiling in microchannels is developed to be very promising, as the attainable heat transfer rate is very favorable when compared to traditional thermal solutions. In this study, numerical simulations are conducted to investigate nucleate boiling in microchannels, with the interface separating liquid and vapor phases tracked by a conservative level set method (LSM). The behavior of bubbles in uniformly superheated liquid, and flow boiling regimes in a microchannel are identified to validate the applicability of the proposed methodology. Boiling mechanisms are found to be strongly dependant on wall surface conditions, simulations are thereby conducted to investigate flow boiling in microchannels with reentrant cavities. Comparisons of the performance of the enhanced and the plain-wall microchannels are performed, and the structured surfaces are demonstrated to facilitate nucleating and enhance critical heat flux (CHF). The identification and quantification of key design parameters of cavities including mouth opening (R), depth (H), diameter (D) and density are conducted, addressing an optimal topology design with R of 9.5 mu m, H of 60 mu m and D of 120 mu m, which nucleates first under a given set of conditions from rather low superheating. To enable a compatible view, two cavity characteristic models are investigated. The stochastic model with randomly sized and located cavities has been proved to hinder the cooling capability by decreasing CHF, as compared to the deterministic model that comprises regular cavities. Nevertheless, it still outperforms the plain-wall microchannel. Finally, heat flux conditions of the cooling target are studied to seek high-performing cooling schemes, considering seven different heating loads. (C) 2013 Elsevier Ltd. All rights reserved.