The Impact of Sorption History and

Hysteresis on Moisture Pattern in a

Wood-Framed Building Envelope
Dominique Derome, Ph.D.

ABSTRACT
Moisture present in envelope assemblies can accumulate within hygroscopic materials such as wood and cellulose insulation.
The amount of moisture adsorbed by the material is dependent on the relative humidity of air to which it is exposed or to the avail-
ability of water. These relationships are available under the form of sorption curves. It is known that wood does not retain the
same quantity of moisture content when exposed to the same relative humidity of the ambient air, whether wood is getting drier
or wetter. This phenomenon, called hysteresis between the adsorption and desorption curves, is rarely considered during modeling
or experimental studies of the building envelope. Because of hysteresis, the prior conditions of moisture exposure of wood have
an impact on its performance under the next set of conditions. Therefore, the sorption history of wood should be taken into account
in the assessment of the moisture performance of envelope assemblies.
This paper reviews the concepts mentioned above and then presents experimental data that exhibit the impact of different
sorption histories on the moisture pattern of wood members in assemblies exposed to the same conditions. It is shown that the
final moisture content depends on the sorption history.

INTRODUCTION MOISTURE STORAGE IN HYGROSCOPIC

Moisture present in envelope assemblies can accumulate MATERIALS
within hygroscopic materials such as wood and cellulose insu- Whereas many materials have little affinity for water and
lation. The amount of moisture adsorbed by the material is let water diffuse through their matrix, hygroscopic materials,
dependent on the relative humidity of air to which wood is such as wood, accumulate the diffusing moisture on their pore
exposed or to the availability of water. These relationships are surface and, at higher moisture content, within their pore.
available under the form of sorption curves. It is known that Moisture moves from high concentration to lower concentra-
wood does not retain the same quantity of moisture content tion areas, and this driving potential can be represented by a
when exposed to the same relative humidity of the ambient air, gradient of moisture content, relative humidity, or energy level
whether wood is getting drier or wetter. This phenomenon, of water. The rate of storage depends on temperature, ratio of
called hysteresis between the adsorption and desorption surface to volume, relative humidity, the diffusivity of the
curves, is rarely considered during modeling or experimental material, and the convective mass transfer coefficient at inter-
studies of the building envelope. Because of hysteresis, the faces. The rate of moisture movement cannot be directly
prior conditions of moisture exposure of wood have an impact measured but can be examined by monitoring changes in
on its performance under the next set of conditions. Therefore, moisture content. The moisture content at equilibrium for a
the sorption history of wood should be taken into account in given relative humidity and temperature can be measured.
the assessment of the moisture performance of envelope Sorption curves relate the equilibrium moisture content as a
assemblies. function of relative humidity and are developed with an exper-

Dominique Derome is an assistant professor in the Department of Building, Civil, and Environmental Engineering, Concordia University,
Montreal.

Buildings VIII/Moisture Control Performance Measurements—Principles 1

imental setup that employs steady-state conditions. In this
manner, a given material is exposed to an environment at
steady temperature and relative humidity. When the material
mass is stable, it indicates that its moisture content is in equi-
librium with the ambient conditions. The moisture content of
the material is expressed as a percentage of its dry mass:

m moist – m o
M = ----------------------------- × 100 (1)
mo

where
M = moisture content (%),
mmoist = mass of the moist material (kg),
mo = mass of the dry material (kg).
The procedure is then repeated for a higher relative Figure 1 Schematic diagram of the sorption of a
humidity. The same method is used for a wet material exposed hygroscopic material showing the hysteresis
successively to dryer conditions. Sorption curves are thus between the adsorption and desorption curves.
developed for the full range of relative humidity for a given
temperature.
The processes of desorption and adsorption of the same
specimen, at the same temperature, yield different equilibrium
curves. During desorption, a wet porous material retains more
moisture than it can adsorb at any given relative humidity. This
difference in moisture content at the same relative humidity
between the adsorption and desorption curves is called hyster-
esis, as shown in Figure 1. There are two common explana-
tions for hysteresis depending on the moisture content.
Green wood generally contains water in three forms—
liquid water partially or completely filling the cell cavities,
water vapor in the empty cell cavity spaces, and water in the
cell wall. The bound water found in cell walls is attracted to
wood with stronger forces than the free water held within the
cell. When wood is dried, the free water is lost first, as it is held Figure 2 Sorption curves function of the relative humidity.
with weaker forces due to capillary action. The moisture
content at which all free water of a cell has exited, but the cell
walls are still saturated with water, is called the fiber-satura- matrix. Spalt (1958) explained that during the cell formation,
tion point (FSP) (Tiemann 1906). It ranges from 20% to 40% cellulose is formed in a glucose solution to precipitate on the
of the dry weight. It follows that moisture movement above the cell walls. Cellulose is thus dispersed in water and there are
FSP is due mainly to capillary action and is a function of the few lateral cohesive bonds between cellulose fibers. During
vapor pressure above the meniscus. During desorption, pores drying, the gel gets more compact and lateral bonds develop.
present bottlenecks full of liquid. The meniscus at the surface The more the bonds, the less the mobility. Further removal of
of the liquid is of smaller radius than the one present at the start water produces distortion due to shrinkage, which also allows
of adsorption at the bottom of the cell. As the radius is smaller, cellulose fibers that are approaching each other to bond.
the partial pressure of vapor above the meniscus is lower and During drying, cellulose has, therefore, exchanged bonds with
less evaporation may take place. As the global vapor pressure water to lateral bonds with cellulose fibers, which results in a
is reduced further, the bottleneck will eventually be cleared reduction of available sites and explains the hysteresis below
and evaporation of the rest of the free water in the bottle- the FSP.
shaped pores will occur. At that point, the desorption and
adsorption isotherms converge. This explanation applies for FULL SORPTION CURVES
hysteresis in the capillary region. Figure 2 shows typical sorption curves (wood moisture
In the hygroscopic region (i.e., below the FSP), a reduc- content vs. air relative humidity) of two well-known indige-
tion of available sites for water explains hysteresis; this reduc- nous species—white spruce and sugar maple. These curves
tion has to do with the structure of a wood cell. The drying cover only the hygroscopic domain of wood (i.e., up to the
process below the FSP continues through water exiting the cell FSP). As generally the case for species with low extractive
walls. This process occurs at the molecular level of the wood content, the sorption curves are very similar. The magnitude of

2 Buildings VIII/Moisture Control Performance Measurements—Principles

the hysteresis is almost the same, and the FSP is around 30% For any pair of desorption and adsorption curves, inter-
M for typical ambient temperature. mediate desorption curves can be determined from any start-
ing moisture content on the full adsorption curve and extended
INTERMEDIARY SORPTION CURVES to join the desorption curves following the pattern determined
by Peralta. Such interpolated intermediate curves could be
The relationship between moisture storage and moisture used to understand the variations in moisture content within
transfer is not completely understood since the moisture stor- envelope assemblies. One example of an analysis referring to
age or release rates are not known for all sets of conditions. intermediate desorption curves follows.
However, transient moisture transfer in the building envelope
is more common than steady-state moisture transfer due to the IMPACT OF SORPTION HISTORY
diurnal cycling of temperature and relative humidity. With
To illustrate the impact of sorption history, the moisture
changing conditions, equilibrium is rarely reached.
content variations of the wood plank and the joist of the same
In most cases, the drying process starts at conditions roof assembly, exposed to similar conditions across the assem-
much lower than saturation and closer to the fiber saturation bly, are presented. The roof assembly, described in Derome
point. Examining the wood drying process from different and Fazio (2000), has been subjected to a full wetting/drying
starting points, Peralta (1995) observed that intermediary cycle of 190 days, and its configuration is presented in Figure
sorption curves fall between the full desorption and adsorption 4. The 4 m long assembly was subjected to air exfiltration
curves. This confirms the finding that the full curves should along its length for 91 days of winter conditions that led to
be considered as the boundary of the hysteresis (Urquhart a gradual wetting of the plank and the cellulose insulation.
1960). Any point between the curves can be met depending Then, spring and summer conditions induced drying. The last
on which conditions the material was subjected to. Therefore, 56 days of the drying cycle are presented here. Figure 5 shows
temperature and relative humidity alone are not sufficient to that relative humidity levels, measured in the middle and at
determine moisture content. The sorption history is necessary. the top of the assembly, are similar and almost constant
Starting on the adsorption isotherm, the intermediary curves throughout the duration of the period being analyzed. The
of Figure 3 almost join the desorption isotherm with a differ- presence of cellulose insulation in the whole assembly plays
ential of the relative humidity of approximately 40%. The a role in unifying the conditions within the assembly. There
92% intermediary curve is the only one with the characteristic is a small temperature gradient across the assembly, between
sigmoidal shape. The other ones have a simple convex form 3º and 6º. Both the plank and the joist were made of spruce.
and, compared to the full desorption curve, demonstrate a The conditions within the roof assembly induced some
smaller reduction of moisture content for the same decrease moisture movement in the wood members of the assembly
of relative humidity. (i.e., the gradual drying of the wood planks located below the
roof membrane and the wetting/drying pattern for the wood
joists located at the bottom of the assembly). The cross

Figure 4 Cross section of the roof assembly composed

(from top to bottom) of elastomeric roof
membrane, wood planks, cellulose insulation,
plaster on wood lath, and alkyde paint. The
Figure 3 Full adsorption and full and intermediary structure is composed of 2-by-8-in. joists and 2-
desorption curves for yellow poplar according to by-6-in. cross members. The diagram shows the
Peralta (1995) as a function of relative humidity, location of the sensors and the gravimetric wood
at 21ºC. samples.

Buildings VIII/Moisture Control Performance Measurements—Principles 3

 Figure 7 Schematic representation of the desorption and
adsorption processes involved in the roof
assembly. The gray arrow represents the
desorption process of the roof plank while the
Figure 5 Temperature and relative humidity conditions black arrows show (a) the adsorption process of
measured across the roof assembly, as well as the joist, followed (b) by a reversal of the moisture
above and below the assembly. movement according to an intermediary
desorption curve.

The continuous release of the moisture from the wood

plank and the cellulose insulation maintains relatively
constant relative humidity conditions across the assembly,
thereby inducing a gain in moisture content in the bottom joist.
The joist had remained dry through the wetting part of the test.
Notwithstanding equilibrium once again, the wetting pattern
of the joist can be understood to follow the full adsorption
curve, as shown also on Figure 7a.
Around mid-September, as drying of the assembly
continues and less moisture is released, the conditions in the
assembly allow the plank to continue drying and reverse the
moisture transfer in the top of the joist. Figure 7b indicates the
Figure 6 Moisture content vs. time for the plank and joist of
intermediary desorption curve of the wood joist when the
the roof assembly measured continuously with
moisture movement reverses. The first part of the intermedi-
moisture content resistive pins and bimonthly
ary desorption curve is not as steep as the full desorption
with gravimetric samples.
curve. The same variation in relative humidity that would lead
to a more important moisture content variation in the wood
members between the wood planks and the joists were not plank along the full desorption curve induces a much smaller
monitored. variation in the moisture content of the joist along the inter-
Figure 6 shows the variations in moisture content of the mediary desorption curve during the drying process. The
wood plank that, after being wetted through the winter condi- resultant moisture contents of the wood plank and joist are not
tions, is drying at a regular pace. The core of the plank dries only dependent on the relative humidity to which they are
from 19% M (moisture content) to below 9% M as measured exposed, but also on the previous conditions to which they
with a pair of resistance-type moisture pins. Gravimetric were exposed, and finally on the direction of the moisture
measurements for the whole thickness of the wood plank indi- movement.
cate variations from 15% M to 11% M. The difference in the
measurement indicates that equilibrium is not attained To illustrate further the sorption history shown on Figure
through the wood plank thickness. While it is understood that 8, arrows have been superimposed on the data presented there.
sorption curves convey equilibrium moisture content relation- Figure 8 clearly highlights the continuous desorbing on the
ships with relative humidity, the continuous drying pattern of wood plank with the gray arrows. The black arrows are dented,
the wood plank can be approximated to follow the full desorp- marking the reversal or change of rate of the moisture transfer.
tion curve of the isotherm as shown schematically in Figure While full and intermediate desorption curves do not cross
7a. over, as the full curve is an upper boundary, Figure 8 shows

4 Buildings VIII/Moisture Control Performance Measurements—Principles

 CONCLUSION

This paper has reviewed the concept of moisture content,

hysteresis, and full and intermediary sorption curves, which
explained that similar conditions can lead to different moisture
content in the wood depending on its sorption history. The
paper has presented experimental data that show the impact of
the sorption history on the moisture patterns of the wood
members exposed to the same conditions within an assembly.
The different sorption history of the plank and the joist led the
wood members to have different rates of variation of moisture
content and to reach different final moisture contents.
Although sorption curves are equilibrium relationships
that cannot fully explain the transient aspect of moisture trans-
fer, taking into account the sorption history can give more
insight into the complex moisture behavior of wood-framed
assemblies exposed to varying conditions, as it provides a
basis for explaining the different moisture contents attained.

Figure 8 Data presented in Figure 6 with an overlay of REFERENCES

arrows. The grey arrows show the decrease of
moisture content in the plank. The black arrows Derome, D., and P. Fazio. 2000. Large scale testing of two
show the change of moisture content in the joist flat roof assemblies insulated with cellulose. ASCE
where wetting is slowed in the gravimetric sample Journal of Architectural Engineering, Vol. 6 No. 1. New
measurements and drying induced as measured York: American Society of Civil Engineers, pp.12-23.
by the moisture pins. Goulet, M. 1968. Phénoménes de second ordre de la sorption
d′humidité dans le bois au terme d′un conditionnemenr
de trois mois á température normale. Notes derecherdes
that the reversal of the direction of the moisture transfer takes
No. 3, Dép. Expl. bois, Universite Laval, Mars.
some time to be effective through the specimen. Thus, the tran-
sient nature of moisture movement explains the crossover of Hedlin, C.P. 1966. Sorption isotherms of twelve woods at
the gray and black arrows (i.e., the fact that at the end of the subfreezing temperatures. Forest Products Journal, Vol.
17, No. 12, pp. 43-48.
test, the joist is more moist than the plank). One additional
reason may be the increase of around 3°C of the outside Peralta, P.N. 1995. Sorption of moisture by wood within a
temperature in the last 15 days of the test. Although other limited range of relative humidities. Wood and Fiber
factors, such as different diffusivity and surface-to-volume Science. Society of Wood Science and Technology, Vol.
27, no. 1, pp. 13-21.
ratios, may explain in part the difference in behavior between
the joist and the plank, the example presented indicates that Spalt, H.A. 1958. The fundamentals of water vapor sorption
wood members within the same assembly, exposed to the same by wood. Forest Products Journal, pp. 288-295.
conditions, have different changes in moisture content Tiemann, H.D. 1906. Effect of moisture upon the strength
depending on the sorption history. Furthermore, intermediary and stiffness of wood. USDA for Serv. Bull. 70.
sorption curves, which result from hysteresis, may explain the Urquhart, A.R. 1960. Sorption isotherms in J.W.S. Hearle
different rate of change in moisture content following a rever- and R.H. Peters, eds., Moisture in Textiles. New York:
sal of moisture movement direction from wetting to drying. Wiley Interscience, pp. 14-32.

Buildings VIII/Moisture Control Performance Measurements—Principles 5