3.2.1. Analysis of the Buildings' Architecture Influence on the Energy Results

Since four test sites were available for this research, a complementary study was performed to analyze the influence of the building's architecture on the previous energy results. The four buildings, which were completely different in terms of the materials, construction systems, thermal mass, and window-to-wall ratio, were simulated with the same weather data (*on-site* and *third-party* weather files). For this study, we selected the most homogeneous weather when comparing the *third-party* to the *on-site* weather data: as shown in the previous analysis, energy demands are very sensitive to *WS*, so the Gedved weather was discarded for this analysis because its *WS* was the one with the worst fit to the reference (see Figure 3). *Temp* was also a sensitive parameter, and the three weathers had similar statistical metrics. Finally, for the solar radiation parameters *DHI* and *DNI*, Pamplona's

weather better matched the reference compared to Lavrion. Therefore, the Pamplona weather file was chosen to develop this theoretical study, and for this reason, the four models were configured to have the same internal loads, HVAC systems, and schedules as the Pamplona office building.

Figure 7 shows the *MADP* results for this study. The two graphs on the top present the results using the *third-party* weather file, which had all the weather parameters provided by the weather service. They show how when the test sites were simulated with their own weather files (graph on the left), the *MADP* results and the trend of the curve were very different for the four test sites (each colored line represents one test site). However, when the test sites were simulated with Pamplona's weather file (graph on the right with dashed lines), the curves became very similar, reducing the differences in the *MADP* values and in the trend of the curve.

Thus, the main value responsible for the variation in the energy demand was the weather dataset employed in the simulations and not the building's characteristics. This effect was also reflected in the results from the sensitivity analysis, which are also presented in Figure 7 for the *Temp*, *DNI*, *DHI*, and *WS* parameters. The curves from the graphs on the right, which are the simulations of all the test sites with Pamplona's weather file, were very similar compared to the curves from the graphs on the left, especially in the case of wind speed. This study demonstrated the great influence of the weather parameters on the variation of the building's energy demand, almost independently of the model, and this showed the importance of the selection of the weather dataset used in the BEM simulations.

**Figure 6.** Representation of the *MADP* (%) of the energy demand analysis for the four test sites and for the different time grains. The dashed black line represents the results using the weather file with all the *third-party* weather parameters. Each colored line represents each one of the weather parameter results from the sensitivity analysis.

**Figure 7.** The comparison between the simulation results when each test site was simulated with its weather (on the left with continuous lines) and when all the test sites were simulated with Pamplona's weather (on the right with dashed lines). Each color represents a test site. The graphs show the *MADP* for the energy demand for the different temporal resolutions. From above to below are shown: results when the *third-party* weather file was used (all the parameters were changed in the weather file) and the sensitivity analysis results for temperature, direct normal irradiation, diffuse horizontal irradiation, and wind speed.
