This thesis aims to quantitatively investigate how two sources of input uncertainty, namely the distribution shape and the input range of model parameters, influence the estimated mean greenhouse gas reduction from retrofit measures applied to a stock...
This thesis aims to quantitatively investigate how two sources of input uncertainty, namely the distribution shape and the input range of model parameters, influence the estimated mean greenhouse gas reduction from retrofit measures applied to a stock of office buildings in South Korea. To achieve this objective, an ANN surrogate model was developed based on building energy simulation results, and empirical cumulative distribution functions of the mean greenhouse gas reduction were derived for ten uncertainty scenarios.
The results show that variations in distribution shape led to relatively limited changes in the mean reduction, remaining within approximately 4%. In contrast, the definition of the input range can exert a more dominant influence on the estimated mean reduction. Under a practical input range reflecting domestic codes and standards, the variation in mean reduction reached 6.7%, which is comparable to the effect of distribution shape. However, when a physically plausible maximum credible range was applied, the mean reduction increased by up to 28.8%. This indicates that overly broad assumptions regarding input ranges may result in substantial overestimation of mitigation potential. Furthermore, the relative importance of distribution shape and input range can vary depending on how uncertainty in the building stock is assumed, with input range potentially becoming the dominant source of uncertainty.
From a policy perspective, even changes of a few percent in estimated mean greenhouse gas reduction can have significant implications for national mitigation target setting. Therefore, a rational and context appropriate definition of input ranges should be prioritized in retrofit impact assessments, while distribution shape should also be considered according to the analysis objective and application scope. This thesis highlights the importance of systematic input uncertainty management in archetype based urban building energy modeling and suggests the need for adopting probability-based performance indicators to support more reliable evaluation of greenhouse gas mitigation policies.