• No results found

There are several constraints and limitations associated with this EngD project and these are listed below:

Single-Zone Case Study in Inter-Model Comparison

The single-zone building case study selected for the inter-model comparison of different BPS tools prevented several important factors related to thermal mass and ICF simulation from being analysed, such as the impact of variable internal gains and air flows, the impact of intermittent occupation, and others. The case study set up was selected in order to reduce the specification and scenario uncertainties as much as possible. The specification uncertainties are associated with incomplete or inaccurate specification of building input parameters (Hopfe & Hensen, 2011). The scenario uncertainties are all the external conditions imposed on the building due to weather conditions, occupants’ behaviour and others (De Wit & Augenbroe, 2002). From that perspective, the case study selection served well the main purpose of analysing the “modelling gap”. Certainly, it was difficult to derive solid conclusions about the actual thermal performance of ICF construction method in such a simplified simulation scenario.

Lack of Real Data in Inter-Model Comparison

Various previous studies analysed the predictive variability found between different BPS tools for the same building (Brun et al., 2009; Raslan & Davies, 2010; Zhu et al., 2012). The inter- modelling comparative analysis performed as part of this EngD was the first one to report on the modelling uncertainties associated to ICF simulation. A significant limitation, however, was the lack of real data that could serve as a validation reference for the accuracy of simulation predictions. In other words, it should be acknowledged that due to the absence of an absolute truth, it was impossible to say what is correct and what is wrong or whether one tool performs closer to reality than the other.

Single Building Case Study

As discussed in Section 3.3.3.1, the case study research method allows for an empirical and in- depth investigation of a specific phenomenon within its real-life context (Yin, 2009). However, the scope of a case study is bounded and care must be taken not to draw generalised conclusions to ensure academic rigor (Brown, 2008). In that respect, it is important to emphasize that the building case study selected in this project was built to achieve close to Passivhaus standards. It is a high-end, low-energy construction, which might not be fully representative of ordinary buildings and more conventional constructions. Furthermore, the impacts of building design and operation on the thermal performance of ICF were not investigated as part of this research.

Source of Experimental Errors in Empirical Validation

There are several advantages when pursuing an empirical validation of BPS predictions especially under realistic conditions of monitoring a real building case study. Empirical validation allows to test the combined effect of all internal errors in a program (Lomas et al., 1997). Moreover, doing it under realistic conditions allows to interpret the impact of occupants’ behaviour instead of focussing only on the effects of the building structure and HVAC systems (Ryan & Sanquist, 2012). However, there are also some disadvantages. Firstly, in empirical validation, it is difficult to interpret the results and to draw conclusions on the possible sources of errors in the simulation because they are all simultaneously in effect. Moreover, there is a fair possibility for experimental errors to occur (Judkoff & Neymark, 1995).

It is generally accepted that there is a level of experimental uncertainty associated with in-situ measurements that may arise from random or systemic errors and could compromise the validity of the measurements (Evangelisti et al., 2018). Systemic errors are standard errors introduced to the system due to inaccuracies and sensitivities of the instrumentation used for the measurements (Coleman, 2009). The range of systemic uncertainty in the recording of the monitoring sensors was considered and included in the analysis of results. Nevertheless, other

sources of experimental errors should be acknowledged. One example is the recording of zone mean air temperatures. The internal air temperature was measured in one location within each room by using HOBO U12 stand-alone loggers. The loggers were placed at the height of 1.5m from the floor, away from heat sources and direct solar radiation, as suggested in literature (Singh et al., 2010; Kumar et al., 2017). However, this decision does not account for the effects of air stratification that may arise in the room due to buoyancy. Another example is the simulation of natural ventilation. Although monitored data were available for windows operation (opening and closing incidents), several assumptions had to be made due to two reasons. First, the set of data was incomplete, including a lot of noise, and, secondly, other critical information such as opening factors were not available.

Empirical Validation of Two BPS Tools

Due to time restrictions, the empirical validation of simulation results based on measured data from the monitoring project was performed for two of the nine tools included in the initial comparative analysis. Although some insights were provided on the accuracy of ICF simulation and the key factors contributing to the modelling uncertainties, these findings concerned just two BPS tools that were chosen as representative examples of the modelling methods employed in whole building simulation. There are, however, several algorithms and calculation methods -for example, the impact of frequency domain conduction solution method, or the impact of combined convective and radiative surface coefficients, among others- that were not included in the analysis.

Limitations of Simulation Models

The internal thermal mass due to furnishing was not included in all simulation models and this was identified as one of the reasons contributing to the divergence between simulation results and measured data. Simultaneously, the comparison of simulation to monitoring results showed

of the study was that, during the monitoring period, only global horizontal radiation was recorded on site. The split between direct normal and diffuse horizontal components was performed in EnergyPlus using the Perez model (Perez, 1992). This, however, introduced a certain level of modelling uncertainty since there were no monitoring data available to use as a reference point for direct and diffuse radiation values used in the simulation.

Assumptions on Range of Uncertainty

Finally, the uncertainty and sensitivity analyses conducted as part of this research, relied on information found in literature and used indicative values for the range of uncertainty in the wall material properties. Quantifying the actual range of uncertainty in the material properties of ICF would definitely improve the rigour and reliability of the findings.

Application to Other Climates

The analysis conducted, as part of this EngD, on the thermal performance of ICF was focused on two climates:

• The DRYCOLD typical meteorological year (TMY) weather file, used in the ASHRAE standard 140 (ASHRAE, 2014), representing a climate with cold clear winters and hot dry summers.

• The weather data, as recorded on site, in the temperate climate of Guildford, UK.

Hence, it is important to highlight that the research findings are highly relevant to these two climatic scenarios. Further investigation is required to assess the thermal performance of ICF is different climatic patterns, in other climates (such as cooling dominating locations) and under future climatic predictions.