Ashrae D RP 1325

Copyrighted material licensed to Elie Baradhy on 2019-05-26 for licensee's use only.
All rights reserved. No further reproduction or distribution is permitted. Distributed by Techstreet for ASHRAE, www.techstreet.com
Shaping Tomorrow’s
Built Environment Today
©2012 ASHRAE www.ashrae.org. This material may not be copied nor distributed in either paper or digital form without
ASHRAE’s permission. Requests for this report should be directed to the ASHRAE Manager of Research and Technical
Services.
SYRACUSEUNIVERSITY
BUILDING ENERGY AND ENVIRONMENTAL SYSTEMS LABORATORY
(BEESL)
263 Link Hall, Syracuse University, Syracuse, NY13244-1240; http://beesl.syr.edu
FINAL REPORT
Environmental Weather Loads for Hygrothermal Analysis and Design

of Buildings
ASHRAE RP-1325
Simulation studies and data-analysis
By
Mikael Salonvaara, Co-PI, project manager

Owens Corning
(formerly Huber Engineered Woods LLC)
(formerly Syracuse University)
Email: mikael@salonvaara.com
Managed by:
Prof. Jensen Zhang, PI, project manager
Syracuse University
Project Contributors:
Lambros Pezoulas and Konstantinos Karagiozis
Wholistic Analysis
January 5, 2011
1
Table of Contents
Table of Contents ................................................................................................................ 2
Acknowledgements ............................................................................................................. 4
Disclaimer of Warranties .................................................................................................... 4
Executive summary............................................................................................................. 5
List of Figures ..................................................................................................................... 6
List of Tables ...................................................................................................................... 9
List of Acronyms and Symbols ........................................................................................ 10
Introduction ....................................................................................................................... 12
Literature Review.............................................................................................................. 13
Reference year and procedures for its determination ................................................... 14
International Energy Agency – Annex 24 Approach .................................................... 15
Determination of the existence of DRY.................................................................... 16
Carsten Rode Method ................................................................................................... 17
Geving Approach .......................................................................................................... 18
The Pi-factor method .................................................................................................... 20
Moisture Index Method................................................................................................. 21
The ANK-ORNL method ............................................................................................. 23
Summary of Advantages and Disadvantages of Each Method ..................................... 24
Methodology to Evaluate Existing Weather Selection Methods ...................................... 26
Evaluation process ........................................................................................................ 27
Available weather file sources and formats .................................................................. 29
Weather parameters included in the analyses ........................................................... 30
Locations used in the analyses ...................................................................................... 30
Orientation effects ......................................................................................................... 31
Solar radiation ........................................................................................................... 31
Wind driven rain effects............................................................................................ 32
Selected simulation models........................................................................................... 33
Wall systems selected for analysis ................................................................................ 34
Procedure for running the simulations .......................................................................... 36
Indoor air and initial conditions .................................................................................... 37
Use of damage functions for selecting weather years ................................................... 37
Critical layers in the walls for investigation ............................................................. 38
Rankings based on damage functions ....................................................................... 38
Simulation Results ............................................................................................................ 41
Analysis of the orientation effects ................................................................................ 41
Results – Damage functions for different orientations ............................................. 42
Simulation results – Moisture contents of OSB in wood frame wall ........................... 43
Simulation results – Moisture contents in CMU........................................................... 44
Rankings of weather years based on RHT-index – Wood frame wall versus CMU wall
....................................................................................................................................... 45
Evaluation of Weather Data Selection Methods ............................................................... 47
Existing methods for selecting weather data for moisture simulations ........................ 48
Using Weather Selection methods to select MDRY ..................................................... 48
Correlations between weather parameters .................................................................... 51
2
Goodness-of-fit of the existing weather selection methods ...................................... 55
Simulated Performance and Correlation with Weather Data - New Method Development
........................................................................................................................................... 58
Statistical parameters .................................................................................................... 58
Simulated versus predicted RHT-index ........................................................................ 63
Goodness-of-Fit of the New Method ................................................................................ 67
Contour maps of weather parameters and damage functions ....................................... 69
Conclusions ....................................................................................................................... 71
References and literature................................................................................................... 72
Appendix A: Moisture content of OSB with 30 years of weather data ............................ 75
Appendix B: Moisture content of CMU with 30 years of weather data ........................... 79
Appendix C: Relative damage function for the light weight wall and CMU wall ........... 84
Appendix D: Simulated vs. predicted damage function values ........................................ 88
Appendix E: Material properties in simulations ............................................................... 92
Appendix F: Equation method to select weather year ...................................................... 96
3
Acknowledgements
The authors would like to thank the many people who have helped them during the
course of this project. Particular thanks are due to the subcontractors Lambros Pezoulas
and Konstantinos Karagiozis (Wholistic Analysis) for persistent attention to details and
their patience during the long course of the project, and to Marcin Pazera (Syracuse
University/Simpson Gumpertz and Heger) for help during different stages of the project.
The Project Monitoring Subcommittee – Chair: Andre Desjarlais (ORNL), Anton Ten
Wolde (USDA Forest Products Laboratory), Andreas Holm (Fraunhofer Institut for
Building Physics), Drury Crawley (DOE)/Steve Cornick (NRC Canada), Martha Van
Geem (CTL Group) and Robert Morris (Environment Canada) – provided much needed
advice and comments during the four years the project was underway.
Disclaimer of Warranties
The weather data and associated documentation (research report) is provided "as is"
without warranty of any kind, either expressed or implied. The entire risk as to the quality
and performance of the data is with you. In no event will ASHRAE or its contractors be
liable to you for any damages, including without limitation any lost profits, lost savings,
or other incidental or consequential damages arising out of the use or inability to use this
data or associated documentation.
4
Executive summary
Different sets of weather data exists for varying purposes such as yearly energy use
calculations, dimensioning HVAC and solar energy systems and so forth. In the past
decade the hygrothermal simulation models have become everyday tools for building
envelope designers, yet there is still no consensus about how to select the weather data
for the design calculations. ASHRAE project RP-1325 analyzed several existing weather
selection methods such as the one in ASHRAE standard 160 (Moisture Design Reference
Years: the 10th-percentile warmest and 10th-percentile coldest years from a 30-year weather
analysis) and compared their performance by simulating two wall systems – one light
weight wall and one massive CMU block wall – with commonly used simulation models.
The analysis showed that the existing methods were not very successful in selecting the
right weather data and therefore a new method was developed. The new method is
construction independent and uses only average yearly weather data as input and predicts
an RHT-index for each year based on these. The orientation effects are taken into account
by calculating the solar radiation and the wind driven rain loads on the wall. The north
orientation was found to produce the highest values for the damage function and was
chosen as the design orientation. The coefficients of the equation were optimized by
using weather and simulation data for eight cities and these were subsequently validated
with another four cities. The new method was found successful in selecting the worst
years in all analyzed locations and to be consistent in its predictions. This new method
was selected as the final method to pick the weather years for hygrothermal designs. The
selected year is the third one out of 30 years in the ranking as provided by the new
equation based method. This selected weather year is assumed to occur approximately
once in every 10 years.
Weather years for hygrothermal analyses were selected for 100 locations in the US and
seven locations in Canada. These weather files are available via a piece of software that
was developed to help print the weather data in different formats as used by different
simulation models.
5
List of Figures
Figure 1. Complexity of selecting representative moisture design year. .......................... 14
Figure 2. Indoor Climate Classes as in the IEA Annex 24 method. ................................. 16
Figure 3. Drying index versus Wetting index for Canadian locations (Cornick 2003). ... 23
Figure 4. Principle of selecting weather years: Use prediction methods to match the
ranking based on ‘real life’ (computer simulated) performance data generated by
simulation models. The methods to select the weather years (Prediction methods) can
produce either predicted damage function values (A) which can be used to rank the
weather years, or they can directly produce the prediction for ranking (B) without
intermediate damage function values................................................................................ 26
Figure 5. Outline of the process to select the method for generating weather data for
hygrothermal analyses. ..................................................................................................... 28
Figure 6. Summary of the tasks included in the process of selecting weather years. ....... 29
Figure 7. Location used in the analyses to investigate the performance of the weather
selection methods. ............................................................................................................. 30
Figure 8. Path of Sun around a building at different times of year in the northern
hemisphere (Canadian Home Builders’ Association, 1995). ............................................ 32
Figure 9. Arctic Circle in the northern hemisphere. (http://athropolis.com/map3.htm) ... 32
Figure 10. The light-weight wood frame wall structure used in the simulations and
weather year selection analyses. ....................................................................................... 35
Figure 11. Schematic view of the CMU-block wall. ........................................................ 36
Figure 12. Damage function values of Time of wetness (TOW), RHT-integral (DT*DRH)
and annual average and maximum mold index in a location as a function of orientation.
Three different weather years were used to produce the data. The majority of driving rain
comes from South-West orientation. ................................................................................ 42
Figure 13. Distribution of normalized damage function (RHT-index) is shown for
different orientations. North/North-East orientation has the highest damage function
values for all 8 locations. .................................................................................................. 43
Figure 14. Atlanta: Moisture content of OSB. .................................................................. 44
Figure 15. Atlanta: Moisture Content of the exterior wythe of the CMU block. ............. 45
Figure 16. Atlanta: Normalized damage function values (RHT-index) for CMU block
wall and for light weight wood frame wall. ...................................................................... 46
Figure 17. IECC (International Energy Conservation Code) climate zones in the United
States. The map has been split horizontally through some of the zones to classes A, B and
C indicating Moist, Dry and Marine climates, respectively. Warm-humid climate
classification line goes through zone three splitting the map vertically. .......................... 47
Figure 18. Average yearly ambient air relative humidity against average yearly ambient
air temperature in 12 locations and for 30 years of weather data. .................................... 51
Figure 19. Average yearly ambient air vapor pressure against average yearly ambient air
temperature in 12 locations and for 30 years of weather data. ......................................... 52
Figure 20. Average yearly ambient air temperature against average yearly total radiation
in 12 locations and for 30 years of weather data. ............................................................. 52
Figure 21. Average yearly ambient air relative humidity against average yearly rain on
the ground in 12 locations and for 30 years of weather data. ........................................... 53
6
Figure 22. Boxplot of average yearly relative humidity (30 years of data) in 12 locations
used in the study................................................................................................................ 54
Figure 23. Boxplot of the yearly damage function value (∆T*∆RH integral aka RHT-
index) in 12 locations used in the study. ........................................................................... 54
Figure 24. Matrix of correlation between parameters in the equation to predict the
damage function with average yearly weather data as input. ........................................... 61
Figure 25. Atlanta (GA): Simulated and predicted damage function data (RHT-index) for
30 years and for the light weight wood frame wall. ......................................................... 64
Figure 26. Fairbanks (AK). ............................................................................................... 65
Figure 27. Memphis (TN). ................................................................................................ 65
Figure 28. Miami (FL). ..................................................................................................... 66
Figure 29. Winnipeg (MB, Canada). ................................................................................ 66
Figure 30. Simulated damage function (RHT-integral) versus the predicted (pred) damage
function using the equation method. All 30 years of weather data and 12 locations are
included. ............................................................................................................................ 67
Figure 31. Damage Function (RHT-integral) is shown here as a contour map in the US
and Canada. Crosses show the locations of the cities with data. Data outside the dotted
area has been extrapolated by the plotting software and should not be considered
accurate. ............................................................................................................................ 69
Figure 32. Annual heavy fog days in the US (NOAA). .................................................... 70
Figure 33. Baltimore: Moisture content of OSB. .............................................................. 75
Figure 34. Chicago: Moisture content of OSB. ................................................................ 76
Figure 35. Minneapolis: Moisture content of OSB........................................................... 76
Figure 36. New Orleans: Moisture content of OSB.......................................................... 77
Figure 37. Portland (ME) : Moisture content of OSB. ..................................................... 77
Figure 38. San Francisco: Moisture content of OSB. ....................................................... 78
Figure 39. Seattle: Moisture content of OSB. ................................................................... 78
Figure 40. Baltimore: Moisture Content of the exterior wythe of the CMU block. ......... 79
Figure 41. Chicago: Moisture Content of the exterior wythe of the CMU block. ............ 80
Figure 42. Minneapolis: Moisture Content of the exterior wythe of the CMU block. ..... 80
Figure 43. New Orleans: Moisture Content of the exterior wythe of the CMU block. .... 81
Figure 44. Portland: Moisture Content of the exterior wythe of the CMU block. ............ 81
Figure 45. San Francisco: Moisture Content of the exterior wythe of the CMU block. ... 82
Figure 46. Seattle: Moisture Content of the exterior wythe of the CMU block. .............. 82
Figure 47. Baltimore: Normalized damage function values (RHT-index) for CMU block
Figure 48. Chicago: Normalized damage function values (RHT-index) for CMU block
Figure 49. Minneapolis: Normalized damage function values (RHT-index) for CMU
block wall and for light weight wood frame wall. ............................................................ 85
Figure 50. New Orleans: Normalized damage function values (RHT-index) for CMU
Figure 51. Portland: Normalized damage function values (RHT-index) for CMU block
Figure 52. San Francisco: Normalized damage function values (RHT-index) for CMU
7
Figure 53. Seattle: Normalized damage function values (RHT-index) for CMU block wall
and for light weight wood frame wall. .............................................................................. 87
Figure 54. Baltimore (MD) : Simulated and predicted damage function data (RHT-
index=DTDRH) for 30 years and for the light weight wood frame wall.......................... 88
Figure 55. Chicago (IL) : Simulated and predicted damage function data (RHT-
Figure 56. Minneapolis (MN) : Simulated and predicted damage function data (RHT-
Figure 57. New Orleans (LA) : Simulated and predicted damage function data (RHT-
Figure 58. Portland (ME) : Simulated and predicted damage function data (RHT-
Figure 59. San Francisco (CA) : Simulated and predicted damage function data (RHT-
Figure 60. Seattle (WA) : Simulated and predicted damage function data (RHT-
Figure 61. Sorption isotherms of materials used in the CMU block wall structure.......... 92
Figure 62. Sorption isotherms of materials used in the light weight wood frame wall
structure............................................................................................................................. 93
Figure 63. Water vapor permeability of materials used in the CMU block wall structure.
........................................................................................................................................... 93
Figure 64. Water vapor permeability of materials used in the light weight wood frame
wall structure. .................................................................................................................... 94
Figure 65. Water vapor permeance of thin materials used in the simulated structures. ... 94
Figure 66. Liquid moisture diffusivity of materials used in the simulations of the CMU
block wall and the light weight wood frame wall. ............................................................ 95
8
List of Tables
Table 1. Durability reference years generate and submitted to International Energy
Agency by the 9 participating members. .......................................................................... 17
Table 2. Rankings related to years 1990-1992 and 1994 (Geving, 1997). ....................... 19
Table 3. Pros and cons of different weather selection methods for hygrothermal analyses.
........................................................................................................................................... 24
Table 4. Weather parameters used in the analyses and included in the weather data....... 30
Table 5. Mold growth index descriptions (Viitanen and Ritschkoff 1991). ..................... 40
Table 6. Top three most severe years as selected by the various damage functions and the
existing weather selection methods. Damage function values are from the simulation
results for the light weight wood frame wall facing north. ............................................... 49
Table 7. Top three most severe years as selected by the various damage functions and the
existing weather selection methods. Damage function values are from the simulation
results for the light weight wood frame wall facing the dominant wind driven rain
orientation. ........................................................................................................................ 50
Table 8. Ranking from the weather selection method is in the left and the ranking of the
weather years resulting from the simulated performance is on the right. The first year
selected by this method is the third year based on the simulation results and this year is at
68 percentile in the normalized range of the damage function. ........................................ 55
Table 9. The selected weather years for Seattle as chosen by different weather selection
methods. Rankings based on simulation results are in parentheses e.g., the most severe
year based on 10% hot method is 20th most severe year based on simulations. ............... 56
Table 10. Average normalized damage function values for the top three weather years as
selected by five different weather ranking methods. The average of all locations per
method is shown in the last row........................................................................................ 56
Table 11. Number of years selected by ANK-ORNL and MI-methods from the top three
worst years as selected based on the simulation results.................................................... 57
Table 12. Regression coefficients and statistic variables with twelve predictors plus
constant for calculating the damage function. .................................................................. 60
Table 13. Stepwise regression analysis of the parameters for the equation to predict the
damage function values as performed by the statistical software. .................................... 62
Table 14. Regression coefficients and statistic variables with eight predictors plus
constant for calculating the damage function. .................................................................. 63
Table 15. Normalized damage function values based on the simulated results of the top
three years for the new equation method and the three other promising methods
(ANK/ORNL, PI-factor and moisture index MI). ............................................................ 68
Table 16. Comparison of the number of years selected from the top three worst years
(simulated data) by the new equation based method versus the top two existing methods.
........................................................................................................................................... 68
Table 17. Basic properties of the materials used in the simulations of the CMU block wall
and the light weight wood frame wall............................................................................... 92
9
List of Acronyms and Symbols
ASHRAE - American Society of Heating, Refrigeration, and Air-Conditioning Engineers
Cloud – Average yearly cloud index, 0-8
CMU – Concrete Masonry Unit
DI – Drying Index, integral of drying out potential over time (year)
DRY – Durability Reference Year
DTDRH – same as RHT index, see RHT
dv,sat – Absolute humidity per unit volume at the outside wall surface for saturated air
dv,out – Absolute humidity per unit volume in the external air
FE - Rain exposure factor
FD - Rain deposition factor
FL - Empirical constant, 0.2 kg·s/(m3·mm)
HAM – Heat, Air and Moisture
HVAC - Heating, Ventilation, & Air Conditioning
IEA – International Energy Agency
IECC – International Energy Conservation Code
k1 – Mold growth delay coefficient 1
k2 – Mold growth delay coefficient 2
M – Mold growth index
mairρair – Airflow due to infiltration/exfiltration
MI – Moisture Index, Ratio of Wetting Index/Drying Index
MDRY – Moisture Design Reference Year
MRY – Moisture Reference Year
N – Number of hours in a year
NCDC – National Climatic Data Center
NOAA –National Oceanic and Atmospheric Administration
OSB – Oriented strand board
PI-factor (or ∏) – Drying out potential, Π = d v , sat (Ts ) − d v ,out
∏ – Drying out potential, kg/m3
QWind-Driven Rain – Moisture due to wind-driven rain
R2 – Coefficient of determination
rh – Rainfall intensity, horizontal surface, mm/h
rbv – Rain deposition on vertical wall, kg/(m2·h)
Rh – Horizontal rainfall, mm/m2h
RH– Relative Humidity, % or – (= range of values 0 - 1)
RHT-index – Relative Humidity Temperature index,
SQ – Surface quality (in mold index equation)
Θ – Angle between wind direction and normal to the wall, °
t – Time
T – Temperature, or °F
TMY – Typical Meteorological Year
TOW – Time of Wetness, number of hours when relative humidity is above 80%
TRY – Test Reference Year
Ts– Surface temperature, °C [F]
10
u – Moisture content of air
U10 – Hourly wind speed at 10m, m/s
U – Hourly average wind speed at 10 m height, m/s
W – Wood species (in mold index equation)
WDR – Wind Driven Rain
WI – Wetting Index, potential for wall to get wet due to wind driven rain
WYEC – Weather Years for Energy Calculations
Parameters in the equation to predict damage function values:

Ypredicted – Predicted damage function
ci – Coefficient for an individual weather parameter
T - Average yearly ambient air temperature, °C
RH – Average yearly relative humidity, %
Rad – Average yearly solar radiation on the wall surface, W/m2
Pv – Average yearly ambient air vapor pressure, Pa
Rain – Average yearly wind driven rain on the wall, mm
Windspeed – Average yearly wind speed, m/s
WindOr – Average yearly wind orientation, degrees
Ni2 – Individual average yearly weather parameter (above) raised to the power of 2
11
Introduction
Advanced hygrothermal simulation models need both the indoor and outdoor
environmental conditions to calculate heat, air and moisture transport through building
envelope parts. Recently the more advanced transient models typically use hourly climate
parameter values such as temperature, relative humidity, solar radiation, wind speed and
orientation, cloud index and rain. It have been demonstrated that the use of steady state
hygrothermal models, Dew Point method (ASHRAE Handbook of Fundamentals 2001)
or Glazer methods (Hens) that use averaged winter and summer conditions do not
provide correct building envelope moisture design guidance. This is particularly true in
envelope systems with high thermal performance.
In a moisture design process, the environmental data should impose a more severe stress
than the average climate in order to provide a level of safety in terms of moisture
performance and durability. Several attempts have been made in the past [Sanders 1996,
Hagentoft 1993, Geving 1997, Karagiozis2003, Cornick 2003, ASHRAE standard 160] to
select the moisture design years for different locations. Most of this work has been
initially carried out in Europe. Indeed a considerable amount of time was spent during the
IEA (International Energy Agency) Annex 24 on developing a method of selecting
weather years for hygrothermal design purposes. A general conclusion at the end of the
IEA Annex 24 and since then is a consensus about the ‘right’ method does not exist while
a number of parallel methods have been developed. It is not known whether any of these
methods can provide proper selections for the North American climate zones that vary
from cold and dry to hot and humid.
This report presents an evaluation, and analysis of four existing methods to select a
hygrothermal design weather year for North American locations (US and Canada). The
methods were evaluated using 30 years ofweather data for eight locations. Two well-
defined structures were simulated with a validated hygrothermal model using each
weather year individually. The results were analyzed using the existing selection methods
to rank the years in the order of severity. The simulation results for the structures were
also analyzedusing so called damage functions to get a ranking of the years based on the
actual hygrothermal performance.Damage functions relate changes in climatic conditions
to changes in specific performance parameter such as mold growth, decay, and corrosion
and can be used as performance predictors. It is assumed here that the selected damage
functions are good predictors for performance ranking that would occur in real life thus
the name ‘true’ ranking. The selection methods for moisture design years and the results
were compared with the rankings based on damage function ranking. The different
selection methods were compared in terms of their consistency in predictions and quality
when compared to the damage function rankings.
In the end none of the existing methods were deemed satisfactory as a consistency was
not achieved for any of these methods, and a new more general methodwas developed.
The proposed method is based on analyzing only the weather data independent of the
structure. In this method the annual averaged value for each weather parameter is used in
12
calculating the corresponding damage function for the wall. The calculation is a simple
equation with no simulations required. The method was consistent in predictions in all
climate zones in the US. For most locations it was capable of predicting the same top
three most severe weather years as determined by results from full hygrothermal
simulations.
Literature Review
Representative and reliable weather data for use in moisture design calculations is as
important as other parameters required for hygrothermal computer simulations. Without
representative weather data realistic boundary conditions cannot be simulated and the real
hygrothermal behavior or response of a building system under evaluation cannot
adequately be addressed. Indeed for example, in the past the creditability of any
hygrothermal analysis was low, as prior to 1991 no wind-driven rain was used. As this
represents a very significant hygrothermal load, and was not included in the weather file
sets, many recommendations and guidelines are outdated. For this purpose selecting and
developing a method to generate representative weather data with all of critical weather
parameters becomes a critical parameter. The following sections provide a state of the art
review of currently available methods towards such a development. The review
examines the benefits (pros) and disadvantages (cons) for each of the approaches listed:
 IEA Annex 24 -10% condensation method,

 PI - Factor method,
 ANK - ORNL method,
 MI (Moisture Index) method.
The main objective of the literature review is to provide a concise overview of the current
knowledge base with a clear focus towards selection and recommendation of a procedure
of choice. The strengths and limitations of each method will be investigated. It must be
highlighted that the project began with several assumptions:
 Despite the availability of several weather data formats (frequencies)

including monthly, bi monthly and daily average, the focus in this project
is on hourly weather data,
 Where possible data sets constructed from measured data were used.
Complexity in selecting representative moisture design year is highlighted in Figure 1

showing a range of parameters that might have to be considered in selecting or
developing a representative moisture design year.
13
Figure 1. Complexity of selecting representative moisture design year.
Reference year and procedures for its determination
The wide range of applicability of reference year data is also reflected by the various
acronyms adopted: Test Reference Year (TRY), Design Reference Year (DRY), which
are used in energy applications, Durability Reference Year (DRY) and Moisture Design
Reference Year (MDRY) used in moisture applications, and Moisture Design Reference
Year (MDRY). In this project we have adopted Moisture Design Reference Year
(MDRY) as an acceptable acronym. Although, review of literature reveals that
significant efforts have been spent in generating reference year data.
Methods for selecting moisture reference years fall into one of two general categories;
construction independent or construction dependent methods. For construction
independent methods, moisture reference years (MDRYs) are identified only on the basis
of weather data analysis. For construction dependent methods moisture reference years
(MRYs) computer simulation models are used to identify representative moisture year in
the context of specific construction type (Djebbar et al, 2001). Moisture conditions
within a construction serve as indicators in selecting (determining) 10% moisture level or
worst weather data. This is a more comprehensive approach as the moisture response of
a construction type for a combination of varying conditions is accounted for. Several
researchers Rode (1993), Sanders (1996), and Geving (1996) developed MDRY selection
methods based on construction dependent approach. Prior to a more comprehensive
review of available methods the concept of reference year is discussedfurther.
A reference year is a summary of external climate conditions for a particular location or

region. The typical climatic parameters considered include: temperature, humidity, solar
radiation, long wave radiation, wind speed and direction, total pressure, and rainfall.The
14
most common use of reference year data has been related to energy calculation such as
energy consumption, air conditioning loads and performance of solar energy systems
(Geving, 1996).
The purpose or the use for which the MRYs were developed imposes some constraints on
the extent of their applicability i.e., the use of reference year data developed for use in
energy simulations might not be applicable to HAM calculations. Two primary reasons
justify this inadequacy (Sanders, 1996):
 The need for additional or different climatic parameters,

 In moisture design process the climatic data should impose a more severe stress
on the structure than the average climate in order to compensate for a factor of
safety (redundancy) regarding moisture damage.
The concept expressed in the latter point is similar to the approach adopted in structural
design. With one specific difference, the margin of safety that can be accommodated in
moisture design is much more tolerant because the risk of structural failure is not as high.
With this in mind, the International Energy Agency has adopted a 10 year return period
as an acceptable safety factor for a (moisture design) reference year (Geving, 1997). The
so called “10%-year” has been denoted as a critical reference year during which
hygrothermal stress imposed onto the construction are severe enough to have occurred
once every 10 years. The following sections review the known approaches utilized in
generating reference year data.
Below we review the existing available approaches for generating representative weather
data. The following methods are reviewed:
- IEA-Annex 24 approach
- Carsten Rode-method
- Geving approach
- Pi-factor method
- Moisture Index method
- ANK/ORNL-method
International Energy Agency – Annex 24 Approach
The Annex 24 approach is based on determination of the 10% level of condensation (90-
percentile) occurring within the construction using one of the three methods: Glaser
method, computer modeling tool (software) such as MATCH or Π-Factor method
(Hagentoft and Harderup, 1993 and Harderup, 1994). The procedure is repeated for five
distinct construction elements: north and south facing walls with impermeable façade,
north and south facing light weight walls and a flat roof. The above mentioned
constructions are simulated to determine accumulation of condensation for each year
starting in October and ending in September for as many years as are available monthly
means of external temperature, relative humidity and solar radiation (not taken into
account with Glaser method). The interior climate is maintained at 21°C and the third
15
climate class indoor vapor pressure is calculated by adding a moisture load of 6 g/m3 to
the outside air moisture content.
9
6 Class 4
3
Moisture load, g/m
Class 3
5 Class 2
Class 1
4
0
-20 -10 0 10 20 30
o
Outdoor air temperature, C
Figure 2. Indoor Climate Classes as in the IEA Annex 24 method.
The mean and the standard deviation of the accumulations for each year is determined.
These values are subsequently used in calculating the 10% level of accumulation for each
construction from the normal distribution function. In this manner an interstitial
condensation due to a 10 year return period is determined. We review in detail how
durabilityreference year (DRY) is determined.
Determination of the existence of DRY
The 10%levelinterstitial condensation can be determined. A possible outcome is that one

specific year of meteorological weather data will meet the 10% level of accumulations in
all five constructions. In that case, one specific year is adopted as the DRY for all five
constructions. This however, is highly unlikely to occur because different types of
constructions i.e., impermeable or lightweight walls, are likely to have a different
hygrothermal response. Different combinations of climatic parameters might contribute
to the 10% level of condensation. The International Energy Agency requires the monthly
averages for the external temperature, relative humidity (or vapor pressure) and solar
radiation to be averaged from the different years for each construction in order to
generate the DRY. For all other scenarios i.e., cases when one or more of the
constructions do not have an identifiable year, which corresponds to the 10% level of
accumulations the following approach must be used to construct the DRY. Begin by
calculating monthly temperature means and standard deviations for each year and for as
16
many years as there is available data. Using the normal distribution function calculate
the 10%, 15%, 20%, 25% and 30% temperatures for each month of the year. Generate
12 regression equations between; external temperature and relative humidity (or vapor
pressure), and external temperature and solar radiation. Use these regressions to calculate
RH and solar radiation from each of the temperatures (30% through 10%) for each
month. This procedure generates a series of different sets of monthly means. Use these
newly developed monthly means (averages) to calculate moisture accumulation in each
of the structures in wall and roofs, starting in October and ending in September.
Determine (select) the year that delivers the 10% level of annual interstitial condensation
for each type of construction. This might or might not be the same. If different years
lead to this level of accumulation for each of the wall, average the monthly values. Table
1lists the durability reference years developed under the framework of Annex 24.
Table 1. Durability reference years generate and submitted to International Energy Agency by the 9
participating members.
Country Station Lat Long Alt Data Method Worst DRY
N E M range years
Belgium Uccle 50.60 4.35 100 1965-1991* Glaser 1965
Glaser+Π-
Netherlands De Bilt 52.10 5.18 3 1961-1980 Factor 63, 65, 66, 67 1977
Germany Potsdam 55.45 13.30 55 1951-1993 Match 56, 69, 63,70, 65 1963
Norway Oslo 59.93 10.73 94 1951-1992 Match 65, 64, 62, 54, 58 1954
Sweden Bromma 59.35 18.07 44 1961-1990 Π-Factor CONST*
Sweden Save 52.70 11.97 41 1961-1990 Π-Factor CONST*
Sweden Lulea 65.66 22.13 10 1961-1990 Π-Factor CONST*
UK Aberporth 52.13 -4.57 133 1951-1991* Match 61, 89, 58, 90, 69 1969
UK Eskdalemuir 55.32 -3.20 242 1951-1991* Match 79, 78, 69, 72, 80 1969
UK Kew 51.46 -0.32 5 1951-1991* Match 53, 54, 68, 69, 82 1968
Slovakia Bratislava 48.17 17.13 153 1971-1990 Match CONST*
Italy Torino 45.07 7.67 138 1969-1992 Match 69, 72, 77, 84, 70 1984
Italy Rome 41.60 12.55 161 1961-1991 Match 78, 89, 76, 84, 72 1976
Denmark Copenhagen 55.68 12.55 9 1959-1989 Match 65, 66, 64, 61, 87 1964
*
The term ‘CONST’ was not defined in the IEA Annex 24 report.
Carsten Rode Method
The Carsten Rode (1993) method is construction dependent method which requires the
use of a reliable HAM simulation tool in determining the hourly MDRY. The MDRY
represents the year characterized by the occurrence of the worst moisture condition i.e.:
highest moisture content integral for a construction or critical layer within a construction.
The procedure involves performing (conducting) HAM simulations for as many years as
there is available hourly data, for several different constructions and with different
orientation. The moisture content integral is calculated for each year, which is then
ranked in accordance with the severity of moisture conditions i.e.: the higher the moisture
content integral the worst moisture conditions in the construction. If a single year is
found to be the worst moisture condition year for all construction, the year is selected as
17
MDRY. If enough data is available, the year with the moisture content integral nearest
the 10% limit is selected as the MDRY (Geving, 1997).
Geving Approach
Geving approach is based on the method developed by Carsten Rode (1993) and the
development within the frame work of the IEA Annex 24 HAMTIE. Geving (1997)
examined the applicability of this approach for different types of climates, varied
construction type, orientation, indoor climate conditions, the duration of simulation, and
initial moisture content of the construction. Geving also included in his evaluation
different approaches to result analysis. For example one of the questions that he posed is
whetherthe average of last calculated value is more appropriate in considering total
moisture content as a ranking criterion?
Geving used 6 constructions with varying vapor permeability, and moisture storage
capacity within the assembly systems. These assemblies are as follows:
 Construction 1 – Lightweight flat roof without a vapor barrier (HAMTIE-Annex

24 construction)
 Construction 2 – Lightweight façade wall without a vapor barrier (HAMTIE-
Annex 24 construction)
 Construction 3 – Lightweight metallic façade wall without a vapor barrier, but
completely vapor tight at the exterior surface, and without hygroscopic materials
(HAMTIE-Annex 24 construction)
 Construction 4 – Compact flat roof with concrete on the inside
 Construction 5 – Wood frame wall having high initial moisture content
 Construction 6 – Concrete wall insulated on the inside
Geving used a 1-D coupled transient heat and mass transfer (the software must consider
mass transfer by diffusion and by fluid flow)computer software to perform calculations.
The model accounted for the hygroscopic inertia of the material. Minimum 10 years of
hourly weather data containing the following parameters: outdoor air temperature,
relative humidity, and global radiation were used for a given location under
consideration. Indoor climate should be maintained constant at 21°C with a moisture
concentration difference between indoor and outdoor at 3 g/m3 (this translates to
approximately 16.4% RH). Other values can be used however; it is recommended that
statistically significant values should be used to reflect the indoor climate (Geving, 1997).
North orientation (in Northern hemisphere) should be used for the wall and horizontal
plane should be used for the roof. Simulation should be carried out for a five year period
starting with October 1st and ending with September 30th, to account for winter time
moisture accumulation effects. Initial moisture content of construction was recommended
to be at equilibrium moisture content at 80% RH. The criteria used in the evaluation
18
includes; (1) the maximum moisture content of the hygroscopic layer just outboard of the
insulation and (2) the average moisture content of the entire construction.
The approach requires carrying out a number of simulations with every year of available
weather data (10 years being the minimum). For each preselected construction types
carry out the simulations and generate maximum and average moisture contents. From
the normal distribution function determine the 10% level 90-percentile) moisture content
for both maximum and average moisture content criteria and calculate mean and standard
deviation. Each construction will have two corresponding values, and a total of 12
MDRYs values are generated for each location.
Geving outlined the following steps in selecting a single MDRY using ranking years.
The requirement in this procedure is that the selected MDRY must give more severe
moisture conditions than the average moisture conditions for every year preferably being
as close to the 10% level as possible. These steps are as follows:
 Rank every year and for each construction in accordance with the average
moisture content value determined with a HAM simulation. The year with the
highest average moisture content is ranked 1st, with the subsequent and
monotonously decreasing years being ranked as 2, 3, 4, 5 and so on. For
example, with 30 years of weather data the 10% level will be represented by the
3rd year.
 Select all the years that occurred in the 5 worst year groups for half or more of
the constructions and calculate the sum of the rankings for every construction.
For example Table 2 shows that 5 years 1989, 1990, 1991, 1992, and 1994 were
each found to be in the 5 worst years at least 5 of the 6 constructions.
Table 2.Rankings related to years 1990-1992 and 1994 (Geving, 1997).

Rankings Avg-MC2 Average
Year Constr 1 Constr 2 Constr 3 Constr 4 Constr 5 Constr 6 ranking
1989 2 4 6 3 4 4 3.8
1990 1 5 5 1 1 1 2.3
1991 4 6 2 4 3 3 3.7
1992 3 7 4 2 2 2 3.3
1994 5 1 1 7 5 5 4.0
 Select two or three years with the lowest sum rankings. Determine the
difference in the sum of rankings for each of these years and the difference in the
sum of rankings with one or several year not selected. If the difference in the
sum of rankings between selected and unselected years is small more years
should be chosen.
 Assuming normal distribution function calculate the average probability level for
each selected year using the average moisture content.
 Repeat the first three steps using the maximum moisture content criteria. Note
that in the second step (step 2) the same years as those selected for the average
19
moisture content must be selected. In this case years 1989, 1990, 1991, 1992
and 1994 must be selected. There could be more or higher number of years
selected (see Table 2)
 Calculate the average probability level for average and maximum moisture
content for each year selected.
 The year with the lowest probability level is selected to be the MDRY for the
specific location. Some exceptions apply:
• For probability levels less than 10% check whether the next lowest
average probability level is closer to 10%. If this is the case that year
should be selected as the MDRY.
• In an instance when the probabilities are very high i.e.: >50%, then two
MDRY should be selected; one for average moisture content and the
other for maximum moisture content,
• In an instance when the probability is in excess of 30% for either of the
average or maximum moisture content, one MDRY for each criteria must
be selected,
• In an instance when the lowest probability is similar for two years, select
one on the basis that average moisture content is more significant than
maximum moisture content, with a preference for uniform probability
across all the construction. Note: avoid selecting a year that has a high
range of probabilities for various constructions.
Geving has also developed and proposed a simplified approach which uses only weather
data as input (temperature, relative humidity and solar radiation) and uses regression
analysis to correlate moisture contents to these. For full description of this approach refer
to his thesis (Geving 1997).
The Pi-factor method
The Π-Factor method is defined by the following equation (Hagentoft and Harderup,
1993):
Π = d v , sat (Ts ) − d v ,out (4)
Where
∏ = Drying out potential, [kg/m3]
dv,sat = Absolute humidity per unit volume at the outside wall surface, kg/m3
dv,out = Absolute humidity per unit volume in the external air, kg/m3
Ts = Surface temperature, °C
The right side of the equation represents the average difference between water vapor
density at the surface of the building i.e.: wall or roof, and that of the air. The difference
in density is thought to be related to drying potential (Geving, 1996). The Π-Factor is
20
small in an instance when the water vapor density difference at the surface and in the air
is small. Thus, the drying out potential is small. With the increase in Π-Factor the drying
out potential likewise increases.
Long period is required for drying wet building materials. The Π-factor is determined for
each year from the available hourly water vapor concentrations. The 10% percentile of
the Π-factor corresponds to the 10% year i.e.: year in which 10% of moisture
accumulation would be expected. Researchers, Karagiozis (1998) have highlighted that
construction type in particular the location of vapor impermeable layers within a
construction could have a significant effect on its hygrothermal performance and must be
accounted for. Additional limitations are that this method only includes surface drying
potential without capillary or other transport phenomena.
Moisture Index Method
The method was developed in MEWS project uses Moisture Index (MI) to select
(determine) representative moisture design year. The moisture index approach (Cornick
et al., 2003) consists of a wetting and a drying function. In this respect the moisture
index can be related to a moisture balance, and it can be visualized as a storage container
in which the level of water changes depending whether wetting or dryingis taking place.
Thus the wetting and the drying functions describe the source and sink components of a
moisture balance. Equation 5describes the general form of moisture index.
MI = function (Wetting Index, Drying Index) (5)
The wetting index (WI) can be defined as the potential for wall to get wet (Cornick et al.,
2002). Two measures have been defined as being suitable (adequate) in determining the
wetting potential; total annual average rainfall, and the annual driving rain index. The
former represents total annual average rainfall on a horizontal surface. The annual
driving rain index is a sum of products of hourly wind speed and hourly horizontal
rainfall (Equation (6)). Surry et al, (1995) and Choi (1994) have shown a strong
correlation between the two parameters.
n
aDRI = ∑ U 10 ⋅ Rh (6)
t =1
Where
U10 = hourly wind speed at 10m, m/s
Rh = horizontal rainfall, mm/m2h
n = number of hours in a year.
The drying index (DI) is defined as the potential for removal of moisture from the wall
via evaporation process (Cornick et al., 2002) and it is similar to the Pi-factor
approachwhich calculates the saturation vapor pressure deficit.
21
Next the wetting and drying indices were normalized in accordance with Equation (7).
Both the wetting and the drying indices were normalized.
INormalized = (I – IMin) / (IMax – IMin) (7)
The moisture index is then calculated in accordance with Equation (8).The annual rainfall
can be selected as the parameter for the wetting function. The wetting index is calculated
from the driving rain only for a specified orientation. The drying function represents the
measure of “headroom in the air” which can store moisture (Cornick et al, 2003). It is
calculated from the dry bulb temperature and relative humidity at hourly interval in
accordance to Equation (4).
MI = WI Normalized
2
+ (1 − DI Normalized ) 2 (8)
Where
MI = Moisture Index
WI = Wetting Index
DI = Drying Index.
The drying index is defined as the sum of hourly values for the year. Figure 3 shows
Wetting and Drying Indexes for Vancouver for 41 years of data. The plot highlights that
with increased wetting there is a corresponding decrease in drying. Using moisture index
each year can be defined as dry(lowest moisture index), average (mean moisture index)
or wet (highest moisture index). In the MEWS project the average years were defined as
being within one standard deviation of the mean, and dry and wet years were defined as
remaining years which were more than one standard deviation from the mean moisture
index value.
22
Figure 3. Drying index versus Wetting index for Canadian locations (Cornick 2003).
The ANK-ORNL method
The ANK-ORNL method was developed by Karagiozis (2003) and was further refined to
include the impact of air flow by Karagiozis (2005). This method includes both static and
dynamic hygrothermal loading. The static components are those loads that include the
amount of hygric potential due to ambient conditions. The moisture load present in the air
is summed over a period of a year and is included as a bulk moisture quantity. The
moisture potential is the key component in this method.
The ANK method is also the only method that includes the potential impact of air flow
through the structure.The method includes the potential for moisture deposition due to
hourly interactions of infiltration and exfiltration. The pressure field due to the wind
speed and orientation is calculated and applied to the wall structure. As such the method
is best evaluated when a hygrothermal model that includes the impact of air flow is
deployed.
∑ Hygric Load = ∑t =0 u + ∑t =0 mair ρ air + ∑t =0

e,w 8760 8760 8760
n,s
QWind − Driven Rain (9)
23
Where:
Hygric load = Yearly moisture load potential kg of water

u = Moisture content of air
mairρair = Moisture due to infiltration/exfiltration
QWind-Driven Rain = Moisture due to wind-driven rain
The hygric load includes all the hygric contributions available for the structure to
accumulate moisture. The method assumes the most absorptive wall system with a very
large moisture capacity, maximizing the impact of the various climatic loads, and
minimizing the impact of the type of wall structures.
The higher the hygric load the greater potential to cause moisture induced damage. The
hygric load provides the net moisture available due to diffusion, capillary transport
(wind-driven rain) and air flow movement in a particular year. The method has been
implemented in the weather file analyzer provided by ORNL, Karagiozis (2002). The
approach is implemented as follows:
1) Sum the hygric loads due to water content of the air, the wind-driven rain and air
transport for each hour.
2) Repeat this summation for each available year
3) Rank the years in terms of the hygric potential
4) Select the worst year, and the 10 % year
5) Deploy these two years in succession, i.e. the 10 % year first and then the worst
year for simulation purposes and repeat if more than two years of analysis is
needed.
Summary of Advantages and Disadvantages of Each Method
The advantages and disadvantages of different methods arediscussed in Table 3.
Table 3. Pros and cons of different weather selection methods for hygrothermal analyses.
Method Pros Cons
IEA-Annex 24 approach Looks at different Indoor climate fixed to
constructions (walls, roofs) 21°C and 3rd climate class
for vapor pressure (Figure
2).
Monthly average weather
parametersT, RH, solar
radiation only as
parameters.
Simple constructions
without mass.
Orientations limited to
North and South
Carsten Rode-method Use of real climatic data It is a rare occurrence for
24
(hourly), which implies that the same year to have worst
true sequence frequencies moisture condition effects
and correlations are for all constructions
represented Criteria other than moisture
The effect of climate on content time integral might
moisture conditions in be more applicable
different constructions is (suitable) for comparison of
evaluated moisture conditions in
Multiple orientations construction
investigated Influence of various factors
including: orientations,
construction, exterior and
interior climate on the
selection method remain
unclear
Time consuming method
(approach), extensive
computational effort is
required in conducting
simulations and analyzing
data
Geving approach Two methods were used: Full method is very labor
Full construction method intensive. Simplified
(1) and a simplified method takes only limited
regression method (2). Full number of weather
method (1) : Six type of parameters into account.
construction used. Driving Minimum 10 years of
rain is not considered in the weather data might not be
selection process. Methods I representative (greater data
evaluated for 1-D cases and pool might be required).
cases with no air flow.
PI-factor method Easy to calculate and does Provides only drying index
not depend on construction. and does not look at wetting
due to rain or drying due to
solar radiation. No
orientation effects
considered.
Moisture Index-method Combines wetting and
drying processes into one
index
Construction independent
ANK/ORNL-method Combines air leakage into Uses a preselected type of
the function. construction, no moisture
storage
25
Methodology to Evaluate Existing Weather Selection
Methods
The hygrothermal design weather years will be used by advanced simulation models to
predict the performance of building envelope parts. It is therefore natural that the
approach to select the weather files for design purposes includes using simulation models
to generate performance data that represents the ‘real life’ performance. This
performance data serves as the reference data set that would then be used to select the
most severe years. Other prediction methods would have to try to match the ranking
produced by this reference data set. The above approach assumes as a prerequisite that
the model that is used in predicting the performance has already been validated and the
model can produce reasonably reliable predictions of the hygrothermal performance of
the building envelope components.
Performance data has to be analyzed by using various methods to provide a ranking of

the weather years in terms of severity. This ranking is done by calculating values for
different damage functions that use the simulated performance data as input. These
damage functions indicate the severity of the weather for the durability and service life of
the structure. Examples of damage functions are e.g., time of wetness (relates to
corrosion) and mold index (indicates conditions favorable for mold growth).
The principle of the method is visualized in Figure 4.
Figure 4. Principle of selecting weather years: Use prediction methods to match the ranking based on
‘real life’ (computer simulated) performance data generated by simulation models. The methods to
26
select the weather years (Prediction methods) can produce either predicted damage function values
(A) which can be used to rank the weather years, or they can directly produce the prediction for
ranking (B) without intermediate damage function values.
Evaluation process
The process to evaluate the existing weather year selection methods against the simulated
performance includes several steps that are outlined in the following.
1. Select
a. Weather data source
b. A simulation model to produce performance data
c. Wall structures and materials to be simulated
d. Locations for analyzing the performance of selection methods
e. Damage functions to rank the simulated performance
i. Select critical locations to investigate in the wall structures
f. Indoor climatic conditions
2. Carry out simulations for the wall structures
a. Use each weather year individually
b. Produce hourly thermal and moisture data for damage functions and
selection methods
3. Study the effect of orientation
a. Select one orientation or a method to select the orientation for main
simulation series
4. Rank weather years in the order of severity based on
a. Existing weather selection methods and
b. Damage functions
5. Compare rankings against each other
a. Do different damage functions pick the same years?
b. Select one damage function for further rankings to be used as reference
performance rankings
c. Do existing weather selection methods pick the same or very different
years with damage functions?
i. If yes, pick an existing method
ii. If no, create a new method
The evaluation process is visualized in Figure 5.
27
28
Figure 5. Outline of the process to select the method for generating weather data for hygrothermal
analyses.
Figure 6 summarizes the major tasks undertaken in the process of selecting the method to
produceweather years.
Task 1 Task 2 Task 3 Task 4
Task 5 Task 6
Review Critical Evaluation Use 5 Climates Characterize Generate Final Report
of of moisture design and Weather data weather and
existing years demonstrate Into Zones data for 100 documentation
data selected locations
approach Reduce hourly
data to weekly
5 Climates or monthly
values
5-different
methods
(16 climates)
Output Output Output Output Output Output

State -of -the - Update on Validated Feasibility, CD’s with Final Report
art review physical method maps Weather data
factors
Figure 6. Summary of the tasks included in the process of selecting weather years.
Available weather file sources and formats
At least 30 years or more of weather data is preferred in order to generate statistically

meaningful hygrothermal design years. Data sets must include complete sets of data or
the missing gaps in data must have been filled with artificial data. The weather data must
also include the important parameters needed by the advanced hygrothermal simulation
models – most importantly these include temperature, relative humidity, solar radiation
and rain with wind speed and orientation.
There are many weather data sets available in the US such as Typical Meteorological
Years (TMY) and Weather Years for Energy Calculations (WYEC). However, many of
these data sets are based on the same source of measured data and are indeed derivations
of the original sets and typically include one year of weather data per location. The
source provider of the raw or base weather data is in most cases the National Climatic
Data Center.
The available weather data sets for the United States are:
- NOAA-NCDC SAMSON data sets (1961-1990)
- NCDC update 1985-2005.
The NCDC SAMSON data set that comprises of 30 years of hourly data collected from
1961 to 1990 was used in the analyses. The update for the NCDC data set including
weather data till 2005 has recently become available but was not used in this project.
29
Weather parameters included in the analyses
The weather parameters included in the weather files and analysed were those listed
inTable 4. These parameters are used and needed by most of the advanced hygrothermal
simulation models.
Table 4. Weather parameters used in the analyses and included in the weather data.
Weather parameter Comments
Air temperature Dry bulb temperature
Relative humidity Relative humidity at dry bulb temperature
Solar radiation (Direct normal, diffuse) Reflected radiation from ground can be
assumed to be 30% unless the ground
reflectance is better known. The effect of
reflected solar radiation would be more
pronounced when snow covers the ground.
Sky radiation Represented by either cloud index or
measured long wave radiation from the
sky.
Wind Speed and direction
Rain Horizontal rain
Locations used in the analyses

The wide ranges of climatic conditions in the North America have been split to eight
climate zones. For the purpose of comparing different weather selection methods eight
(8) locations were selected to represent the different IECC climate zones.
Figure 7. Location used in the analyses to investigate the performance of the weather selection
methods.
30
Orientation effects
Effects of weather on the performance of building enclosure components can be very

different depending on the orientation they are facing. Solar radiation, rain and wind
loads vary widely depending on the building orientation. Orientation thus affects the
hygrothermal loading imposed on the building enclosure components.
Solar radiation
Solar radiation has three main components, direct, diffuse and reflected radiation. Direct
radiation depends on the angle of incidence which is the angle between the surface
normal (a line from the centre of the object directly outwards at 90° to the surface) and a
line from the object centre running out towards the sun. The intensity of solar radiation
on the surface is the beam radiation (direct radiation when measured towards the sun)
times the cosine of the angle of incidence.
In the northern hemisphere the north facing wall gets usually the least amounts of solar
radiation (Figure 8) due to the Sun’s path around the building. In the southern
hemisphere, the south facing orientation would receive the least radiation. There may be
some exceptions to this rule: In the middle of the summer in the latitudes north of the
Arctic Circle (in 2009, it is the parallel of latitude that runs approximately 66° north of
the Equator. 1) the sun never sets. In the middle of the ‘night’ the Sun is facing the north
facing wall at almost perfect perpendicular angle. The Arctic Circle marks the southern
extremity of the polar day of the summer solstice in June and the polar night of the winter
solstice in December. Within the Arctic Circle, the arctic Sun is above the horizon for 24
continuous hoursat least once per year, in conjunction with the Arctic's Summer Solstice
- this is often referred to as “Midnight Sun”. In the United States only some parts of
Alaska are inside the Arctic Circle (Figure 9).
1
The position of the Arctic Circle is not fixed, but instead depends on the Earth's axial tilt which again is
not a constant but fluctuates approximate 2° over a 40000 year period.
31
Figure 8. Path of Sun around a building at different times of year in the northern hemisphere
(Canadian Home Builders’ Association, 1995).
Figure 9. Arctic Circle in the northern hemisphere.(http://athropolis.com/map3.htm)
Wind driven rain effects

Rain will fall on the walls primarily only when the wind drives the droplets and moves
them horizontally thus the name Wind Driven Rain (WDR).
32
Wind driven rain will impose a rain load on the wall in proportion to the wind speed,
wind direction, and rain intensity. Many authors have developed equations to estimate the
rain loads using the average hourly weather data (Lacy, Künzel, Straube). Most of these
methods take one value for the wind direction when averaged over time (hour). Wind
often varies in direction during the hour and it is possible that all orientations will receive
wind driven rain. However, since the available weather data typically only has hourly
averaged data (sometimes even longer e.g. 3 hour averaged data) the rain loads due to
wind fluctuations at shorter than one hour intervals are not taken into account in this
study.In the ASHRAE Standard 160 the linearization of Lacy’s formula was selected to
calculate the wind driven rain (Equation 10).
rbv = FE ⋅ FD ⋅ FL ⋅ U ⋅ cos θ ⋅ rh (10)
where
FE = Rain exposure factor, -
FD = Rain deposition factor, -
FL = Empirical constant, 0.2 kg·s/(m3·mm)
U = hourly average wind speed at 10 m height, m/s
θ = Angle between wind direction and normal to the wall, °
rh = Rainfall intensity, horizontal surface, mm/h
rbv = Rain deposition on vertical wall, kg/(m2·h)
The exposure factor, FE, is influenced by the topography surrounding the building and
height of the building. The deposition factor takes into account the roof slope and
exposure to rain runoff.
In addition to applying the wind driven rain load on the exterior surface of the wall a
portion of the rain deposition on the wall is assumed to penetrate past the exterior surface.
ASHRAE standard 160 uses 1% as the default value for water penetration rate, and the
exterior surface of the water-resistive barrier as the deposit site.
Selected simulation models
MOISTURE-EXPERT (Oak Ridge National Laboratory) and WUFI-Pro were selected to

be used as the simulation tools for the analyses. MOISTURE-EXPERT was used as the
main tool; WUFI-Pro was used to investigate possible model interdependencies of the
results and to carry out some sensitivity analyses. Currently it is the USA industry
standard tool used in hygrothermal design analysis.
MOISTURE-EXPERT was used in the ASHRAE Research Project 1091 and has been
extensively validated against experiments. It is a two-dimensional model that allows for
taking into account factors such as air and water leakage that were mostly not available in
one-dimensional tools at the time when the project was started.
WUFI-Pro is the professional version of widely used WUFI-ORNL model. Since this
model is wide spread and is known to be consistent with MOISTURE-EXPERT it was
33
decided that this model could be used for some sensitivity analyses. The simulations were
carried out in one-dimensional domain because of the expected large number of runs
needed for the investigations.
Wall systems selected for analysis
Two wall systems selected for comprehensive analysis to evaluate the proposed
methodology for weather year selection include a “light weight” wood frame wall and a
“heavy weight” concrete masonry block wall.
The two types of walls were selected to evaluate whether the proposed methodology in
selecting weather data is independent of wall structure.
Wall 1 –“Light-weight” wood frame wall

The light-weight wood frame wall system has an absorptive cladding and includes awater
resistive barrier (WRB). Properly designed wall systems for each climatic location were
chosen. However, instead of changing the wall structure depending on the location
(climate zone), it was chosen to fix the material layers and always have a vapor retarder
in the wall structure. A semi-permeable or a vapor retarder with variable permeance
allows for some vapor movement through the interior layers while still resisting vapor
ingress into the wall. Therefore a kraft paper vapor retarder was chosen instead of a
polyethylene vapor barrier, because it is an appropriate choice for all climates considered.
The selected light-weight wall structure is presented inFigure 10. It consists of the
following layers listed from outside to inside: conventional stucco with an acrylic finish,
60-minutesasphalt impregnated paper based WRB, oriented strand board (OSB), 2x4
wood frame wall with kraft paper faced fiberglass batt insulation (R-13) and drywall
(gypsum board) with primer and latex paint.
34
2” x 4” framing
(Kraft paper)
R-13
Figure 10. The light-weight wood frame wall structure used in the simulations and weather year
selection analyses.
Wall2 – Concrete Masonry Unit (CMU)-block wall

The CMU-block wall consists the following layers listed from outside to inside:3-5/8 in
brick cladding, 1 in air cavity (non-vented), 8 in CMU-block, 4 in metal frame wall with
kraft paper faced fiberglass batt insulation (R-13), kraft paper, drywall and one coat latex
primer and one coat paint layer.
35
Figure 11. Schematic view of the CMU-block wall.
Procedure for running the simulations
In order to find out the effect of individual years on the performance of the structures the
simulations can be performed in different ways. The initial conditions of the material
layers may affect the way the wall responds to the environmental conditions and loads.
The buildings are exposed to the weather years in a sequential fashion in the course of
their service life and any year is an appropriate starting point for the simulation..
1. Simulate each year individually beginning inOctober (ASHRAE Standard 160).

2. Run a sequence of years (i.e.. 30 continuous years) and each subsequent year
starting with initial conditions in the wall corresponding to the conditions at the
end of the previous year.
36
The orientation is a well-known factor that affects moisture performance of walls.
Solar radiation and Wind Driven Rain (WDR) loads that impact the wall can be
different depending on the orientation. The orientation should be selected before
running the simulations for all locations using 30 years of weather data. Wall
orientation effects were investigated running the simulations using eight (8) different
orientations for a number of locations. The Wind Driven Rain dominating direction
was also investigated as part of this sensitivity study.
Indoor air and initial conditions

Interior conditions were created following the ASHRAE Standard 160 intermediate
method with air conditioning. Initial conditions were set to the equilibrium moisture
content equivalent to 80% relative humidity for all material layers. This corresponds to
criteria set in ASHRAE Standard 160 when measures have been taken to protect building
materials from water during construction.
Use of damage functions for selecting weather years
The simulation results were analyzed using damage functions with the aim to create the
ranking of weather years based on each function.
A whole range of physio-chemical and mechanical processes as well as improper design

and implementation practices are responsible for building envelope deterioration. The
ability of construction materials to transfer loads depends on the nature and magnitude of
the applied stresses, moisture contents and temperature. Moisture retained in the building
envelope is a function of both the ambient environmental conditions as well as the
microstructure of the materials. Damage functions are used for estimating hygrothermal
damage for the materials in the assemblies. Damage mechanisms can be very different for
different materials: wood can rot and grow mold at high moisture content and relative
humidity, frost is a typical cause of damage in bricks, steel corrode, and paints and
coatings can crack due to various reasons. “Damage” can be either structural or
aesthetical without the expectance of life threatening failures or diminished service life.
Damage modes can be very different depending on the climate. In cold climates freeze-
thaw cycles and condensation in the winter can cause severe problems, whereas in hot
climates high temperatures and UV-radiation (in all climates) can weaken the structures,
allowing for crack formation and subsequent water leakage due to wind driven rain. In
both climates excessive moisture and temperature fluctuations are a cause for undesirable
deformations.
Different “damage functions” have been developed over the years to estimate the
deterioration or damage on the materials and structures. These are used to quantify the
conditions in such a way that would allow for predicting the durability or service life.
37
In this study the damage functions were not used to predict the long term performance of
the structures but rather to provide criteria data to allow for ranking the years in terms of
the stresses they cause to the materials or assemblies. A high damage function value
based on the response of the structure in the simulations would indicate a severe year
with respect to hygrothermal performance and therefore it should be used in the design of
well performing and durable building envelopes.
Critical layers in the walls for investigation

The critical layers were selected based on the simulation results for further analyses with
damage functions. These layers were chosen to be the exterior sheathing (Oriented Strand
Board) in the stucco clad wall and the outer wythe of the CMU block.
Rankings based on damage functions

Based on the simulation results and damage function analyses the critical moisture design
years for the locations were selected by ranking the damage function values in the
decreasing order.
The damage functions selected for the purpose of this study were
1. Time Of Wetness (TOW)
2. RHT-index
3. Mold Index
4. Maximum moisture content
The first three damage functions are of integral type. In many cases the hygrothermal
loading – e.g. wetting and drying cycles – causes permanent changes in the material
allowing for more and more damage in the subsequent cycles. Wood for example has to
be subjected to excess moisture over extended periods of time. Therefore, the integral
damage functions are likely to be more prescriptive of the severity of the whole year.
Maximum moisture content may not necessarily express the loads over the whole year,
but may in some cases be an indicator of intense short term loads only.
The rate of corrosion can be severely affected by the proximity to salt water and
pollutants in the air. The weather files do not include pollutants or chemical contents in
the air streams and therefore these potential effects were not included in this study.
Time of Wetness
Time of wetness (TOW) is calculated as time in hours when both the temperature and the
relative humidity are above prescribed critical levels. The values that a TOW function
takes are either 0 or 1. Commonly used reference values are 0°C for temperature and 80%
for relative humidity. For a full year the value of time of wetness can range between 0
and 8760. Alternatively the TOW can be expressed as a percentage of the potential time
of wetness, 50% being wet 50% of the time.
38
RHT-index
RHT-index (Mukhopadhyaya et al., 2003) (or DTDRH integral) is similar to the time of
wetness. Instead of counting only the hours when the conditions are satisfied the actual
value of the following integral is calculated (if T>TL and RH>RHL) as in Equation (12):
RHT = ∑ (T − TL ) ⋅ ( RH − RH L ) (12)
The limiting values are typically chosen to be the same as for the time of wetness: TL =
32F (0°C) and RHL = 80%. The limiting value for the relative humidity was however
later changed to a lower 70% for practical reasons – namely because in some climates the
conditions inside the walls happened to stay so dry that the damage function would not
show any results that would allow for proper ranking of performance.
Mold growth index

Presence of mold fungi indicates increased humidity and moisture levels in buildings.
Mold fungi are able to grow on practically any building material, including wood. They
do not significantly affect the strength of wood or other materials but cause mostly only
discoloration. The most severe effects are odor, poor indoor air quality and health
problems.
Viitanen et al. (1997) measured critical time for mold fungi germination and the growth
rates at different temperatures and humidity on pine and spruce samples. The tests were
conducted in steady-state and dynamic conditions. The results revealed that the initial
time needed for mold growth follows an exponential relationship in terms of relative
humidity and temperature. Time needed for growth of mold fungi to start is longer at low
temperature than at higher temperature. Similarly low relative humidity delays the
growth. Growth of mold fungi requires temperature to be above freezing (0°C/32F) and
below 50°C/122F. Very high temperature can destroy the micro-organisms and stop mold
growth. The relative humidity needs to be above 75% or higher depending on the
temperature. The growth initiation times have been proven to be short (from a few days
to a few weeks) in pine sapwood in conditions favorable to the growth of micro-
organisms and long (from a few months to a year) in conditions close to the minimum
and maximum moisture or temperature levels. In the final model the minimum critical
relative humidity was set to 80%. Availability of nutrients on the surface was shown to be
more significant than the effect of wood species. This also indicates that mold can grow
on any surface and mold index can therefore be used as a comparative indicator for
performance review.
Under the favorable conditions mold growth may start and proceed at different rates
depending upon the interrelationship between humidity and temperature and other factors
such as the organisms and the properties of the materials.
The present VTT model consists of a mathematical model that also takes into account the
delay in mold growth rate due to the unfavorable conditions. There are different
mechanisms for situations with delay:
39
a) The early stages (germination of spores)
b) Too dry and too cold conditions and
c) The late stages of mould growth.
The model also accounts for the growth under fluctuating humidity conditions. Equation
(13) solves the mould growth index M as a function of time (weeks) t, temperature T,
relative humidity RH, the wood species W (0 =pine and 1 = spruce) and the surface
quality SQ from the drying process. The coefficients k1 and k2 represent the delay in early
and late stages of growth, respectively. These parameters and the model itself is
described more detailed in (Hukka and Viitanen 1999) and (Viitanen et al. 2000).
dM 1
= k1k 2 (13)
dt 7 ⋅ exp(−0.68 ln T − 13.9 ln RH + 0.14W − 0.33SQ + 66.02)
Hukka and Viitanen developed the mathematical model for mold growth with regression
analysis of the measured data, which is expressed as mold index (Hukka and Viitanen
1999). The index is defined as in Table 5. The index values relate to mold growth that
can be detected visually or using microscopy. The presented mold index is based on the
growth activity of different mixed mold species.
Table 5. Mold growth index descriptions (Viitanen and Ritschkoff 1991).

Index Growth rate Description
0 No growth Spores not activated
1 Small amounts of mold on surface (microscope) Initial stages of growth
2 <10% coverage of mold on surface
(microscope)
3 10-30% coverage mold on surface (visual) New spores produced
4 30-70% coverage mold on surface (visual) Moderate growth
5 > 70% coverage mold on surface (visual) Plenty of growth
6 Very heavy and tight growth Coverage around 100%
Maximum moisture content

Moisture content of the material layers in the simulated wall assemblies are recorded for
every hour of the simulation period. The maximum moisture content is the highest
moisture content of the critical layer at any hour of the. The higher the maximum
moisture content is in the structure at any given year, the greater the likelihood of high
moisture loads and/or poor drying capability.
40
Simulation Results
The two walls – the light weight wood frame wall and the CMU-block wall – were to be
simulated in eight locations in the US using 30 years of weather data. Since solar
radiation and wind driven rain depend heavily on orientation it became evident early on
in the project that the orientation effects would have to be investigated first prior to
conducting simulations with 30 years of weather data.
Analysis of the orientation effects

Orientation can play an important role on the performance of the wall. This can be due to
the difference in the amount of solar radiation and wind-driven rain that the wall is
exposed to. In order to find out which orientation should be used in the final analyses to
select the weather years a series of simulations were carried out.
The light weight wood framed wall was simulated facing eight (8) orientations at 45
degree intervals: North, North East, East, South East, South, South West, West and North
West. The performance of the wall was then analyzed using damage functions to rank the
years in order of worst performance. Eight locations from different climates zones were
analyzed. The results of simulations with the light-weight wall in New Orleans climate
are shown in Figure 12.Three different years are shown for each damage function. It is
apparent that for this location regardless of the year or damage function that the North is
clearly the orientation of interest. In this location the bulk of the driving rain comes from
South-West.
41
TOW DT * DRH
9000 7000
8000
6000
7000
5000
6000
Year1L 4000 Year1L
5000
Year2L Year2L
4000 3000
Year3L Year3L
3000
2000
2000
1000
1000
0 0
N NE E SE S SW W NW N NE E SE S SW W NW
Average Mold Max Mold
1.4 4.5
1.2 4
3.5
1
3
0.8 Year1L Year1L
2.5
Year2L Year2L
0.6 Year3L 2
Year3L
0.4 1.5
1
0.2
0.5
0
0
N NE E SE S SW W NW
N NE E SE S SW W NW
Figure 12. Damage function values of Time of wetness (TOW), RHT-integral (DT*DRH) and annual
average and maximum mold index in a location as a function of orientation. Three different weather
years were used to produce the data. The majority of driving rain comes from South-West
orientation.
Results – Damage functions for different orientations

The desired result of the simulations would have been that one orientation would
consistently be the deciding factor and could be used as the worst case scenario. The
rankings of the weather for different years were slightly different depending on the
damage function – mold index typically generated the highest damage function values in
the North-East sector. However, in some instances the orientation with the highest wind-
driven rain could be the worst. Overall, damage functions ‘time of wetness’, RHT-
integral and mold index all showed same trends in terms of orientation. Everything
indicated that the north orientation would be the best candidate for the worst orientation.
Figure 13 shows the normalized damage function values for lightweight walls facing
eight different orientations. The data generated on this graph was normalized by scaling
the RHT-integral from 0% being the minimum and 100% being the maximum.
42
Design year - Relative Damage Function for orientations
N
100%
90%
80%
NW 70% NE
60%
50%
40% Atlanta
30%
Baltimore
20%
Chicago
10%
Minneapolis
W 0% E
NewOrleans
Portland
SanFrancisco
Seattle
SW SE
Figure 13. Distribution of normalized damage function (RHT-index) is shown for different
orientations. North/North-East orientation has the highest damage function values for all 8 locations.
Simulation results – Moisture contents of OSB in wood frame

wall
The moisture contents of the exterior sheathing layer (Oriented Strand Board) are shown
in the following figures as a function of time when 30 years of weather data were used
individually in eight locations in the US. The initial moisture content in the OSB
corresponds to 80% relative humidity. The course of moisture content is shown in the
simulation results for Atlanta, GA when each individual weather yearwas used ( Figure
14). The results for the remaining 7 locations are included in Appendix A. The damage
function (RHT-index) was used to rank the performance of the wall in the weather
location. The ‘Least severe’ and ‘Most severe’ refer to the moisture contents in the OSB
in those (three) years that resulted in the lowest and highest damage function values
(maximum moisture content in this case), respectively.
43
Moisture Content, kg/kg [lb/lb] 0.2
0.15
0.1
Least severe
Easiest 3
Most
Worstsevere
3
0.05
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 14. Atlanta: Moisture content of OSB.
Simulation results – Moisture contents in CMU

The following figures show the moisture contents in the CMU block wall with the results
marked black and red for those years that were ranked worst and easiest years based on
ranking of the wood frame wall simulations data. Data here is for CMU walls and the
moisture content in the CMU block’s exterior layer. The black lines in Figure 22
represent the worst weather years for the CMU block wall simulated with Atlanta, GA
weather data. The same three worst years in CMU block wall are appear to be among the
top worst yearsfor the wood frame wall in previous section (Figure 14). For the two
types of constructions examined the same three years with weather data represented the
top worst years. Figure 22 is shown as a sample and the results for the remaining 7
locations are included in Appendix B. It is important to highlight that for San Francisco
weather data the agreement between the three worst weather data years selected did not
predict the highest moisture content in the CMU wall. Simulations conducted with
Seattle weather data showed that two out of the three worst weather years predicted
highest moisture contents in simulations. This trend was not observed in simulation
results with the wood-framed wall.
44
0.04
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 15. Atlanta: Moisture Content of the exterior wythe of the CMU block.
Rankings of weather years based on RHT-index – Wood frame

wall versus CMU wall
The method to select design weather years for moisture simulations would preferably be
independent of the structural design. In this project lightweight wood-framed and CMU
block walls were simulated and the question that arises is how well does the performance
of the two systems compare with selection of the weather years? To answer this, the
damage function RHT-index was used in the comparison by normalizing the damage
function values for each wall in individual locations. Both walls have the normalized
values ranging from 0% to 100% in every city which allows for visually comparing the
performance during different years and between the two walls.
The following chart Figure 16shows the relative damage function value for the RHT-
index for the 30 simulated years in Atlanta, the data for the rest of the locations is shown
in Appendix C.
The graphs show that the normalized results for the damage function values for the
CMU-block wall are similar to the values in the wood frame wall with the exception of
San Francisco. We therefore limited the further analysis to the wood frame wall only.
45
46
LWW
CMU
Figure 16. Atlanta: Normalized damage function values (RHT-index) for CMU block wall and for
30
25
20
Year, -
15
10
light weight wood frame wall.

5
0
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Relative Damage Function, %
Evaluation of Weather Data Selection Methods
Five existing weather data selection methods for the development of moisture weather
design years have been evaluated based on the results from the simulations in eight (8)
US locations (1 from each IECC Zone).
The selected locations were (IECC climate zone in Figure 17):
1. Seattle, Washington (4c)

2. New Orleans, Louisiana (2)
3. Minneapolis, Minnesota (7)
4. Chicago, Illinois (5a)
5. Atlanta, Georgia (3a)
6. Portland, Maine (6a)
7. Baltimore, Maryland (4)
8. San Francisco, California (3c)
Figure 17. IECC (International Energy Conservation Code) climate zones in the United States. The
map has been split horizontally through some of the zones to classesA, B and C indicating Moist, Dry
and Marine climates, respectively. Warm-humid climate classification line goes through zone three
splitting the map vertically.
The weather data for this part of the work is 30 years (1961-1990) from National
Climatic Data Center’s (NCDC) SAMSON database. The NCDC update for 1985-2005
became recently available but was not used in the analyses. A light weight wood frame
wall was used in the simulations for each location.
47
Analyses were conducted to determine estimates of sensitivity of weather data in a
similar manner to Crawley (1998) by employing the MOISTURE-EXPERT hygrothermal
simulation model. The purpose of these analyses was to confirm the applicability of
various methods for determining moisture reference years for moisture design
calculations – and to select a method for further weather analyses.
Possible problems were foreseen to arise from the fact that different construction types
can have different critical weather parameters and the same weather file may not be the
most critical year for different wall types. However, a decision was made early on that
only the selected two wall systems were to be used in this project. The weather data is not
the only deciding factor on the moisture performance. Engineers still need knowledge
and understanding of the fundamentals of moisture movements in addition to just
knowing how to run simulation models.
Existing methods for selecting weather data for moisture

simulations
The following methods were investigated by using them to select design weather years
and comparing their selections against the simulated performance of the light weight
wood frame wall. The methods were:
 IEA Annex 24
o 10% percentile hot and cold years
 PI-factor method
o Accumulation of drying potential in a selected critical location in the
structure
 ANK-ORNL method
o Combination of wind driven rain, air flow transport, indoor humidity loads
and drying capability
 MI (Moisture Index) method
o Wetting/Drying ratio
 Construction evaluation method
o Full method (multiyear simulations)
Using Weather Selection methods to select MDRY
Once the critical moisture design years had been selected with the use of simulation tools
and damage functions, the Moisture Design Reference Years for these locations were
selected using the weather selection methods (IEA Annex 24, PI-factor, ANK-ORNL,
Moisture Index). The rankings were created using the simulated conditions of the OSB
layer in the wood frame wall to determine the time of wetness, maximum moisture
content, RHT-integral and mold index (annual average mold index for the year (AvMold)
and the maximum index value of any hour (MaxMold)). The simulated rankings and the
selections from weather selection methods were then compared and compared to see
whether there is any agreement between them. Agreement was chosen to be satisfactory
48
if the selections were within 10% in ranking e.g., when 30 years of weather data is used
and both the simulations and the weather selection methods select a year that is among
the 3 top years in ranking. The results for north are shown in Table 6 and the predominant
wind direction in Table 7. The agreement among the methods is poor.
Table 6. Top three most severe years as selected by the various damage functions and the existing
weather selection methods. Damage function values are from the simulation results for the light
weight wood frame wall facing north.
Ranking criteria Atlanta Baltimore Chicago Minneapolis NewOrleans Portland SanFrancisco Seattle
TOW 1964 1983 1990 1983 1964 1983 1962 1988
1966 1989 1974 1982 1961 1979 1986 1990
1983 1961 1983 1961 1963 1984 1961 1981
DTDRH 1964 1983 1990 1983 1961 1962 1962 1990
1966 1989 1983 1986 1986 1961 1986 1989
1983 1961 1973 1982 1963 1983 1961 1988
AvMold 1966 1983 1990 1983 1961 1962 1986 1989
1964 1989 1973 1986 1986 1983 1962 1990
1983 1961 1983 1982 1963 1979 1961 1988
MaxMold 1966 1983 1990 1983 1961 1962 1986 1989
1964 1989 1973 1986 1986 1961 1962 1990
1983 1961 1983 1982 1963 1983 1961 1988
MaxMC 1966 1983 1983 1983 1961 1962 1982 1989
1964 1961 1990 1982 1963 1983 1989 1990
1990 1989 1979 1985 1986 1961 1986 1974
MI 1964 1979 1983 1983 1961 1962 1982 1989
1975 1961 1990 1982 1964 1983 1961 1990
1982 1989 1978 1984 1989 1961 1962 1980
10% Cold 1966 1963 1985 1972 1968 1972 1975 1964
1976 1962 1963 1962 1983 1976 1965 1985
1968 1967 1972 1967 1966 1962 1971 1971
10% Hot 1990 1990 1987 1987 1990 1990 1983 1987
1986 1975 1973 1990 1982 1973 1987 1967
1980 1977 1990 1973 1980 1983 1988 1979
ANK/ORNL 1979 (E) 1979 (E) 1983 (E) 1983 (E) 1973 (S) 1983 (E) 1983 (S)1990 (S)
1961 1983 1990 1986 1975 1977 1982 1968
1975 (S) 1989 (N) 1972 (E) 1984 (E) 1972 (E) 1973 (E) 1986 (S)1972 (S)
PI yearly 1975 1989 1965 1962 1975 1972 1962 1964
1966 1974 1967 1965 1964 1977 1963 1962
1965 1979 1972 1986 1961 1975 1986 1980
PI Winter 1965 1961 1961 1961 1961 1961 1973 1989
1969 1973 1973 1965 1975 1978 1961 1967
1973 1984 1985 1984 1976 1962 1963 1985
PI Summer 1975 1989 1967 1962 1975 1972 1963 1964
1966 1974 1981 1961 1964 1975 1965 1962
1967 1979 1990 1965 1989 1973 1974 1980
49
Table 7.Top three most severe years as selected by the various damage functions and the existing
weather selection methods. Damage function values are from the simulation results for the light
weight wood frame wall facing the dominant wind driven rain orientation.
Ranking criteria Atlanta Baltimore Chicago Minnesota NewOrleans Portland SanFrancisco Seattle
TOW 1966 1961 1984 1983 1985 1990 1962 1988
1972 1983 1982 1982 1964 1983 1983 1983
1975 1987 1985 1986 1986 1979 1982 1981
DTDRH 1983 1983 1982 1983 1964 1962 1983 1988
1972 1961 1985 1982 1963 1983 1963 1990
1966 1989 1984 1986 1985 1979 1962 1984
AvMold 1983 1983 1982 1983 1961 1962 1983 1990
1972 1961 1975 1982 1964 1983 1963 1984
1966 1989 1984 1986 1986 1990 1962 1988
MaxMold 1983 1983 1982 1983 1961 1983 1963 1990
1972 1961 1985 1982 1964 1961 1983 1984
1966 1989 1975 1986 1986 1962 1982 1988
MaxMC WDR 1983 1983 1982 1983 1964 1983 1983 1984
1966 1961 1975 1982 1961 1962 1982 1989
1982 1962 1985 1975 1963 1990 1963 1988
MI 1975 1979 1982 1990 1988 1979 1982 1980
1973 1989 1990 1984 1961 1990 1986 1972
1961 1975 1975 1983 1989 1981 1962 1988
10% Cold 1966 1963 1985 1972 1968 1972 1975 1964
1976 1962 1963 1962 1983 1976 1965 1985
1968 1967 1972 1967 1966 1962 1971 1971
10% Hot 1990 1990 1987 1987 1990 1990 1983 1987
1986 1975 1973 1990 1982 1973 1987 1967
1980 1977 1990 1973 1980 1983 1988 1979
ANK/ORNL 1979 (E) 1979 (E) 1983 (E) 1983 (E) 1973 (S) 1983 (E) 1983 (S)1990 (S)
1961 1983 1990 1986 1975 1977 1982 1968
1975 (S) 1989 (N) 1972 (E) 1984 (E) 1972 (E) 1973 (E) 1986 (S)1972 (S)
PI yearly 1975 1989 1965 1962 1975 1972 1962 1964
1966 1974 1967 1965 1964 1977 1963 1962
1965 1979 1972 1986 1961 1975 1986 1980
PI Winter 1965 1961 1961 1961 1961 1961 1973 1989
1969 1973 1973 1965 1975 1978 1961 1967
1973 1984 1985 1984 1976 1962 1963 1985
PI Summer 1975 1989 1967 1962 1975 1972 1963 1964
1966 1974 1981 1961 1964 1975 1965 1962
1967 1979 1990 1965 1989 1973 1974 1980
50
Correlations between weather parameters
Figure 18 to Figure 21 show some yearly average weather parameters for 12 locations
and 30 years plotted against each other. Trends can be found such as ambient vapor
pressure following ambient temperature, and ambient air temperature following the
amount of solar radiation reaching the surface of the earth. However, within a city or
location the trend may not be so clear. For example, the average yearly outdoor air
temperature in Miami has hardly any correlation with the average yearly solar radiation
in Miami, but when looking at all locations together the average yearly outdoor
temperature seems to be correlated with the average yearly solar radiation.
Scatterplot of RH vs T
0.85 Location
Atlanta
Baltimore
0.80 Chicago
Fairbank s
Memphis
0.75 Miami
Minneapolis
New Orleans
Portland
RH
0.70
San Francisco
Seattle
Winnipeg
0.65
0.60
0.55
-5 0 5 10 15 20 25
T
Figure 18. Average yearly ambient air relative humidity against average yearly ambient air
temperature in 12 locations and for 30 years of weather data.
51
Scatterplot of PV vs T
2500 Location
Atlanta
Baltimore
Chicago
2000 Fairbank s
Memphis
Miami
Minneapolis
New Orleans
1500 Portland
PV
San Francisco
Seattle
Winnipeg
1000
500
-5 0 5 10 15 20 25
T
Figure 19. Average yearly ambient air vapor pressure against average yearly ambient air
temperature in 12 locations and for 30 years of weather data.
Scatterplot of T vs Totrad
Location
25 Atlanta
Baltimore
Chicago
20 Fairbank s
Memphis
Miami
15 Minneapolis
New Orleans
Portland
10
T
San Francisco
Seattle
Winnipeg
5
-5
100 120 140 160 180 200 220

Totrad
Figure 20. Average yearly ambient air temperature against average yearly total radiation in 12
locations and for 30 years of weather data.
52
Scatterplot of RH vs RainHor
0.85 Location
Atlanta
Baltimore
0.80 Chicago
Fairbank s
Memphis
0.75 Miami
Minneapolis
New Orleans
Portland
RH
0.70
San Francisco
Seattle
Winnipeg
0.65
0.60
0.55
0 500 1000 1500 2000
RainHor
Figure 21. Average yearly ambient air relative humidity against average yearly rain on the ground in
12 locations and for 30 years of weather data.
Figure 22 compares the outdoor air relative humidity (yearly average) between different
cities. The damage function values (RHT-integral) are similarly shown for the cities
inFigure 23.The cities with higher relative humidity tend to have higher damage
functions, but this correlation does not always occur. For example, Miami is fairly humid
on average, but the damage function value (RHT-integral) is low. Figure 23shows that
the damage function value can be expected to vary in Portland and New Orleans, whereas
in Fairbanks all years have similar damage function values.
How to read the boxplots:

1. Outlier (*) - Observation that is beyond the upper or lower whisker
2. Upper whisker - Extends to the maximum data point within 1.5 box heights
from the top of the box
3. Interquartile range box - Middle 50% of the data
a. Top line - Q3 (third quartile). 75% of the data are less than or equal to
this value.
b. Middle line - Q2 (median). 50% of the data are less than or equal to this
value.
c. Bottom line - Q1 (first quartile). 25% of the data are less than or equal to
this value.
4. Lower whisker - Extends to the minimum data point within 1.5 box heights
from the bottom of the box
53
Boxplot of RH
0.85
0.80
0.75
RH
0.70
0.65
0.60
0.55
a e o s is i is
nt or ag nk m ol ns nd isc
o
ttl
e
pe
g
lt a c a ph ia p a la a i
l ti
m i
ir b em M a r le r t
an
c
Se
n
A Ch ne O Po in
Ba Fa M in w Fr W
M Ne Sa
n
Location
Figure 22. Boxplot of average yearly relative humidity (30 years of data) in 12 locations used in the
study.
Boxplot of DTDRH
16000
14000
12000
10000
DTDRH
8000
6000
4000
2000
0
a e i
nt or ag
o
nk
s is m ol
is
an
s nd isc
o
ttl
e
pe
g
lt a m ic a ph ia p le tla c a ni
A lti Ch i rb em M
ne
a
O
r
Po
r an Se in
Ba Fa M in w Fr W
M Ne Sa
n
Location
Figure 23. Boxplot of the yearly damage function value (∆T*∆RH integral aka RHT-index) in 12
locations used in the study.
54
Goodness-of-fit of the existing weather selection methods
A method was created to numerically compare the goodness-of-fit of the different
methods for the purpose of picking the years with the highest damage function values.
The method has the following steps and demonstrated in Table 8:
1. Rank the weather years in the decreasing order of the damage function
2. Normalize the damage function values to have a range 0-100%
3. Take the top three years as selected by a weather selection method and find the
corresponding normalized damage function values as given by the simulation
results
4. Calculate the average of the normalized damage function values of the three years
5. Compare the average normalized damage function values of the methods. Which
method picks the highest percentile for the top three years?
Table 8. Ranking from the weather selection method is in the left and the ranking of the weather
years resulting from the simulated performance is on the right. The first year selected by this method
is the third year based on the simulation results and this year is at 68 percentile in the normalized
range of the damage function.
Rank Year Rank Year Damage
by by function
Method Simulations value
1 1983 1 1985 2501 100%
2 1986 2 1977 1998 68%
3 1961 3 1983 1765
30 1973 30 1972 230 0%
The simulation results were used to calculate the RHT-integral in the OSB of the wood
frame wall.These results were then considered to represent the real performance of the
wall.The rankings of the weather years served as the reference for all other methods.
The different weather selection methods found the following weather years to be the most
severe ones for Seattle (Table 9). The corresponding ranking of these years are listed
below the table in respective rows.
55
Table 9. The selected weather years for Seattle as chosen by different weather selection methods.
Rankings based on simulation results are in parentheses e.g., the most severe year based on 10% hot
method is 20th most severe year based on simulations.
Method 10% hot 10% cold ANK/ORNL PI Yearly PI winter PI summer MI

1st year 1964 (20) 1987 (15) 1990 (1) 1964 (20) 1989 (2) 1964 (20) 1990 (1)
2nd year 1985 (9) 1967 (5) 1968 (11) 1962 (7) 1967 (5) 1962 (7) 1972 (23)
3rd year 1971 (29) 1979 (27) 1972 (23) 1980 (10) 1985 (9) 1980 (10) 1988 (3)
The results for rankings with different methods were normalized for all cities. Table 10
shows the average of the normalized damage function values for the top three years that
the methods selected. For some cities the methods selected reasonably good years
(average normalized damage function value above average 50%) whereas the methods
selected some easy years for some locations (such as PI-factor method for Seattle with
only 31% average normalized damage function value for the most severe years).
Table 10. Average normalized damage function values for the top three weather years as selected by
five different weather ranking methods. The average of all locations per method is shown in the last
row.
Eq ANK/ORNL PI Yearly PI winter PI summer MI
Atlanta 82% 53% 68% 26% 63% 42%
Baltimore 84% 45% 63% 21% 60% 37%
Chicago 82% 73% 35% 62% 59% 69%
Minneapolis 69% 65% 42% 37% 38% 53%
NewOrleans 89% 13% 66% 45% 55% 68%
Portland 94% 53% 29% 67% 26% 74%
SanFrancisco 85% 53% 85% 69% 61% 80%
Seattle 91% 76% 31% 72% 31% 83%
85% 54% 52% 50% 49% 63%
When taking the top three worst years from the simulations and comparing those to the
top three highest ranked years from the two best performing methods as per the
normalized damage function ranking (ANK/ORNL and Moisture Index (MI)), we end up
with the Table 11. The match between the rankings is not very good with Atlanta having
no match at all either by ANK/ORNL- or MI-method and only one or two matches in
other cities at best.
56
Table 11. Number of years selected by ANK-ORNL and MI-methods from the top three worst years
as selected based on the simulation results.
Location ANK- MI
ORNL
Atlanta 0 0
Baltimore 2 1
Chicago 2 1
Minneapolis 1 1
New Orleans 0 1
Portland 1 0
San Francisco 1 2
Seattle 1 2
The performance of the existing weather selection methods was found not to be
satisfactory and the development of a new method was initiated.
57
Simulated Performance and Correlation with Weather
Data - New Method Development
A new method was developed to improve the weather selection capabilities to match the
simulated performance better. The idea to be tested was to use yearly average weather
data parameters and use regression analysis to fit the parameters of an equation that
would be used to calculate the predicted damage function value for each year. A second
order polynomial equation (includes nonlinear effects)was tested with all major weather
parameters. The selected function type is shown in the following equation:
Y predicted = c0 + c1 ⋅ T + c2 ⋅ RH + c3 ⋅ Rad + c4 ⋅ Cloud + c5 ⋅ Rain

+ c6 ⋅ Pv + c7 ⋅ WindSpeed + c8 ⋅ WindOr
(14)
+ c9 ⋅ T 2 + c10 ⋅ RH 2 + c11 ⋅ Rad 2 + c12 ⋅ Cloud 2 + c13 ⋅ Rain 2
+ c14 ⋅ Pv 2 + c15 ⋅ WindSpeed 2 + c16 ⋅ WindOr 2
Where
Ypredicted = Predicted damage function
ci = Coefficient for an individual weather parameter
T = Average yearly ambient air temperature, °C
RH =Average yearly relative humidity, %
Rad =Average yearly solar radiation on the wall surface, W/m2
Cloud =Average yearly cloud index, -
Rain =Average yearly wind driven rain on the wall, mm
Pv =Average yearly ambient air vapor pressure, Pa
Windspeed =Average yearly wind speed, m/s
WindOr =Average yearly wind orientation, degrees, and
Ni2 =Individual average yearly weather parameter (above) raised to the power
of 2.
Simulated results for the light-weight wall were converted to damage function values by
taking the temperature and relative humidity in the exterior sheathing panel and by
integrating the (T-0)*(RH-70%) over time. We refer to this later either as RHT-
index/integral or DTDRH-value.
Statistical parameters
Statistical analyses were carried out to investigate the possibility to determine the
hygrothermal design weather file based on the weather data only. Damage functions were
used as key indicators of weather severity. They were created by post-processing the
simulation results and by calculating the representative damage function value for the full
year. Weighing parameters were generated for each weather parameter (temperature,
58
vapor pressure/dewpoint temperature, solar radiation, precipitation, etc.) for the selected
function to predict the damage function value for each year by minimizing the error
between the damage function values resulting from simulations and the predictions from
the equation.
A widely used statistic to gage the goodness-of-fit of the model is the coefficient of
determination R2. A value of R2 = 1 indicates a perfect correlation between actual data
and the regression equation; a value of R2 = 0 indicates no correlation. As a rule of thumb
the value of R2 should never be less than 0.75.
For more than one independent variable in the regression, R2 is not sufficient to
determine the goodness-of fit. The standard error (SE) of the estimate of the coefficients
becomes more important. The smaller the standard error compared to the coefficient’s
magnitude, the more reliable the coefficient estimate. T-statistics (or t-values) are used to
identify the significance of individual coefficients. The t-values are the ratio of the
coefficient estimate divided by the standard error of the estimate.
The coefficient of each variable included in the regression has a t-statistic. For a
coefficient to be statistically meaningful, the absolute value of its t-statistic should be at
least 2.0. In other words, a variable should not be included in a regression if the standard
error of its coefficient estimate is greater than half the magnitude of the coefficient (even
when including a variable that increases the R2). Generally, including more variables in a
regression results in a higher R2, but the significance of most individual coefficients is
likely to decrease. A high t-value (absolute value without sign) typically results in low p-
value. P-value is the probability value of the null hypothesis H0 (if H0 is true, the result
can occur by chance). If p-value is low, the result is significant meaning that – in our case
– the coefficient has an effect (example of values are given in Table 14).
The coefficients of the equation to estimate the damage function values were fitted using
commercially available statistical software (Minitab).
First the regression equation used all weather parameters (temperature, solar radiation on
the wall, cloud index, relative humidity and rain on the wall; wind speed and orientation
were neglected, see reasons later) and their second order (n2) values.
The regression equation now becomes:
DTDRH(70) = 112530 - 194 Rad - 4514 Cloud + 38.6 T - 305794 RH + 205108 Rain
+ 14.1 Pv - 0.39 Rad*Rad + 341 Cloud*Cloud + 24.3 T*T (15)
+ 255461 RH*RH - 644993 Rain*Rain - 0.00914 Pv*Pv
Where the input values are the year average data of

Rad = Solar radiation intensity on the wall, W/m2
Cloud = Cloud index, 0-8
T = Air temperature, °C
RH = Relative Humidity, - (0-1)
Rain = Wind-driven rain on the wall, kg/h,m2
Pv = Vapor pressure, Pa.
59
Table 12. Regression coefficients and statistic variables with twelve predictors plus constant for
calculating the damage function.
Predictor Coef SE Coef T-value P-value
Constant 112530 14487 7.77 0.000
Rad -194.5 308.5 -0.63 0.529
Cloud -4514 2134 -2.12 0.035
T 38.58 71.94 0.54 0.592
RH -305794 36219 -8.44 0.000
Rain 205108 27597 7.43 0.000
Pv 14.078 2.210 6.37 0.000
Rad*Rad -0.390 2.839 -0.14 0.891
Cloud*Cloud 341.3 235.2 1.45 0.148
T*T 24.314 4.554 5.34 0.000
RH*RH 255461 26573 9.61 0.000
Rain*Rain -644993 829082 -0.78 0.437
Pv*Pv -0.0091393 0.0005873 -15.56 0.000
S = 1100.14 R-Sq = 87.3%

R-Sq(adj) = 86.9%
PRESS = 472732818 R-Sq(pred) = 85.73%
When looking at the t-values of individual predictors we can see that some of them have
a low absolute value (<2) and they are recommended not to be included in the regression.
Before eliminating these parameters with low t-values, stepwise regressions methods
were used to isolate the meaningful parameters.
By looking at the individual correlations of weather parameters it is clear that a single

weather parameter alone is not a good predictor of the severity of the weather, but rather
a combination of weather parameters is needed to precisely rank the years in terms of
damage potential to the building envelope. Figure 24 shows the correlations between the
parameters and the DTDRH(70) (=RHT-index) damage function in a matrix format. For
example, Pv (vapor pressure) and T2 are almost linearly correlated.
60
Matrix Plot of DTDRH(70), Cloud, RH, Rain, Pv, Drad*Drad, T*T, ...
00 02 04 00 00 00 30 45 60
4 5 6 0. 0. 0. 20 30 40 0. 0. 0.
16000
8000
DTDRH(70)
0
6
5
Cloud 4
0.8
0.7
RH
0.6
0.04
0.02
Rain
0.00
2400
1600
Pv
800
4000
3000
Drad*Drad
2000
800
400
T*T
0
0.60
RH*RH 0.45
0.30
5000000
2500000 Pv*Pv
0
0 00 0 6 7 8 0 00 00 0 0 0 0 0 0
00 0. 0. 0. 80 25 50 00 00
80 16 16 24 00 00
25 50
Figure 24. Matrix of correlation between parameters in the equation to predict the damage function
with average yearly weather data as input.
Stepwise regressions methods were carried out to select the best subsets and statistically
meaningful parameters and to eliminate the coefficients that may not explain or improve
the precision of the regression. Parameters such as Mallows’ Cp (Minitab, 2006) can be
used to estimate the quality of the regression parameters. The software used for the
analysis does the stepwise regression for all parameters automatically allowing for the
user to select the final regression.
61
Table 13. Stepwise regression analysis of the parameters for the equation to predict the damage
function values as performed by the statistical software.
C
L
o W W R
u S D a
R d P I i
A * D R n
C d C R * * * P
l W W R * l H W W R v
R o S D a R o T * S D a *
a u R P I i P a u * R P I i P
Vars R-Sq R-Sq(adj) Mallows Cp S d d T H D R n v d d T H D R n v
1 11.5 11.3 2108.7 2861.2 X
1 11.5 11.2 2110.2 2862.0 X
2 45.1 44.8 1176.1 2257.5 X X
2 42.3 42.0 1253.7 2314.0 X X
3 66.5 66.2 581.7 1766.0 X X X
3 65.7 65.4 604.5 1787.4 X X X
4 74.8 74.5 351.9 1533.3 X X X X
4 73.9 73.6 377.4 1560.9 X X X X
5 79.0 78.7 238.4 1403.5 X X X X X
5 78.5 78.2 249.9 1417.1 X X X X X
6 84.0 83.8 99.1 1224.5 X X X X X X
6 82.7 82.4 136.0 1274.2 X X X X X X
7 84.8 84.5 79.6 1196.2 X X X X X X X
7 84.8 84.4 80.8 1198.0 X X X X X X X
8 87.2 86.9 15.1 1099.8 X X X X X X X X
8 87.2 86.9 15.1 1099.9 X X X X X X X X
Mallows' Cp is a statistic used as an aid in choosing between competing multiple

regression models. Mallows' Cp compares the precision and bias of the full model to
models with the best subsets of predictors. It helps strike an important balance between
the accuracy and the number of predictors in the model. A model with too many
predictors can be relatively imprecise while one with too few can produce biased
estimates. A Mallows' Cp value that is close to the number of predictors (plus the
constant) indicates that the model is relatively precise and unbiased in estimating the true
regression coefficients and predicting future responses.
The analysis indicates that for example wind speed and wind direction are not good
predictors in the equation. These parameters are already indirectly taken into account in
the wind driven rain loads that were already calculated on the wall facing a given
orientation. It can be seen that the Mallows’ Cp becomes close to the number of
predictors with only eight (8) predictors. The two eight predictor sets are equally good,
the last one suggests using solar radiation, cloud index, relative humidity, rain, vapor
pressure, temperature*temperature, relative humidity*relative humidity and vapor
pressure*vapor pressure as predictors.
The regression equation with these parameters is:
62
DTDRH(70) = 108307 - 241 Rad - 1391 Cloud - 312326 RH + 183308 Rain + 15.2 Pv
+ 27.3 T*T + 261079 RH*RH - 0.00972 Pv*Pv (16)
Table 14. Regression coefficients and statistic variables with eight predictors plus constant for
calculating the damage function.
Predictor Coef SE Coef T P
Constant 108307 11643 9.30 0.000
Rad -241.30 27.80 -8.68 0.000
Cloud -1390.6 170.5 -8.16 0.000
RH -312326 35156 -8.88 0.000
Rain 183308 9748 18.80 0.000
Pv 15.193 1.079 14.08 0.000
T*T 27.340 2.775 9.85 0.000
RH*RH 261079 25556 10.22 0.000
Pv*Pv -0.0097163 0.0002887 -33.66 0.000
S = 1099.91 R-Sq = 87.2%

R-Sq(adj) = 86.9%
PRESS = 451996042 R-Sq(pred) = 86.36%
Now all the predictors have absolute t-values higher than 2 indicating that these are
proper predictors in the equation. The R-squared with eight predictors 87.2% is still about
the same as the 87.3% with 12 predictors.
Simulated versus predicted RHT-index

Equation (14) was statistically analyzed and coefficients were found (Equation 15) to find
an overall best fit for the damage function values for each year in eight locations in the
US. The fitting of the coefficients was carried out using the damage function values in the
OSB in the light weight wood frame wall.Figure 25 shows the performance of the
predicted damage function against the damage function data from simulations in Atlanta
(the figures for other locations are in Appendix D). The performance of the predictions is
good in all locations and the predicted values follow the same trends and patterns as the
simulated ones.
63
Variable
11000
DTDRH simulated_Atlanta
DTDRH Predicted_Atlanta
10000
9000
Y-Data
8000
7000
6000
0 5 10 15 20 25 30
Year_Atlanta
Figure 25. Atlanta (GA): Simulated and predicted damage function data (RHT-index) for 30 years
and for the light weight wood frame wall.
Four other locations were used to verify that the model would work even with locations
that were not part of the parameter fitting.Figure 26 to Figure 29 show the comparison of
the results for Fairbanks (AK), Memphis (TN), Miami (FL) and Winnipeg (MB, Canada).
64
7000 Variable
DTDRH simulated_Fairbank s
DTDRH Predicted_Fairbank s
6000
5000
Y-Data
4000
3000
2000
0 5 10 15 20 25 30
Year_Fairbanks
Figure 26. Fairbanks (AK).
12000 Variable
DTDRH simulated_Memphis
11000 DTDRH Predicted_Memphis
10000
9000
Y-Data
8000
7000
6000
5000
4000
0 5 10 15 20 25 30
Year_Memphis
Figure 27. Memphis (TN).
65
5000 Variable
DTDRH simulated_Miami
DTDRH Predicted_Miami
4000
3000
Y-Data
2000
1000
0 5 10 15 20 25 30
Year_Miami
Figure 28. Miami (FL).
12000 Variable
DTDRH simulated_Winnipeg
11000 DTDRH Predicted_Winnipeg
10000
9000
8000
Y-Data
7000
6000
5000
4000
3000
0 5 10 15 20 25 30
Year_Winnipeg
Figure 29. Winnipeg (MB, Canada).
66
Figure 30 combines all the simulated damage function values (RHT-integral) and the
predicted values into one chart. If an ideal fit were found the data would collapse into a
straight line. These results show that the fit is not perfect but instead some deviations
from the line exist.
Scatterplot of Pred vs DTDRH

16000 Location
Atlanta
14000 Baltimore
Chicago
Fairbank s
12000
Memphis
Miami
10000 Minneapolis
New Orleans
8000 Portland
Pred
San Francisco
6000 Seattle
Winnipeg
4000
2000
0 2000 4000 6000 8000 10000 12000 14000 16000

DTDRH
Figure 30. Simulated damage function (RHT-integral) versus the predicted (pred) damage function
using the equation method. All 30 years of weather data and 12 locations are included.
Goodness-of-Fit of the New Method

The previously used method to compare the goodness-of-fit of different methods was
used with the newly developed equation based method. Different techniques to look at
the performance of the weather selection methods were:
1. Visual comparison of rankings

2. Counting how many of the top three worst years do the methods pick (vs.
simulations)
3. Normalizing a selected damage function value and counting the average of the
three selected weather years.
67
Table 15. Normalized damage function values based on the simulated results of the top three years
for the new equation method and the three other promising methods (ANK/ORNL, PI-factor and
moisture index MI).
Eq ANK/ORNL PI Yearly PI winter PI summer MI
Atlanta 82% 53% 68% 26% 63% 42%
Baltimore 84% 45% 63% 21% 60% 37%
Chicago 82% 73% 35% 62% 59% 69%
Minneapolis 69% 65% 42% 37% 38% 53%
NewOrleans 89% 13% 66% 45% 55% 68%
Portland 94% 53% 29% 67% 26% 74%
SanFrancisco 85% 53% 85% 69% 61% 80%
Seattle 91% 76% 31% 72% 31% 83%
85% 54% 52% 50% 49% 63%
The normalized damage function values were on average 85% for the eight locations
investigated ranging from 69% to 94%. The average values for the other methods were
between 49-63% with a range from 13% to 85% for individual locations. Normalized
damage function values of 50% would indicate more or less average years and not severe
years. Only the equation based method produced consistent performance for all locations
with average damage function values well over 50% in any location (Table 15).
Table 16 lists again the match between the top three worst years in the simulations and
those selected by the methods. The order of the ranking within the top three was not
considered i.e., if for example a year was the worst year in the simulations and the third
year in the ranking by MI-method it was considered a match. The new equation based
method was able to select at least two years out of three possible in all eight locations.
Table 16. Comparison of the number of years selected from the top three worst years (simulated
data) by the new equation based method versus the top two existing methods.
Location ANK- MI New

ORNL equation
method
Atlanta 0 0 2
Baltimore 2 1 2
Chicago 2 1 2
Minneapolis 1 1 2
New Orleans 0 1 2
Portland 1 0 3
San Francisco 1 2 2
Seattle 1 2 3
68
Contour maps of weather parameters and damage functions
The method developed to select hygrothermal design years was utilized and the design
weather years were determined for 100 US locations and 7 Canadian locations. The 10-
percentile weather year was chosen for each location out of 30 years of weather data. The
selection was based on order of years, i.e. the selected year was the year having the third
highest damage function value for the location. The damage function data (RHT-values)
were used to create a contour plot of the values on the North American map – similar to
IECC climate zone classification. The contour map is shown inFigure 31. The map shows
similarity to the IECC classification in the way that the mid-western US shows low
damage function values which is due to the dryness of the area, the eastern part of the US
shows higher damage function values corresponding to the ‘moist’ area of IECC. The
east and west coast in the northern part of US show the highest damage function values.
In the IECC classification e.g. the north-western coast is classified as ‘marine’ climate
zone.
It appears that even though none of the single weather parameters (temperature, relative
humidity, solar radiation, rain, vapor pressure) appeared to correlate well with the
calculated damage function values and the ranking resulting from those, the contour map
showing the number of heavy fog days (Figure 32) follows a very similar pattern as the
contour map for the damage function values of the selected years (Figure 31).
Figure 31. Damage Function (RHT-integral) is shown here as a contour map in the US and Canada.
Crosses show the locations of the cities with data. Data outside the dotted area has been extrapolated
by the plotting software and should not be considered accurate.
69
Figure 32. Annual heavy fog days in the US (NOAA).
A piece of software was developed as part of this project that allows for printing out
hygrothermal design weather data in the format needed for various simulation models.
The basis of this model (WeatherFile Analyzer) was developed by Karagiozis at ORNL.
70
Conclusions
The purpose of this project was to investigate a number of methods to select a weather
year for hygrothermal design analysis. The choice of the weather year is critical for
proper hygrothermal design analysis. The use of weather years for energy calculation
purpose were found not be acceptable, especially as the key durability influencing
parameters are not based on exterior temperature.
A number of critical parameters were assessed, and the orientation of the building
envelope was shown to affect the wall performance the most. Indeed, the north
orientation was found to be worst factor with the highest damage function values and
accumulated moisture contents. A number of previously proposed weather selection
methods were tested based on simulated wall performance. The analysis included 30
years of hourly weather data for 12 locations in the US and Canada. From the extensive
analysis, none of the existing methods was found satisfactory and a new method was
developed to rank the years in terms of hygrothermal loads.
A simple approximate method was developed. An equation based method – as shown by

Equation (16) –predicted the best performance of all the analyzed methods. This load
based approach was chosen as the final method for selecting the weather years for
hygrothermal designs. The method was proved to be the most consistent and accurate out
of all analyzed methods in selecting the most severe years in terms of hygrothermal
performance in all locations.
The method uses average weather parameters for a north facing wall and calculates an
estimate for the damage function RHT-index (integral of (T-0)*(RH-70%) in the OSB
layer of the light weight wood frame wall. The year with the third highest RHT-index
value is proposed as the year to be selected for hygrothermal designs.
The chosen method that is based on an equation that uses yearly average weather
parameters as input was shown to perform most reliably and to be consistent in predicting
the most severe years. A CD was created with 100 US location and 7 Canadian locations
including the selected weather years for those locations. Appendix F explains how the
equation is to be used when selecting the weather year hygrothermal design for a location
with many years of weather data.
71
References and literature
Blocken, B. 2004. Wind-driven Rain on Buildings; measurements, numerical modeling
and applications, Ph.D. dissertation, Leuven, K.U. Leuven.
Blocken, B. and Carmeliet J. 2002. Spatial and temporal distribution of driving rain on a
low-rise building. Wind & Structures 5(5): 441-462.
Briggs, R.S., Lucas, R.G. and Taylor, Z.T. 2002. Climate Classification for Building
Energy Codes and Standards, Technical Final Paper Draft, Pacific NW National
Laboratory, March 26.
Burch, D.M. and Chi, J. 1997. MOIST: A PC Program for Predicting Heat and Moisture
Transfer in Building Envelopes. NIST Special Publication 917, Gaithersburg, MD.
Canadian Home Builders’ Association. 1995. Builders’ Manual. ISBN 0-86506-054-1.

Choi, E. 1994. Characteristics of the co-occurrence of wind and rain and the driving-rain
index. Journal of Wind Engineering and Industrial Aerodynamics 53, 49-62.
Cornick, S., Dalgliesh, A., Said, N., Djebbar, R., Tariku, F. and Kumaran, M. K., (2002),
"Report from Task 4 of MEWS Project - Environmental Conditions, Final Report", IRC-
RR-1130, Institute for Research in Construction, National Research Council Canada,
Ottawa, Canada
Cornick, S., Djebbar, R. and Dalgliesh, W. A. 2003. Selecting moisture reference years
using a Moisture Index approach. Building and Environment, 38, 1367-1379.
Crawley, Drury B. 1998, “Which Weather Data Should You Use For Energy Simulations
of Commercial Buildings”, ASHRAE Transactions 104, part 2, ASHRAE, Atlanta, GA .
Crow, L W. 1981. “Development of hourly data for weather year for energy calculations
(WYEC), including solar data, at 21 stations throughout the US”. ASHRAE Transactions,
Vol. 87, Part 1
Djebbar, R., van Reenen, D. and Kumaran, M.K, (2001), "Environmental boundary
conditions for long-term hygrothermal calculations", Proceedings for Performance of
Exterior Envelopes of Whole Buildings VIII: Integration of Building Envelopes,
December 2-7, Clearwater Beach, Florida.
FSEC 3.0, 1992 Florida Software for Environmental computation, FSEC-GP-47-92,

Florida Solar Energy Center.
Geving, S., "Moisture Design of Building Constructions: Hygrothermal analysis using

simulation models - Part I and II" Ph.D. Thesis, Norwegian University of Science and
Technology, Faculty of Civil and Environmental Engineering, June 1996: 41-55.
72
Hagentoft, C.E., and E. Harderup. 1993. Reference for moisture calculations. Report T2-
S-93/01, IEA Annex 24, HAMTIE.
Harderup, 1994. Durability reference year: Using the Π-factor method to find the DRY
for three locations in Sweden. Report T2-S-94/01, IEA Annex 24, HAMTIE.
Hukka A. & Viitanen H. (1999). A mathematical model of mould growth on wooden

material. Wood Science and Technology 33(6): 475-485.
Künzel, H. M., Simultaneous Heat and Moisture Transport in Building Components:

One- and two-dimensional calculation using simple parameters. Fraunhofer-Institut für
Bauphysik, IRB Verlag, Stuttgart, 1995.
Lacy, R. E., "Driving-Rain Maps and the Onslaught of Rain on Buildings", Proceedings
of the RILEM/CIB Symposium on Moisture Problems in Buildings, Helsinki Finland,
1965.
Karagiozis, A.N.; Hadjisophocleous, G.V.; Cao, S. "Wind-driven rain distributions on

two buildings," Journal of Wind Engineering and Industrial Aerodynamics, 67-68, April-
June, pp. 559-572, April 01, 1998
Karagiozis, A.N., Salonvaara, M. H. 1995. Influence of material properties on the

hygrothermal performance of a high-rise residential wall. ASHRAE Transactions,
ASHRAE Symposium, pp. 647-655, 1995.
Karagiozis, A.N., 2003.WeatherFile Analyzer Software, Oak Ridge National Laboratory.
Minitab® Statistical Software (version 15.1.1.0).
Rode, C. 1993. Reference years for moisture calculations, Denmark. Report T2-DK-
93/02, IEA Annex 24, HAMTIE.
Salonvaara, M.H.; Karagiozis, A.N. "Influence of waterproof coating on the

hygrothermal performance of a brick facade wall system," ASTM Special Technical
Publication, 1314, ASTM Symposium on Water Leakage Through Building Facades
(Orlando, Florida, U.S.A. 3/17/1996), pp. 295-311, 1998
Sanders C., 1996, “Task 2, Environmental Conditions”, IEA Annex 24.
Stoffel, T.L., and M.D. Rymes. 1998. “Production of the Weather Year for Energy
Calculations Version 2 (WYEC2) Data Files.” ASHRAE Transactions, Vol. 104, Part 2.
Straube, J. F., Moisture Control and Enclosure Wall Systems, Ph.D. Thesis, Civil
Engineering Department, University of Waterloo, 1998.
73
Surry, D., Skerlj, P.F. and Mikitiuk, M.J, (1995), An exploratory study of the climatic
relationships between rain and wind, Boundary Layer Wind Tunnel report BLT-2230-
1994 for Canada Mortgage and Housing Corporation.
TenWolde, A., Walker, I.S. 2001 Interior Moisture Design Loads. Proceedings of
Performance of Exterior Envelopef of Whole Buildings VIII: Integration of Building
Envelopes, ASHRAE, Atlanta.
Viitanen H. (1996). Factors affecting the development of mould and brown rot decay in
wooden material and wooden structures. Effect of humidity, temperature and exposure
time. Doctoral thesis. Uppsala. TheSwedish University of Agricultural Sciences,
Department of Forest Products. 58 p.
Viitanen H., Hanhijärvi A., Hukka A. & Koskela K. (2000). Modelling mould growth and
decay damages. Healthy Buildings. Espoo, 6 - 10 August 2000. Vol. 3. FISIAQ, 2000, p.
341–346.
Wilson, A.G. 1966 Condensation in Insulated Masonry Walls in the Summer. Proc. Of
RILEM/CIB Symposium, Helsinski DBR Tech Paper 226.
"MINITAB® and all other trademarks and logos for the Company's products and services
are the exclusive property of Minitab Inc. All other marks referenced remain the property
of their respective owners. See minitab.com for more information."
74
Appendix A: Moisture content of OSB with 30 years of
weather data
Simulated moisture contents of the OSB layer in the light weight wall are shown in
Figure 33 to 39 for Baltimore, Chicago, Minneapolis, New Orleans, Portland (ME), San
Francisco and Seattle. Results for Atlanta are shown in Figure 14.
0.2
Moisture Content, kg/kg [lb/lb]
0.15
0.1
Least severe3
Easiest
Most severe
Worst 3
0.05
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 33. Baltimore: Moisture content of OSB.
75
0.15
0.1
Least severe3
Easiest
Most severe
Worst 3
0.05
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 34. Chicago: Moisture content of OSB.
0.2
0.15
0.1
Least severe3
Easiest
Most severe
Worst 3
0.05
0
0 100 200 300
Time, d (Oct 1 st -> )
Figure 35. Minneapolis: Moisture content of OSB.
76
0.15
0.1
Least severe3
Easiest
Most severe
Worst 3
0.05
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 36. New Orleans: Moisture content of OSB.
0.2
0.15
0.1
Least severe
Worst 3
Most severe3
Easiest
0.05
0
0 100 200 300
Figure 37. Portland (ME): Moisture content of OSB.
77
0.15
0.1
Least severe
Easiest 3
Most severe
Worst 3
0.05
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 38. San Francisco: Moisture content of OSB.
0.2
0.15
0.1
Least severe
Easiest 3
Most severe
Worst 3
0.05
0
0 100 200 300
Figure 39. Seattle: Moisture content of OSB.
78
AppendixB: Moisture content of CMU with 30 years of
weather data
Simulated moisture contents of the CMU block in the CMU wall are shown in Figure 40
to 46 for Baltimore, Chicago, Minneapolis, New Orleans, Portland (ME), San Francisco
and Seattle. Results for Atlanta are shown in Figure 15.
0.05
0.04
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 40. Baltimore: Moisture Content of the exterior wythe of the CMU block.
79
0.04
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 41. Chicago: Moisture Content of the exterior wythe of the CMU block.
0.05
0.04
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
Figure 42. Minneapolis: Moisture Content of the exterior wythe of the CMU block.
80
0.04
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
st
Time, d (Oct 1 -> )
Figure 43. New Orleans: Moisture Content of the exterior wythe of the CMU block.
0.05
0.04
0.03
0.02
0.01
Least severe
Worst 3
Most severe
Easiest 3
0
0 100 200 300
Figure 44. Portland: Moisture Content of the exterior wythe of the CMU block.
81
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
Figure 45. San Francisco: Moisture Content of the exterior wythe of the CMU block.
0.03
0.02
0.01
Least severe3
Easiest
Most severe
Worst 3
0
0 100 200 300
Figure 46. Seattle: Moisture Content of the exterior wythe of the CMU block.
82
83
Appendix C: Relative damage function for the light
weight wall and CMU wall
Damage function values (RHT-index) are shown for each of the 30 years weather data
when normalized by setting the lowest damage function value of all 30 years to 0% and
the maximum to 100%.Figure 47 to 53 are for Baltimore, Chicago, Minneapolis, New
Orleans, Portland (ME), San Francisco and Seattle. Results for Atlanta are shown in
Figure 16.
100%
90%
80%
70%
60%
CMU
50% LWW
40%
30%
20%
10%
0%
0 5 10 15 20 25 30
Year, -
Figure 47. Baltimore: Normalized damage function values (RHT-index) for CMU block wall and for
100%
90%
80%
70%
60%
CMU
50%
LWW
40%
30%
20%
10%
0%
0 5 10 15 20 25 30
Year, -
Figure 48. Chicago: Normalized damage function values (RHT-index) for CMU block wall and for
84
100%
90%
80%
70%
60%
CMU
50%
LWW
40%
30%
20%
10%
0%
0 5 10 15 20 25 30
Year, -
Figure 49. Minneapolis: Normalized damage function values (RHT-index) for CMU block wall and
for light weight wood frame wall.
100%
90%
80%
70%
60%
CMU
50%
LWW
40%
30%
20%
10%
0%
0 5 10 15 20 25 30
Year, -
Figure 50. New Orleans: Normalized damage function values (RHT-index) for CMU block wall and
85
100%
Relative Damage Function, % 90%

80%
70%
60%
CMU
50%
LWW
40%
30%
20%
10%
0%
0 5 10 15 20 25 30
Year, -
Figure 51. Portland: Normalized damage function values (RHT-index) for CMU block wall and for
100%
90%
80%
70%
60%
CMU
50%
LWW
40%
30%
20%
10%
0%
0 5 10 15 20 25 30
Year, -
Figure 52. San Francisco: Normalized damage function values (RHT-index) for CMU block wall and
86
87
LWW
CMU
Figure 53. Seattle: Normalized damage function values (RHT-index) for CMU block wall and for
30
25
20
Year, -
15
10

5
0
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Appendix D: Simulated vs. predicted damage function
values
Simulated RHT-index (DTDRH) is compared with the predicted RHT-index (Equation
method) in Baltimore, Chicago, Minneapolis, New Orleans, Portland (ME), San
Francisco and Seattle. Results for Atlanta are shown in Figure 25.
13000 Variable
DTDRH simulated_Baltimore
12000 DTDRH Predicted_Baltimore
11000
10000
9000
Y-Data
8000
7000
6000
5000
4000
0 5 10 15 20 25 30
Year_Baltimore
Figure 54. Baltimore (MD): Simulated and predicted damage function data (RHT-index=DTDRH)
for 30 years and for the light weight wood frame wall.
88
12000 Variable
DTDRH simulated_Chicago
DTDRH Predicted_Chicago
11000
10000
9000
Y-Data
8000
7000
6000
5000
0 5 10 15 20 25 30
Year_Chicago
Figure 55. Chicago (IL): Simulated and predicted damage function data (RHT-index=DTDRH) for
30 years and for the light weight wood frame wall.
13000 Variable
DTDRH simulated_Minneapolis
12000 DTDRH Predicted_Minneapolis
11000
10000
9000
Y-Data
8000
7000
6000
5000
4000
0 5 10 15 20 25 30
Year_Minneapolis
Figure 56. Minneapolis (MN): Simulated and predicted damage function data (RHT-index=DTDRH)
89
V ariable
D TDRH simulated_N ew O rleans
15000 D TD RH P redicted_N ew O rleans
12500
Y-Data
10000
7500
5000
0 5 10 15 20 25 30
Year_New Orleans
Figure 57. New Orleans (LA): Simulated and predicted damage function data (RHT-index=DTDRH)
14000 Variable
DTDRH simulated_Portland
13000 DTDRH Predicted_Portland
12000
11000
10000
Y-Data
9000
8000
7000
6000
5000
0 5 10 15 20 25 30
Year_Portland
Figure 58. Portland (ME): Simulated and predicted damage function data (RHT-index=DTDRH) for
90
16000 V ariable
D TDRH simulated_S an F rancisco
DTD RH P redicted_S an F rancisco
14000
12000
Y-Data
10000
8000
6000
0 5 10 15 20 25 30
Year_San Francisco
Figure 59. San Francisco (CA): Simulated and predicted damage function data (RHT-
index=DTDRH) for 30 years and for the light weight wood frame wall.
14000 Variable
DTDRH simulated_Seattle
DTDRH Predicted_Seattle
13000
12000
11000
Y-Data
10000
9000
8000
7000
0 5 10 15 20 25 30
Year_Seattle
Figure 60. Seattle (WA): Simulated and predicted damage function data (RHT-index=DTDRH) for
91
Appendix E: Material properties in simulations
The properties of the materials that were used in the simulations of the CMU block wall
and the light weight wood frame wall are listed and graphed in Table 17 and in Figure 61
to Figure 64.
Table 17. Basic properties of the materials used in the simulations of the CMU block wall and the
Material Density, Thermal conductivity, Volumetric heat capacity,
kg/m3 W/mK J/m3K
Gypsum 620 0.2 520800
Fiberglass 30 0.041 25200
OSB 641 0.12 538440
Concrete 2200 1.7-2.2 1848000
Brick 1600 0.4-0.62 1344000
Stucco 1683 0.47 1413720
0.06
0.05
0.04
Moisture Content, kg/kg
0.03
0.02
0.01
0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
RH, -
Gypsum Fiberglass Kraft Brickwork Concrete
Figure 61. Sorption isotherms of materials used in the CMU block wall structure.
92
0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
RH, -
Gypsum Fiberglass Felt OSB Stucco
Figure 62. Sorption isotherms of materials used in the light weight wood frame wall structure.
0.06
0.05
Water Vapor Permeability, kg/msPa
0.04
0.03
0.02
0.01
0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
RH, -
Gypsum Fiberglass Kraft Brickwork Concrete
Figure 63. Water vapor permeability of materials used in the CMU block wall structure.
93
1.40E-10 7.00E-12
1.20E-10 6.00E-12
1.00E-10 5.00E-12

8.00E-11 4.00E-12
6.00E-11 3.00E-12
4.00E-11 2.00E-12
2.00E-11 1.00E-12
0.00E+00 0.00E+00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Relative Humidity, -
Gypsum Fiberglass OSB (Y2) Stucco (Y2)
Figure 64. Water vapor permeability of materials used in the light weight wood frame wall structure.
30
25
Water Vapor Permeance, perm
20
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative Humidity, -
Felt Kraft
Figure 65. Water vapor permeance of thin materials used in the simulated structures.
94
1.0E-06
1.0E-07
1.0E-08
1.0E-09
Liquid Diffusivity, m 2/s
1.0E-10
1.0E-11
1.0E-12
1.0E-13
1.0E-14
1.0E-15
1.0E-16
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Gypsum OSB Stucco Brickwork CMU
Figure 66. Liquid moisture diffusivity of materials used in the simulations of the CMU block wall and
the light weight wood frame wall.
95
Appendix F: Equation method to select weather year
If many years of weather data is available for a given location or new weather data
becomes available for the locations with already selected weather years, the newly
developed equation method can be used to select the most severe years with the following
procedure:
Calculate average yearly weather parameters for

- Air temperature (°C),
- Relative humidity (-),
- Cloud index (values ranging between 0 and 8),
- Water vapor pressure (Pa) (calculate values for every hour using temperature and
relative humidity unless vapor pressure is available in the weather file, then take
the average over the whole year),
- Solar radiation on a north facing wall (W/m2) and
- Wind driven rain on a north facing wall (kg/h,m2)
Use equation 16 to calculate the damage function values for each year.
Rank the weather years in the order of descending damage function values.
The most severe years will have the highest damage function values. The selected
weather year for hygrothermal analyses would be the 10%-tile year in ranking, i.e. if 30
years of weather data is used the weather year would the year with third highest damage
function value.
Equation 16 is repeated here for easy reference. The damage function value represents
RHT-index (or DTDRH as shown in the equation):
DTDRH(70) = 108307 - 241·Rad - 1391·Cloud - 312326·RH + 183308·Rain + 15.2·Pv

+ 27.3·T*T + 261079·RH*RH - 0.00972·Pv*Pv (16)
96

Ashrae D RP 1325

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Ashrae D RP 1325

Transféré par

Droits d'auteur :

Formats disponibles

Copyrighted material licensed to Elie Baradhy on 2019-05-26 for licensee's use only.

Environmental Weather Loads for Hygrothermal Analysis and Design

Simulation studies and data-analysis

Mikael Salonvaara, Co-PI, project manager

Parameters in the equation to predict damage function values:

 IEA Annex 24 -10% condensation method,

 Despite the availability of several weather data formats (frequencies)

Complexity in selecting representative moisture design year is highlighted in Figure 1

Reference year and procedures for its determination

A reference year is a summary of external climate conditions for a particular location or

 The need for additional or different climatic parameters,

International Energy Agency – Annex 24 Approach

Figure 2. Indoor Climate Classes as in the IEA Annex 24 method.

Determination of the existence of DRY

The 10%levelinterstitial condensation can be determined. A possible outcome is that one

Carsten Rode Method

 Construction 1 – Lightweight flat roof without a vapor barrier (HAMTIE-Annex

Table 2.Rankings related to years 1990-1992 and 1994 (Geving, 1997).

The Pi-factor method

Π = d v , sat (Ts ) − d v ,out (4)

Moisture Index Method

MI = function (Wetting Index, Drying Index) (5)

INormalized = (I – IMin) / (IMax – IMin) (7)

The ANK-ORNL method

∑ Hygric Load = ∑t =0 u + ∑t =0 mair ρ air + ∑t =0

Hygric load = Yearly moisture load potential kg of water

Summary of Advantages and Disadvantages of Each Method

The advantages and disadvantages of different methods arediscussed in Table 3.

Performance data has to be analyzed by using various methods to provide a ranking of

The principle of the method is visualized in Figure 4.

The evaluation process is visualized in Figure 5.

Output Output Output Output Output Output

Available weather file sources and formats

At least 30 years or more of weather data is preferred in order to generate statistically

Locations used in the analyses

Effects of weather on the performance of building enclosure components can be very

Figure 9. Arctic Circle in the northern hemisphere.(http://athropolis.com/map3.htm)

Wind driven rain effects

rbv = FE ⋅ FD ⋅ FL ⋅ U ⋅ cos θ ⋅ rh (10)

Selected simulation models

MOISTURE-EXPERT (Oak Ridge National Laboratory) and WUFI-Pro were selected to

Wall systems selected for analysis

Wall 1 –“Light-weight” wood frame wall

Wall2 – Concrete Masonry Unit (CMU)-block wall

Procedure for running the simulations

1. Simulate each year individually beginning inOctober (ASHRAE Standard 160).

Indoor air and initial conditions

Use of damage functions for selecting weather years

A whole range of physio-chemical and mechanical processes as well as improper design

Critical layers in the walls for investigation

Rankings based on damage functions

Mold growth index

Table 5. Mold growth index descriptions (Viitanen and Ritschkoff 1991).

Maximum moisture content

Analysis of the orientation effects

Average Mold Max Mold

Results – Damage functions for different orientations

Simulation results – Moisture contents of OSB in wood frame

Figure 14. Atlanta: Moisture content of OSB.

Simulation results – Moisture contents in CMU

Rankings of weather years based on RHT-index – Wood frame

light weight wood frame wall.

The selected locations were (IECC climate zone in Figure 17):