تجزیه و تحلیل حساسیت از عملکرد خنک کننده شب پیش بینی شده برای مدل سازی انتقال حرارت داخلی

The potential of night cooling, a passive cooling technique of growing interest, is typically investigated by numerical means. In particular multi-zone energy simulation is currently appraised for building design. Unfortunately, in addition to the inaccurate approximation of an ideally mixed room, the implemented empirical convective heat transfer coefficients (CHTCs) only apply to specific flow regimes – forcing to use arbitrary correlations and, thus, possibly limiting the usefulness of the simulation results. Therefore, the authors of this paper investigate the sensitivity of the night cooling performance to convection algorithms. First, the authors examine the applicability of convection correlations for real building enclosures, extracted from literature. Subsequently, simulations of a night cooled office room during summertime of a moderate climate (Belgium) are carried out in TRNSYS, using different convection correlations in addition to varying design parameters. The results show that the choice of the convection algorithm strongly affects the energy and thermal comfort predictions. More importantly, the convection algorithm is of the same importance as the design parameters – making an exact definition of the CHTC crucial. Therefore, additional research by experiments or airflow codes, based on fluid dynamics, is regarded necessary.

Developments in information technology, since the mid-1960s, have led to opportunities to study thermal processes in buildings dynamically. Until then only simple hand calculation methods were available. Today, multi-zone energy simulation (BES) software coupled with airflow codes, based on fluid dynamics, can provide information on all locations in and outside a building of any desired variable. However, high computational costs of computational fluid dynamics (CFD) limit its usability to the domain of academics or specialists. Moreover, competency in handling complex airflow codes brings about a steep learning curve for practicing engineers. Therefore, stand-alone BES is currently appraised for building design. Alongside the implementation of more advanced heat transfer models such as ray-tracing for view factor calculation together with radiosity models, the radius of BES has widened substantially, integrating not strictly thermal physical processes, but also e.g. illumination [1] or moisture transport in fabric [2]. Unfortunately, the prediction of the interior convective heat transfer still remains ill-defined. Some BES programs, such as DOE-2 and SUNREL, combine convection with inter-surface radiation by modelling the two processes with a combined transfer coefficient. More recently (re)developed dynamic whole-building simulation programs – e.g. ESP-r, EnergyPlus, TRNSYS – model radiation and convection separately, increasing the adaptability and, thus, their accuracy. These programs employ the so-called well-mixed assumption. This treats the room air as uniform and characterizes the surface convective heat transfer by a convective heat transfer coefficient (CHTC) and by the temperature difference between the room air and the internal surface. Predominantly, these convection coefficients rely on correlations derived from experimental data. However, the convective heat transfer algorithms can also be based on boundary layer theory, admitting a pure theoretical development [3]. More recently, first cautious steps are taken towards modelling the convective heat transfer by CFD – as performed by, amongst others, Awbi and Salmerón et al. [4]. Anyhow, comparison of convection correlations reveals strongly differing convective heat transfer predictions. Alongside the investigated flow regime and the mechanism of fluid flow, the dimensionality of the experimental/simulation model itself plays an important part in deriving convection correlations. For example, the majority of the correlations recommended by CEN [6] are derived from data mainly based on experiments with isolated horizontal and vertical surfaces. Although these studies treat an important class of problems with many practical engineering applications, the suitability of these coefficients for building energy analysis is at best questionable. By neglecting the inherent three-dimensionality, complexities in real building enclosures, which significantly affect the flow pattern and the heat transfer, are not taken into consideration – as shown by Clarke [7]. Bearing this in mind, numerous researchers have shown the sensitivity of thermal predictions to the modelling of internal convection. The worthy effort of the IEA [8], solely focusing on heating and free-floating conditions, acknowledged the dominant role of convective heat transfer in the building's energy balance, next to more detached attempts [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20] and [21]. Moreover, the importance is believed to significantly increase in case of high ventilation rates, such as night cooling – as explored in e.g. Artmann et al. [11]. Unfortunately, before-mentioned authors based their investigation on arbitrary values of the CHTC – i.e. independent of the interior and ventilation system design, limiting their authority. Note that this last-mentioned passive cooling technique constitutes a valuable alternative to energy-wasting mechanical cooling using an air conditioning system – as experimentally shown by, amongst others, Blondeau et al. [12] and Høseggen [13]. In fact, at night – when the outdoor air temperature is usually lower than the building temperature, natural or mechanical ventilation cools down the building fabric. The following day, the thermal mass absorbs the heat by solar and infrared radiation and room air convection, affecting the internal conditions by day as follows: reducing peak air temperatures and the operative temperature while creating a time lag between the occurrence of external and internal maximum temperatures [14] and [15]. As a result, building users enjoy an improved thermal comfort while clients can possibly build smaller mechanical cooling plants – or even leave out – and operate their buildings more efficiently. Finally, reduction of the peak cooling demand would be of great interest to the power generating industries. Thus, as the design and performance analysis of this passive cooling technique of growing interest is typically based on numerical modelling, the authors of underlying paper believe that further analysis of the CHTC is crucial. Therefore, the authors of this paper assess the impact of empirically derived convection algorithms on the night cooling performance, relative to the most meaningful design parameters. First, based on an extensive literature review, this paper reports on the applicability of convection correlations for BES. Extending the work of Khalifa [16] and Beausoleil-Morrison [10], this study focuses on algorithms derived for real building enclosures. Subsequently, numerical simulations of a night cooled office during summertime of a moderate climate (Belgium) are carried out in TRNSYS. In this, the authors implement different convection correlations – which are not necessarily apt for the flow regime at hand – to assess the impact on the predicted night performance. Finally, this sensitivity analysis also varies the design parameters to investigate the relative importance of the choice of convection algorithms.

This paper reports an extensive analysis of the applicability of convection correlations, primarily derived for real building enclosures. The study shows that the boundary conditions of the experiments – i.e. the dimensionality, the flow regime and the mechanism of fluid flow – on which the correlations are based, limit to a large extent the applicability. Secondly, the sensitivity of the night cooling performance to these correlations, relative to the most relevant design parameters, was investigated, leading to the following conclusions. In case no night cooling is applied, the choice of the convection correlation is of minor importance. On the other hand, for night cooled spaces, a correct description of the convective heat transfer is regarded necessary – given the large spread of the predicted performances. The operation time mainly depends on the convective heat transfer during the night, while the temperature excess hours and the cooling demand depend to a larger extent on the ability to store heat during the day. Therefore, correlations producing high convective heat fluxes all the time, evidently provide a good overall performance. However, surprisingly, not necessarily the mixed convection correlations lead to the best performance. Additionally, as observed in the sensitivity analysis, the choice of the CHTC can alter the design decisions drawn from a BES-based analysis. Clearly, a more exact definition of the CHTC in particular for night cooling – for both day and night regime – is necessary, which can be obtained by experiments or CFD.