Abstract

The Nationwide House Energy Rating Scheme (NatHERS) in Australia governs the software programs that model energy use in residential houses. These programs all utilise the Chenath engine that models energy and temperature dynamically. With the inclusion of brand new condensation provisions into the 2019 version of the National Construction Code (NCC), there is a need to expand the functionality of NatHERS software to undertake dynamic hygrothermal analysis.

A plugin has been developed to apply the Glaser method of vapour transport in the building’s fabric as a step subsequent to building simulation. Although this approach is simplified and has been known to err on the side of caution, the undertaking not only assists in the valuation of the building fabric’s vapour management, but also opens up ways of further inquiry into the physics of condensation and biology of fungal growth.

The extension is scripted in R and produces graphical outputs for dewpoint analysis and vapour transport. This allows building practitioners who are new to condensation analysis to be cognisant of where and how much condensation is occurring, and whether recovery is possible before mould germination commences.

Introduction

In May 2019, condensation provisions were introduced for the first time into Australia’s National Construction Code (ABCB, 2019). This requirement, comprising three specific clauses was implemented across domestic detached houses and multi-residential buildings (Volume Two 3.8.7.1 and Volume One F6 respectively, ABCB, 2019). In summary, for the cooler climate zones (Zones 6-8 being temperate, cool temperate and alpine) the code requires a vapour permeable membrane on the side of the drained cavity in external walls. In addition, humid area mechanical ventilation rates were specified, and if these exhausts were discharged into the roof, a minimum amount of ventilation openings in the eaves and ridge was specified. A verification method (V2.4.7, Volume Two; or FV6, Volume One) stipulates functionality requirements for hygrothermal modelling, which needs to meet the following requirements:

FV6 Software verification method in NCC 2019

Although condensation assessment software is far from commonplace in the Australian Building sector, two programs have been gaining a following: WUFI (by the Fraunhofer Institute for building physics) and JPA Designer. WUFI has had decades of development and represents the specialist end of sophisticated condensation analysis. Australia, having only recently included condensation provisions to the construction code, will for the time being need a simpler method of analysis, something closer to what JPA Designer provides, the software screenshot as seen in Figure 1.

Figure 1.Condensation analysis presentation from JPA Designer.

The take up of condensation software is a far cry from that for energy efficiency. In Australia the three software used for evaluating projected energy use (AccuRate, First Rate and BERS) are essentially different front ends which all utilise the same simulation engine: Chenath. The name derives from a combination of the previous software, Cheetah, which featured quick computation of dynamic simulation based on algorithms developed by Muncey (1979); and the suffix “-nath” being short for NatHERS (Nationwide House Energy Rating Scheme), the Australian organisation responsible for developing, maintaining and promoting the program.

Figure 2. NatHERS certificates in comparison to Australian Bureau of Statistics Building approvals (May 2016-Jan 2020) https://ahd.csiro.au/other-data/certificates-vs-building-approvals/

Figure 2 shows the NatHERS certificates in comparison to Australian Bureau of Statistics building approvals (May 2016-Jan 2020). It can be seen that in some Australian states such as Victoria, virtually all houses that are built have been evaluated utilising the Chenath engine, virtual none opt to meet the elemental deem-to-satisfy prescriptions of the construction code. Also noteworthy is that even though a software accreditation protocol is available to validate other programs that can simulate energy consumption (NatHERS, 2012 and 2019), to the authors’ knowledge no other software is being used for regulatory approval.

The domination of the market by NatHERS software, and the lack of extensibility of it to undertake condensation risk assessments presents a natural opportunity. The common use of the Chenath engine is to generate the house energy consumption based on the climate type, reflected as a star rating. The national mandated minimum is 6-stars (with 10-star being close to, or effectively, no heating and cooling requirement with location variations). However, the Chenath engine does much more, in particular it produces a detailed hourly projection of internal temperatures which can be extracted to run a hygrothermal assessment.

The impetus to develop this hygrothermal extension came from a few quarters. The process of software introduction to the Australian construction industry is typically a long process, thus it would be far better to embed a plugin into the NatHERS workflow for which energy assessors were already familiar with. Software like JPA Designer utilises a Glaser (1958) method as published in ISO13788 (2012) which simulates condensation and evaporation on a monthly basis. However, this resolution can be too coarse to identify mould-associated condensation risks, seeing it was only a matter of days that mould germination could occur as shown in the isopleths of Figure 3 and Figure 4, obtained from AIRAH (2016) Application manual DA20: Humid Tropical Air Conditioning and the WHO (2009) Guidelines for indoor air quality: dampness and mould. WUFI on the other hand was highly specialised and new users would be faced with a learning curve steeper than any of the NatHERS accredited programs.

Figure 3. Mould germination and growth isopleths from AIRAH (2016) Application manual “DA20 Humid Tropical Air Conditioning”.
Figure 4. Mould germination and growth isopleths from WHO (2009) “Guidelines for indoor air quality: dampness and mould” (bottom).

Instead of developing a commercial program, the motivation was to make a condensation tool available to other building scientists for further development. Thus instead of working around a ‘black box’ of proprietary code, the aspiration here was to create an open-source hygrothermal simulation engine other scientists could ‘peek under the hood’ and modify the engine for their own purposes, and share their code snippets. The present version has been written in the open statistical programming language R (R Core Team, 2016), for which many scientists would be familiar with. By keeping the code legible to other building scientists, the programming script, though slower than object-oriented programming languages, could be better distributed and co-developed. At this nascent stage, development of vapour migration principles into simulation and graphical presentation was deemed more important than computer runtime efficiency.

Chenath can be run with any climate set to the Australian Climate Data Bank Reference Meteorological Year (ACDB-RMY). The RMY picks, on a monthly basis, a characteristic month in the past 10-25 years and concatenates 12 of these months into a year (NIWA, 2017). Similar to a TMY (Typical Meteorological Year), the RMY is a selection of averages rather than extremes, resulting in an underestimating of the frequency of condensation.

This new program we have called Hyenath as a working name, being a hy-grothermal function added to Chenath, and because a hyena is ostensibly slower than a cheetah. Thus the hope is that with this disclosure, users will be more forbearing in their expectations. shows the workflow of Chenath and how it can be extended with Hyenath. The steps are as follows:

  1. BOM2RMY: acquire BoM (Bureau of Meteorology) data for location, convert to RMY format using Excel.
  2. Chenath simulation: individual room profiles: Run AccuRate simulation of house building model using the new BOM2RMY.
  3. Hyenath simulation: from AccuRate output, temperature profile for a selected room is used for simulation in the Glaser engine to develop diffusion characteristics and rates of condensation and evaporation.
Figure 5. Workflow with Chenath and Hyenath.

To illustrate the important difference between the annual record and a reference year, both BOM and RMY data have been taken for Hobart, the southernmost capital city of Australia. In Figure 6, the boxplots show how measured weather station data (BOM) have wider fluctuations and are situated further away from the human thermal comfort bandwidth as opposed to RMY. The RMY does not adequately acknowledge that overnight temperatures are frequently below dewpoint; and in fact, at frost point on some nights. The disparity can exceed 10°C on occasion as seen in Figure 7, which shows a snapshot during winter.

Figure 6. Boxplot of recorded temperatures acquired from BOM against NatHERS RMY for Hobart, by month. Note the larger extremes of measured temperatures as opposed to milder (i.e. closer to human thermal comfort) and narrower fluctuations of RMY.
Figure 7. Time series plot of dry bulb temperatures during winter.

As such, a more robust approach will be to apply a recent record acquired for a weather station near the subject site. Such data can be purchased from the Australian Bureau of Meteorology (BOM). However, as BOM does not use the RMY format, an Excel template was created for the conversion of formats (such as global solar radiation and direct beam solar radiation) and calculation of extra-terrestrial radiation (based on Boland, 2008) to synthesise unrecorded data (such as diffuse solar radiation). The resulting RMY is plotted in R to support graphical checking that the data is in a reasonable ballpark, shown in Figure 8.

Figure 8. Console and graphical output of BoM to RMY conversion.

General analysis in R

Working in R taps on its native functionality to run statistical operations and have these displayed with its comprehensive and powerful graphing capability. This allows the data to be easily visualised and checked to see if anything is immediately out of the ordinary, and flagged out for closer inspection.

Using a reference house acquired from the software accreditation protocol (NatHERS, 2019), monthly boxplots are generated from AccuRate simulated using a BOM climate file. The boxplots give a quick graphical presentation of key statistical distribution intervals, enabling one to check one’s code by observing if the data conforms to familiar patterns. Figure 9 plots internal against external temperatures. In terms of condensation risk, the months where there is the largest difference (May-Aug) would deserve closer scrutiny. plots internal against external dewpoints. In this instance, daytime ventilation is a viable strategy to admit the dryer outdoor air into the house. In this case, the low external dewpoint would support the use of ventilation strategies, especially with heat recovery, as appropriate measure to mitigate condensation risk.

Figure 9. Boxplot of internal vs external (BOM) temperatures.

Figure 10 plots internal dewpoints against outdoor temperatures. The preference for translating humidity data to dewpoint is to enable a direct comparison with temperature. It has been found with houses studied in Tasmania (Law & Dewsbury, 2018) that thermal bridging was a common cause for condensation. Where outdoor temperature is below internal dewpoint (May-Aug), special care needs to be given to parts of the building envelope at risk of thermal bridging.

Figure 10. Internal dewpoint vs outdoor temperatures.

Figure 11 plots the temperature difference (dT = internal dewpoint – outdoor temperature) where a negative value indicates a condensation risk. Delta T frequency distribution of Month of May (bold line) is plotted against that for Jun, Jul and Aug. The month of May has the largest frequency of negative values, and also most extreme values (most negative delta T) of it.

Figure 11. Frequency distribution of delta T (internal dewpoint – outdoor temperature) Month of May (bold line), Jun-Aug (light lines).

Hyenath Simulation

As a proof of concept, the wall specification in the ISO13788 example, seen in Table 1, has been used for running the simulation. To assist users in understanding the graphical output, two graphs are generated in Figure 12. The graph on top, commonly referred to as a dewpoint diagram (familiar to designers), shows the thickness of materials and where the temperature profile intersects with the dew point profile, indicating where condensation will occur. The lower graph shows the vapour pressure against saturated vapour pressure and the quantity of moisture generated, or, when no condensation occurs, the amount that can be evaporated . A series of hourly outputs is collated in Figure 13.

Table 1. Wall properties used for simulation (ISO13788,2012).
Figure 12. Hyenath simulation graphical output. Dewpoint analysis (top) and Glasser method (bottom).
Figure 13. Hyenath hourly simulation showing condensation occurring in hours 3177-3179, and evaporation from hours 3180-3182.

By looping the hourly simulation in Hyenath, the output (condensation, or evaporation) is concatenated and graphed over a year of 8760 hours to show the overall moisture accumulation trend. Due to the presence of a vapour control layer, it is not surprising that there were only very few instances where condensation occurred, and when that happened it was only for a fairly short duration over a few hours, and very occasionally persisting for a couple of days, as graphed in Figure 14.

Figure 14. Hyenath hourly simulation for the whole year (above) and a detailed snapshot (below).

Currently the vapour permeability data of many Australian building materials is not available. However, there is some limited information on vapour permeable pliable building membranes, as these have to meet broad vapour permeability classifications of the standard AS4200.1 (2017) in Table 2.

Table 2. Vapour control classification (AS4200.1, 2017).

The same simulation was repeated with small modifications: the vapour control layer was replaced with a Class 2 vapour barrier from Table 2, using a middle value of 0.0726 μg/N.s (or of 2.7567 m). In addition the vapour permeable membrane has been set to the middle value of a Class 4 vapour permeable membrane of 1.1403 μg/N.s (or of 0.1754 m).

In this case, a very different profile emerges. Not only is condensation more prevalent, the persistence of it is across a much longer period. To illustrate the implications of this hourly record, an arbitrary threshold of 48 hours continuous moisture (i.e. when condensation is not fully evaporated) is set as a criteria for the onset of mould growth and used to determine how often this threshold has been breached as an indicator of mould risk. Figure 15 (top) marks the points at which the threshold criteria was breached (red lines), together with a more detailed view, Figure 15 (bottom). Looking closer, we can observe the germination phase of mould during the period when moisture is continuously availability, and the hyphal growth phase following it. It is evident that a much higher sensitivity to the conditions for mould onset is achieved with the combination of (1) the use of measured site data and (2) undertaking hourly condensation-evaporation simulation.

Figure 15. Hyenath hourly simulation where the VCL replaced with Class 2, and the vapour permeability adjusted to Class 4 membrane. Shaded regions show the 48h threshold (arbitrary) where mould germination is initiated, and the subsequent hyphal growth stage following.

Working in the same R environment, one can revert to the climate file and house simulation to take a closer look at those conditions for which thresholds were breached, in this instance from hours 3300-3500, illustrated by way of a psychrometric chart in Figure 16 revealing a large number of instances where external air was at dew point conditions.

Figure 16. Psychometric chart of outdoor and indoor conditions at the onset of persistent condensation.

The key differences between the ISO13788 implementation of Glaser method, and the Chenath-Hyenath variation of it is consolidated in Table 3.

Table 3. Summary of ISO13788 core implementation and Chenath-Hyenath features.

Limitations of the ISO 13788 method

Within the ISO13788 standard, mention is made under its scope that the method does not take into account some physical phenomenon and needs to be used in conditions where the following factors are deemed sufficiently negligible:

· Variation of material properties with moisture content;

· Capillary suction and liquid moisture transfer within materials;

· Air movement from within the building into components through gaps or within air spaces;

· Hygroscopic moisture capacity of materials.

Variable material properties

As earlier alluded, there is a lack of permeability data in Australian building materials. More work certainly needs to be undertaken here. Once this data is available, the updating of the material database in the engine is a trivial matter. The ability to allow permeability rates to fluctuate with moisture content is slightly more complex but can be easily accommodated. This is not a limitation of the Glaser method itself, but the particular ISO publication’s implementation of the Glaser method. It is also worth referencing that thermal testing which has been long undertaken in Australia still uses a static R-value at 23°C (AS4859.1, 2002), and variable material properties tests will likely be a long time away.

Capillary suction, hygroscopic moisture capacity

In the context of Australian domestic buildings, where the majority of houses are of framed construction, it is reasonable to assume that there is a low mass of materials with corresponding low capacity for absorption and distribution. Care still needs to be applied when applying this method to materials with high hygric-buffering capacity such as brick. In any case, the lack of such functionality places the analysis on the conservative side of condensation risk.

Air movement

This is a limitation that cannot be easily overcome. In an Australia-wide study involving 129 houses, the average air tightness was 15.4ACH@50Pa (Ambrose & Syme, 2015). Although this infiltration rate is very high compared to many developed countries, it is still within the upper bounds of the NatHERS methodology. So the high replacement of indoor air with relatively drier outdoor air would be factored into the internal relative humidity from the Chenath engine.

Conclusion

Hyenath is a script written in R that allows a Glaser type analysis in accordance with ISO13788. Hyenath extends the functionality of the Australian legislated Chenath engine by taking the simulation with a changed context of utilising BOM records in place of default weather files for better accuracy and currency. Although the ISO13788 method is simplified, the condensation-evaporation measurements can be extended by making a slight modification so as to undertake hourly simulations rather than monthly ones, providing the nuance needed to ascertain if a threshold period is breached for the germination of mould.

As a closing remark, this project derives much inspiration from the book The Cathedral & the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary (Raymond & O’Reilly, 2001), extolling the benefits of a vibrant and open exchange of programming ideas in the open-source world akin to a bazaar, as opposed to the cloistered world of secretive software development no different from artisans working on a cathedral away from the public gaze. One of the principles, and the one most pertinent here, is:

“Given a large enough beta-tester and co-developer base, almost every problem will be characterized quickly and the fix obvious to someone.”

It is the software developers hope and invitation that any others interested in the physics of condensation and biology of mould will find the software script useful for further development.