mjwitte
commented
3 years ago See #8785
Open msfer86 opened 3 years ago
mjwitte
commented
3 years ago See #8785
eugeniorodrigues
commented
3 years ago @mjwitte This new issue is different from #8785. In this one, the issue is with the SurfaceConvectionAlgorithm:Inside object when using AdaptiveConvectionAlgorithm option (we have tested all options).
In #8785, the issue is related to how the outside heat convection coefficients are calculate in SurfaceConvectionAlgorithm:Outside according to wind direction (no implementation issue here, just rather are a result of the type of model; actually, Emmel model has an implementation error when calculating theta).
In this new issue, the spikes seem random and are significantly larger (up to 45 % of the total HVAC energy consumption just by changing few degrees in the building orientation).
eugeniorodrigues
commented
3 years ago @mjwitte We believe we might have found one of the reasons for the spikes in the 360° rotation of the building (see Figure).
We added to the IDF the following Outputs:
Output:Surfaces:List, ViewFactorInfo; !- Report Type
Output:Surfaces:List, Details; !- Report Type
We have used the same IDF to run two simulations (case 1 and case 2) but with the following changes:
Then, we compared the EIO files to find the diferences. The only important difference is the Perimeter value in the Roof Surface Convection Parameters (line 142 in Figure below), which is incorrectly determined for the building north 112° (the spike case). In this case, the surface perimeter is larger (279.38) than in the building north 292° (231.26). We identified the case by comparing the Roof Perimeter with the Floor Perimeter (231.26).
We believe it has something to do with how the Roof Perimeter is calculated in the ConvectionCoefficients.cc starting in line https://github.com/NREL/EnergyPlus/blob/9e55c4ccb7c70c5c4b3fb4e1d987023927e97183/src/EnergyPlus/ConvectionCoefficients.cc#L4287
The used IDF and the output IEO files are below. The weather file is the same as in the original post.
Note: See also a similar issue but with the calculation of the building façade perimeters (#8881).
Issue overview
Assessing the annual HVAC cooling consumption for different building orientations, covering the 360º-span (using the ideal loads air system object), while applying the TARP algorithm for the 'SurfaceConvectionAlgorithm:Inside', results in the expected yearly evolution, although with some spikes in the curve (see TARP image attached). However, applying the AdaptiveConvectionAlgorithm results in very high consumption peaks in (what seems) random orientations. Using different convection models in the 'SurfaceConvectionAlgorithm:Inside:AdaptiveModelSelections' only alters the orientations where the peaks appear (see some examples with different models in the Adaptive images attached).
This occurs while using a specific weather file (PRT_LB_Lisboa.Portela.AP.085360_TMYx.2004-2018.epw attached), but not with others (we have tested also in other 2 locations). However, by slightly altering the building dimensions (slightly different depth, width and height, while maintaining the same building volume), the problem does not occur in any of the 3 locations.
The building (Building.idf attached) consists of a 10000 m3 parallelepiped single zone, where the internal partitions are considered by adding some internal mass; which seems not to be related with the problem, since removing it does not solve the issue. The building does not contain any windows, which may simplify the analysis of this issue.
Details
Some additional details for this issue (if relevant):
Checklist
Add to this list or remove from it as applicable. This is a simple templated set of guidelines.