تجزیه و تحلیل حساسیت از مدلسازی گرما و رطوبت غیر همدما CFD توأم

CFD (Computational Fluid Dynamics) is a useful tool to study air flow patterns in a room. Current CFD models are able to simulate air flow combined with temperature distributions and species distributions. In this paper a coupled CFD–HAM model is discussed. This model combines CFD with a HAM model (Heat, Air and Moisture) for hygroscopic materials. This coupled model is able to simulate air flow around a porous material and combines this with heat and moisture transport in the porous material. Validation with a small scale experiment in which gypsum board was used as a hygroscopic material showed good results. In this paper a further validation of the model is discussed based on a sensitivity analysis of some model parameters. Especially hygrothermal parameters like sorption isotherm and water vapour permeability proved to have a non negligible influence on the modelling outcome. Adding a hysteresis model showed improvement of the model during desorption. The model was also used to compare two modelling strategies. In one strategy the gypsum board was modelled as a uniform material, in a second approach the material was modelled as being layered. The difference between the two approaches showed to be negligible.

Temperature and relative humidity are two important parameters for damage risk assessment of buildings, e.g. too high levels of indoor relative humidity can cause mould growth on the inside surfaces of the building envelope. When moisture migrates through the building envelope, interstitial condensation can occur which can lead to rot, deterioration of outside surfaces or other damage phenomena. Even if humidity levels are kept low enough, damage can still occur due to too strong variations, e.g. paintings and artefacts can show cracks when exposed to fluctuating temperatures and humidity levels [1]. Having a good knowledge of the heat, air and moisture transport in a building is also of great importance for many other applications. Moisture buffering by hygroscopic materials levels out indoor relative humidity fluctuations. This can reduce the energy use of HVAC systems [2] and improve the perceived indoor air quality at the same time [3]. In literature some examples are found where the importance of knowing the relative humidity in the design stage of an HVAC system is highlighted [4] and [5]. Buildings are complex systems and can be studied at different levels (whole buildings, rooms, building components…). Therefore, depending on the application, Heat, Air and Moisture (HAM) transfer in buildings is modelled through different approaches and a lot of different modelling tools are being developed. Overviews of recent developed HAM models are found in Refs. [6] and [7]. A new trend in HAM modelling is the coupling of these models to BES (Building Energy Simulation) models or CFD (Computational Fluid Dynamics) models depending on the application aimed at. Both modelling approaches were evaluated by Steeman et al. [8]. Coupling HAM models with BES is useful when the impact of moisture on energy use in a building is studied. Examples of such modelling approaches are found in Refs. [9] and [10]. Kwiatkowski et al. [9] used an isothermal modelling approach and neglected the latent heat in the porous material, where Steeman et al. [10] added this to their model. Combining CFD with a HAM model is interesting when a detailed study of the air flow field around a hygroscopic object is needed. For example microclimates can occur near artefacts. A detailed study of these microclimates is necessary for the assessment of damage risks [11] and [12]. Most BES programs are typically multizone models: they represent a room as one node and have the assumption that state variables (e.g. temperature, relative humidity…) are uniform for the entire zone (well-mixed air assumption). The coupling between the HAM model and the BES model is accomplished by using transfer coefficients. The heat transfer coefficient is used to model the heat transfer (convective and radiant) between the environment and the surface of the porous material (walls, furniture…). The mass transfer coefficient models the moisture transfer between the air and the porous material [13]. They have to be determined indirectly through empirical or analytical correlations or from CFD calculations. Often the heat and mass transfer analogy is used to convert heat transfer coefficients into mass transfer coefficients. Steeman et al. [14] however showed that this analogy does not always apply. CFD on the other hand does not require transfer coefficients to model the interaction between the fluid and solid interface. At the same time CFD allows the analysis of complex geometries and provides detailed information on temperature and humidity distributions in the air. One major drawback of CFD is the high computational cost. Therefore, up till now, applications are limited to the study of microclimates and building details. Whatever coupling approach is used, HAM models still need proper input data like boundary conditions, initial conditions and material property data. Extensive databases for these material properties can be found in literature [6], [15] and [16]. However, recent studies revealed a large spread of some of these material properties when the same material was measured by different laboratories [16] and [17]. It is often not clear how this will affect the model outcome. This paper highlights the importance of a sensitivity analysis for newly developed coupled HAM models. These models need a lot of material property data as input which introduce uncertainty to the model outcome. Therefore an elaborate sensitivity analysis is performed on a recently developed coupled CFD-HAM model [12]. The first part of this paper gives a brief description of the model. This model is then used to simulate air flow over a gypsum board surface. In the reference case predefined material properties are used. Afterwards different input parameters are evaluated based on a so-called One-at-a-Time sensitivity analysis. Also the air velocity, inlet temperature and humidity and the impact of hysteresis modelling are evaluated. Finally uniform modelling of gypsum board is investigated. Gypsum board is built up out of layers (finishing paper and gypsum) but modelled with averaged material properties. This modelling approach is compared to an approach where each layer is modelled separately.

An extensive sensitivity analysis was performed on a recently developed coupled CFD–HAM model. This model uses CFD to calculate the indoor air distributions around a porous material and combines this with a HAM model to incorporate the heat and mass transfer between air and the porous material. By using a direct coupling method, no external data exchange between the two models is needed which increases the computational speed of the model. Data from a benchmark transient heat and mass transfer experiment performed during IEA Annex 41 were used as a reference case for the sensitivity analysis. The material data used for this case were the average values found in a round robin test which was also performed during IEA Annex 41. This test showed that large discrepancies could occur between material properties measured at different laboratories. In this paper it is shown that the coupled CFD–HAM model is rather insensitive to deviations of most of the material properties. For density, heat capacity and thermal conductivity of the porous material no significant effect on simulated temperature and relative humidity was found when these properties were changed by 5%. The impact of sorption isotherm and vapour resistance factor was more severe. These properties are often harder to measure, resulting in large uncertainties. Deviations in the simulated relative humidity up to 2% RH were found for the different isotherms and resistance factors. These hygroscopic properties also have an impact on the calculated temperature although this effect is limited. Increasing the air velocity from laminar to turbulent had no effect on the relative humidity inside the porous material because the diffusive mass transfer dominates over the convective mass transfer. This is however not the case for temperature, where the impact of the convective heat transfer coefficient is of more importance. The effect of hysteresis modelling was also investigated in this paper. Including hysteresis in the model improved the model outcome, though it is not sufficient to get a full agreement with measurement data. These deviations can now be explained by the sensitivity analysis. A large uncertainty of the material property data, especially of the sorption isotherm and the vapour resistance leads to deviations in the modelling outcome. The combined effect of an uncertainty on the sorption isotherm and on the vapour resistance can lead in some cases to even greater deviations in the model. These effects are large enough to explain why the model does not fully match the measurements. Finally, modelling gypsum board as a layered material showed to have no impact on the results. This paper shows that the model works well and that its accuracy is to a great extent determined by the accuracy of the input data and material properties. For the moisture transfer modelling this data is still accompanied by large inaccuracies and even discrepancies between laboratories which means that more research is certainly needed.