(A) study on the design and operation support of energy reducing building remodeling utilizing simulation & artificial neural network

Abstract

Energy use in the building sector accounts for a large percentage of the world's total energy consumption, which has a significant impact on $CO_2$ emissions. Therefore, there is a growing doubt about remodeling existing buildings from an energy efficient point of view.

The purpose of this study is to provide an example of remodeling design which shows energy efficiency compared to the existing building for existing domestic buildings. For remodeling design alternatives analysis, we modeled the testbed building using the design builder, a dynamic building simulation tool. In addition, a window element, which is a building element that has a great influence on the energy efficiency of a building, is set as a case study analysis item and 11 cases (including initial models) of window cases generally used in Korea are classified through four criteria. The analysis of annual heating and cooling energy for alternative buildings modeled by these eleven window cases was conducted and Case 11, which saved 2736.06 kW of cooling and heating energy per year compared to the initial target building, was selected as the final alternative.

This study also developed an energy prediction algorithm by applying ANN (Artificial Neural Network) technique to the final selected alternative building data. This is the basic process for designing optimal HVAC operation by detecting the peak load of the alternatively selected building. For performance evaluation of ANN model, this study used , MBE, Cv(RMSE) which is a performance evaluation measurement of prediction model. The which is correlation coefficient was 0.9592, MBE was 7.4115, and Cv (RMSE) was 15.18. This proved that the proposed model satisfies the ASHRAE Guideline 14-2002 and IPMVP which is the building performance criterion and proved that it is suitable as the load prediction model.