Construction and Building Materials 160 (2018) 211–222

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Multicriteria analysis demonstrating the impact of shiv on the properties

of hemp concrete
César Niyigena ⇑, Sofiane Amziane, Alaa Chateauneuf
Université Clermont Auvergne, Institut Pascal, Polytech’ Clermont-Ferrand, 63174 Aubière Cédex, France

h i g h l i g h t s

 A multicriteria analysis was used to classify the types of shiv into three groups.
 Three mechanical responses were demonstrated for the compression test on hempcrete.
 Large specific surface area of shiv is the source of weak binder–aggregate bonding.
 The shiv which optimises all characteristics, offers concrete with good performance.
 A multicriteria analysis allows also to classify hempcrete into three groups.

a r t i c l e i n f o a b s t r a c t

Article history: The shiv derived from the shredded stem of the hemp plant [1], when mixed with a mineral binder and
Received 25 August 2017 water, produces hemp concrete (or ‘‘hempcrete”), which is considered as an eco-material because of its
Received in revised form 8 November 2017 remarkable environmental properties (favourable CO2 balance, easy management and disposal at the end
Accepted 9 November 2017
of the product life, etc.) [2]. However, much like any agricultural product, its quality may be affected by
Available online 29 November 2017
the soil where the hemp has been grown, the weather conditions, and a variety of other factors. As these
parameters are rather variable, the produced plant matter tends to be somewhat irregular, which can
Keywords:
hamper the development of this type of building solution. The study at hand tests the characteristics
Shiv
Hempcrete
of the hemp (particle size distribution, water absorption capacity, etc.), to assess the impact of the vari-
Multicriteria analysis ability of these factors on the performance of the hempcrete made with the shiv. To begin with, the thir-
Mechanical performance teen types of shiv used in the study were compared by multicriteria analysis, and then classified into
Elastic modulus three different groups.
Apparent modulus Next, nine types of shiv were selected, so that each of the three groups was represented. In identical
Mechanical response conditions (i.e. the only variable being the type of shiv), specimen of hempcrete were made up in the
lab for experimentation. The aim here was to validate the results of the study characterising the different
types of shiv. The results show that there are three main types of mechanical response when the hemp-
crete is subjected to a compression test, where the material behaviour is characterised by small, medium
or large strains. The large specific surface area of shiv is known to be responsible for weak binding at the
interface between the binder and the plant particles, leading to total failure of the test piece. In addition,
its high water absorption capacity, which is the likely cause of the ‘‘false set effect” associated with the
elongated form of the particles, is identified as being responsible for the high rates of deformability found.
It should be noted that in the latter case, the specimens behave like a succession of layers, which accounts
for the significant levels of strain.
By multicriteria analysis of the results for the hempcrete specimens, we found that they can be also
classified into three groups, matching the three detected by the multicriteria analysis of the types of shiv
from which they were made. Unlike the previously published literature, this study leads to the conclusion
that shiv types with small particle sizes and high water absorption capacities do not produce the best
performance.
Ó 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.
E-mail addresses: cesar.niyigena@uca.fr (C. Niyigena), sofiane.amziane@uca.fr
(S. Amziane), alaa.chateauneuf@uca.fr (A. Chateauneuf).

https://doi.org/10.1016/j.conbuildmat.2017.11.026
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
212 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222
1. Introduction
Owing to its high degree of porosity, hemp concrete is sensi-
tive to a number of factors, which can be classified into three
categories: (1) methods of manufacture or installation [3–5];
(2) methods and techniques for measuring the properties of the
hempcrete [6–8]; and (3) nature and quantity of the components
[9–13]. In light of the objectives of this study, and in view of
the three categories mentioned above, we shall focus on the
nature and quantity of the components – more specifically, the
shiv.
Certain properties of the shiv have been identified as being
responsible for problems observed during the setting and harden-
ing of hempcrete [11,14]. Such problems prevent strong connec-
tions between the binder and the shiv, leading to poor
mechanical performances of the material. These properties are
often due to the morphology of the shiv (the particle size)
[15,16]; to its composition – in particular, its fibre and dust content
[17], to its chemical composition [18,19] or to its water absorption
capacity [20].
A literature review concerning the morphological influence of
the shiv on the properties of hempcrete shows discrepancies
between various results, highlighting the need for further research
to shed light on the matter. In a study published in 2012 [15], three
types of shiv (N°1 to N°3) – of equal thickness but different widths,
lengths and densities – were used. To exclude any influence by the
binder, a single type of binder was used: a mixture of 75% aerated
lime, 15% hydraulic lime and 10% pozzolana. The test pieces made
up with the finest-grained shiv (N°3) exhibit poor strength at 28
days, but much better strength at 4 months, which improves sig-
nificantly beyond that point. The authors state that shiv N°3 is less
porous because its particles are finer than those of the other two
samples. Consequently, the CO2 which is responsible for setting
and hardening is less easily diffused, which slows the kinetics of
drying of the test pieces, leading to poor mechanical strength in
the short term. Conversely, as the particles are fine, they are well
coated with binder when the concrete is mixed, which accounts
for the better mechanical performances observed in the long term.
In conclusion, the use of fine-grained shiv helps improving the
hempcrete’s long-term mechanical performances. This conclusion
is not entirely consistent with the work of Nguyen [21]. Like
Arnaud and Gourlay [15], Nguyen also observed the better
mechanical performances at 28 days for hempcrete made with lar-
ger particles, but noted no significant difference at 90 days. How-
ever, the results found by Stevulova et al. [10] are in keeping
with Arnaud and Gourlay’s [15] results. Note, also, that the study
on panels manufactured with binderless hemp shiv demonstrated
the improvement of the mechanical properties with smaller parti-
cles of hemp [22].
In the work of Stevulova et al. [10], six samples of shiv (two
original samples and four fractions), with different particle sizes,
were studied in the same conditions (same binder, formulation,
etc.). The results at 28 days show that an increase in the average
length of a particle decreases the compression stress. At the same
time, the compression stress differs depending on the origins of the
used shiv. On the latter point, the authors merely stated that the
origin of the shiv can impact the mechanical behaviour of hemp-
crete, but do not go into further detail. The origin is not, in itself,
a quantifiable parameter, which makes it difficult to take it into
account. However, it is possible to study its impact on the irregu-
larity of the hempcrete, based on other characteristics such as,
say, the chemical composition or water absorption. This, as we
shall see later on, is the reason why other characteristics are taken
into account in our study. In spite of the contradictions, many
authors agree that decreasing the particle size leads to improve-
ments in the mechanical performance.
Consequently, for this study, we consider the assumption that it is
better to minimise the particle size
Besides the size of the shiv particles, the fibre content can also
influence the final properties of hempcrete. Chamoin [23] looked at
two types of shiv: one without fibre (F1) and the other containing
fibres (F2). The production conditions were exactly the same – in
particular, the nature of the binder, the proportion of binder to shiv
and the method of installation. The only difference lies in the
amount of water used, as F2 requires more mixing water. The
author concludes that F1 has better mechanical performances,
and thus the use of F2 affects the setting of the binder. Indeed,
the large amount of mixing water required by the fibrous shiv
means that the binder is diluted, lessening its mechanical strength
when set. Finally, the use of F2 increases the hempcrete
apparent density, but noticeably decreases its performance under
compression.
In an earlier study [21], it was found that fibres helped to pre-
serve the structure of the test specimens, preventing them from
collapsing. A priori, this would seem to indicate an improvement
of the mechanical performance. However, according to the analysis
of the experimental results, this is not the case: the test specimens
made up with fibrous shiv exhibit low values of compression stress
and Young’s modulus.
The author accounts for these results by saying that there is
greater inter-granular porosity with the hempcrete made from
fibrous shiv. In addition, the adhesion between the fibres and the
binder may be poorer than that between the shiv and the binder,
which further degrades the mechanical performance. Also, De
Bruijn [24] observed no difference in terms of mechanical beha-
viour, between the use of fibrous and pure shiv.
In view of these contradictions, and in light of the non-fibrous nat-
ure of the shivs used in our study, we assume that fibres do not have
significant impact on the mechanical performance of the hempcrete.
Therefore, fibre content is not taken into account.
Recently, studies have been carried out to demonstrate the
effect of the chemical compounds of shiv on the properties of the
resulting hempcrete [11,14]. When the shiv is put in water, mole-
cules which are weakly bound to the cell walls dissolve; these
water-extractible compounds are called ‘‘lixiviates”. When the
hempcrete is being made, they combine with the mixing water
needed to hydrate the cement, which can affect its setting and
hardening. It has been shown that these lixiviates are made up pri-
marily of ash, lignin, sugars, proteins and other compounds which
have to be identified. The ash may derive from the minerals mak-
ing up the plant, or from dust which has not been entirely elimi-
nated during the defibration process [25]. The shivs used contain
lignin in varying quantities, but lignin has a significant impact on
the time taken for the cement to set [26]. Most likely, it is that lig-
nin content which is to blame for the phenomenon of ‘‘false set”
observed during this study. In conclusion, the nature of the
water-extractible compounds depends on the type of shiv. They
greatly delay the setting, and hence have a negative impact on
the performances of the hempcrete.
If lime is considered, there are a number of elements missing
from the literature regarding the impact of the chemical com-
pounds on the properties of hempcrete.
Therefore, the chemical compound aspect is not taken into account
in the context of this study – especially as the binder used is lime.
Although the impact of shiv on hempcrete properties has been
studied [7,10,15,21], where conflicting results are strongly
observed, we can highlight the interest and the need to investigate
this issue further.
The first step in conducting an in-depth study is to characterise
the type of shiv on the basis of several properties at once. The char-
acteristics taken into account are: the bulk density, the water
absorption capacity and the particle size distribution (PSD). Using
 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222 213

multicriteria analysis, we are able to compare different types of Table 1

hemp. Following this phase, experimental tests are conducted on Nomenclature of the different types of shiv.

the hempcretes made with different types of shiv investigated ear- Types of hemp Nomenclature
lier. It should be noted that the properties of these hempcretes are Fibrous Auvergne hemp C1
also analysed using the multicriteria approach. Fresh hemp (TREMARGAT) C2
The results of this study, unlike the pre-existing literature, show English hemp C3
that the types of shiv with small particle sizes and high water Fresh Auvergne hemp (Standard) C4
LCDA hemp C5
absorption capacity do not necessarily produce the best mechani- ISOcanna hemp (CESA) C6
cal performances. It is evident that the size alone (length, width, Swedish hemp C7
etc.) of the type of shiv is not sufficient to predict the hempcrete KANABAT hemp, Vicat (shiv 1) C8
performance. Other characteristics need to be taken into account KANABAT hemp, ENTPE (shiv 2) C9
Standard Combraille hemp C10
(e.g. water content, etc.), and the type of shiv that optimises these
Standard Lézoux hemp C11
characteristics produces concrete with the best mechanical Standard Brioude hemp C12
performance. CVF hemp C13

2. Materials and methods Table 2

2.1. Types of shiv samples
The materials used for this study are differentiated on the basis of, firstly, the
types of shiv, and secondly the resulting hempcretes. In the former case, we exper-
imentally tested and classify thirteen different types of shiv, as shown in Table 1;
they can be identified by the name of their manufacturer and/or origin. Among
the types of shiv used in the study, six are produced in the Auvergne-Rhône-
Alpes region in France, and are not specifically certified for use in construction;
the others are certified and commercially available, and are produced either in
France or elsewhere in Europe. They are distinguished by their characteristics (size,
water absorption capacity, etc.). Hereinafter, for simplicity, the different types of
shiv are referenced as C1–C13 – see Table 1.
Note that C8 and C9, which are produced in France and are available on the
market, were also used in a detailed RRT13 study in [6]. Among these types of shiv,
nine types were used to make up samples of hempcrete for the study. They were pro-
duced in exactly the same conditions (formulation, drying, etc.).
2.2. Composition of hempcrete specimens
The used formulation is walling hempcrete, with 250 kg/m3; to make up
100 litres, we need 25 kg of binder, 12 kg of shiv and 30 kg of water. The used cylin-
drical moulds are 11  22 cm. To limit the number of samples, a distinction is
drawn between the primary and secondary tests, Table 2. In the former case, a large
number of specimens were made: nine per timescale (four timescales in total, at 30,
60, 90 and 180 days after pouring), whereas in the latter, they are control tests, in
which three specimens are tested at 30 and 180 days. In total, 196 specimens were
mixed for this study, and classified as described below.
For instance, the label ‘‘P-C2-11-2” would denote: primary test, TREMARGAT
fresh hemp (C2), 11  22 cm cylindrical mould, and specimen N°2 in this
configuration.
2.3. Mixing and testing of hempcrete sample pieces
Overview of codes used for numbering specimens.
Type of test Type of Type of mould Number of
hemp specimens
P: Primary S: C1 to C13 11: Cylindrical 1 to n in each
Secondary (11  22 cm) case
2.4. Methods for characterisation and analysis
We can distinguish between the methods used to characterise the shiv material
and those used to characterise the hempcrete. In the literature, there are a number
of methods for characterising hempcrete: density [27], water absorption capacity
[20] and PSD using image analysis [28]. An inter-laboratory study [27] has focused
on methods of characterising the shiv. By comparing different protocols between
different labs, the results have shown to provide good repeatability, and recommen-
dations have been put forward. In our study, for quantifying the density and water
absorption capacity, we used the recommended protocols [29]. Granulometric anal-
ysis consists of studying the distribution of different types of grains contained in a
sample, based on their characteristics such as weight or size. Such an analysis can
be conducted either by mechanical sieving or by image analysis. In this study, the
image analysis method [28] was applied using the software ‘‘image J” [30]. With
this method, it is possible to obtain the number of particles contained in a given
sample of shiv, and also the projected area of each shiv particle. Therefore, the
specific area is considered to be equal to the sum of the projected areas for all shiv
particles contained in a sample. It should be noted that, for the study presented
herein, a sample shiv of 3 grams in mass was used.
The characteristics of hempcrete to be determined are the density, the initial
and apparent moduli, the compression stress and the heat conductivity. In the latter
case, the so-called ‘‘hot wire” method is used [31]. In addition, the protocols and
methods used to measure the other properties (moduli and compression stresses)
are the same as those used in the previous study [6].
Beyond these methods, multicriteria analysis is favoured. Such an approach
enables us to carry out a comparative study, taking account of several criteria. It
is valid for the type of shiv and the hempcrete, and details are given below.

To make the test specimens, for a given type of shiv, the moulds were filled
freely by hand, up to around four layers. The upper surface of each layer was
scratched to help the next layer to adhere properly. In these conditions, a reference
mould was filled and then weighed, the other specimens are calibrated to give the
same weight of this reference specimen. This strategy enables us, indirectly, to con-
trol the compacting energy within the same type of shiv. The same operator filled
the reference moulds, in order to help control the compacting energy between
the different types of shiv.
When filled and weighted, the specimens were left in their moulds for 72 h.
After removal, two different types of drying were used: open-air drying and oven
drying.
Open-air drying took place in a test room for around 6 months – see Fig. 2(a). As
the climate conditions in that room are not precisely controlled, the pieces were
exposed to variations of temperature and relative humidity, respectively between
14 and 30 °C and between 31 and 66% RH. Average conditions were 21 °C and
48% RH – see Fig. 1.
As it is not possible, with open-air drying, to eliminate all of the water con-
tained in the sample, it is necessary to further dry the pieces in a kiln. They were
kiln dried at 60 °C for 48 h, which corresponds to a variation in mass of less than
0.1% when weighed twice at an interval of 24 h.
1
RRT: Round Robin Test.
2.5. Multicriteria analysis: Comparative scale
By analysing the results obtained in the literature [10,15,21], we were able to
identify the characteristics of the types of shiv likely to influence the performances
of hempcrete. However, the characteristics are taken into account separately. By
taking account of several different characteristics, we are better able to appreciate
the impact of the type of shiv.
For all of the characteristics being analysed, a representation on an unidimen-
sional scale of 1–10 was chosen. That unidimensional scale allows us to represent
all the characteristics on a single plot, thus providing an overall view of the results.
2.5.1. Principle of the scale
For a given characteristic, the maximum value observed among the different
types of shiv or hempcrete corresponds to 10. The values of the other types of hemp
or hempcrete are then expressed by simple linear interpolation. It should be noted
that:
 the scale from 1 to 10 is chosen and considered for a given characteristic and for
all types of shiv or hempcrete;
 the representation on the same graph enables us to observe the correlation
between the different characteristics under study.
214 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222

Fig. 1. Temperature and HR where the test pieces were dried.

Fig. 2. Conditions of drying of the pieces: (a) open-air drying and (b) kiln drying.

2.5.2. Limitations of the scale

This representation, although it has the advantage of showing all the character-
istics for all types of shiv or hempcrete, it gives the same weight and importance to
all the characteristics being studied. In the case of the different types of shiv, the
characteristics are considered as though they had the same influence on the hemp-
crete performance which is not the case in reality. The lacking in terms of the
weight attached to each of the characteristics is due to missing data in literature
about this issue. To the best of our knowledge, there is no method of quantitatively
predicting the impact of a combination of given characteristics on hempcrete per-
formance. In conclusion, the comparative scale should allow such a comparison,
but it cannot be performed for the moment, because of a lack of information on
the subject.
70.83 kg/m3 to 158/85 kg/m3. The initial and final water content
ranges in 112.23%–242.59% and 293.05%–432.49%, respectively.
For the granulometric analysis, also variability can be seen; for
example, the length of the particle ranges from 0.64 mm to
5.88 mm. For the time being, these degrees of variability are
too large for comparative analysis. For this reason, the present
paper considers a multicriteria study taking account of all the
characteristics.
3.2. Multicriteria comparison

3. Results and analysis
3.1. Characterisation of the type of shiv
The characteristics considered in herein are listed in Table 3;
they are essentially associated with the bulk density, the water
absorption capacity and the granulometry. To simplify the nota-
tion, their abbreviations are given below and are used hereinafter
in this article.
The data in Table 3 show high variability for all of the
characteristics. For example, the bulk density varies from
The results are illustrated in Fig. 3 according to the unidimen-
sional scale discussed above. It can be seen that the hemp materi-
als occupy almost the entire from the centre to the boundary,
which represents the significant variability of the properties of
the studied shiv. The curves which are located around the centre
of the chart naturally correspond to types of shiv with low values
for the considered characteristics. Conversely, those which are
around the outskirts exhibit high values. In the interests of read-
ability in the representation, it is more sensitive to represent the
curves separately, distinguishing those at the centre, in the middle
and on the outside.
 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222 215

Table 3
Nomenclature and overview of results of characterisations of types of shiv.

Nomenclature BD: Bulk IWC: Initial FWC: Final MPS: Mean Particle MP: Mass of L: W: El: FD: Feret
Density Water Content1 Water Content2 Surface3 Particle Length Width Elongation Diameter
[kg/m3] [%] [%] [mm2] [mg] [mm] [mm] [mm]
C1 70.8 159.8 293.1 0.91 0.18 0.64 0.19 2.62 0.32
C2 89.7 194.2 379.2 4.89 0.84 3.40 1.04 3.47 1.79
C3 118.0 242.6 432.5 1.57 0.28 1.11 0.32 2.63 0.58
C4 118.3 233.8 358.3 1.94 0.35 1.45 0.44 2.60 0.76
C5 125.7 154.0 351.3 8.10 1.77 5.88 1.40 4.97 2.78
C6 128.2 181.4 358.5 5.31 1.14 4.20 1.27 3.95 2.23
C7 129.9 163.6 321.9 3.25 0.79 1.93 0.42 3.19 0.84
C8 143.6 152.7 328.0 6.96 1.51 5.59 1.23 5.52 2.53
C9 147.5 211.8 381.5 6.95 1.50 5.11 1.38 4.47 2.58
C10 130.7 112.2 307.3 0.82 0.15 0.77 0.25 2.28 0.41
C11 95.4 165.9 344.6 1.72 0.23 1.46 0.34 3.32 0.66
C12 103.9 162.9 338.7 1.18 0.18 1.02 0.28 2.77 0.50
C13 158.9 226.2 375.1 1.36 0.22 1.11 0.52 2.27 0.80
Minimum 70.8 112.2 293.1 0.82 0.15 0.64 0.19 2.27 0.32
Maximum 158.9 242.6 432.5 8.10 1.77 5.88 1.40 5.52 2.78
Mean 120.0 181.6 351.5 3.46 0.70 2.59 0.70 3.39 1.29
Standard deviation 24.7 38.0 36.4 2.64 0.60 1.96 0.48 1.05 0.94
1
Quantity of water absorbed by a 20 g sample of shiv after 1 min of immersion.
2
Quantity of water absorbed by a 20 g sample of shiv after 48 h of immersion.
3
Obtained by dividing the specific surface of a 3 g sample by the number of particles in that sample.

Fig. 3. Comparison of the different characteristics for all the types of shiv. Fig. 5. Types of shiv classified in group 2.

analysis and separate representation, they can be classified into

three groups.

3.2.1. Group 1 shiv

Fig. 4. Types of shiv classified in group 1.
It is to note that, this representation is not, in itself, sufficient. It
is important to explain how exactly the types of shiv occupying
different areas of the chart differ from one another. Thus, with
Earlier studies [15,32,33] have shown the negative impact of
water absorption capacity and size of the hemp particles on the
mechanical performance of hempcrete. Moreover, it has been
shown that bulk density increases when the size of particles
decreases. Indeed, in the case of non-fibrous types of shiv, the lar-
ger particle size is, the more difficult it becomes for the particles to
rearrange. Hence, we see an increasing number of empty spaces
between the particles, resulting in a large volume and a low den-
sity. For instance, C2 hemp has large particles, measuring 1.04
mm and 3.4 mm in width and length, respectively, and it has a
low density in comparison to C13 hemp. The latter has small par-
ticles: 0.52 mm and 1.11 mm in width and length, respectively –
see Table 3. However, as the low bulk density corresponds to a high
volume and therefore a high degree of inter-particle porosity, one
might imagine that the quantity of water absorbed should be high.
The results found show that this is not strictly true, Fig. 4. Indeed,
the method of curing, which is used to measure water absorption,
216 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222
is capable of eliminating a significant amount of the inter-particle
water. Thus, it is clear that the bulk density is not strictly propor-
tional to the amount of absorbed water. Consequently, bulk density
is a characteristic which should be maximised. On the other hand,
particle size and water absorption capacity are two characteristics
which should be, a priori, minimised.
It should be stressed that fibrous shiv C1 is a special case, hav-
ing a low density of 70.83 kg/m3 but falling into this category. This
low apparent density is linked to the fact that the bulk fibres have
difficulty in rearranging themselves, which increases the amount
of inter-particle voids, leading to a large volume and therefore a
low density. The characteristics identified as being harmful – nota-
bly water absorption capacity (initial and final water content) and
particle size (length, width, etc.) are low (except for C13), on one
hand and the characteristic identified as beneficial (density) is
higher, on the other hand. Consequently, these types of shiv, in
group 1, will produce mechanical performance that is relatively
favourable in comparison to groups 2 and 3.
3.2.2. Group 2 shiv
Group 2, Fig. 5, includes types of shiv whose characteristics are
largely similar to those of group 1 – particularly regarding the bulk
density and water absorption. The results range from 164% to 243%
and from 322% to 432% for the initial and final water absorption,
respectively. The bulk density ranges from 89.74 kg/m3 to
129.91 kg/m3. However, we must point out significant differences
regarding the size of the particles. In this group, the particle size
increases, with values of around 2.60–3.47 and 0.58 mm–1.79 m
m of elongation and Feret diameter. It has been shown earlier that
this characteristic needs to be minimised.
According to our analyses and to references [10,15], increasing
particle size has a negative effect on the mechanical performance
of hempcrete. Therefore, the types of shiv in this group will prob-
ably produce hempcretes with poorer mechanical performance
than those of group 1 shiv types.
3.2.3. Group 3 shiv
The types of shiv in group 3 tend to occupy the outside of the
radar chart, corresponding to maximum values for certain charac-
teristics such as particle size. We observe density values of around
135 kg/m3, on average, with a minimum of 125.56 kg/m3 and a
maximum of 147.5 kg/m3 for C5 and C9 hemp, respectively. The
analyses presented earlier show that this is a characteristic which
it is best to maximise.
In view of the high values of the particle size and water content,
and given the unfavourable effects these properties are known to
have, this group is likely to yield results of poor quality. It is
assumed herein that harmful characteristics are considered to be
predominant over the density. Furthermore, the strong correlation
between the characteristics for this group (Fig. 6) may lend these
types of shiv a particular behaviour giving rise to unexpected
results. For this reason, a study characterising the hempcrete made
with the types of shiv tested here is needed in order to better
understand the impact of the type of shiv on the hempcrete
materials.
3.3. Characterisation of hempcrete
3.3.1. Mechanical response
Three mechanical behaviours of hempcrete are observed in
relation to the type of shiv used. They are characterised by differ-
ent levels of strain:
 Low level of strain: there is low strain, of around 3%, so that the
maximum compression stress is reached, in certain cases, in the
second cycle of loading before the end of the test.
Fig. 6. Types of shiv classified in group 3.
 Moderate level of strain: there is an intermediate behaviour,
characterised by medium strain levels, with maximum stress
at about 5% of strain.
 High level of strain: there are specimens which exhibit high
strain; they are compressible to beyond 20%, which is the max-
imum allowed level.
The Figures below give (left to right) the specimen photos
before and after, and the stress–strain curve (Figs. 7–9), for the
three behaviours. Low strain is encountered when using hemp
with small particles (1.36 mm2 average particle surface area) and
a high specific surface area (18822 mm2/3 g), such as the C13 shiv.
The particles prevent good bonding with the binder matrix, which
weakens the assembly and results in poor resistance. Failure is
characterised by the test piece being totally crushed – see Fig. 7b.
In the case of high strain, the shiv particles rearrange to a series
of stacked layers. This rearrangement is aided by the size of the
particles, which means there is a high degree of overlap between
the particles. It is also possible that the high water absorption
capacity (340% after 48 h) for this type of shiv increases the level
of strain. Indeed, as the water needed for setting and hardening
of the binder is absorbed by the shiv, we are likely to see the phe-
nomenon known as false set. As a result of this phenomenon, the
binder matrix no longer performs its intended mechanical func-
tion, and so the hempcrete exhibits behaviour similar to that of
hemp aggregate, characterised by high strain (Fig. 9c). It must be
stressed that these findings cannot necessarily be generalised
across all the studied shiv types. This suggests that the chemical
interaction between the binder and the types of shiv can also con-
tribute to the mechanical response.
Studying the chemical composition and the molecules weakly
bound to the ‘extractable’ plant cell walls may help to better
understand the impact of the binder–shiv chemical interaction
on the setting and the hardening of hempcrete. In an earlier study
[34], these ‘extractable’ chemical compounds were shown to sig-
nificantly affect the hydraulic setting of binders.
3.3.2. Properties of the hempcrete
3.3.2.1. Density. The density results obtained after drying the test
pieces in a kiln are presented in Fig. 10. Values range from 385
kg/m3 to 480 kg/m3 for the C3 and C10 shiv types, respectively.
The previous density study [7] showed slight dispersion for trials
conducted in the same context (on a single formulation). This
seems to be the case in the current context. Nevertheless, disper-
sions are observed depending on the particular type of shiv.
 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222 217

Fig. 7. Behaviour of hempcrete with low strain.

Fig. 8. Behaviour of hempcrete with moderate strain.

Fig. 9. Behaviour of hempcrete with high strain.

3.3.2.2. Thermal conductivity. Thermal conductivity was measured
by the ‘‘hot wire” method as cited above (3.4). The test pieces used
were the same as for the compression test. Thus, for a thermal con-
ductivity measurement, two test pieces are needed, with the hot
wire being placed between them during the test, as shown in
Fig. 11.
To ensure good repeatability of results, three measurements
were carried out by coupling the test pieces, two by two. For exam-
ple, for tests with only three test pieces available per age (case of
60 days and 90 days), the coupling was as follows: (1 with 2); (2
with 3) and (1 with 3). Then, an average of three measurements
is calculated, and results at different ages are shown in Fig. 12
below.
The results show that the thermal conductivity of hemp con-
crete varies depending on shiv type. At 30 days, a minimum of
0.093 W/m
K and a maximum of 0.124 W/m
K are observed for
C12 and C5, respectively. Note that over time, the thermal conduc-
tivity decreases, with values under 0.1 W/m
K being obtained at
218 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222
Fig. 10. Density of different types of hempcrete after drying.
Fig. 11. Experimental device for thermal conductivity test.
Fig. 12. Thermal conductivity for different hemp concretes at different ages.
180 days. This is consistent; in fact, the more the material dries, the
more the water contained in the pores evaporates; these pores are
then either empty or filled with air, which increases the insulating
capacity of the material. At this same age, results confirm again,
the change in thermal conductivity, as a minimum of 0.075 W/
m
K is observed for both C4 and C13 with a maximum of 0.092
W/m
K for C3. Later in this paper, a comparative study between
hemp shivs and hemp concrete will illustrate how and why these
values differ depending on hemp type.
3.3.2.3. Maximum compression stress. Unlike density, the compres-
sion stress results reveal considerable variability on two levels.
Firstly, variability is observed with respect to the type of shiv, with
a factor of 10 between the minimum and maximum observed val-
ues – between the C13 and C6 shiv types, respectively (Fig. 13).
Secondly, no-insignificant dispersions are observed within a
given category of shiv. For example, for shiv types C2 and C5,
respectively, we see minimum values of 0.47 MPa and 0.65 MPa,
and maximum values of 0.65 MPa and 0.99 MPa (Fig. 13). Over
time, the values are seen to improve at 60 days and stabilise at
180 days. The lower values are observed at 90 days. The cause of
this variation has not been identified, but it is likely related to
the change in the testing used machine. Indeed, at 90 days, a differ-
ent machine was used for reasons relating to time management.
However, it has been shown that the compression stress results
for hempcrete can be affected by the test machines [7].
3.3.2.4. Apparent and elastic moduli. The variations observed for the
stress are confirmed in the case of the modulus. The results of the
apparent modulus show that the results are less dispersed for cer-
tain types of shiv than for others. This is particularly the case for
the C12 shiv type, characterised by low dispersions in comparison
to other shiv types (Fig. 14). Furthermore, this low dispersion of
C12, although less marked, remains observable for the compres-
sion stress with values of the order of 0.4 MPa (Fig. 14).
The same trends are observed for the elastic modulus, Fig. 15.
Here, also low dispersions are observed for C2 shiv. These results
show how the variability is sensitive to the property under consid-
eration and, in this respect, are in agreement with the existing lit-
erature [7].
In addition to presenting the results with the individual charac-
teristics, the multicriteria approach offers an in-depth analysis
drawing the link with the properties of the shiv type.
3.4. Demonstration of the impact of the type of shiv on the hempcrete
by multicriteria analysis
The characteristics considered in this section are listed in
Table 4; they are essentially associated with the density after dry-
ing, the maximum stress, and stress at 5% strain, both apparent and
elastic modulus and also thermal conductivity. Only results at 30
and 180 days are presented and analysed, since all nine selected
shiv particles were tested at those ages. It must be remembered
that, in order to reduce the number of samples, at 60 and 90 days,
not all shiv particles are tested as explained in 2.2.
The above results 3.3.2 enable us to evaluate the quality of the
hempcrete by examining one characteristic. For better understand-
ing, the multicriterion classification takes into account several
characteristics. From this analysis, it will be possible to compare
the classification of types of shiv seen earlier and the classification
of the hempcrete.
To recap, the characteristics of thirteen different types of shiv
were studied and recorded, and those shiv types were the subject
of a classification study. Subsequently, nine types were selected
and used to make the test pieces. During the selection, samples
from all three groups were included so that each was represented.
The types of shiv selected are: group 1 (C10, C12 and C13); group 2
(C2, C3, C4 and C11); and group 3 (C5 and C6).
The results of the study show the sensitivity of hempcrete prop-
erties to the type of shiv used. Some are less impacted by the type
 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222
Fig. 13. Maximum compression stress for different types of hemp.
Fig. 14. Apparent modulus for different types of hemp.
Fig. 15. Elastic modulus for different types of hemp.
of shiv, while others may be significantly altered. On the same
comparison scale, the results in Fig. 16 show dispersions of the
order of 1 to 10 – particularly in regard to mechanical perfor-
mances. In addition, dispersions of the order of 8–10 are observed
for dry density and heat conductivity. For these two properties, the
values obtained for the different types of hempcrete are very sim-
ilar. This confirms our earlier choice to base the classification on
mechanical performance, as it is this performance which is most
heavily impacted by shiv type. Beyond the graphical representa-
tion (Fig. 16), it is important to understand the connections
between the observed differences and the used shiv type. By a sep-
arate representation, we are better able to distinguish the hemp-
cretes occupying the centre, the middle and the boundary of the
radar chart, corresponding to the three groups which are detailed
below.
219
3.4.1. Group 1 hempcrete
The hempcretes in this group are those made with shiv types
classified as group 1, Fig. 17, which is characterised by small parti-
cle size with a low and medium water absorption capacity. Accord-
ing to the assumptions and given the low values of such properties,
which are considered to be detriment to mechanical performance,
it was concluded that these shiv types would produce concretes
with good mechanical performance. These hempcretes offer poor
mechanical performance. Indeed, while the water absorption
capacity is considered detrimental, this is not necessarily so in
reality. The beneficial or detrimental nature probably depends on
the type of binder used. In a study carried out by Nguyen [21] on
four types of binder, those containing more pozzolan offered better
mechanical strength, and cement-based hydraulic binders yielded
poor resistances.
Furthermore, the study by Arnaud and Gourlay [15] predicted
that a reduction in particle size should improve the mechanical
performance of hempcrete. However, for our results, this is not
the case. Moreover, this finding was evidenced in the work of Ste-
vulova et al. [10], the reduction of particle size leading to an
improvement provided that the types of shiv are of the same ori-
gin. It should be emphasised that using the Tradical PF70 binder,
Nguyen [21] also found an improvement in mechanical perfor-
mance as particle size increases.
Regarding the particle size, our results demonstrate that the
higher the specific surface area is, the poorer is the mechanical per-
formance. This finding is partly in agreement with the results
found by Arnaud and Gourlay [15], since, the more the surface area
is, the more the binder–aggregate interface increases. However, it
has been demonstrated by Nozahic [28] that a naturally bad bin-
der–aggregate interface contributes to poor mechanical perfor-
mance of the hempcrete. It is clear that the beneficial or
detrimental nature of the particle size and the water absorption
capacity depends on the type of binder used.
3.4.2. Group 2 hempcrete
The hempcretes in this group, Fig. 18, are characterised by an
improvement in mechanical performance with respect to those
in group 1. The difference observed for the mechanical perfor-
mance of this group gives rise to two subgroups; subgroup A con-
taining C2 and C3 and subgroup B containing C4 and C11.
Subgroup B seems to belong to or be at the boundary of group 1.
These subgroups result from the characteristics of the shiv types,
are explained below.
Firstly, it should be noted that C11 hempcrete can pass from
group 1 to group 2; in addition, the C12 shiv classified in group 1
can also pass to group 2. Indeed, when classifying these shiv types,
their positions on the radar chart could allow them to be classified
in either group 1 or 2. The decision was made by analysing the geo-
metrical characteristics. C11 is characterised by a higher geometry
than C12; hence its classification in group 2.
Secondly, in the same subgroup, classification of C4 shiv places
it at the boundary between groups 1 and 2. Owing to its slightly
larger particle size and relatively lower water absorption capacity
with respect to the group 1 shiv types, it was classified in group
2. When the group 1 shiv types demonstrated mechanical perfor-
mance values of the range of 1–4 on the rating scale, an improve-
ment is observed in group 2, this time with values of the order of
4–7. This justifies their belonging to this group and corresponds
perfectly to the prediction.
3.4.3. Group 3 hempcrete
These hempcretes are derived from shiv types which are also
positioned on the boundary of the radar chart (group 3), Fig. 19.
According to our assumptions, these types of shiv were predicted
to offer poor performance. Nevertheless, a reservation was
220 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222

Table 4
Overview results for hemp shiv characterisation at 30 and 180 days.

Nomenclature qafter drying (kg/m3) rmax (MPa) r5% of e (MPa) Eapparent (MPa) Eelastic (MPa) Thermal
conductivity k
(W/m.K)
Age in days 30 180 30 180 30 180 30 180 30 180 30 180
P-C2-11 407 430 0.55 0.57 0.51 0.55 29.49 29.09 54.22 58.84 0.12 0.08
S-C3-11 390 410 0.61 0.66 0.60 0.61 33.49 30.99 56.81 55.98 0.11 0.09
P-C4-11 393 399 0.40 0.42 0.34 0.31 15.00 15.09 29.45 33.12 0.09 0.08
P-C5-11 426 452 0.84 0.87 0.79 0.82 46.77 40.42 87.41 88.97 0.12 0.09
S-C6-11 460 480 1.07 1.07 1.05 1.02 61.00 42.79 114.1 103.4 0.12 0.09
S-C10-11 478 493 0.18 0.25 0.16 0.18 4.89 7.68 15.59 20.94 0.10 0.09
S-C11-11 417 418 0.44 0.40 0.22 0.21 6.17 10.91 18.45 20.08 0.10 0.08
P-C12-11 431 438 0.44 0.42 0.40 0.36 16.39 14.12 37.66 35.22 0.09 0.08
S-C13-11 449 449 0.13 0.13 0.05 0.10 5.68 12.27 18.99 20.72 0.10 0.08
Minimum 390 399 0.13 0.13 0.05 0.10 4.89 7.68 15.59 20.08 0.09 0.08
Maximum 478 493 1.07 1.07 1.05 1.02 61.00 42.79 114.1 103.4 0.12 0.09
Mean 428 441 0.52 0.53 0.46 0.46 24.32 22.60 48.08 48.59 0.11 0.08
Standard deviation 30 31 0.30 0.30 0.32 0.31 19.89 13.38 34.03 30.76 0.01 0.01

Fig. 16. Comparison of the different hempcretes, taking account of various characteristics.

Fig. 17. Hempcretes in group 1.

 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222 221

Fig. 18. Hempcretes classified in group 2.

Fig. 19. Hempcretes classified in group 3.

expressed as to the type of shiv in this group due to the close cor-
relation between the results for their characteristics, which was
not the case for groups 1 and 2.
Understanding the reasons for the good correlation between
these characteristics can help us to better understand why these
types of shiv produce concretes with good mechanical perfor-
mance. One of the possible ways is to analyse the internal structure
of these shiv types on a microscopic scale. This will help better
understand their levels of porosity and elucidate the link between
particle size and water absorption capacity.
Compared to groups 1 and 2, these hempcretes offer perfor-
mances of the order of 8–10 on the same comparison scale.
In conclusion, it is clear that the size (length, width, etc.) of the
shiv particles alone is not sufficient to predict the performance of
the hempcrete. Other characteristics must be taken into account,
such as water content, and the type of shiv which optimises these
characteristics, offers concrete with good performance. However, it
is entirely possible that a change in the type of binder will also
change the observed trends. There are probably chemical interac-
tions between the binder and the shiv which affect the mechanical
performance.
4. Conclusions and prospects
In the first step of this study, a multicriteria analysis based on a
comparison scale was used to classify the types of shiv into three
groups. In the second step, three types of mechanical response
were demonstrated during the compression test on hempcrete.
They are characterised by low, medium and high levels of strain.
The large specific surface area of the shiv was identified as the
source of weak binder–aggregate bonding at the interface, leading
to the complete failure of the specimen. Furthermore, the high
water absorption capacity, which is the probable cause of the ‘false
set effect’ associated with considerable elongation of particles, was
identified as being behind the significant strain observed. It has
been pointed out that in this latter case, the specimen behave like
a stack of layers, which induces great strains. The type of shiv has
been identified as the source of the significant dispersions
observed in the results – particularly in terms of mechanical per-
formance. The results differ from one type of shiv to another
regardless of the property, but the most remarkable change is in
the modulus and the maximum compression stress, with factors
of 10 between the minimum and maximum values.
222 C. Niyigena et al. / Construction and Building Materials 160 (2018) 211–222
A multicriteria analysis made it possible to classify hempcrete
into three groups corresponding to those obtained in the multicri-
teria classification of the shiv type. Unlike the literature, the results
indicate that small-grained shiv types with high water absorption
capacities do not necessarily yield the best performance.
The characteristics taken into account in the multicriteria anal-
ysis of the type of shiv did not fully explain the variability observed
in the results of the hempcrete. Insofar as the nature of the shiv
type was emphasised as the source of the observed differences
[10], and taking account of the effect of the chemical compounds
in the type of shiv on the setting and hardening of the hempcrete
[34], we propose:
(1) to carry out a microscopic analysis to understand the inter-
nal structure (porosity of the shiv). This will provide a better
understanding of the relationship between the physical, geo-
metric and water absorption characteristics of the types of
shiv. In particular, it will provide an understanding of the
trade-off or close correlation observed for the group 3 shiv
types with the best mechanical performances;
(2) to use other types of binder to study the role of the binder–
shiv interaction on the observed properties;
(3) to analyse the chemical composition of the shiv and espe-
cially its interaction with the binder for a better understand-
ing of the binder–aggregate bond, which contributes
significantly to the mechanical strength of the material.
Acknowledgements
The authors would like to express their sincere thanks to the
regional authority of Auvergne-Rhône-Alpes for funding this
research.
References
[1] Pierre Bouloc, Le chanvre industriel: production et utilisations, France Agricole
Editions, 2006 (ISBN2855571308).
[2] Luca Zampori, Giovanni Dotelli, Valeria Vernelli, Life cycle assessment of hemp
cultivation and use of hemp-based thermal insulator materials in buildings,
Environ. Sci. Technol. 47 (13) (2013). ISBN 1520-5851.
[3] S. Elfordy, F. Lucas, F. Tancret, Y. Scudeller, L. Goudet, Mechanical and thermal
properties of lime and hemp concrete (‘‘hempcrete”) manufactured by a
projection process, Constr. Build. Mater. 22 (10) (2008) 2116–2123. Elsevier.
[4] Tai Thu Nguyen, Vincent Picandet, Patrick Carre, Thibaut Lecompte, Sofiane
Amziane, Christophe Baley, Effect of compaction on mechanical and thermal
properties of hemp concrete, Eur. J. Environ. Civil Eng. 14 (5) (2010) 545–560.
Taylor & Francis.
[5] Florence Collet, Julien Chamoin, Sylvie Pretot, Christophe Lanos, Comparison of
the hygric behaviour of three hemp concretes, Energy Build. 62 (2013) 294–
303. Elsevier.
[6] César Niyigena, Sofiane Amziane, Alaa Chateauneuf, Laurent Arnaud, Bessette
Laetitia, Florence Collet, Gilles Escadeillas, C. Lanos, Mike Lawrence et al.,
RRT3: Statistical analysis of hemp concrete mechanical properties variability,
in: First International Conference on Bio-based Building Materials, ICBBM vol.
2015, 2015.
[7] César Niyigena, Sofiane Amziane, Alaa Chateauneuf, Laurent Arnaud, Laetitia
Bessette, Florence Collet, Christophe Lanos, Gilles Escadeillas, Mike Lawrence,
et al., Variability of the mechanical properties of hemp concrete, Mater.Today
Commun. 7 (2016) 122–133.
[8] César Niyigena, Sofiane Amziane, Alaa Chateauneuf, Etude de la variabilité des
caractéristiques de granulats de chanvre. Rencontres Universitaires de Génie
Civil, 2015.
[9] Youen Diquélou, Etienne Gourlay, Laurent Arnaud, Bernard Kurek, Influence of
binder characteristics on the setting and hardening of hemp lightweight
concrete, Constr. Build. Mater. 112 (2016) 506–517. Elsevier.
[10] Nadezda Stevulova, Lucia Kidalova, Julia Cigasova, Jozef Junak, Alena Sicakova,
Eva Terpakova, Lightweight composites containing hemp hurds, Procedia Eng.
65 (2013) 69–74. Elsevier.
[11] Y. Diquélou, E. Gourlay, L. Arnaud, B. Kurek, The impact of hemp shiv on
cement setting and hardening: The influence of extracts and study of the
interface, in: IIBCC2012 Int. Inorganic-Bonded Fiber Composites Conference
Proceedings, 2012, pp. 41–50.
[12] Garci Juenger, C. Maria, Hamlin M. Jennings, New insights into the effects of
sugar on the hydration and microstructure of cement pastes, Cem. Concr. Res.
32 (3) (2002) 393–399. Elsevier.
[13] K. Luke, G. Luke, Effect of sucrose on retardation of Portland cement, Adv. Cem.
Res. 12 (1) (2000) 9–18. Thomas Telford.
[14] Youen Diquelou, Interactions entre les granulats de chanvre et les liants à base
de ciment et de chaux, Mécanismes de la prise et propriétés des interfaces
formées dans les agrobétons, Reims (2013).
[15] L. Arnaud, E. Gourlay, Experimental study of parameters influencing
mechanical properties of hemp concretes, Constr. Build. Mater. 28 (1) (2012)
50–56. Elsevier.
[16] V. Nozahic, S. Amziane, G. Torrent, K. Saïdi, H. De Baynast, Design of green
concrete made of plant-derived aggregates and a pumice–lime binder, Cem.
Concr. Compos. 34 (2) (2012) 231–241. Elsevier.
[17] Tai-Thu Nguyen, Vincent Picandet, Sofiane Amziane, Christophe Baley,
Influence of compactness and hemp hurd characteristics on the mechanical
properties of lime and hemp concrete, Eur. J. Environ. Civil Eng. 13 (9) (2009)
1039–1050. Taylor & Francis.
[18] Nadezda Stevulova, Julia Cigasova, Adriana Estokova, Eva Terpakova, Anton
Geffert, Frantisek Kacik, Eva Singovszka, Marian Holub, Properties
characterization of chemically modified hemp hurds, Materials 7 (12)
(2014). ISBN 4215560241.
[19] A. Arizzi, G. Cultrone, M. Brümmer, H.A. Viles, Chemical, morphological and
mineralogical study on the interaction between hemp hurds and aerial and
natural hydraulic lime particles: implications for mortar manufacturing, Cons.
Build. Mater. 75 (2015). ISBN 0950-0618.
[20] M. Fourmentin, P. Faure, P. Pelupessy, V. Sarou-Kanian, U. Peter, D. Lesueur, S.
Rodts, D. Daviller, P. Coussot, NMR and MRI observation of water absorption/
uptake in hemp shives used for hemp concrete, Cons. Build. Mater. 124 (2016)
405–413. Elsevier.
[21] Tai Thu Nguyen, Contribution à l’étude de la formulation et du procédé de
fabrication d’éléments de construction en béton de chanvre, Université de
Bretagne Sud, 2010.
[22] A. Almusawi, R. Lachat, K.E. Atcholi, S. Gomes, Proposal of manufacturing and
characterization test of binderless hemp shive composite, Int. Biodeterior.
Biodegrad. 115 (2016). Elsevier.
[23] Julien Chamoin, Optimisation des propriétés (physiques, mécaniques et
hydriques) de bétons de chanvre par la maîtrisede la formulation, INSA de
Rennes, 2013.
[24] Paulien Brigitte de Bruijn, Knut-Håkan Jeppsson, Kenneth Sandin, Christer
Nilsson, Mechanical properties of lime–hemp concrete containing shives and
fibres, Biosyst. Eng. 103 (4) (2009) 474–479. Elsevier.
[25] Rahime Bag, Johnny Beaugrand, Patrice Dole, Bernard Kurek, Viscoelastic
properties of woody hemp core, Holzforschung 65 (2) (2011) 239–247.
[26] Ketty Bilba, Marie-Ange Arsene, A. Ouensanga, Sugar cane bagasse fibre
reinforced cement composites. Part I. Influence of the botanical components of
bagasse on the setting of bagasse/cement composite, Cem. Concr. Compos. 25
(1) (2003) 91–96. Elsevier.
[27] Sofiane Amziane, Florence collet, Mike Lawrence, DR. Camille, Magniont,
Vincent Picandet, Laurent Arnaud, Pete Walker, Etienne Gourlay, Sandrine
Marceau, et al., Round Robin Test for hemp shiv characterization: Committee
report of TC – 236 Bio-Based Building Materials part II: bulk density and
particle size distribution, Mater. Struct. (2014).
[28] Vincent Nozahic, Vers une nouvelle démarche de conception des bétons de
végétaux lignocellulosiques basée sur la compréhension et l’amélioration de
l’interface liant/végétal : application à des granulats de chenevotte et de tige
de tournesol associés à un liant ponce/chaux, Université Blaise Pascal-
Clermont-Ferrand II (2012).
[29] Sofiane Amziane, Florence Collet, Bio-aggregates Based Building Materials:
State-of-the-Art Report of the RILEM Technical Committee 236-BBM, Springer,
2017. ISBN 9402410317.
[30] Tiago Ferreira, Wayne Rasband, The Image J User Guide, National Institutes of
Health, USA, 2011.
[31] J.J. Healy, J.J. De Groot, J1 Kestin, The theory of the transient hot-wire method
for measuring thermal conductivity, Physica B+C 82 (2) (1976) 392–408.
Elsevier.
[32] Etienne Gourlay, De la chènevotte au béton de chanvre, optimisation des
propriétés mécaniques et hygrothermiques, Vaulx-en-Velin, 2009.
[33] Paulien de Bruijn, Hemp concretes: mechanical properties using both shives
and fibres: Department of Rural Buildings, SLU/Sveriges Lantbruksuniversitet i
Alnarp, 2008 (ISBN 9185911879).
[34] Etienne Gourlay, Caractérisation expérimentale des propriétés mécaniques et
hygrothermiques du béton de chanvre, Détermination de l’impact des
matières premières et de la méthode de mise en oeuvre. Lyon, 2014.