The Modified Energy Cascade (MEC) crop model was originally developed to predict the edible biomass production of bioregenerative life support systems (BLSS) along with BLSS consumption and production of O2 and CO2. Three distinct MEC versions support this original goal and controlled environment agriculture (CEA) on Earth. Cavazzoni built the first MEC for predicting crop growth, transpiration, and productivity of BLSS. Boscheri et al. and Amitrano et al. each developed versions building off Cavazzoni's work. While each of these model versions builds off each other, differences in methodology and assumptions of plant physiology impact the outputs of the model, necessitating a comparison between versions. To describe the effects of input variability and model structure on the outputs of the MEC versions before further development for BLSS and CEA production facilities, four research questions were chosen to guide this evaluation. 1) How much variation in transpiration and yield predictions can be attributed to the model version? 2) How are input variations propagated through the cascading nature of the models? 3) Which model components are highly sensitive or uncertain to which environmental conditions? 4) How well does each model version predict the outcome of lettuce yield and transpiration outcomes of data sets independent from model development? To answer the first three questions, a series of global sensitivity and uncertainty analyses were performed. They revealed that 1) for daily transpiration rate and edible biomass model version alone can explain between 69% and 82% with Amitranos representing the lowest values and Boscheris the highest typically. 2) Even in sequences of identical equations, where each subsequent calculation is identical, variability is gradually reduced with final output variations between 40% - 55% that can be attributed to the prior upstream differences. 3) The Cavazzoni and Boscheri edible yield predictions are highly sensitive to photosynthetic photon flux density (PPFD) and CO2 across calculations while Amitrano’s is more responsive to photoperiod rather than PPFD. 95% of Boscheris transpiration output is driven by relative humidity while the other two utilize a combination of that and photoperiod. Lastly, these models and their performance were evaluated using environmental and yield data from an indoor vertical farming facility and growth chamber experiments. Together these analyses provide the information necessary to continue the development of the MEC for the prediction of resource flows and yield of CEA and BLSS supporting the optimization of electricity usage and circularity processes within closed-loop agriculture.