Construction and Building Materials 67 (2014) 321–337

Contents lists available at ScienceDirect

Construction and Building Materials

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

Assessment of the sustainability potential of concrete and concrete

structures considering their environmental impact,
performance and lifetime
Harald S. Müller, Michael Haist ⇑, Michael Vogel
Institute of Concrete Structures and Building Materials (IMB) and Materials Testing and Research Institute (MPA Karlsruhe), Karlsruhe Institute of Technology (KIT), Germany

h i g h l i g h t s

 Sustainability potential influenced by environmental impact, service lifetime and performance.

 Improved environmental impact with reduced cement content.
 Cement usage efficiency increases with concrete strength.
 UHPC provides high sustainability potential.

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

Article history: Sustainable structural engineering is based on the basic principle that the energy and resources con-
Available online 26 February 2014 sumption due to the construction and operation of a structure must be minimized. Relating to concrete
structures this principle can be realized by the use of the material in the most efficient way considering
Keywords: its strength and durability within the service life of the structure. On this background this paper outlines
Sustainability the three basic possibilities to increase the sustainability of a structure, i.e. methods to assess and reduce
Durability the environmental impact of concrete, means to increase the performance of concrete and design con-
Service life design
cepts which increase the possible lifetime of a concrete structure. The presented concept is applied to
Green concrete
Ultra high performance concrete
the concretes with the greatest potential in sustainability, i.e. green concretes as well as ultra high per-
formance concretes. Thereby, the basic principles of mix design of green concrete are introduced and a
systematic study of the influence of the cement content of these concretes on the fresh and hardened
concrete properties is presented. With regard to UHPC, the sustainability potential of this concrete type
is analyzed within a literature review. From the results it can be seen, that green concretes possess a very
high sustainability as long as they do not underlie any durability requirements. The sustainability of
green concrete subjected to corrosive exposures such as carbonation or frost, however, still has to be pro-
ven. UHPC, on the other hand, does not underly this constraint, as it provides superior mechanical and
durability performance compared to the environmental impact resulting from concrete production.
Ó 2014 Published by Elsevier Ltd.

1. Introduction cement – is highly energy intensive and prone to large emissions

More than any other industry, the building industry is affected
by the ongoing sustainability debate. Primarily, this is due to the
pronounced environmental impact going along with the produc-
tion of building materials, the erection of buildings and structures
and the subsequent use of these structures [1,2]. This is especially
true for structures made out of concrete, as the production of this
material – and here especially the production of its raw material
⇑ Corresponding author. Tel.: +49 721 608 44884; fax: +49 721 608 48400.
E-mail address: Michael.Haist@kit.edu (M. Haist).
of CO2 [3]. Reducing the environmental impact during concrete
production without looking at the impact on the performance
and durability of the material however would fall short, as the re-
quired lifetime of concrete structures normally ranges between 50
and 100 years and thus normalizes the environmental impact over
a long time span. Increasing the sustainability of building struc-
tures therefore requires both a reduction of the environmental im-
pact going along with the erection, maintenance and operation
processes as well as an increase of the durability of the structure
at maximum technical performance. This relation is detailed in
Eq. (1).

http://dx.doi.org/10.1016/j.conbuildmat.2014.01.039
0950-0618/Ó 2014 Published by Elsevier Ltd.
322 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
Building material sustainability potential
Lifetime performance
 ð1Þ
Environmental impact
Even though the definition given above differs from standard
definitions of the term sustainability, it is well in line with the la-
ter, as it addresses the three basic pillars of sustainability – i.e.
environmental aspects (by introducing the environmental impact),
social and economical aspects (hidden in the lifetime and the per-
formance parameters). As social and economical aspects however
are extremely difficult or even impossible to evaluate during the
concrete development process (i.e. the mix design), the definition
given in Eq. (1) provides the engineer with a simple way to quan-
tify the advantages and disadvantages of a certain concrete type
with regards to its potential as a sustainable material. Whether this
potential is used later on in the design and construction process is
up to the designer of the building or structure.
According to Eq. (1), there are three basic approaches to achieve
a sustainable use of concrete: (1) the optimization of the composi-
tion of the concrete in view of its environmental impact while
maintaining an equal or better performance and lifetime; (2) the
improvement of the concretes performance at equal environmen-
tal impact and lifetime; (3) the optimization of the lifetime of
the building material and the building structures at equal environ-
mental impact and performance. This paper deals with the combi-
nation of the aforementioned approaches (1)–(3) and primarily
outlines the procedures applied during this optimization process.
In the following this combination of both life cycle assessment
techniques (often called eco balance) and service life design tech-
niques is defined as assessment of sustainability potential.
In a first step, the principal methods for evaluating the environ-
mental impact of concrete production is shortly summarized in
this paper (see Section 2). Building up on that, in Section 3, the
mix design procedure of concretes with extremely low cement
contents as well as of concretes with ultra high performance is pre-
sented. Both types of concretes possess a very high potential in
increasing the sustainability of concrete structures. From Eq. (1),
however, it is clear, that all changes in environmental impact and
performance must be evaluated regarding possible changes in
the lifetime of the concrete or the concrete structure. Therefore,
a service life design must be carried out in order to optimize the
concrete composition and execution (e.g. the concrete cover) in
view of the designed lifetime of the structure. Within this design
process, both the performance of the concrete as well as the vari-
ous loadings acting simultaneously on the concrete must be con-
sidered (see Section 4). As a matter of principle, all efforts will
however only guarantee a maximum in sustainability unless both
the lifetime and the performance are fully utilized by the user of
the structure. As the actual lifetime by definition cannot be calcu-
lated, the service life of the material, component or structure can
only be considered as auxiliary parameter in this calculation.
2. Evaluation of the environmental impact of concrete using the
eco-balance method
The starting point in optimizing the sustainability potential of
building materials and structures lies in determining the environ-
mental impact in a standardized manner within a so-called eco-
balance as described e.g. in the European standards EN ISO
14040 and EN ISO 14044 [4,5]. The impact of every substance emit-
ted into the environment hereby can be attributed to one out of 7
impact categories, which have been internationally agreed upon
(see Table 1).
The environmental impact resulting from the production of
1 m3 of concrete can be calculated by multiplying the impact
Table 1
Impact categories classifying the environmental impact of industrial processes [4,5].
Category Designation Unit
CED-fossil Cumulative Energy Demand (non- Joule
renewable)
CED- Cumulative Energy Demand (renewable) Joule
renewable
GWP Global Warming Potential kg CO2-
equivalent
ODP Ozone Depletion Potential kg R11-
equivalent
AP Acidification Potential kg SO2-
equivalent
NP Nutrification Potential kg PO4-
equivalent
POCP Photochemical Ozone Creation Potential kg C2H4-
equivalent
resulting from the production of each raw material with the
amount of raw material used in the concrete and by summing up
the individual impacts. Table 2 gives an overview on the environ-
mental impact of typical concrete raw materials. In many countries
exact data on specific raw materials can be obtained from the raw
materials producer using so-called environmental product declara-
tions (EPD, [6]).
As can be seen from Table 2, cement has a very decisive influ-
ence on the environmental impact of concrete, taking into account
the large amounts of cement needed in the production of a stan-
dard concrete. The replacement of cement by reactive additives
such as fly ash or finely grounded slag sand however is only bene-
ficial to the environment, as long as these materials are treated as
industrial wastes and are therefore not attributed to the eco-bal-
ance of the concrete (see Table 2). Respect must also be given to
the sometimes limited availability of these materials [11,12],
which has led to a worldwide search for new hydraulic binders
with reduced environmental impact but unlimited availability.
Examples for this development are Celitement [13] and calcined
clays [14].
3. Mix development of concretes with low environmental
impact and high performance
3.1. Concrete with minimum environmental impact – green concrete
Besides the search for binders with reduced environmental im-
pact a very promising approach in designing concretes with mini-
mum environmental impact lies in reducing the amount of binder
in the production of concrete. Great progress with regard to this
approach could be achieved in Denmark in the last decade. Glavind
et al. [15,16] for example presented concretes with mean compres-
sive strengths around 30 MPa in which cement was systematically
replaced by reactive additives such as fly ash or slag sand as well as
stone powders. Proske and Graubner et al. [17] report about the
development of low strength concretes, where Portland cement
was successfully replaced by limestone powder, fly ash and com-
posite cements and by which the environmental impact of the con-
crete was significantly lowered while maintaining sufficient
strength and durability. These findings were confirmed by own re-
sults of the authors presented in [18]. A systematic investigation
on the mix design principles of ecological concretes – i.e. concretes
with a strongly reduced cement clinker content – was presented by
Fennis [19]. By using packing algorithms such as the Compressible
Interaction Packing Model (CIPM) Fennis was able to significantly
reduce the amount of cement clinker needed for the production
of concrete. With regard not only to the building material but also
to the building structure fib bulletins 28 and 67 [20,76] give a good
overview on the state of the art in sustainable building design.
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337 323

Table 2
Life cycle assessment data of important concrete raw materials according to [7–10].

CED GWP ODP AP EP POCP Source

Non-renewable Renewable
(MJ/kg) (MJ/kg) (kg CO2/kg) (kg R11/kg) (kg SO2/kg) (kg PO4/kg) (kg C2H4/kg)
Cement
Portland cement CEM I 32,5 5.650 8.74  102 0.951 1.64  108 5.31  104 3.30  105 2.20  106 [7]
Portland cement CEM I 52.5 5.800 9.71  102 0.476 1.79  108 5.74  104 3.50  105 2.36  105 [7]
Cement (in general) 2.451 6.58  102 0.691 1.50  108 8.30  104 1.2  104 1.0  104 [8]
Fly ash
Treated as industrial waste 0 0 0 0 0 0 0 [–]
Treated acc. to mass fraction generated 49.70 4.180 4.06  108 3.2  102 1.76  103 1.1  103 [9]
Treated acc. to added value generated 4.84 0.350 8.45  109 2.67  103 1.52  104 9.34  105 [9]
Slag sand
Treated as industrial waste 0 0 0 0 0 0 0 [–]
Treated acc. to mass fraction generated 22.20 1.390 2.72  108 5.39  103 7.52  104 9.32  104 [9]
Treated acc. to added value generated 3.54 0.149 6.76  109 8.59  104 8.18  105 0.10 [9]
(inert) Fillers and aggregates
Limestone powder 0.350 2.10  102 1.72  102 5.72  109 1.24  104 9.22  106 8.71  106 [7]
Limestone sand 0.114 4.64  103 6.11  103 3.19  109 5.31  105 6.05  106 6.10  105
Quartz powder 0–0.22 mm 0.820 3.16  102 2.34  102 4.98  109 1.58  104 6.75  106 5.57  106
Quartz sand 0.539 1.29  102 1.02  102 2.10  109 7.54  105 3.00  106 2.58  106
Sand 0.022 1.49  103 1.06  103 2.30  1010 6.57  106 2.99  107 2.39  107
Crushed sand 0.113 2.21  103 7.02  103 6.31  109 8.35  105 1.24  105 1.34  105
River gravel 0.022 1.49  103 1.06  103 2.30  1010 6.57  106 2.99  107 2.39  107
Recycled concrete aggregates 0/16 0.084 2.00  104 6.00  103 n.a. 5.70  105 9.00  106 8.00  106
Admixtures
Superplasticizer PCE 27.95 1.20 0.944 3.29  108 1.19  102 5.97  103 5.85  104 [10]
Plasticizer 14.30 1.70 0.739 n.a. 9.04  103 4.55  104 9.91  104 [7]

Fig. 1. Schematic sequence of the mixture development procedure for green concretes.

3.1.1. Mix design of green concrete
The key challenge in the development of concretes with mini-
mum environmental impact lies in maintaining a sufficient work-
ability of the concrete at very low water contents. As in the fresh
state water is needed to fill the inter-granular voids in the mix con-
sisting of aggregates and cement particles and to lubricate the
deformation of this granular system, a reduction of the water con-
tent at constant packing density would inevitably result in a loss of
workability and in the formation of voids. Before this background,
methods for optimizing the packing density of the granular mix
constituents form a key step in the mix development process.
The mix development algorithm applied in this paper is detailed
in Fig. 1 and refers to several diagrams and equations given in this
paper (the order of appearance thus is not continuous). The pro-
posed scheme hereby builds up on the principles by Fennis [19].
In the first step, the desired concrete properties and the ce-
ment content have to be defined. Further the properties of the
raw materials have to be experimentally determined. Based on
the desired cement content and the strength requirements, the
w/c-ratio can be assessed using Fig. 4 and the minimum packing
density /req required to cope with this reduced water content can
be calculated using Eq. (3). In a next step, the grain size distribu-
tion of all granular constituents has to be optimized using a com-
bination of the grain size models by Funk and Dinger [21] and the
CIPM-Modell proposed by Fennis [19], so that the packing density
of the mix in the compacted state /comp exceeds the required
packing density /req (for details see [22]). Based on the achieved
packing density the fresh concrete properties of the mix can be
evaluated using Fig. 3 and the compressive strength can be pre-
dicted using Fig. 4.
324 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
3.1.1.1. Optimization of grain size distribution. The packing density
of granular systems is strongly interlinked with the grain size dis-
tribution of the particles. Therefore, by choosing the fractions of
the available raw materials in a manner as to ideally fit an ideal
grain size distribution for example as proposed by Funk and Dinger
(see Eq. (2); [21]) or Andreasen [23] is supposed to yield optimally
packed mixes.
n n
d  dmin
AðdÞ ¼ n n
dmax  dmin
In Eq. (2) A(d) designates the passing volume fraction of the par-
ticles as a function of the grain diameter d. The parameters dmin
and dmax denote the minimum and maximum grain diameter. In
order to achieve a maximum packing density for cement based sys-
tems the fit parameter n is recommended to be chosen between 0.3
and 0.4 [21].
Even though Eq. (2) provides a very simple way of optimizing
the packing density of a granular mix, this method is afflicted with
a major deficiency, i.e. it does not allow for a calculation of the ac-
tual value of the packing density. Further, the outcome of this opti-
mization process is strongly influenced by the fit parameter n, for
which only empirical data is available. Within the research pre-
sented in this paper, the Funk and Dinger Model was therefore
combined with the Compressible Interaction Packing Model (CIPM)
proposed by Fennis [19], which allows for the calculation of the
absolute packing density of a given particle mix.
3.1.1.2. Calculation of the packing density with the CIPM model. The
Compressible Interaction Packing Model was proposed by Fennis
[19] and is based on the Compressible Packing Model first pre-
sented by de Larrard [24]. In contrast to other packing models it
recognizes the fact, that the packing density of a granular system
is not only a function of the grain size distribution of the particles
but also of the amount of energy used to compact the material.
Further the packing density is significantly influenced by the grain
shape. Instead of incorporating the later influence mathematically
into the model – which was done by various authors before and is
highly prone to errors – both de Larrard as well as Fennis consider
this influence by including the experimentally determined packing
density of each raw material into the model. The calculation of the
packing density in the CIPM model relies on the assumption, that
one grain fraction assumes a dominant role and that the other frac-
tions are used to fill the voids between the grains of the dominant
fraction. By limiting the span been the minimum and maximum
grain size dmin/dmax to values between 0.5 and 0.9 the dominating
fraction however is unknown, thus requiring an iterative calcula-
tion of the packing density of one specific fraction when interacting
with all other fractions. As all of these interactions contribute to
the packing density of the final mix, the later can thus be deter-
mined by summation of the packing densities of all fractions while
paying respect to the different compactabilities of each fraction. A
detailed description of the CIPM model can be found in [19,24].
As the CIPM model is mathematically highly complex, its appli-
cation requires a computerized approach. In order to find mix com-
positions with maximum packing density it is thus suitable to
optimize the grain size distribution of the investigated mixes using
the model by Funk and Dinger (see Section 3.1.1.1) prior to the
application of the CIPM model. Fig. 2 shows the influence of the
fit parameter n (Funk and Dinger model; Eq. (2)) on the packing
density of concrete mixes with dmax = 16 mm, dmin = 0.0001 mm
and cement contents between 100 and 300 kg/m3 as calculated
by the CIPM model.
As can be seen, an optimal packing density of the mixes can be
achieved, when the parameter n in Eq. (2) is set to values between
0.37 and 0.40. This is in good agreement with the recommenda-
tions by Funk and Dinger [21] and Andreasen [23]. The pronounced
scattering in Fig. 2 (left) however also shows, that the packing den-
sity can be strongly improved beyond the possibilities of these
simple models when applying the CIPM Model and actually calcu-
lating the real packing density while varying the fines composition.
Further, the maximum achievable packing density is influenced
by the cement content in the dry particle mix (see Fig. 2, right).
Hereby the highest packing densities of approx. 0.92 were obtained
for mixes with the lowest cement content of 4 vol.%. This can be
ð2Þ
explained by the fact that ordinary Portland cements with a high
Blaine value – such as used in this study – strongly disturb the
grain size distribution of the concrete with regard to an optimal
packing density. Reducing the cement content however mitigates
this effect.
3.1.2. Influence of mix design on green concrete properties
Based on a total of 1500 mix designs calculated with the CIPM
model, 15 mixes with maximum packing density were chosen
and tested for their fresh and hardened concrete properties in
the lab. The composition of selected mixes is detailed in Table 3
and is characterized by cement contents between 4 and 10 vol.%
of the dry mix, corresponding to approx. 110–265 kg/m3 in the
fresh concrete. As for the cement a Portland cement CEM I 52.5 R
and a microcement with strongly reduced particle size were cho-
sen, respectively. As fillers, quartz powders were used to compen-
sate for the reduced cement content. Details on the properties of
the raw materials and on the mixing regime applied in this study
can be found elsewhere (see [22]).
As can be seen from Fig. 3, the fresh concrete properties (i.e. the
degree of compactability) essentially are a function of the w/c-ratio
applied in the mix design. By increasing the w/c-ratio the degree of
compactability decreases yielding mixes with an improved work-
ability. Despite their nearly identical grain size distribution (see
Eq. (2), n = 0.37 and 0.34, respectively), mixes with reduced cement
content in general require a higher w/c-ratio to ensure sufficient
workability. This tendency seems especially pronounced for mixes
with cement contents below 6 vol.% of the dry mix, even though
ongoing tests still have to verify this finding.
Further, the workability of the mixes is closely interlinked with
the packing density ratio, i.e. the packing density of the mix / re-
lated to the maximum possible packing density /comp (see Fig. 3,
right). With reducing packing density ratio /=/req the lubricant
film between the particles increases leading to a significant in-
crease in workability (i.e. a reduction in compactability c). Hereby
the actual packing density / of the mix can be calculated from the
volumes of water Vw and particles Vp used in the mix according to
Eq. (3). The maximum possible packing density /comp must be cal-
culated using the CIPM model (see Section 3.1.1.2).
vp
/¼ ð3Þ
vw þ vp
With regard to the performance of the investigated concretes at
the hardened state Fig. 4 shows that despite their strongly reduced
cement content (compare Table 3), all investigated concretes show
an equal or significantly higher concrete strength than ordinary
Portland cement concretes at equal w/c-ratio especially at very
early ages (line referring to standard concrete 7 day strength as a
function of cement type and w/c-ratio; compare [28]). A pro-
nounced increase in compressive strength could be found for con-
cretes where ordinary Portland cement was replaced by a so-called
microcement, i.e. a Portland cement with an extremely high Blaine
value of approx. 6900 cm2/g compared to 5500 cm2/g for the CEM I
52.5 R used in the tests. Based on these findings, it can be con-
cluded, that a cement content of 4 vol.% on the dry mix obviously
generates a sufficient amount of cement gel to coat all inert raw
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337 325

Fig. 2. Particle packing density /comp as calculated using the CIPM Model for granular mixtures with Portland cement CEM I 52.5 R and various quartzite aggregates as a
function of the fit parameter n (see Eq. (2); left) and as a function of cement content of the dry granular mixture for varying fit parameters n (right).

Fig. 3. Fresh concrete compactability c (tested acc. to EN 12350-4 [26]) in relation to the w/c-ratio (left) and to the packing density ratio /=/comp (right) of mixtures
containing Portland cement CEM I 52.5 R (CEM I; Blaine value 5500 cm2/g) and microcement (lCEM; Portland cement; Blaine value 6900 cm2/g) with a cement concentration
of 4%, 6% and 10% by volume in the dry mix, respectively, and fit parameters n of the particle size distribution curve between 0.34 and 0.37; compactability index classes
according to [27].

Fig. 4. Left: compressive strength fcm,cube,7d of mixtures at an age of 7 days as a function of w/c-ratio and Walz-Curves for standard concrete ([28]; conversion to 150 mm cube
strength and 7 day strength according to fib Model Code 2010 [29]); right: binder intensity related to compressive strength fcm,cube,7d of mixtures with varying cement content
and particle size distribution (fit parameter n) compared to literature results by Fennis [19], Proske et al. [17] and Daminelli [30] (all literature values 28-days strength).

materials and thus generates enough bond-strength between the
aggregates and the cement paste to provide compressive strengths
up to 65 MPa at the age of 7 days.
As explained in Section 1, the reduction in cement content de-
scribed above only leads to an increase in the sustainability poten-
tial as long as both an equal performance and durability are
ensured. Unfortunately investigations regarding the durability of
the investigated concretes are still under way. Fig. 4 (right) shows
the binder intensity bi, i.e. the amount of cement needed for 1 m3
of concrete to generate 1 MPa of strength for the developed low-
cement concretes in comparison to standard concretes as reported
on by Daminelli [30] and other green concretes proposed by Proske
326 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337

Table 3
Composition and characteristics of tested concrete mixtures (selected mixes; w/c considering the specific water adsorption of the aggregates).

Raw material/characteristic value Dimension Concrete

Cement content in dry mix (Vol.-%) 10.0 6.0 4.0 4.0 4.0 4.0
Grain size distribution (fit parameter n acc. to Eq. (2)) (–) 0.37 0.37 0.37 0.34 0.37 0.34
Cement type (–) CEM I CEM I CEM I CEM I lCEM lCEM
Cement content (kg/m3) 264 165 113 109 110 111
Quartz powder 1 44 46 48 92 96 96
Quartz powder 2 22 70 143 138 72 145
Sand 0.1/1 (mm) 429 482 479 478 499 502
Sand 1/2 (mm) 357 372 383 414 384 435
River gravel 2/8 (mm) 535 557 575 460 576 483
River gravel 8/16 (mm) 491 511 527 483 528 507
Water 158 124 96 129 94 89
Superplasticizer (PCE based) 3.96 3.37 3.70 4.10 3.30 4.20
w/c-Value (–) 0.538 0.643 0.691 1.049 0.683 0.656
Compressive strength fcm,7d (EN 12390-3 [25]) (MPa) 52.8 41.9 41.4 20.1 63.2 66.3
Degree of compactability c (EN 12350-4 [26]) (–) n.d. 1.19 1.26 1.16 1.26 1.27

et al. [17] and Fennis [19]. The reduction in cement content hereby
leads to a pronounced reduction in binder intensity, hinting for a
strongly increased sustainability potential especially of mixes with
none or low durability requirements. Fig. 4 (right) however also
shows, that none of the available mix design approaches for green
concretes – including the one presented in this paper – succeed to
de-couple the binder intensity from the strength requirements.
With reducing strength requirement a pronounced increase in bin-
der intensity is observed for all concretes. This interdependency
can be explained by the fact, that for low strength requirements
the cement content required according to the w/c-ratio would
eventually go towards zero. However, in order to coat all inert par-
ticles with a sufficient amount of cement gel, a minimum cement
content is required, leading to the observed increase.
Calculating the Global Warming Potential (GWP) of the investi-
gated concretes using the environmental impact data provided in
Table 1 and referring this to the compressive strength – i.e. the per-
formance of the concrete – shows, that the normalized GWP of the
mixes significantly increases with increasing compressive strength
of the concrete. At equal compressive strength, however, the im-
pact on the GWP can be significantly reduced by reducing the Fig. 5. Global Warming Potential (GWP) related to the compressive strength
amount of cement used in the mix design (see Fig. 5). fcm,cube,7d of mixtures with varying cement content and particle size distribution (fit
parameter n).

3.2. Concretes with maximum performance and durability – ultra high

performance concrete (UHPC)
Ultra high strength concrete is usually defined as a concrete
having a characteristic compressive strength above the strength
grade C100/115. One may distinguish between two types:
 reactive powder concrete (RPC), characterized by a maxi-
mum aggregate size of approximately 0.5 mm and
 ultra high performance concrete (UHPC), with coarse
aggregates normally up to an aggregate size of 8 mm.
The water/cement ratios of such concretes are usually below
0.25 and highly reactive additions, such as silica fume, have to be
added to the mix. The workability of the concrete can only be en-
sured by applying large amounts of superplasticizer. Thus a com-
pressive strength of approximately 200 MPa may be achieved
[31]. Depending on the type of curing, e.g. with additional heat
treatment without or with mechanical pressure, this concrete
may reach a compressive strength of up to 800 MPa, provided a
proper selection of the aggregates is carried out [32].
No standard has been adopted yet for this type of concrete,
which significantly exceeds the strength range specified in EN
206-1 [27]. Nevertheless UHPC might become interesting for a
broader application in practice, as the very high strength allows
a significant reduction of the cross-section of members, while
retaining a high load bearing capacity. This is accompanied by a
significant mass reduction, allowing in principle much longer
spans compared to those realized when ordinary structural con-
crete is used. Mass reduction also decreases the loads on the foun-
dation, as well as inertia and seismic forces with respect to
earthquake loads. These savings in concrete consumption highly
contribute to an increased sustainability potential of UHPC struc-
tures compared to conventional structures, despite the signifi-
cantly higher cement content of UHPC (see below).
3.2.1. Mix design of UHPC
The mix design procedure of UHPC basically is identical to the
approach already presented for green concretes (see Section 3.1)
except for the fact that UHPC requires a significantly higher binder
content, having a negative effect on the environmental impact.
Commonly, binder contents between 600 kg/m3 and 1000 kg/m3
have been reported in the literature for this type of concrete
[31]. The grain size distribution of UHPC normally is chosen to cor-
respond to a distribution by Funk and Dinger [21] or Andreasen
[23] with the fit parameter n being between 0.35 and 0.40, i.e.
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
identical to green concretes (see Section 3.1). However, due to the
dosage of the cement and the fineness, the actual grain size distri-
bution of typical UHPC strongly deviates from the models men-
tioned above in the fine particle region of d < 0.1 mm [34,35,31].
Further, in order to increase the packing density and improve the
contact zone between the hardened cement paste and the aggre-
gates, large amounts of silica fume between 10% and 30% by mass
of cement are added to the concrete also disturbing the particle
size distribution in the low particle diameter range. Regarding
the cement type used in UHPC, Portland cements with a low C3A
content are recommendable, as these cements have a low water
demand [36]. This is also advantageous in view of the risk of a sec-
ondary ettringite formation in case of curing at high temperatures.
The strength classes of the cements normally applied is 42.5 or
52.5 according to European standard EN 197-1 [33]. The fineness
of the cement should be between 3000 and 4500 cm2/g. Blast fur-
nace slag cement of strength class 52.5 has also been successfully
used [37].
Aggregates that may be used in the production of UHPC include
the materials referred to as ‘‘aggregates for concrete’’ in EN 12620
[38] or similar standards. The particle strength should be suffi-
ciently high and the grain size distribution should guarantee a high
packing density. Quartz powder with particle sizes similar to ce-
ment is predominantly used for heat-treated UHPC. The quartz
particles are inert at room temperature but, together with Ca(OH)2,
they react at high temperatures to form CSH phases.
Fresh UHPC is not workable without the addition of large quan-
tities of high performance plasticizers (up to 5% by mass of ce-
ment). Only third generation plasticizers (polycarboxylate-ethers,
PCE) allow saving a sufficient amount of water while making UHPC
workable. With regard to the environmental impact of the mix, it
must be kept in mind, that superplasticizers have a pronounced
detrimal effect on the environmental impact of the concrete und
should therefore be limited in dosage.
The same is true for fibers and especially for steel fibers. As
UHPC shows an elastic and brittle behavior, fibers are added to
the concrete to compensate for this disadvantage in construction
practice. It has been proven that approximately 2.5% of steel fibers
by volume, at an aspect ratio l/d between 40 and 60, lead to best
results, both in view of fresh and hardened concrete properties
[32,34,35]. The fiber length should be adjusted to the maximum
aggregate diameter. With regard to the sustainability potential of
UHPC it should however be kept in mind, that the fiber content sig-
nificantly influences the environmental impact of the concrete and
should thus be minimized.
3.2.2. Properties of the fresh UHPC
UHPC is characterized by high powder contents, in excess of
1000 kg/m3 in combination with high dosages of superplasticizers.
Its properties at the fresh state are thus comparable to self-com-
pacting concrete of the powder type. For this reason, the workabil-
ity of this type of concrete should rather be measured with the test
set-up used for SCC, i.e. slump flow and V-funnel flow test accord-
ing to EN 12350-8 and EN 12350-9 [39,40]. The slump flow of
UHPC normally ranges from 650 to 800 mm and the V-funnel flow
time from 15 to 40 s (also see [31]). Superplasticizers strongly re-
duce the flow resistance of the concrete and lead to a normaliza-
tion of the dynamic viscosity over the whole range of shear rates
(Fig. 6).
As can be seen from Fig. 6, the dynamic viscosity of UHPC re-
mains nearly constant – even though at a very high level – no mat-
ter whether the concrete is at rest or it is agitated, for example by a
poker-vibrator. For UHPC, the de-airing process therefore cannot
be enhanced significantly by vibrating the concrete. At rest, how-
ever, for concretes with a very low dynamic viscosity, sedimenta-
tion may be observed, which must be prevented in UHPC by
327
optimizing the grain size distribution of the aggregates, thus
ensuring an interlocking of the particles. At high shear rates, UHPC
normally shows a dilatant material behavior, i.e. an increase of the
dynamic viscosity.
The air content in UHPC ranges from 1% to 5% by volume, with
the lower ratios observed in low viscosity mixes. In addition, there
is a trend towards an air volume increase that occurs in line with
the increasing length of the steel fibers used.
Plastic (capillary) shrinkage and autogenous deformation are
promoted by high cement paste content and a low water/cement
ratio. For this reason, the immediate curing of UHPC is basically
very important.
3.2.3. Properties of hardened UHPC
In the subsequent sections the mechanical, physical and chem-
ical properties of UHPC are outlined and also focused in view of the
consequences with regard to environmental considerations and
sustainability. This is in line with a more global look on the build-
ing material sustainability potential as indicated also by Eq. (1).
3.2.3.1. Strength characteristics. The compressive and the tensile
strength which may be achieved by UHPC strongly depend on
the concrete composition, in particular with regard to the type
and amount of binders and the fine aggregates (micro-fillers) as
well as the type and duration of curing. If ordinary curing at room
temperature of 20 °C is applied, a maximum compressive strength
of UHPC of approximately 200 MPa can be achieved [34,41]. If the
curing temperature is increased to 100 °C, the strength will reach
approximately 250 MPa [42]. A further increase of the curing tem-
perature up to 250 °C is accompanied by a strength gain up to al-
most 400 MPa. A compressive strength value of approximately
800 MPa may be achieved, if mechanical pressure is applied in
addition [32]. With regard to the sustainability potential these
examples show, that the performance of UHPC can be significantly
increased at slightly increased environmental impact (i.e. by heat
treatment).
Similarly to ordinary fiber reinforced concrete, the addition of
fibers (mostly steel fibers) causes a small improvement of the com-
pressive strength of UHPC, but may significantly affect its strain
capacity. The investigations on the effect of the volume content
of fibers of up to 6% by volume, point to an optimum fiber content
of approximately 2.5%. As fibers significantly deteriorate the envi-
ronmental impact of UHPC, the dosage should be closely focused at
the individual needs regarding the tensile strength and strain
capacity of each building element or structure.
It should be noted that, unlike the compression characteristics,
the tensile strength of UHPC may be strongly increased when fibers
(1.5–3.0% by volume) are added to the mix. This may be attributed
to a large extent to the reduction of the brittleness of non-rein-
forced UHPC, where minor flaws and cracks may tremendously re-
Fig. 6. Dynamic viscosity of fresh UHPC, conventional concrete (NC) and self-
compacting concrete (SCC), as a function of the externally applied shear rate.
328 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
duce the tensile strength. The uniaxial tensile strength and the
flexural tensile strength of fiber-reinforced UHPC reach values
ranging approximately from 10 to 60 MPa. When special binders
and techniques are applied, a tensile strength of 150 MPa may be
attained [43].
Concerning the fracture energy and the characteristic length of
UHPC, the available knowledge is still very insufficient. It appears
that for UHPC made without fibers the values are somewhat lower
than for high strength concrete. Of course, the addition of fibers
may increase the fracture energy up to a factor of approximately
100 [44].
3.2.3.2. Deformation behavior. The stress–strain behavior of non-
reinforced UHPC shows even under compression a linearly elastic
pattern almost up to the ultimate load level [31]. This must be con-
sidered as a detrimal effect with regard to the sustainability as the
lack of prior notice of failure of the structure requires a higher
safety margin in turn lowering the useable performance of the con-
crete. In a fiber reinforced UHPC, the fibers prevent macro-crack-
ing, because they hold the crack edges together already at the
micro-cracking stage and thus hinder the cracking progress. As a
result, a significant additional strain capacity is obtained while
the increase in compressive strength is minor. Similar to ordinary
fiber reinforced concrete, the fiber content determines whether
this strain results in strain softening or in strain consolidation.
The modulus of elasticity of UHPC reaches approximately
50–60 GPa when the compressive strength is 200 MPa [31]. High
values for the modulus of elasticity hereby contribute to the sus-
tainability potential, as they permit to reduce the building member
dimensions – and therefore the mass of the material used – with-
out significantly increasing the deformation of the structure.
The autogenous shrinkage of UHPC is very pronounced and may
reach values beyond 0.0012 (1.2 lm/mm) [45]. Often much lower
values are reported in the literature. However, this is mostly due to
the applied measurement technique, in connection with the age of
concrete when the measurement starts. When it starts e.g. at a
concrete age of one day, most of the autogenous shrinkage, which
develops mainly between an age of 8–24 h, has already taken place
and is consequently not recorded. As may be seen from Fig. 7 (left),
autogenous shrinkage observed after a concrete age of one day
reaches strain values of approximately 0.3 lm/mm for an UHPC
with a compressive strength of 170 MPa (C170).
Drying shrinkage of UHPC, obtained from the difference of total
shrinkage and autogenous shrinkage, is much lower, even when
compared to that of high strength concrete [45]. This may be
traced back to the extremely low porosity of UHPC. In studies on
autogenous and drying shrinkage of different UHPC starting at
the concrete age of one day, an autogenous shrinkage of 0.3 lm/
mm and a drying shrinkage of 0.1 lm/mm (at a maximum) have
been observed (see Fig. 9, at the top; [46]). The test results shown
in Fig. 7 have been obtained on concrete cylinders (with diameter
100 mm and height 300 mm) stored at an ambient temperature of
20 °C, where the specimens were either sealed (water loss pre-
vented) or remained unsealed in an ambient environment of 65%
relative humidity.
Fig. 7 (right) exemplarily shows the influence of different con-
crete ages at loading on the specific creep (creep per unit stress)
of an UHPC (C170). The stress level was 30% of the compressive
strength and the specimens were stored unsealed in a climate of
20 °C and 65% relative humidity. As already observed in experi-
ments on normal strength and high strength concretes, the magni-
tude of creep is significantly decreasing with increasing concrete
age. However, this effect is much more pronounced for UHPC than
for high strength concrete or normal strength concrete. If creep
coefficients are calculated from the results shown, for an age at
loading of t0 = 1 day or t0 = 3 days, final creep coefficients between
approximately u = 3 and u = 2, respectively, are obtained. Even for
an age at loading of t0 = 28 days the final creep coefficient reaches
approximately u = 1.5. This means that the pronounced decrease
of the creep capacity of concrete with the increase in strength of
the hardened cement paste is no longer true for UHPC. This is to
be seen as a significant drawback on the performance of the mate-
rial, as an increased creep requires greater amounts of steel rein-
forcement and eventually a reduction of the loading.
The application of high sustained loads showed no significant
difference in the sustained load strength of UHPC compared to nor-
mal strength concrete.
3.2.3.3. Physical characteristics. The physical characteristics of
UHPC are to be considered as the major advantage of this concrete
type, as the strongly reduced porosity of UHPC guarantees a pro-
nouncedly increased durability and superior performance.
Whereas conventional concrete (C35) and high strength concrete
(C100) show marked peaks in the pore size distribution between
0.01 and 0.1 lm, these peaks are nearly eliminated in the case of
UHPC (C200) and disappear completely in RPC (C500). Absolute
porosities then become so low, that transport processes almost
come to a halt [47,48]. As a consequence carbonation depths mea-
sured on UHPC are almost negligible and range between 0.1 and
0.2 mm after 3 years [49,50]. This value is equivalent to a CO2 dif-
fusion coefficient below approx. 0.07  109 m2/s [50]. Water va-
por diffusion was investigated in several types of concrete
applying the dry-cup method specified in EN ISO 12572 [51]. A va-
lue of 5  109 m2/s was documented for a C190 UHPC [52]. Chlo-
ride penetration was investigated in migration tests where a
diffusion coefficient of 0.02  1012 m2/s was measured [50],
which is about 100 times lower than the value for a C30 concrete.
Similar values were reported by Scheydt for heat-treated and non-
heat-treated UHPC [49]. Scheydt further emphasises the fact, that
the chloride migration is increased by the addition of steel fibers,
as the contact zone between the fibers and the matrix constitutes
a migration path with increased porosity.
The nitrogen permeability of UHPC decreases to a level that is
ten times lower than in a high performance C100 concrete and
100 times lower than in a C30 concrete [49]. The water permeabil-
ity of a C190 UHPC was found to be in a range between 4 and
5  1015 m2/s, which corresponds to the permeability of dense
natural stone. The same behavior was found for capillary water
absorption [47].
3.2.3.4. Durability. Structural concrete usually containing capillary
pores is often destroyed by mechanisms triggered by frost or
freeze–thaw impact. If no capillary pores are present, or if only a
minimal amount of such pores exists, only a small amount of water
can be absorbed and no saturation will occur [53]. Correspond-
ingly, the concrete shows a very good behavior when exposed to
frost or freeze–thaw cycles, as demonstrated in tests [47,49].
Tests in sodium sulfate solution demonstrated a very high
chemical resistance [50,49]. No conclusive findings have been
established yet with regard to the ASR issue. However, the tight-
ness of the system appears to be one of the crucial factors that
determine resistance.
Concerning the risk of the formation of secondary ettringite,
investigations in [47] showed that independent of the C3A content
of the cement, no adverse change of the microstructure of UHPC
could be observed.
Even when UHPC is exposed to a strong acid attack (pH = 1 and 3)
it shows a significant better performance than ordinary structural
concrete [49,54]. Of course also UHPC will be destroyed by a long
term attack of strong acids as the CSH component of the hydrated
cement paste is principally dissolved by acids as a consequence of
its chemical nature. However, due to the extreme low porosity of
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
Fig. 7. Shrinkage of UHPC under different storage condition (left), and creep of UHPC for different ages at loading (right) [46].
UHPC compared to ordinary structural concrete and even high
strength concrete, the rate of damage is much slower leading to a
significant higher durability or life time, respectively. UHPC there-
fore is highly suitable to improve the sustainability e.g. of sewage
pipes, etc.
The abrasion resistance of a C200 UHPC was measured by
underwater storage of the concrete. At regular intervals, the con-
crete was sand-blasted at an angle of 45° and a pressure of
250 kPa. This test simulates the transport of sediments in water
above a concrete base. A wear resistance was found that is about
four times higher than that of a C30 concrete, and roughly corre-
sponds to the wear resistance of added hard aggregate [55].
Initially, UHPC had not been fire-resistant since the tested
material bursted and the fibers in the fiber reinforced material oxi-
dised under heat exposure. Fire resistant UHPC could be success-
fully produced by adding 0.3–0.6 vol.% of polypropylene fibers
[56,57]. It is also helpful to replace quartz with basalt aggregate,
which prevents the occurrence of the deleterious quartz
conversion.
3.2.3.5. Environmental impact. For the evaluation of the environ-
mental impact and thus the sustainability, various impact catego-
ries are considered (see Section 1). When entering the values of the
material’s impact categories in a radar chart, a closed polygon can
be drawn for UHPC, see Fig. 8. The closer the polygon lines ap-
proach the point of origin, the more favorable is the environmental
impact of the considered material.
In a direct comparison, i.e. when a volume-based analysis is
made, the considered UHPC (C190) has a significantly more ad-
verse impact than the normal strength concrete NPC (C30), see
Fig. 8. This is caused by its higher cement and superplasticizer con-
tent and especially by the addition of steel fibers, compared to NPC.
When, in contrast, these environmental impact parameters are re-
lated to durability values, e.g. to the permeability, the UHPC dem-
onstrates a clear superiority (Fig. 9). Similar diagrams result if the
impact categories are related to other performance parameters,
such as the resistance against chemical attack, or also to the con-
crete strength.
4. Service life design as a key to sustainable buildings and
structures
As illustrated by Eq. (1), maximizing the lifetime of a building or
a structure is a very efficient way to improve the sustainability of
our built environment. Methods allowing to predict the service life
of a structure and to design the structure accordingly are therefore
essential tools in the sustainability assessment process for
329
sustainable buildings and structures. However, this aspect is often
neglected in the current life-cycle assessment debate, leading to a
single sided focus on a pure reduction of environmental impact
while neglecting the durability and thus the sustainability of the
designed structures.
The service life design process is dominated by assessing the
alteration – i.e. ageing and often deterioration – of the material
on the one hand and the varying environmental exposures on the
other hand. Moreover the various deterioration mechanisms usu-
ally occur combined and interact with each other. Hence, for a real-
istic service life prediction these highly varying influences and
interactions have to be modeled by means of probabilistic tools.
As a sustainable structure by definition has the least impact on
our environment and considering that impacts normally go along
with costs, such structures also help to minimize construction
and especially maintenance costs. The latter goal requires an effec-
tive inspection and assessment management in order to upgrade
repair strategies and trigger preventive repair work. In the follow-
ing, the procedure for the service life design for concrete structures
as well as a new approach to model combined deterioration mech-
anisms in a realistic way will be presented. The focus will be placed
on the influence of relevant deterioration mechanisms, such as
chloride and carbonation induced corrosion on the durability and
sustainability of concrete structures.
4.1. Service life design concepts
4.1.1. Descriptive concept
The existing procedure for durability design of civil engineering
structures made out of concrete is primarily based on empirical
evidence. The available national and international standards imply
special descriptive limits in connection with rough environmental
classifications to ensure a defined minimum service life of e.g.
50 years according to valid standards [27,58]. Key measures in this
design concept are limiting the w/c-ratio and ensuring a minimum
concrete cover to protect the reinforcement from damaging effects
resulting e.g. from carbonation, frost attack or chloride ingress. As
neither the real environmental impacts nor the actual deteriora-
tion mechanisms are incorporated in this concept, this design ap-
proach gives only a rough estimation of the durability. As a
consequence the difference between the action S and the resis-
tance R of the structure, being a measure of the failure safety, relies
on experience and cannot be quantified (see Fig. 10, top). The effec-
tive safety margin and thus the expected service life are unknown
to the designer. If the lifetime is assumed to be greater or smaller
than 50 years (e.g. 10 years or 200 years), this approach does not
allow the structural engineer to evaluate the sustainability of dif-
330 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
Fig. 8. Impact assessment results for 1 m3 of concrete when normal strength concrete (NPC) and ultra high performance concrete (UHPC) are compared.
Fig. 9. Relative impact assessment results for 1 m3 of normal strength concrete
(NPC), when NPC and UHPC are compared, referring to the individual coefficients of
water absorption w24, NPC = 0.5 kg/(m2 h0.5) and w24, UHPC = 0.05 kg/(m2 h0.5).
ferent competing designs, as essential differences in the expected
lifetime of the structure cannot be calculated.
4.1.2. Performance concept
In contrast to the descriptive concept, the performance concept
– which is based on a probabilistic approach – allows quantitative
estimations of the durability and thus of the expected service life of
concrete structures. Hereby, the increasing damage with time, i.e.
the development of the action S and the resistance R with time is
modeled by means of appropriate deterioration time laws
([59,60]; see Fig. 10, bottom). Since there are various uncertainties
to be coped with both on the action as well as on the resistance
side, it is necessary that the variability and the observable scatter
of both functions (e.g. for the material parameters) are described
by means of related statistical parameters. As a consequence, the
safety margin between the well defined functions for the action S
and the resistance R can be expressed in terms of the failure
probability.
With the use of the performance concept the time-dependent
increase of damage (see overlap area in Fig. 10, bottom) and the
failure probability with regard to a pre-defined unintended condi-
tion of the structure can be calculated (see e.g. [61–63]). Based on
such calculations also the intended service life of the structure can
be estimated and the environmental impact resulting from rehabil-
itation works can be quantified.
4.2. Basics and principles of service life design
It is obvious from the preceding remarks that the prediction
quality of the service life design is highly dependent on the quality
and the related variability of the time functions for both the actions
S and the resistance R. Further, statistical methods to quantify the
safety margin between the S and R functions have to be applied.
Whereas the later are already well developed and usually imple-
mented in commercial statistical software tools, deterioration-
time laws both on the building material level as well as on the
building member or structural level are missing in many cases.
One process which however is already well understood is the dete-
rioration due to carbonation and subsequent carbonation induced
corrosion. Hence, in the subsequent paragraphs the essential ele-
ments and design steps for the service life design of civil structures
are briefly summarized using this process as an example.
4.2.1. Deterioration processes
The increasing deterioration with time, i.e. the gradual loss of
durability due to environmental actions, has to be described by
means of deterioration-time laws. Such laws should preferably
take into consideration real physical or chemical mechanisms. This
holds true e.g. for the degradation process caused by carbonation.
The process of carbonation is described using the material law in
Eq. (4) (see [59]).
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffi
xc ðtÞ ¼ 2  ke  kc  ðkt  R1 ACC;0 þ et Þ  C S  t  WðtÞ ð4Þ
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
Fig. 10. Action S and resistance R in view of the durability of concrete members:
Scheme of the descriptive concept (top) and the performance concept including a
probabilistic approach (bottom).
In Eq. (4) xc (t) describes the carbonation depth at the time t in
[mm], the parameters ke, kc and kt take into account environmental
conditions, curing and test effects in [–], R1
ACC;0 is the inverse effec-
tive carbonation resistance of concrete and et is the corresponding
error term in [(mm2/years)/(kg/m3)], CS describes the CO2-concen-
tration in [kg/m3] and W(t) is the dimensionless weather function,
see [59].
Of course, carbonation itself does not damage the concrete or
the concrete cover, respectively, but if the carbonation front
reaches the reinforcement, depassivation takes place which initi-
ates corrosion of the reinforcement in the presence of moisture
and oxygen. Considering this process in terms of action S and resis-
tance R, the action is described by means of the material law for the
progress of the carbonation front in concrete taking into account
environmental and material parameters. The resistance is for
example given by the thickness of the concrete cover.
4.2.2. Model parameters
The parameters included in the models for the action S and the
resistance R are not exact values because they scatter around aver-
age values (see Fig. 10, bottom). This can be easily observed for the
carbonation depth (action S) as well as for the concrete cover
(resistance R) in a concrete member in practice. Hence, the varying
parameters are considered as random variables, also called basic
variables. The following parameters have to be measured in order
to carry out a reasonable service life design: mean value of carbon-
ation depth, corresponding coefficient of variation and type of the
distribution function.
4.2.3. Limit states
A so-called limit state defines a condition at which a structure
or a structural component ceases to comply with its intended ser-
viceability. In the case of carbonation induced corrosion of the
reinforcement a limit state may be defined by the condition that
331
the carbonation front reaches the reinforcement. Correspondingly,
for chloride induced corrosion a limit state is reached when the ac-
tual chloride content is equal to the critical chloride content in the
depth of the reinforcement. It is obvious that further limit states
may be defined, e.g. the initiation of cracks or any higher level of
chloride content.
4.2.4. Service life of structures
The loss of durability, i.e. the increase of the deterioration with
time, reduces the reliability and/or the safety of a civil structure. In
order to be able to evaluate this reliability or this safety at any age
of the structure, a reference period for the service life has to be
specified. Reference values of the service life of buildings and
structures are listed in relevant standards and guidelines. As an
example, the intended service life of residential buildings and
other simple engineering structures is 50 years, for complex engi-
neering structures it is 100 years [64]. Based on the given material
laws (see e.g. Eq. (4)) the point in time of expectable damages at
concrete structures can be determined and an appropriate mainte-
nance management can be established which leads to a significant
increasing of the intended service life.
4.2.5. Failure probability and limit state function
The failure probability pf is defined as the probability for
exceeding a limit state within a defined reference time period.
When this occurs, an unintentional condition (limit state) of a
building component is reached.
The magnitude of the failure probability is closely connected
with the safety margin between the resistance and the action func-
tions and varies with time (see Fig. 10, bottom). For a fixed point in
time the safety margin may be calculated by means of the so-called
limit state function Z (see Eq. (5)).
Z ¼RS ð5Þ
The function Z represents the elementary form of a limit state
function in which R and S are random variables. If the value of Z
turns to zero, the limit state is reached. The statistical properties
of the function Z can be expressed in the form of a distribution
function, if this function is considered to be normal distributed
and the resistance R as well as the action S are expressed including
related mean values l and standard deviations r.
By means of the introduction of the so-called reliability index b,
a direct correlation between this reliability index b and the failure
probability pf is obtained. In case of a normal distributed limit state
function Z, the failure probability pf can be directly determined
using Eq. (6).
pf ¼ pfZ < 0g ¼ UðbÞ ð6Þ
Above, the variable U() denotes the distribution function of the
standardized normal distribution (see [65]). The correlation be-
tween various values for the failure probability pf and the reliabil-
ity index b is shown in Table 4. Note e.g. that the often used 5%
quantile in civil engineering is equal to a failure probability of
5  102 which corresponds to a reliability index b = 1.645.
The definitions and derivations given above are generally valid,
i.e. for mechanical as well as for physical and chemical actions and
resistances which are related to durability. As the durability of con-
crete is pronouncedly dependent on time t, the functions for S, R
and subsequently of Z are also time-dependent. As a consequence
also the reliability index b is obtained as a function of time, where
the value of b = b(t) decreases with time as the durability decreases
and the failure probability increases.
The calculation of the failure probability pf for a building com-
ponent considering a particular mechanism related to durability
(e.g. carbonation induced corrosion of the reinforcement) may be
performed by the use of Eq. (7).
332 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
Table 4
Values for the failure probability pf and the related reliability index b [64].
pf 101 102 103 104 105 106
b 1.28 2.32 3.09 3.72 4.27 4.75
pf ¼ pfZ < 0g 6 pt arg et
As the failure probability increases with time, pf = pf(t) ap-
proaches ptarget = const. Finally, pf(t = tcrit) = ptarget is obtained,
where tcrit is the time when the failure probability of the member
becomes equal to the target failure probability. Using the calcu-
lated value for tcrit as a measure for the expected service life, finally
the sustainability potential of the structure can be calculated using
Eq. (1). In practical applications the determination of the service
life tcrit is done by means of the reliability index b as pf and ptarget
may be easily expressed as the reliability indices b and btarget.
4.2.6. Target reliability indices
The target values of the reliability index btarget depend on the
consequences of failure (loss of serviceability) and the relative cost
of safety measures. Table 5 indicates target values of the reliability
index b for building components in the serviceability limit state
(SLS), see [66,67].
Considering the case of depassivation of the reinforcement due
to carbonation or chloride ingress the target reliability index is rec-
ommended to be b = 1.3 [59].
4.3. Reliability analysis of components – importance related to
sustainability
Within the following example, a probabilistic service life design
is carried out on the inner shell of a tunnel (see Fig. 11). Hereby, the
main focus will be placed on the question, how the designed ser-
vice life changes in case strong deviations in the concrete cover oc-
cur due to an insufficient quality management system. Similar
calculations could have been also carried out e.g. for a bridge deck,
see Fig. 12.
With regard to the sustainability of the structure and the life cy-
cle analysis, it is necessary to implement a quality management
system already at the design stage of the structure. Quality man-
agement comprises the inspection of the planned and built struc-
ture with regard to the workmanship. This means that after the
construction phase the corresponding material properties or struc-
tural parameters – e.g. the concrete strength or the concrete cover
– have to be measured. By means of the determined data and based
on the service life design of the structure a verification of the
planned reliability and sustainability at the end of the structure’s
service life can be conducted, or a more precise (updated) service
life design is obtained.
Especially the concrete cover is subjected to several material
and production dependent influences. Among them, the most
important are the form and quality of the re-bar spacer, the form
and quality of the formwork and the placing and compaction of
Table 5
Target values of the reliability index b depending on the relative cost of safety
measures.
Relative cost of safety Reliability index b Reliability index b
measures [66] [67]
High 1.3 (pf  10%) 1.0 (pf  16%)
Moderate 1.7 (pf  5%) 1.5 (pf  7%)
Low 2.3 (pf  1%) 2.0 (pf  2%)
107
5.20
ð7Þ
Fig. 11. Tunnel structure subdivided in its basic elements.
the concrete. For the inner shell of a tunnel construction – here
the tunnel wall – in particular the concrete cover is in the focus
of consideration. Deviations from the planned cover thickness ex-
ert a pronounced effect on the long term durability. Within the
performed study, the planned service life of the tunnel is assumed
to be 100 years. The target value of the reliability index is set to
b = 1.7 (pf  5 %). The limit state is defined as the depassivation
of the reinforcement of the tunnel wall. Thus, when the carbon-
ation front reaches the reinforcement, the intended service life of
the wall ends. During the design process the concrete cover – i.e.
the designed cover thickness – has been specified in view of its
mean value and its related standard deviation. By means of non-
destructive testing the realized cover and its variation may be
determined.
Table 6 shows the corresponding parameter study and the re-
sults of the reliability analysis, which was performed using the
data in [68] and by applying the software code STRUREL [69].
Case A represents the design situation. At the end of the in-
tended lifetime of the structure the calculated maximum failure
probability – the probability of depassivation – is approx. 5%.
For case B – which might represent the results of an investiga-
tion after the completion of the construction – it is assumed that
during the construction process the mean value of the concrete
cover was correctly performed but the intended standard deviation
got doubled (from 8 mm to 16 mm) due to poor workmanship. The
effect of this deviation on the service life and thus on the sustain-
ability potential of the structure is significant. The probability of
depassivation, i.e. the failure probability is more than doubled
(from 5% to 11%). Accordingly, the expected lifetime is reduced
from 100 years to 60 years. In turn, rehabilitation measures might
be required which induce a significant environmental impact and
further reduce the sustainability of the structure.
Case C considers the case that the workmanship was in accor-
dance with the assumption at the design stage but a wrong re-
bar spacers were used (mean cover 45 mm instead of 55 mm).
For this case the failure probability is tripled (from 5% to 15%) com-
pared to the design assumptions. As a result, after approx. 60 years
of service life repair measures have to be conducted. Again, the
sustainability of the structure is significantly reduced.
This simple study reveals two main aspects. First, by the appli-
cation of a probabilistic based performance concept, deterioration
effects are quantified. The designer is not only able to design a
structural member for durability but he is also able to quantify
changes in the durability behavior due to deviations from the
planned conditions. Second, and more important with regard to
the topic of this paper, it is evident from this study that small
changes or errors in the construction process or the building mate-
rials used have a very significant influence on the sustainability po-
tential of the built structures. This effect is especially pronounced
in the considered example, as poor workmanship does not lead
to a reduction of the environmental impact (i.e. the denominator
of Eq. (1)) but strongly reduces the expected lifetime (in this par-
ticular case by 40%) and thus the sustainability potential.
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337 333

Fig. 12. Fault tree of the series system modeled for a bridge superstructure.

Table 6 4.5. Modeling of degradation interactions including singular risks

Service life design depending on concrete cover.

Case Mean value (mm) Standard deviation (mm) In the framework of a system analysis the investigation of inter-
Concrete cover c actions relating to durability relevant degradation mechanisms can
A 55 8 be realized by means of a fault tree analysis. In the context of a
B 55 16 fault tree analysis two possible procedures can be distinguished:
C 45 8
Case Reliability index Failure probability Achievement of the limit 1. Fault tree analysis using failure probabilities, which are
b (–) pf (%) state (years) results of a statistical analysis or knowledge-based. Models
Service life prediction for the description of the time-dependent deterioration
A 1.7 5 100 progress are generally not used.
B 1.3 11 60
2. Fault tree analysis using time variant and limit state defined
C 1.0 15 62
failure probabilities. Models for the description of the time-
dependent deterioration progress are generally used.
4.4. Reliability analysis of systems
These two procedures basically enable the frame conditions for
In the previous section the procedure for service life design was the modeling of combined deterioration interactions based on
shown only for structural components considering a single limit existing mathematical rules, see Eqs. (8) and (9). In the case of
state. However, it has to be kept in mind that typical civil engineer- the first approach, the material-dependent interactions cannot be
ing structures are complex systems which have to satisfy more considered due to unavailable deterioration-time laws. The second
than one limit state criterion according to the different environ- approach of fault tree analysis offers the possibility of modeling the
mental exposures. Therefore, it is necessary to differentiate be- material-dependent interactions. Here, by means of deterioration-
tween the reliability of components and the reliability of systems. time laws in connection with appropriate parameters which
In view of a system reliability analysis there are two basic describe the material and environmental conditions the material-
elementary systems: the series system termed also weakest link dependent interactions between the relevant deterioration-time
system and the parallel system termed also redundant system. laws (see e.g. Eq. (4)) can be considered.
By definition, a serial system failure, requires only one element Considering complex concrete structures there are also
to fail, whereas in a parallel system, failure is definitively triggered so-called singular risks (e.g. leakage of the sealing or cracks in the
by a failure of all elements in parallel. Using means of mathemat- concrete) which are interrelated with existing deteriorating mech-
ical rules one can define the lower and upper bounds of the failure anism relevant to the structures durability. In the course of a fault
probability of such systems [65]. The simple bounds for the failure tree analysis these singular risks have to be considered. Basically
probability of a series system can be calculated by means of Eq. (8): it should be kept in mind that all effects on service life due to inter-
acting degradation processes and singular risks, as is subsequently
shown, will directly affect the sustainability potential.
Y
n
max½pfi  6 pf;series 6 1  ð1  pfi Þ ð8Þ
i¼1
4.5.1. Material-dependent interactions not considered
The simple bounds for the failure probability of a parallel sys- The following deterioration mechanisms and singular risks are
tem can be calculated using Eq. (9): considered: carbonation and chloride induced corrosion, alkali-sil-
ica reaction (ASR) and an insufficient grouting of the tendon ducts
in a pre-stressed structure which leads to corrosion of tendons.
Y
n Fig. 12 shows the fault tree of a bridge element superstructure
pfi 6 pf;parallel 6 min½pfi  ð9Þ modeled for the mentioned deterioration mechanisms. The super-
i¼1
structure of the bridge represents a series system. In the first step,
The bounds for the failure probability of civil engineering struc- the interaction between the different degradation processes is not
tures depend on the statistical dependencies of the identified fail- considered, and therefore the factor considering this interaction
ure events. Examples related to probabilistic lifetime assessment of gcarbo equals 1.0. The limit state related reliability indices b were
entire concrete structures are given in [70,71]. calculated on the basis of the Model Code for service life design
334 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
[59]. Appropriate values for the model parameters were selected
from the literature [60,68,72]. For the deterioration caused by alka-
li–silica reaction and the corrosion of steel tendons corresponding
failure probabilities were taken from an example of Zhu [73]. The
related value for an alkali–silica reaction (ASR) is pf,ASR = 0.5%,
and for the corrosion of tendons the failure probability was as-
sumed to be pf,corr = 2.0%.
The prediction of the system failure probability of the bridge
superstructure was performed for a service life of 100 years. The
target value of the reliability index is set to be b = 1.6. Here, only
the upper bounds of the series system were calculated, see
Fig. 12 and Eq. (8).
The results displayed in Fig. 13 show that the reliability of the
superstructure of the bridge decreases when the number of simul-
taneously acting deterioration mechanisms increases. For example
the limit state (here b = 1.6) is reached after 95 years if only the
deterioration mechanisms carbonation and chloride induced corro-
sion take place. In the case that all exposures and damaging actions
are considered (carbonation, chloride, ASR and corrosion of ten-
dons), the limit state is already reached after 50 years, leading to
a reduction of the lifetime of 50%.
4.5.2. Material-dependent interactions considered
In order to consider the interactions between defined actions,
the factor gcarbo is introduced, see Fig. 12. By means of this interac-
tion factor gcarbo it is possible to account for the interrelated effects
of carbonation and chloride ingress. There are different possible ef-
fects which might be caused by this material-dependent interac-
tion. On the one hand due to the carbonation the concrete will
have an increased density and a lower porosity which in turn will
impede the further ingress of substances from the environment. On
the other hand the chloride binding capacity of the concrete is low-
ered due to the carbonation process. The total chloride concentra-
tion might be higher than in non-carbonated concrete since the
bound chlorides are released. To take into consideration the influ-
ence of carbonation on the chloride ingress the chloride diffusion
process is modified by introducing the factor gcarbo. Hereby the
chloride migration coefficient may either be increased or de-
creased. The chronology in which the different actions occur plays
an important role within this context, which is not further dis-
cussed here (compare [74]).
In the following example the interaction is limited to the influ-
ence of carbonation on the chloride ingress. Other effects are ne-
glected. Therefore, the corrosion of the tendons does not have an
impact on the chloride ingress nor does the carbonation. Further,
the ASR can be considered to be independent from the other
Fig. 13. Reliability index bsys vs. age t of the superstructure of the bridge.
Table 7
Parameter study on reliability bg, probability pf and expected lifetime depending on
the factor gcarbo.
gcarbo Upper bound
bg (–) pf (%)
0.5 1.8 3.6
1.0 1.6 5.5
1.5 1.4 7.7
2.0 1.3 10.0
accompanying factors. In Table 7 the varying ranges of the reliabil-
ity index bg and the failure probability pf according to the varying
factor gcarbo are given.
Within this study the factor gcarbo was varied between 0.5 and
2.0. The parameter study was performed for a fixed service life of
50 years. The influence of an increasing (>1.0) or a decreasing
(<1.0) factor gcarbo on the development of the reliability is shown
in Table 7. If the average chloride diffusion coefficient is increased
by the factor gcarbo being higher than 1.0 then the reliability de-
creases. Correspondingly the reliability increases with a decreasing
chloride diffusion coefficient as a result of a factor gcarbo being low-
er than 1.0.
The results being indicated in Table 7 and Fig. 13 show that
within the range of the varying factor gcarbo the reliability of the
series system varies from b = 1.3 (pf = 10%) to 1.8 (pf = 3.6%). If
the factor gcarbo is 0.5 the allowed safety level, related to the reli-
ability index b = 1.6 (gcarbo = 1.0), is reached after 115 years. Other-
wise if the factor gcarbo is 2.0 the allowed safety level is reached
already after 15 years.
It is clearly evident that the magnitude of the factor gcarbo, and
therefore the extent of the interaction of chloride ingress and car-
bonation, has a very pronounced effect on the reliability and sus-
tainability of the series system of the superstructure. Hereby it
must be kept in mind, that the interaction factor also depends on
the concrete quality and thus on the environmental impact during
concrete construction.
This brief example shows the necessity for the realization of a
system reliability analyses for concrete structures underlying dif-
ferent and complex deterioration mechanisms and singular risks.
For further information in relation to system reliability and risk
analysis see [74]. In view of sustainability considerations the pro-
nounced effect of interactions – lifetime or safety level being
reached either after 15 or 115 years as shown in the given example
– must be imperatively reflected, see Eq. (1).
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
5. Discussion and conclusions
The evaluation of the sustainability potential of building mate-
rials as well as of entire buildings or structures is highly complex,
as it involves the quantification of three interacting and interde-
pendent parameters, i.e. the environmental impact, the technical
performance and the lifetime. From this it is highly obvious, that
a sole reduction of the environmental impact e.g. during the con-
struction process will not benefit the environment as long as an
equivalent technical performance and the lifetime cannot be guar-
anteed. This interdependency can be evaluated using the definition
of the so-called sustainability potential proposed in Eq. (1).
As was shown in this paper, the introduction of the aforemen-
tioned definition is especially useful with regard to the compara-
tive evaluation of the sustainability of building materials such as
concrete. Whereas existing sustainability evaluation programs
such as DGNB, LEED or BREEAM are exclusively focused on the
evaluation of residential or industrial buildings and are highly
complex in their application (see e.g. [75]), simple indexes allow-
ing a fast comparison of the sustainability potential of different
concretes are missing. This becomes especially relevant, when
the focus of the design process is placed on infrastructure buildings
such as bridges and tunnels. For these structures, both the techni-
cal performance as well as the durability are decisive parameters
which are directly linked to the material behavior. The sustainabil-
ity potential then has a dimension of time multiplied for example
by strength or loading capacity divided e.g. by the GWP. Other
dimensions are possible, when different properties or impact cate-
gories are considered. However, as the compressive strength at
28 days is the decisive design parameter in concrete design, it ap-
pears reasonable to express the performance of the concrete by the
characteristic strength at 28 days fck. By choosing the characteristic
instead of the mean strength, safety aspects resulting from an in-
creased scattering of the material parameters can be considered.
Similarly, we propose to incorporate the GWP as lead parameter
for the environmental impact into Eq. (1), as the environmental
impact of the concrete production is mainly characterized by pro-
nounced emissions of CO2 resulting from the cement production.
Finally, with regard to the lifetime of the material, an approach
has been presented in Section 4, which allows for the calculation of
the service life tSL of a concrete subjected to one or more environ-
mental loadings. By introducing weighting factors and fault tree
approaches, also interdependencies between the various loadings
can be correctly considered. These influences and interdependen-
cies are not necessarily limited to durability actions. Basically
any time dependent influence can be incorporated in the calcula-
tions as long as it can be described mathematically and the stan-
dard deviation and corresponding distribution function of the
data is available. From the explanations above, the sustainability
potential X of a particular concrete can be calculated as
fck tSL
X¼ ð10Þ
GWP
and possesses the dimension [MPa s/kg] = [s1 m1].
Based on this definition, today basically two different ap-
proaches are being pursued to improve the sustainability of con-
crete: the development of concretes with minimum
environmental impact on the one hand and concretes with maxi-
mum performance and durability on the other hand.
Recognizing the substantial influence of the binder cement on
the environmental impact of concrete, major research is directed
at developing concretes with strongly reduced cement content.
The mix design principles as well as the systematic influence of
the cement content on the concrete properties were presented in
Section 3.1. As can be seen, concretes with approx. one third of
the usual cement content can be realized without substantial
335
losses in compressive strength. Hereby the environmental impact
can be reduced by up to approx. 50% depending on the impact class
considered. According to the presented results, the binder intensity
– i.e. the mass of binder in 1 m3 of concrete being necessary to gen-
erate 1 MPa of compressive strength – can be reduced from approx.
10 to 15 kg/(MPa m3) (average value for ordinary concrete today)
to values between 0 and 5 kg/(MPa m3). However, from the results
it becomes also evident, that the necessary binder intensity to gen-
erate sufficient strength strongly increases with decreasing design
strength, thus limiting the possibilities in a further reduction of the
cement content (compare Fig. 4). A final evaluation of the sustain-
ability potential of green concretes today is limited to pure com-
parisons between performance and environmental impact
parameters, as reliable information of the durability and potential
lifetime of such concretes with regard to standard expositions is
missing so far.
The second approach to increase the sustainability potential of
concrete described in this paper consists in the use of ultra high
performance concretes. This type of concrete highly benefits from
the fact that with increasing strength, the binder intensity strongly
reduces, providing a more efficient cement usage. Unfortunately,
the environmental benefit from this increase in efficiency is partly
consumed by the addition of steel fibers with a very high environ-
mental impact, needed to cope with the extreme brittleness of the
material. Thus, with regard to future research, ways have to be
found to reduce the necessary fiber dosage and to control the
materials brittleness by using mineralogical instead of technical
approaches. The key advantage of ultra high performance con-
cretes therefore can be seen in their superior durability, allowing
lifetimes far beyond the ones of ordinary concrete structures.
Before this background, the key to evaluating the sustainability
potential of concrete is to be seen in a correct evaluation of the life-
time of the material. As a reference for the expected lifetime, the
service life can be calculated by comparing the influences (environ-
mental or mechanical) acting on the concrete with its resistance to
these actions. In contrast to the descriptive procedure which is gi-
ven in national and international standards the advantage of the
service life design method is based on the fact that the durability
of concrete structures can be quantified in terms of failure proba-
bilities. With the use of deterioration time laws in connection with
predefined limit states and safety levels the point in time of
expectable damages at concrete structures can be calculated.
References
[1] O’Brien M, Wallbaum H, Bleischwitz R, editors. Eco-innovation observatory
thematic report, April 2011. World Wide Web: <http://wupperinst.org/
uploads/tx_wupperinst/EIO_WP4_ResEff_Constr_Report.pdf> [last access:
October 2013].
[2] ECORYS SCS Group, Editor. FWC sector competitiveness studies N° B1/ENTR/
06/054 – Sustainable competitiveness of the construction sector. World Wide
Web: <http://ec.europa.eu/enterprise/sectors/construction/files/compet/
sustainable_competitiveness/ecorys-final-report_en.pdf> [last access:
October 2013].
[3] Malhotra VM. Making concrete ‘‘greener’’ with fly ash. Concr Int
1999;21(5):61–6.
[4] EN ISO 14040:2009-11. Environmental management – life cycle assessment –
principles and framework. Beuth Publishers, Berlin.
[5] EN ISO 14044:2006-10: Environmental management – life cycle assessment –
requirements and guidelines. Beuth Publishers, Berlin.
[6] EN 15804:2012-04: Sustainability of construction works – environmental
product declarations – core rules for the product category of construction
products. Beuth Publishers, Berlin.
[7] Bundesverband der Deutschen Ziegelindustrie e. V, editor. Green building
challenge handbuch. World Wide Web: <http://www.ziegel.at/gbc-
ziegelhandbuch/default.htm> [last access: October 2013].
[8] Institut für Bauen und Umwelt, editor. Umwelt-Produktdeklaration nach ISO
14025 für Zement; Deklarationsnummer: EPD-VDZ-2012111-D, Inhaber:
Verein Deutscher Zementwerke e. V., Düsseldorf, Ausstellungsdatum:
16.03.2012.
336 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337
[9] Chen C, Habert G, Bouzidi Y, Jullien A, Ventura A. LCA allocation procedure
used as an initiative method for waste recycling – an application to mineral
additions in concrete. Resour Conserv Recycl 2010;54(12):1231–40.
[10] Schießl P, Stengel Th. Nachhaltige Kreislaufführung mineralischer
Baustoffe. Abteilung Baustoffe, München: Forschungsbericht der Technischen
Universität München; 2006.
[11] Backes H-P, Brandenburger D, Meißner M. Flugasche – Verfügbarkeit, Logistik,
Potential. VGB PowerTech 2012; (6): 45–51.
[12] Merkel Th. Daten zur Erzeugung und Nutzung von Eisenhüttenschlacken. In:
Report des FEhS Institut für Baustoff, Forschung 2013; 20(1): 12.
[13] Stemmermann P, Schweike U, Garbev K, Beuchle G, Möller H. Celitement – a
sustainable prospect for the cement industry. Cement Int 2010;8(5):52–66.
[14] Fernandez R, Martirena F, Scrivener KL. The origin of the pozzolanic activity of
calcined clay minerals – a comparison between kaolinite, illite and
montmorillonite. Cement Concr Res 2011;41(1):113–22.
[15] Glavind M, Munch-Petersen C. Green concrete in Denmark. Struct Concr
2000;1(1):1–7.
[16] Glavind M. Green concrete structures. Struct Concr 2011;12(1):23–9.
[17] Proske T, Hainer St, Jakob M, Garrecht H, Graubner C-A. Stahlbetonbauteile aus
klima- und ressourcenschonendem Ökobeton. Beton- und Stahlbetonbau
2012;107(6):401–13.
[18] Haist M, Müller HS. Nachhaltiger Beton – Betontechnologie im Spannungsfeld
zwischen Ökobilanz und Leistungsfähigkeit. In: Müller HS, Nolting U, Haist M,
Kromer M, editors. Nachhaltiger Beton – Werkstoff, Konstruktion und
Nutzung, 9. Symposium Baustoffe und Bauwerkserhaltung. KIT Scientific
Publishing, Karlsruhe, Germany; 2012. p. 29–52.
[19] Fennis SAAM. Design of ecological concrete by particle packing optimization.
PhD Thesis. Technical University of Delft, The Netherlands. Gildeprint
Publishers, The Netherlands; 2010. ISBN 978-94-6108-109-4.
[20] International Federation for Structural Concrete fib, editor. Guidelines for
green concrete structures. fib-bulletin 67, Lausanne, Switzerland; 2012.
[21] Funk JE, Dinger DR, Funk JE, Jr. Coal grinding and particle size distribution
studies for coal-water-slurries at high solids content. Final report. Empire
State Electric Energy Research Corporation, New York, USA, 1994.
[22] Haist M, Moffatt JS, Breiner R, Müller HS. Development principles and
technical boundaries of concrete production with low cement content. In:
Beton- und Stahlbetonbau 2013; 18(12) [accepted for publication].
[23] Andreasen AHM, Andersen J. Über die Beziehung zwischen Kornabstufung und
Zwischenraum in Produkten aus losen Körnern (mit einigen Experimenten).
Kolloid-Zeitung 1930;50:217–28.
[24] de Larrard F. Concrete mixture proportioning – a scientific approach. London,
Great Britain: E & EN Spon; 1999.
[25] EN 12390-3:2009-07: Testing hardened concrete – compressive strength of
test specimens. Beuth Publishers, Berlin.
[26] EN 12350-4:2009-08: Testing fresh concrete – degree of compactability. Beuth
Publishers, Berlin.
[27] EN 206-1:2001-07 (incl. A1:2004–10 and A2:2005–09): Concrete –
specification, performance, production and conformity. Beuth Publishers,
Berlin.
[28] Walz K. Beziehung zwischen Wasserzementwert, Normfestigkeit des Zements
(DIN 1164 Juni 1970) und Betondruckfestigkeit. Beton 1970;20(11):499–503.
[29] International Federation for Structural Concrete fib, editor. fib Model Code
2010 – Final Draft. fib-bulletin 65, Lausanne, Switzerland; 2012.
[30] Damineli BL, Kemeid FM, Aguiar PS, John VM. Measuring the eco-efficiency of
cement use. Cement Concr Compos 2010;32:555–62.
[31] Schmidt M, Bunje K, Dehn F, Droll K, Fehling E, Greiner St, Horvath J, Kleen E,
Müller Ch, Reineck K.-H, Schachinger I, Teichmann Th, Teutsch M, Thiel R, Tue
NV. Sachstandsbericht Ultrahochfester Beton. Deutscher Ausschuss für
Stahlbeton (Hrsg.) No. 561, Beuth Publishers, Berlin; 2008.
[32] Richard P. Reactive powder concrete – a new ultra-high-strength cementitious
material. In: De Larrard F, Lacroix R, editors. Proceedings of the 4th
international symposium on utilisation of high-strength/high-performance
concrete. Paris, France; 1996. p. 1343–49.
[33] EN 197-1:2011-11: Composition, specifications and conformity criteria for
common cements. Beuth Publishers, Berlin.
[34] Bornemann R, Schmidt M, Fehling E, Middendorf B. Ultra-Hochleistungsbeton
UHPC – Herstellung, Eigenschaften und Anwendungsmöglichkeiten. Beton-
und Stahlbetonbau 2001;96(7):458–67.
[35] Scheydt JC, Herold G, Müller HS, Kuhnt M, Fehling E, Schmidt M, Stürwald S,
editors. Development and application of UHPC convenience blends. In: 2nd
International symposium on ultra-high-performance concrete. Kassel
University Press, Kassel, Germany; 2008. p. 69–76.
[36] Siebel E, Müller C. Geeignete Zemente für die Herstellung von UHFB. In: König
G, Holschemacher K, Dehn F, editors. Ultrahochfester Beton – Innovationen im
Bauwesen. Berlin: Bauwerk Publishers; 2003. p. 13–24.
[37] Long G, Wand X, Xie Y. Very-high-performance concrete with ultrafine
powders. Cement Concr Res 2002;32:601–5.
[38] EN 12620:2013-07: Aggregates for concrete. Beuth Publishers, Berlin.
[39] EN 12350-08:2010-12: Testing fresh concrete – Self-Compacting Concrete –
Slump flow test. Beuth Publishers, Berlin.
[40] EN 12350-09:2010-12: Testing fresh concrete – self-compacting concrete – V-
funnel test. Beuth Publishers, Berlin.
[41] Rossi P, Renwez S, Guerrier F. Les bétons fibrés à ultra-hautes performances –
L’expérience actuelle du LCPC. Bullet LCPC 1996; 204: 87–95 [Juillet-Aout].
[42] Scheydt JC, Herold G, Müller HS. Ultrahochfester Beton. In: Müller HS, Nolting
U, Haist M, editors. Proceedings of the 3. Symposiums Baustoffe und
Bauwerkserhaltung. Universitätsverlag Karlsruhe, Germany; 2006. p. 33–44.
[43] Richard P, Cheyrezy M. Composition of reactive powder concretes. Cement
Concr Res 1995;25(7):1501–11.
[44] Karihaloo BL, de Vriese K. Short-fibre reinforced reactive powder concrete. In:
Reinhardt HW, Naaman AE, editors. Proceedings of the third international
workshop on High Performance Fiber Reinforced Cement Composites (HPFRCC
3). RILEM Publications, Cachan, France; 1999. P. 53–63.
[45] Burkart I, Müller HS. Creep and shrinkage characteristics of ultra high strength
concrete (UHPC). In: 2nd Int. symposium on ultra high performance concrete.
Kassel, Germany, March 05–07, 2008. p. 469–76.
[46] Müller HS. Time dependent behaviour of ultra high performance concrete
(UHPC). In: Proceedings of the 3rd fib international congress, Washington,
USA; 2010.
[47] Fehling E, Schmidt M, Teichmann T, Bunje K. Entwicklung, Dauerhaftigkeit und
Berechnung Ultra-Hochfester Beton (UHPC). Schriftenreihe Baustoffe und
Massivbau. In: Schmidt M, Fehling E, editors. Kassel University; 2004, No. 1.
[48] Müller HS, Scheydt J. Dauerhaftigkeit und Nachhaltigkeit von ultrahochfestem
Beton. Beton 2011; (9): 336–43.
[49] Scheydt JC. Mechanismen der Korrosion bei ultrahochfestem Beton. PhD
Thesis. Karlsruhe Institute of Technology (KIT). Karlsruhe, Germany, KIT
Scientific Publishing; 2013.
[50] Roux N. Durabilité des bétons de poudres réactives. Centre d’assistance
technique et de documentation, St-Rémy-lès-Chevreuse, France; 1997; (8):
46–58.
[51] EN ISO 12572:2001-09: Hygrothermal performance of building materials and
products – determination of water vapour transmission properties. Beuth
Publishers, Berlin.
[52] Jooss M, Reinhardt H-W. Permeability and diffusivity of concrete as function of
temperature. Cement Concr Res 2002;32(9):1497–504.
[53] Fagerlund G. Freeze-thaw resistance of high performance concrete. In: König
G, Dehn E, Faust T, editors. 6th Int. symp. of high strength/high performance
concrete, Leipzig, Germany; 2002. p. 45–62.
[54] Franke L, Kiekbusch J. Behaviour of high performance concrete under acid
attack. In: Proceedings of the international conference on durability of high
performance concrete and final workshop of CONLIFE, September 2004,
University of Essen, Germany, 2004, p. 311–20.
[55] Roux N, Andrade C, Sanjuan MA. Experimental study of durability of reactive
powder concrete. J Mater Civil Eng 1996;8(1):1–6.
[56] Bornemann R, Schmidt M, Vellmer C. Brandverhalten ultra-hochfester Betone.
In: Beton 2002;52(9):418–22.
[57] Heinz D, Dehn E, Urbonas L. Fire resistance of ultra high performance concrete
(UHPC) – testing of laboratory samples and columns under load. Ultra high
performance concrete (UHPC). Schriftenreihe Baustoffe und Massivbau. In:
Schmidt M, Fehling E, Geisenhanslüke C, editors. University of Kassel,
Germany; 2004. p. 703–16.
[58] DIN 10415-2:2008-08: Concrete, reinforced and prestressed concrete
structures – Concrete – Specification, properties, production and conformity
– Application rules for DIN EN 206-1. Beuth Publishers, Berlin.
[59] International Federation for Structural Concrete, editor. Model Code for Service
Life Design. fib bulletin 34, Lausanne, Switzerland, 2006. ISBN 2-88394-074-6.
[60] The European Union – Brite EuRam III: Modelling of Degradation. DuraCrete:
Probabilistic Performance based Durability Design of Concrete Structures.
Contract BRPR-CT95-0132. Project BE95-1347. Document BE95-1347/R4-5,
December 1998.
[61] The European Union – Brite EuRam III: Design Framework. DuraCrete:
Probabilistic Performance based Durability Design of Concrete Structures.
Contract BRPR-CT95-0132. Project BE95-1347. Document BE95-1347/R1,
March 1997.
[62] The European Union – Brite EuRam III: Probabilistic Methods for Durability
Design. DuraCrete: Probabilistic Performance based Durability Design of
Concrete Structures. Contract BRPR-CT95-0132. Project BE95-1347.
Document BE95-1347/R0, January 1999.
[63] Gehlen C, Mayer TM, von Greve-Dierfeld S. Lebensdauerbemessung. In: Beton-
Kalender 2011. Part 2. Ernst & Sohn Verlag; 2011. p. 231–78.
[64] EN 1990:2010-12: Eurocode – Basis of structural design; German version EN
1990:2002 + A1:2005 + A1:2005/AC:2010. Beuth Publishers, Berlin.
[65] Melchers RE. Structural reliability analysis and prediction. John Wiley & Sons;
2002.
[66] Joint Committee on Structural Safety (JCSS): Probabilistic Model Code – Part I:
Basis of Design; 2001.
[67] Rackwitz R. Zuverlässigkeitsbetrachtungen bei Verlust der Dauerhaftigkeit von
Bauteilen und Bauwerken. Bericht zum Forschungsvorhaben T 2847.
Fraunhofer IRB Verlag; 1999.
[68] Durable and Reliable Tunnel Structures (DARTS) – The Reports (CD Rom). CUR
Gouda, May 2004.
[69] RCP GmbH: STRUREL – A Structural Reliability Analysis Program System
(STATREL Manual 1999; COMREL & SYSREL Manual. 2003). RCP Consulting
GmbH, Munich, Germany.
[70] Müller HS, Vogel M. Lebenszyklusmanagement im Betonbau. Beton 1999; (5):
206–14.
[71] Müller HS, Vogel M. Lebensdauerbemessung im Betonbau – Vom
Schädigungsprozess auf Bauteilebene zur Sicherheitsanalyse der
Gesamtkonstruktion. Beton- und Stahlbetonbau 2011;106(6):394–402.
 H.S. Müller et al. / Construction and Building Materials 67 (2014) 321–337 337

[72] The European Union – Brite EuRam III: Statistical Quantification of the [75] Deutsche Gesellschaft für Nachhaltiges Bauen e. V., editor. Das DGNB System
Variables in the Limit State Functions. DuraCrete: Probabilistic Performance im internationalen Vergleich – ein Überblick. Presentation, Stuttgart,
based Durability Design of Concrete Structures. Contract BRPR-CT95-0132. Germany, 24.06.2013. available on: <https://www.google.de/
Project BE95-1347. Document BE95-1347/R9, January 2000. search?client=opera&q=vergleich+leed+dgnb&sourceid=opera&ie=utf-
[73] Zhu W. An Investigation into Reliability based Methods to include Risk of 8&oe=utf-8&channel=suggest>.
Failure in Life Cycle Cost Analysis of Reinforced Concrete Bridge Rehabilitation [76] International Federation for Structural Concrete, editor. Environmental design
(Master Thesis). School of Civil. Environmental and Chemical Engineering. State-of-art report. fib bulletin 28, Lausanne, Switzerland, 2004. ISBN 978-2-
Science. Engineering and Technology Portfolio. RMIT University; 2008. 88394-068-0.
[74] Müller HS, Vogel M, Neumann T. Quantifizierung der Lebensdauer von
Betonbrücken mit den Methoden der Systemanalyse. Berichte der
Bundesanstalt für Straßenwesen (BASt), Brücken- und Ingenieurbau 81, 2011.