THE ENERGY FUTURE OF EXISTING BUILDINGS IN BRUSSELS: BETWEEN PRESERVATION AND PERFORMANCE

old buildings can consume less energy than buildings from the 1960s or 1970s.

The propagation of basic uncertainties is relatively similar to those from the first case study. Nevertheless, it should be noted that in this case there are a greater number of glass walls, meaning that there is greater uncertainty in relation to this item. The opaque walls also play an important role here (though less than in the previous example) with a 15% input difference implying a 10% output difference for the heating. One reason for this could be the solar factor of the opaque walls, which we do not completely understand. This can produce a difference of up to 1.5% in consumption. All of these small errors, when put together, ultimately produce significant output errors. The annual curves representing the overall uncertainties again evidence in the same difficulties outlined above in relation to matching the calculations with the measurements in the field. For this building, the bias arises from our lack of information about the adjoining apartments, which is limited to temperature measurements.

I would like to mention an interesting bias relating to the calculation engine. The apartment is southeast facing. The first version of the calculation engine that we used took the general orientation into account as either due north, due south, due east or due west. However, by using a second, improved version of the calculation engine which enabled the exact orientation to be taken into account, divergences emerged in the results. When it was due south, consumption fell by 8%; when due east, it increased by 5%. This showed once again the importance of inputting precise data in the calculation engine, including orientation.

CONCLUSION

In conclusion, for our two buildings, the input parameters with the biggest impact on the results of these dynamic simulations were the U value of opaque walls, the average temperature measured in the apartment (a deviation of less than one degree from the set-point temperature has a significant impact on consumption) and site measuring. This fact is often forgotten, especially with these buildings. However, not having the architect's blueprints, drawing them up in situ, or making estimates based on inaccurate documents can have major effects on the calculations. Finally, the closing of sunscreens scenario also plays its role, in particular in the extreme cases of where they are closed during the day or not used at all.

The annual heating projection graphs (fig. 10 and 11) were produced with very poor simulation data because the extremes of the input uncertainties had been used, which, statistically, have a very low chance of occurring. The blue bell curves, which are the outputs obtained with the simulations, are very wide. This means that the average measurement, which according to my simulations should be 125 kWh/m²/year for the Parisian apartment with regards to heating, may increase to a maximum of 175 kWh/m²/year and decrease to a minimum of 75 kWh/ mVyear.

Once again, it is very difficult to “tune” the measurement to the calculation. Clearly, the data used could be refined. However, we deliberately put ourselves in a situation which consultancy firms frequently encounter, namely a lack of time and difficulty in collecting information. A consultancy firm working on dynamic thermal simulations is not going to spend three weeks collecting input data; there will therefore be a lack of precision in its results. We therefore worked within this realistic constraint.

Fig. 9

Annual monitoring of apartment's consumption (source: Cerema).

61