Selection of a tritium accident consequence analysis (ACA) computer model can present challenges and uncertainties in many deterministic and probabilistic safety analyses (PSAs). This complexity is mainly due to the multifaceted, biochemical nature of tritium in the environment. In addition, there can be misalignment of the computer model with the expectations, inputs, and assumptions appropriate for the tritium consequence application of interest, leading to results that are misleading and often difficult to defend.

The present study reviews the minimum required analysis to model acute and late phases following a postulated tritium release to the environment. Based on considerations for supporting deterministic analyses and PSAs, a decision tree logic model is developed that identifies contemporary tritium ACA computer models that are most applicable to one of four types of tritium safety applications, including (1) deterministic/generic, (2) deterministic/site specific, (3) probabilistic/generic, and (4) probabilistic/region specific. It is determined that there is not one single ACA code that meets the requirements of each of the four categories. Alignment of the selected software with corresponding input templates is outlined, followed by examples to illustrate linking the decision tree logic model software with appropriate input/assumption templates.

While a representative set of ACA computer models can address most contemporary ACA needs, there are opportunities for improvement. Five main improvement areas for consideration include development of PSA-specific input templates, update of tritium plume depletion data sets, enhancement of food ingestion model flexibility, incorporation of options for aqueous releases, and cooperative development of one computer model for both mixed source and tritium source analyses.