Cement and Concrete Research 157 (2022) 106802

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Slow penetration for characterizing concrete for digital fabrication

Lex Reiter a, Timothy Wangler a, Nicolas Roussel b, Robert J. Flatt a, *
a
Institute for Building Materials, ETH Zürich, Switzerland
b
Navier Laboratory, IFSTTAR, East Paris University, France

A R T I C L E I N F O A B S T R A C T

Keywords: Digital fabrication processes with concrete require a specific yield stress evolution after placing. This paper
Penetration shows that this goal can be achieved with displacement controlled continuous slow penetration methods, and
Yield stress gives a model understanding and specific suggestions for measurement. It is found that the force recorded in
Structural build-up
continuous and point-wise penetration tests with a conical tool moving at a slow rate scales linearly and over
Thixotropy
multiple decades with independent measurements of yield stress carried out using uniaxial compression and vane
test. A model adapted from soil-mechanical stability predicts the bearing capacity factor that relates the pene­
tration force to yield stress. The experimental value agrees with the predicted one for an undrained/plastic
material. The measurements indicate a yield stress validity range of 1–200 kPa and as low as 0.1 kPa with
consideration of depth dependency. This range is the one of greatest interest for structural build-up in digital
fabrication during production.

1. Introduction account, measuring techniques of yield stress for concrete at rest are
necessary.
Digital fabrication is opening new horizons for architecture and Characterization of fresh concrete extensively deals with flow
construction. In the case of concrete however, this field is challenged by behaviour, with a focus on the onset and end of flow, using methods as
the need to control the material from its fresh to its hardened state in a flow spread and slump [10,11]. In recent decades, the development of
much more demanding way than for traditional construction [1–4]. self-compacting concrete and the concern for formwork pressure have
Especially in layered extrusion [5] and digitally controlled casting sys­ led to developments in characterizing thixotropic effects [12,13]. Fewer
tems for slipforming or weak formworks [6–8] the yield stress of con­ methods have been employed to measure yield stress of fresh concrete
crete needs to be assessed throughout production. At the time of placing, after long periods at rest. Notable examples of this are compressive
yield stress needs to be controlled within a well-defined and rather strength [9,14], vane tests [15–17] or penetration [18] measurements
narrow window. After this, when the concrete is at rest, it needs to in­ carried out at fixed times after casting. A common point between these
crease its strength at a specific rate. methods is their (at least partially) destructive nature, requiring the
The required precision in terms of initial yield stress and build-up is need for an individual sample for each measurement. The reason for this
substantially higher than in traditional processes. Indeed, a too high is that failure in such measurements is based on a transitional flow state
yield stress upon placing can lead to weak layer interfaces, to filament with the development of a surface on which shear strain is concentrated.
tearing or to insufficient form filling. In contrast, a too low yield stress This leads to partial breakdown of the supporting structure of the ma­
can lead to flow off or segregation. After placing, a too slow yield stress terial [19,20], the recovery of which takes time, if possible at all [21].
increase can lead to flow off and at later ages to buckling, while a too fast A continuous and simple method to measure the build-up of strength
increase can cause filament tear-off in slip-forming [2,9]. at rest by mechanical loading would be ideal. Such a method would only
An important distinction from traditional casting processes is the require one sample and would, in turn, reduce labour time and offer
emphasis on so-called “green” strength evolution or structuration of measurements in real time. Among non-mechanical methods, there are
fresh concrete at rest. Variability in hydration and hardening kinetics non-destructive indirect methods. One of these is calorimetry [22],
can be a result of compositional and environmental variability that can which gives (indirect) information on growth of hydration products. Its
compromise process robustness. In order to take such variations into drawback for the problem at stake is that it does not directly measure

* Corresponding author.
E-mail address: [email protected] (R.J. Flatt).

https://doi.org/10.1016/j.cemconres.2022.106802
Received 27 July 2021; Received in revised form 23 March 2022; Accepted 31 March 2022
Available online 11 April 2022
0008-8846/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

physical interactions and their mechanical consequences. Mechanical theoretically and in Section 4 experimentally. Finally, we show that this
non-destructive methods such as ultrasound travel speed [18,23,24] or method can also be applied to mortars (Section 4.4), although more care
small amplitude oscillatory strain measurements [20,25–27] and must be taken to define the validity range. Overall, this paper serves to
settling [28] also exist. Ultrasound can also measure dynamic shear establish slow penetration tests as a viable means to characterize the
moduli [24], but, for yield stress calculation, a constant critical defor­ early strength development of concrete in digital fabrication. Recom­
mation needs to be assumed and the measurement does not occur at the mendations for measurement and analysis are summarized in Section
onset of flow. It therefore also presents a number of limitations. 5.3 with the main analysis step is summarized in Eq. (10).
Continuous methods measuring at the onset of flow are mainly
limited by the transitional flow state on a fixed shear surface. Examples 2. Materials and methods
of this are compression, vane and penetration tests [29]. In compression
tests, sample layers can be directly loaded at the rate of the building The cement pastes used in this study consist of Ordinary Portland
process, detecting failure and relating it to yield stress at the time of Cement CEM I 52.5R (details in Supplementary materials) mixed with
failure [14]. Vane measurements involving multiple discontinuous tap water. They are prepared without admixture additions and have a
measurements have been studied [15]. However, the structural break­ w/c ratio of 0.35. The goal of the study being to relate physical char­
down from one such measurement to the next may not be fully recov­ acterization methods at a macroscopic level, only one cement paste
ered. In addition, vane geometries are prone to other phenomena such as composition is used. Materials are mixed for 1 min at slow speed and 2
wall slip [30], shear localisation and fracturing in the shear zone, which min at medium speed using a Hobart mixer in batches of 4 L.
leads to a permanent change in measuring conditions [19]. Attempts at The conical penetration tip used in this study has 30 mm height and
continuous penetration measurements related different penetration ge­ 10 mm radius and is attached to a load cell of 1kN on a Zwick universal
ometries and methods, to yield stress have been reported. Among these, testing machine. Cylindrical containers of 9 cm diameter and 12 cm
some could measure continuously, but their application was limited to height filled with cement paste are used unless specified otherwise
cement pastes [18]. (Fig. 1). The penetration tip is inserted into the sample at a speed of 1
In the field of digital fabrication, the processes in most direct need of mm/s before measurement until the conical part is level with the sample
a real-time measuring technique are layered extrusion and digitally surface for measurements in Figs. 6, 7, 9 and 10. For fast penetration
controlled casting processes. In both cases, their overall processing, and measurement, the tip keeps penetrating at a rate of 1 mm/s. For slow
especially their pumping, defines that they must be matrix-rich mortars. penetration measurement, the penetration rate is 20 mm/h. For results
Additionally, as for most other digital fabrication methods, they have an in Fig. 5, the measurement starts immediately upon contact between tip
important requirement to offer good quality geometrical detailing, and concrete.
which generally limits the maximum grain sizes to 4 mm at most. Vane tests are carried out independently on a rotational rheometer
Therefore, an adequate measuring technique should target this kind of (Anton Paar, MCR501, Graz) using a Vane tool with 4 blades, being 40
materials. mm high, and 11 mm wide. The cement paste is poured in a cylinder of
Finally, it should be noted that, for these new building processes, the diameter 27 mm immediately after mixing and the vane tool is intro­
strength upon placing should evolve from values as low as 0.1 kPa – duced immediately. After a specified resting time, a 0.05 s− 1 shear rate is
sufficient to self-support a single placed layer – up to several 100 s of kPa applied and the peak shear stress is considered as yield stress, assuming a
– sufficient to self-sustain a structure several meters high [3,31]. This cylindrical shear surface (i.e. the so-called “Couette analogy”) [32].
sets therefore the range of values that an ideal continuous measurement Uniaxial compression tests are carried out on cylinders 18 cm high
method should cover. and 9 cm in diameter. The samples are cast immediately upon mixing.
This paper examines various test methods, showing, in particular, They are later demoulded and placed for testing 10 min before the
that penetration tests allow for a direct measurement of yield stress. This measurement. The loading is displacement-controlled at a rate of 0.2
is established through soil mechanics considerations detailed in Section mm/s. Between the top surface and the steel calotte, a foam sheet is
3. Importantly for digital fabrication, we show that slow penetration added. The yield stress is calculated considering the von Mises plasticity
tests provide a reliable means of obtaining a continuous measurement of criterion (elongational flow) [10].
yield stress and its evolution. Selected geometries allowing this are re­ As a complementary test, a flat penetration tip with diameter 18.8
ported in Section 2 and their relevance is confirmed in Section 3.3 mm and height 4 mm is used as studied by Mettler et al. [29].

Fig. 1. Penetration test setup. a) Photograph of sample container and penetration tip, b) dimensions, c): Starting point of penetration tests with respect to concrete
surface and approximation of the outermost characteristic (dashed line) with increasing penetration depth d.

2
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

3. Background: penetration experiments in concrete and soil space, reaching eventually the solid percolation threshold and leading to
a state, in which the material rheology is governed by frictional particle
In this section, the theoretical frame for yield stress measurement interactions [37,41]. This was demonstrated by Mettler et al. [29] in the
from penetration tests is established. Section 3.1 qualitatively links the case of self-compacting concretes (SCC) for digital fabrication, showing
mechanical behaviour of early age concrete to soil mechanics. Section that, at very early age, failure is localised at shear bands and that tensile
3.2 introduces the full force-yield stress relationship in analogy to sta­ and compressive strengths are of similar magnitude. The same authors
bility of foundations in soils. Section 3.3 analyses the main factors also reported that over the course of hydration the ratio of tensile to
affecting penetration measurements and simplifies the relationship compressive strength evolves towards the value of about 1:10, which is
down to its main influencing factors. Section 3.4 relates compressive typical for hardened concrete. This takes place along with the occur­
strength to penetration stress, compares to literature values and predicts rence of distributed cracks upon failure and an increase in the internal
the validity range of yield stress calculations from penetration tests. friction angle φ [29].
In terms of soil mechanical analogy, the change in the ratio between
tensile and compressive strength marks the transition to an undersatu­
3.1. A soil mechanics approach to fresh concrete rated or drained soil. In SCC, the same general phenomenon is observed
for increasing aggregate contents, as, above a critical aggregate content,
As for any material, the stability of soils or concrete depends on the yield stress is dominated by friction between aggregates [35]. Moreover,
relation of applied load to resistance. In soil stability, the main points of this transition also takes place at a similar time as final set by Vicat
interest are drainage state, consolidation degree, saturation with water, needle measurements and can be evidenced by the occurrence of
pore water pressure and microstructure [33,34]. In soils, the presence of negative pore water pressure, showing positive normal stress between
water is an essential factor for mechanical behaviour. It is expressed by particles in the concrete and implying friction between particles
the concept of saturation and is an environmental variable. In concrete, [42,43].
the water content is defined at mixing and decreases as hydration pro­ While the cementitious matrix is water saturated, it can be consid­
gresses. On the timescale of interest for this study, this water content ered as an undrained soil, with its key properties being yield stress and a
does not depend on environmental conditions. plastic behaviour that can be described by a Tresca criterion (Fig. 2).
For processing reasons, the type of concrete used in digital fabrica­ With the progress of hydration, a drained state is reached, in which
tion requires a microstructure that is saturated with matrix material, and failure is dominated by cohesion and the angle of internal friction, which
a matrix that is itself saturated with water [35–37]. Consequently, the can be described by Mohr-Coulomb. In this case, the resistance to failure
mechanical behaviour of such concrete is governed by colloidal in­ depends on the effective normal stresses between the particles. Gener­
teractions, hydration products growing at the contact points of particles ally, the material is stable if shear stresses in all directions remain within
at rest [20] and hydrodynamic interactions under flow [35]. This is the space delimited by the Mohr-Coulomb failure criterion (Fig. 2),
usually approximated as a yield stress fluid using the Bingham or the defined by the angle of internal friction, cohesion and effective normal
Herschel-Bulkley model [38,39]. stresses.
In the field of soils, the most comparable materials are fine-grained In what follows, we will use these general definitions to analyse the
clay and silt, exhibiting similar colloidal interactions and a flow nature of penetration tests applied to mortars suitable for digital
behaviour linked to their saturation degree and classified by Atterberg fabrication.
limits [40]. Soils consolidate, losing water through drainage and in­
crease their shear resistance in the process, which is, especially for clays
and silts, a long-term process. 3.2. Penetration in soil mechanics
Drainage in concrete can occur by bleeding (usually prevented by
proper mix design) and by hydration. During hydration, free pore water The stability of soils has been studied extensively in the field of soil
is consumed to form hydration products. This increases the solid content mechanics and is of great importance for many practical applications,
and strengthens interparticle bonds/cohesion [21]. It also fills the pore such as the verification of the bearing capacity of foundations and piles

Fig. 2. Mechanical behaviour of un­

drained (left) and drained (right) soils.
Tresca criterion for plastic material and
Mohr-Coulomb criterion for frictional
material, with the relationship of normal
stresses to shear stress in the Mohr cir­
cle. In frictional materials, shear failure
stress depends on effective normal
stresses. The angle of shear failure in
compressive strength measurements
(top) and the shape of failure zone in
penetration tests (bottom) depends on
the angle of internal friction φ, and is
qualitatively indicated in red. The incli­
nation of the shear failure surface is
related to internal friction angle via
Mohr's circle. In penetration tests in
frictional materials high effective
normal stress occurs.
with:
φ: Internal friction angle
c0: Cohesion
σ1: Normal stress
σv, a, σv, p: Active and passive vertical stress. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

3
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

[34,44,45]. The resistance of concrete to a penetration tip can be For such undrained materials, the cohesion bearing capacity factor
calculated by analogy to the stability of foundations. As such, the Nc for a conical penetration tip as used in this study is approximately 9 as
penetration stress σf of a penetration tip of cross-section area Apen into a found by Houlsby et al. [47]. This factor assumes adhesion between the
mortar sample may be expressed using bearing capacity factors Ni: penetration tip and the material [49]. Houlsby also finds that the shape
/ of the outermost characteristic (i.e., the volume in which shearing oc­
Fpen
= σf = Nc ⋅c0 + Nq ⋅ρgd + Nγ ⋅1 2bρg (1) curs) has approximately the shape of a cone with an apex angle of 90◦
Apen
extending from the penetration tip towards the surface, in a similar way
as shown in Fig. 3.
where
The relative contributions of cohesion and surcharge can be
compared for this case of undrained material. In the penetration setup of
Fpen: Penetration force
this study, the depth may reach up to 8 cm, which gives a maximum
Apen: Penetration tip cross-section area /
surcharge ρgd of approximately 1.8 kPa (ρ = 2300 kg m3 ). At the lowest
σ f: Penetration stress
c0: Cohesion yield stress of interest τ0 = c0 = 0.1 kPa, cohesion contributes to eq. (2)
Nc: Cohesion bearing capacity factor with 0.9 kPa. Therefore, a yield stress in the lower range has a high
Nq: Surcharge bearing capacity factors surcharge dependence on penetration depth. A higher yield stress, such
Nγ: Soil-weight bearing capacity factor as 1 kPa, gives a cohesion contribution of 9 kPa, and the effect of sur­
ρ: Mortar density charge is lower than 20% and can be neglected.
g: Gravity The material, however, transitions to a drained material after
d: Penetration depth placement, when frictional contacts between grains become dominant.
b: Penetration tip width With the defined maximum friction angle of 40◦ , the expected contri­
butions of surcharge and soil weight (which becomes nonzero) can be
The cohesion, surcharge and soil-weight bearing capacity factors in calculated relative to cohesion. The surcharge contribution (Nq = 64)
Eq. (1) (Nc, Nq and Nγ respectively) are analytically or empirically becomes 115.2 kPa, and the soil weight contribution (Nγ = 100) be­
derived correlation factors between applied penetration stress and comes 22.5 kPa. This is in comparison to a cohesion contribution of
resisting stress in the soil [46] (see Table 1). They define the contribu­ 7530 kPa (Nc = 75.3) when the lowest cohesion with friction contacts is
tion of the different mechanisms resisting penetration and depend considered (100 kPa), using results by Mettler et al. [29]. For this case,
strongly on the angle of internal friction. Eq. (1) has been introduced for the surcharge and soil weight contributions are an even smaller per­
2D-wedges of infinite length, but correction factors, or even bearing centage (<2%) compared to cohesion, and can be neglected for the
capacity factors calculated for other geometries exist, including conical property range of materials considered here.
and flat penetration tips [47–49]. Penetration test approaches using this The above suggests that, if yield stress is larger than 1 kPa, only the
framework such as the fall-cone test [50] or the cone penetration test for cohesion contribution in mortars needs to be considered in both drained
measurements underground [51,52] have been used to characterize or and undrained conditions, and the following simplified equation holds:
classify soils (foremost clays). Despite the broad range of publications, FPen
assumptions needed to idealise material behaviour for calculations limit σf = = Nc c 0 (3)
APen
the quantitative precision that can be expected from these methods.
However, for the study of parameters influencing the bearing capacity of It is important to note that, for yield stress values below approxi­
a penetrating geometry at a certain depth and for different material mately 1 kPa, the effect of surcharge will have to be taken into account
types, soil mechanical studies can offer valuable insight. as penetration depth increases. However, when performing slow pene­
tration for most mortars of interest in digital fabrication, the yield stress
3.3. Bearing capacity factors for conical penetration tests is expected to anyway increase at least up to 1 kPa before any substantial
penetration depth is reached.
Fresh digital fabrication concrete in a penetration test is expected to Concerning the range of possible bearing capacity factors Nc,
be an undrained material and the lowest yield stress of interest is τ0 = different approaches can be followed. Using a kinematically possible
0.1 kPa. As it hardens, a transition to a drained material with an internal
friction angle of 40◦ is expected to occur at a cohesion of 100 kPa for
OPC mortars [29]. These parameters set the upper and lower boundaries
for what one penetration test should be able to capture and are used in
the following calculations.
Digital fabrication mortars are matrix saturated, and thus can be
defined as undrained according to soil mechanics when they are initially
placed, i.e. not dominated by frictional contacts between grains. This
gives a friction angle φ = 0, which from Table 1 defines the surcharge
bearing capacity factor Nq = 1, and the soil weight bearing capacity
factor Nγ = 0, which gives Eq. (2). Eq. (1) can thus be simplified in the
case of undrained materials to:
Fig. 3. Kinematic shear failure surface used for the calculation of the depth
σ f = Nc c0 + ρ⋅g⋅d (2) dependency factor αd.

Table 1
Bearing capacity factors according to Foundation Engineering Handbook [46] for a 2D-wedge used for the calculations in Section 3.3 and in Fig. 4. Values are given as a
function of the internal friction angle φ.
φ[◦ ] 0 5 10 15 20 25 30 35 40

Nc 5.14 6.50 8.35 11.0 14.8 20.7 30.1 46.4 75.3

Nq 1 1.57 2.47 3.94 6.40 10.7 18.4 33.6 64.2
Nγ 0 0.45 1.22 2.65 5.39 10.9 30.2 48.7 109

4
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

failure surface, an upper limit of the bearing capacity factor can be [ ]

αcomp Pa
computed. For the conical penetration tip used in this study, a simple τ0 [Pa] = √̅̅̅ ⋅Fpen [N] = 774 [Pa/N]⋅Fpen [N] (7)
approximation of such a failure surface may be a conical shear surface αpen 3π R2 N
with an apex angle of 90◦ starting from the cone tip (cf. Fig. 1.c). This The bearing capacity factor in this case can also be derived using the
gives a shear failure surface of 4050 mm2, which can be expressed as the undrained penetration stress and neglecting penetration depth:
Fpen
surface S, on which a shear stress of c0 applies, giving S = c0 = Nc ⋅Apen . √̅̅̅
αpen 3
This leads to Nc being approximately 13. The lower limit can be derived Nc,Mettler = = 4.65
αcompr
directly from the main stress difference applied twice in the material,
giving Nc = 4. The proposed bearing capacity factor Nc = 9 is thus be­ The theoretical value for an undrained soil with a flat penetration tip
tween the boundary limits. In this sense, soil mechanical approaches is 5.69 [49], but this is only applicable for a material with a strain-
contrast with currently documented approaches for yield stress scaling independent cohesion and for shallow penetration. Measurements as
in concrete, where the contact surface is considered, giving in this case done in [29] occurred below the surface, slightly increasing the effective
(993mm2) a value Nc = 3.2 [18]. Nc can also be calculated based on bearing capacity factor. Under such conditions, the formed shear surface
Wuensch [53] as used for penetration test analysis in [54]. Indeed for has to be considered as slightly pre-sheared at the time of measurement,
very high Bingham numbers as expected here (quasi-static), the rela­ slightly decreasing the bearing capacity factor. Given these un­
1 Fpen certainties, the 20% difference between the experimentally found factor
tionship established by Wuensch simplifies to τ0 = 13.8 Apen . This yields
the same equation and a factor Nc = 13.8 for spherical penetration tips. and the theoretical one can be overlooked. In other words, we can
A common limitation of wedge stability models is the extension of consider that the theoretical frame presented above captures well the
failure surfaces into the surcharge material as penetration depth in­ experiments done by Mettler et al. and consequently can also be well
creases (Fig. 3). For conical penetration tips, depth dependent bearing applied to concrete that is suitable for digital fabrication.
factors exist [47]. Alternatively, we propose to use a depth dependency The ratio Ncomp of penetration stress to compressive strength can be
correction factor αd based on a kinematically possible failure surface. used to assess the reliability of penetration tests as a measurement of
This considers the relative shear surface increase due to the surcharge cohesion or compressive strength. In general, for a frictional material
material by considering the minimum surface created by a truncated using Eq. (3) it writes as follows:
cone with apex angle 90◦ and cylinder reaching from the lower cone to σf N ⋅c
the surface. This is shown as dashed lines in Fig. 3: Ncomp = = ( c 0 1 ) (8)
σ comp 2c0 tanφ + cosφ
( √̅̅̅ )
αd =
min 2 dH, H 2 − R2
+1 (4) Table 2 shows the increase of Ncomp with increasing internal friction
H2 angle for a 2D-wedge. This indicates that penetration stress and
From this perspective, the effect of depth dependency is expected to compressive strength do not scale linearly, with a change of internal
be most pronounced up to a depth of H − R, while most soil mechanical friction angle. It confirms the prediction that with the transition from
approaches expect a significant change up to a depth of 4R [47,55,56], undrained to drained state, a linear relationship of penetration to
twice the depth for this geometry. It will be shown in Section 4.1 (Fig. 6) compressive stress cannot be expected. For a given friction angle the
that our samples indeed comply with the proposed depth dependency ratio is however independent of cohesion. Thus, if the internal friction
correction factor αd. angle is known, penetration tests may still be used as a measure of
Houlsby also considers other effects such as that of the cone apex cohesion and by extension of compressive strength. For this purpose, a
angle α, as well as heave and cone bluntness (slightly truncated cone) friction amplification factor αφ maybe be introduced:
[49]. The apex angle should be chosen in a range where the bearing Ncomp
capacity is not strongly affected by it, thus between 40◦ and 70◦ . Blunt αφ = (9)
Ncomp (φ = 0)
cones and the effect of heave increase penetration resistance to a minor
extent (0.5% and 9% respectively). For the purpose of this paper, αφ indicates the expected over­
estimation of compressive strength by penetration measurements when
3.4. Penetration and compressive strength in fresh concrete neglecting the internal friction angle. Given the difficulty to measure
internal friction angles especially in the transition phase from plastic to
Previous studies have found an excellent linear agreement between frictional behaviour, the penetration resistance may be difficult to relate
compressive strength measurements in the experimental timeframe [29] to yield stress in the transition zone. Thus, the penetration test seems
up to a penetration stress of 520 kPa and compressive stress of 190 kPa, most suited for undrained conditions.
equivalent to approximately 110 kPa yield stress. Using the documented An overview of the expected overestimation of cohesion (deter­
experimental results from compressive strength and penetration tests, mining yield stress and compressive strength) from penetration tests
the bearing capacity factor of the penetration tool used in that study can when considering a plastic material in the conversion and neglecting
be calculated. In that work, measurements with exponential functions of surcharge and soil weight is shown in Fig. 4 as lines of a given over­
time: σpen = αpeneβt and σcomp = αcompeβt with factors αpen = 0.188kPa and estimation factor. This is computed by taking the following ratio: For the
αcomp = 0.070kPa and the same factor β for the time dependency eβt are numerator the full Eq. (2) with the value of NC taken for the given
used. In this specific case, the ratio of normal stresses can be expressed as friction angle (from Table 2). For the denominator the simplified Eq. (3)
the ratio of fitting factors: with NC = 9 is taken. In Fig. 4 a factor close to 1 therefore indicates that
the simplifications in Eq. (3) can be done for a given combination of
σ pen αpen
= (5) material properties (internal friction angle and cohesion) without
σ comp αcomp
introducing a substantial bias. The figure shows that materials with in­
To relate uniaxial compression to yield stress, the von Mises criterion ternal friction angles larger than 10◦ (overestimation ratio larger than
is commonly used: 1.5) cannot be accurately captured by Eq. (3) and thus knowledge of the
√̅̅̅ internal friction angle is needed. As the penetration test alone cannot
σ comp = fy = 3 τ0 (6) determine internal friction angle, large angles should be avoided.
Using this, we can calculate a conversion factor for yield stress from Similarly, for cohesion below 1 kPa taken at a penetration depth of 8 cm,
the penetration test (flat penetration tip with 2R = 18.8mm): the overestimation factor is larger than 1.5 for all internal friction an­
gles. This shows that for shallow depth acceptable results can be

5
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

Table 2
Bearing capacity factor NC, ratio of penetration stress to compressive stress Ncomp for a 2D-wedge and amplification factor αφ indicating the overestimation of
compressive strength from penetration tests, when considering a plastic material behaviour.
φ[◦ ] 0 5 10 15 20 25 30 35 40

NC 5.14 6.50 8.35 11.0 14.8 20.7 30.1 46.4 75.3

Ncomp 2.6 3.0 3.5 4.2 5.2 6.6 8.7 12.1 17.6
αφ 1.0 1.2 1.4 1.6 2.0 2.6 3.4 4.7 6.8

Fig. 4. Overestimation of cohesion for a set of internal friction angle and material cohesion, when considering Eq. (3) with NC = 9. a): tip in 1 cm depth, b): tip in 8
cm depth. For very low cohesion between 0.1 and 0.5 kPa, penetration test overestimates yield stress if surcharge is not considered. For materials with high internal
friction angle penetration tests strongly overestimate compressive strength.

Fig. 5. Fast penetration. Dependence of depth and container geometry for not fully immersed penetration tip. a, b, c) Penetration force increases with the square of
penetration depth. d) Force at end of measurement depending on the sample container diameter. The extent of the outermost characteristic of the shear zone is
indicated as a dashed line. The container diameter needs to be substantially larger than the outermost characteristic.

6
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

expected for cohesion as low as 0.1 kPa, while 1 kPa is needed at the full
penetration depth and internal friction angle should always remain
smaller than 10◦ .

4. Discontinuous and continuous methods on cement pastes

The previous section established the analysis of penetration tests in

order to compute yield stress from measurements. This section describes
the measurement results of fast and slow penetration as well as the yield
stress from compressive strength and vane measurements. In particular,
we show the linear scaling of penetration tests with yield stress
measured with other methods in Sections 4.3 and 4.4 shows penetration
tests carried out on mortars.

4.1. Fast penetration

Fig. 5 shows the penetration forces of a conical penetration tip

pushed into cement pastes having the same age and placed in cylindrical
containers having the same height (12 cm), but different diameters (6, 7,
9 and 11 cm). The measurements start when the tip touches the cement
paste surface and end once it is fully inserted, with the displacement
corresponding to the insertion depth of the penetration tip (Fig. 5.a). In
this case, the penetration force scales with the square of penetration
depth and therefore linearly with the cross-sectional area AC in contact
with the paste since AC = πrC2 (Fig. 5.b). The measured penetration
stress σ f thus remains constant throughout the measurement once the
contact surface is large enough (Fig. 5.c). In Fig. 5.d, it can also be seen
that the measured force is substantially smaller for samples in containers
with diameter 7 cm and smaller.
At this stage, a comment should be made about the effect of the
container diameter on the measured force (Fig. 5.d), because it condi­
tioned the container choice for subsequent measurements presented in
this paper. This effect cannot be directly accounted for by analytical
approaches and rather appears to be associated with wall slip on the Fig. 6. Fast penetration for the same geometry at different times and starting
container wall. Indeed, in case of wall slip, penetration force is expected from a fully submerged penetration tip. a. Without depth correction, b. with
to scale with increasing sample container diameter and this agrees with depth correction. The penetration force increases with the age of the material
and with the depth up to a depth of 20–30 mm. The depth dependency between
the measurements reported in Fig. 5.d. This effect could be more pro­
0 and 20 mm depth follows the same characteristic between all measurements
nounced for materials forming a lubrication layer, such as mortars
and can be approximated using the shear failure surface introduced in Eq. (4).
suitable for pumping [36]. Measurements conducted with a roughened
container wall in mortars (coating with 2 mm sand grains) however
between the measurements being constant). This is even better evi­
showed no significant difference between smooth and roughened
denced in Fig. 9, where the log of penetration forces is seen to increase
container walls [57]. From a practical perspective thus a minimum
linearly with time.
container diameter appears to be the preferred solution to counteract
For single point measurements at larger depth, the force increases
difficulties of wall-slip. As shown in Fig. 5.d, the plateau penetration
with depth independently of the material age, as shown in Fig. 6.a. Over
stress does not level up over the range of diameter studied, but the di­
the probed depth of 8 cm, the ratio of the maximum measured force to
ameter's influence decreases from container diameter 9 cm onwards and
the initial measured force is generally in the range of factor 1.7 to 2,
hence this diameter is used as a compromise between sample size and
which is substantially less compared to the depth dependency observed
measurement quality for subsequent measurements in this study. The
in Fig. 5.a. This coincides with the calculated factor αd that converges to
lower limit for the container radius is equivalent to the penetration tip
1.9 for this geometry (Eq. (4)). This indicates that starting the mea­
length. Indeed, at this limit, the analytically predicted kinematic shear
surement from full insertion (as in Fig. 6) is more adequate as a char­
failure zone in Fig. 3 reaches the container wall, implying wall slip in
acterization method. In addition, and to further improve
containers smaller than this limit.
characterization quality, the depth dependency can be well captured in a
Fig. 6 shows the penetration force of a fully submerged conical
single factor (Eq. (4)) by considering the size of the shear failure zone
penetration tip for fast penetration as it is further pushed into samples of
forming during penetration (Fig. 6.b).
different ages. In the first centimetres the force increases rapidly with
While the configuration in Fig. 5 is adequate for single point mea­
depth and then levels off not quite reaching a plateau, but rather
surements, such a configuration is problematic for continuous mea­
showing a slow steady increase (Fig. 6.a). This trend is observed for all
surements with a yield stress increase over time because initially both
cement pastes independently of their age. Also, since this force increase
the contact surface and yield stress are at their lowest, which reduces the
with depth is similar for all measurements, it indicates that these mea­
sensitivity of such a test. For this reason, the setup demonstrated in Fig. 6
surements capture the same physical phenomenon with respect to depth
including a depth correction is preferred for continuous measurements.
dependency. Indeed, by using the depth correction factor of Eq. (4)
Considering the scaling with contact surface (Fig. 5), the increase
introduced in Section 3.3, we find that the corrected force becomes
with resting time (Fig. 6) and previous studies by Lootens et al. [18] and
constant for each individual test (Fig. 6.b).
Mettler et al. [29], these penetration tests are expected to be capturing
Apart from this, we note that the penetration force increases expo­
yield stress. Both the scaling with penetration tip cross-section area and
nentially with material age. This is evident from the roughly constant
the depth dependency observed in Figs. 5 and 6 are also in agreement
intervals between the lines on the log-log scale in Fig. 6 (time intervals

7
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

with soil mechanics, which also predicts that yield stress, respectively upper yield stress range, they display a single exponential trend, which
cohesion, is the main contributor to penetration force. additionally matches the one obtained from the penetration tests. For
this we note that the contribution of self-weight to compressive stress is
smaller than 2% in the weakest samples and is neglected. This suggests
4.2. Slow penetration
that the nature of the mechanism causing strength gain remains the
same and that all these tests are properly capturing it. Importantly, also
In Fig. 7.a, the penetration force of a fully submerged conical tip is
in the scope of this paper, the good correlation between the penetration
shown as it is further pushed into the sample at a very slow rate of 20
measurements and the vane tests demonstrates that the penetration test
mm/h. The force evolution is approximately exponential from the
is indeed reliably measuring a yield stress.
measurement start. A good agreement between repetitions of the
experiment in the force range 1 N–1000 N is obtained. The largest dif­
ference between repetitions is 30%. Overall, the penetration force 4.4. Penetration tests on mortars
measured with the flat tip geometry is similar to the one of the conical
tip. Using the soil mechanical approach, the force ratio between flat and Using this experimental setup, measurements conducted on mortars
conical tips should be 5.69/9 = 0.63 based on the theoretical bearing are reported in the thesis of Reiter [59] and a conference paper of Szabo
capacity factor (see Section 3.4). While results vary with time and depth, et al. [60]. A summary of these measurements conduced on mortars of
the conical tip generally maintains a higher penetration force different compositions is presented in Table 3. The bearing capacity
throughout the experiment, and the measured ratio at 200 min is very factor predicted in Section 3.4 appears to also be valid for mortars,
close to the theory, with a value of 0.76 (indicated by red arrow in Fig. 7. although a higher variability is observed for mortars. The exceptions to
b). this are compositions 5 and 6 that show higher NC, but were likely un­
The exponential increase of force and yield stress observed in all stable due to initially low yield stresses.
measurements agrees with the discontinuous measurements in Fig. 6.
Ongoing formation of hydration products after the onset of hydration 5. Discussion
lead to the resulting strength increase expected in the acceleration phase
of hydration [21,58]. The previous two sections respectively established the theoretical
basis for penetration test interpretation (Section 3) and provided mea­
surements (Section 4). Here we discuss in further detail the applicability
4.3. Yield stress from complementary methods
of penetration tests as yield stress measurement in Sections 5.1 and 5.2,
give measurement recommendations in Section 5.3.
Shear stresses measured in Fig. 8 by vane rheometry are divided into
two phases of linear increase. The first is characterized by high stiffness
up to 0.3–0.5× of the maximum shear stress, giving the vertical seg­ 5.1. Yield stress from penetration tests
ments in Fig. 8.a (and with a narrower range of deflection angle in Fig. 8.
b). It is followed by a second linear increase to the maximum shear stress In order to calculate the yield stress from penetration tests, a first
(yield stress). Beyond this, a residual resistance is measured. So, the analysis consists in calculating a linear correlation factor based on both
cement paste cannot be described by a simple linear elastic behaviour. y-axes of Fig. 9. In this approach, the penetration force is converted into
Fig. 8 also shows that the shear stress for onset of deformation, the a stress by dividing it by a shear zone area to reach a yield stress. The
yield stress and the residual shear stress all increase with material age. agreement with the other methods is obvious from Fig. 9 and the con­
The critical deformation at which the yield stress is reached, decreases in version factor between penetration force and yield stress is approxi­
the first minutes at rest and then appears constant beyond that (from 45 mately 0.2 kPa/N. The shear surface area is therefore approximately
min on based on Fig. 8, marked by arrows). 5000 mm2. It is also possible to directly calculate the bearing capacity
In Fig. 9, a summary of the previous experiments is given with factor according to geomechanics (Section 3.4) by rearranging Eq. (3)
measurements displayed with respect to time since water addition. The as:
left and right vertical axis respectively give the penetration force and the
σf Fpen 1000N
yield stress for vane and compression tests. First, we note the excellent Nc = = = = 15.9
c Atip ⋅τ0 314mm2 200 kPa
agreement between fast (evaluated at a depth of 1 mm) and slow
penetration tests regardless of hydration time. Secondly, it can be seen This experimental value from this analysis is significantly larger than
that the vane and uniaxial compressive strength measurements com­ the theoretical one (Nc = 9), but does not account for depth dependency
plement each other perfectly. Covering respectively the lower and the in the continuous penetration test. In Fig. 10, the yield stress from slow

Fig. 7. Slow penetration a) Repetitions of the same composition using a conical tip. b) Comparison of conical and flat tip. Exponential increase of penetration force
with time.

8
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

Fig. 8. a) Vane measurements at constant strain rate for cement pastes resting for increasing durations before measurement. b) Zoomed in section of low defor­
mation range.

Fig. 9. Early strength measurements on cement paste by slow penetration, fast Fig. 10. Fast penetration, vane test and uniaxial compression plotted versus
penetration, vane test and uniaxial compression. The later are converted to penetration stress. The discontinuous line is a 1:1 ratio of calculated yield stress
yield stress. Results reported versus time after water addition show an excellent of other methods compared to slow penetration, the continuous lines stand for
agreement among each other, in particular the continuous penetration captures +50%/− 33%. Master curve showing, yield stress from all methods compared to
the evolution of all others. the slow penetration test and considering the depth dependency from Eq. (4).

the fast penetration test, but yield stresses are overestimated by about
Table 3
50% (slope of 1.5 in Fig. 10).
Best fit NC* for mortars with different aggregate volume fraction ϕAggr and
While both fast and slow penetration lead, with our method, to the
superplasticizer (SP) content. Apart from M48_S1.0, the best fit NC* is within the
range of approximately 1.6× of the theoretical value. Compositions 5&6 are same yield stress, they each require for the analysis a significantly
suspected to have been unstable owing to a low initial yield stress (Hagermann different bearing factor. A variety of reasons may explain this. For the
cone > 25 cm), which would explain the substantially higher values of Nc for fast penetration tests, this discrepancy could be linked to the high
these two mixes. loading rate not providing sufficient time for relaxation to capture quasi-
Composition ϕAggr SP [%] Best fit NC
static conditions. It is also possible that wall slip on the container walls –
shown to have an effect in Fig. 5.d - affects the measurements differently
1. M20_S0.0 20% 0% 13.8
at low and high loading rate. Nevertheless, we show here that the
2. M40_S0.0 40% 0% 6.4
3. M40_S0.6 40% 0.6% 9.8 consideration of depth dependency together with the theoretical bearing
4. M48_S0.6 48% 0.6% 9.7 factor Nc give a quantitative agreement with Vane and compression
5. M48_S1.0 48% 1.0% 23.2 measurements.
6. Slip-forming 52% 0.8% 24
Aside from having to use a pre-factor to account for fast penetration
7. Layered Extrusion mix 48% 1.0% 14.4
8. UHPFRC for slip-forming [60] 23% 3.65% 14.6 results, it can be stated that continuous and discontinuous penetration,
9. Mortar for slip-forming [60] 52% 0.8% 12.2 as well as vane and compressive strength all coincide in this measure­
ment range of 1–200 kPa yield stress. It is concluded that, for cement
pastes with initial yield stresses similar to the one of this study, pene­
penetration accounting for depth correction is instead compared directly tration experiments are adequate to assess yield stress, either pointwise
to the yield stress from vane, compression and fast penetration (depth 1 or continuously. Slow penetration measurements seem to be repeatable
mm) where the theoretical bearing factor Nc = 9 is chosen. Doing this from as low as 0.2 kPa. Soil mechanics however mandates that such low
gives an excellent quantitative agreement between vane, uniaxial yield stress measurements should only be acquired at low penetration
compression tests and slow penetration tests as can be seen from the depth (see Section 3.3).
direct proportionality shown in Fig. 10. The correlation is also good with Whether depth dependency is accounted for (Nc = 9) or is neglected

9
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

(Nc = 15.9), the bearing capacity factor is significantly higher than monitoring, and this will be an interesting challenge for future research.
considering a shear surface on the penetration tip (Nc = 3.2) as previ­ The analysis shows that yield stress can be directly calculated for
ously proposed in [18]. This indicates that the yield stress is measured plastic materials. For materials beyond the range of plastic behaviour,
inside the material bulk volume (along a surface similar to the one the bearing capacity factor should be determined experimentally as it is
asserted in Fig. 3). This is an important consideration to avoid the bias of strongly influenced by internal friction angle. Thus, for any unknown
wall slip in mortars as it allows measurement in a homogenous bulk mortar matrix, preliminary measurements relating compressive strength
material section. The existence of such a shear zone is required when a and yield stress computed from penetration should be conducted to
mass flow balance is considered. Indeed, the material displaced by the verify the bearing capacity factor and define the yield stress, from which
penetration tip can only be displaced towards the free surface and a zone penetration deviates from compressive stress due to increasing internal
for this to occur is necessary. With the measurement occurring in the friction angle. Beyond this point, penetration tests will no more be
material volume, the surface of interest is linked to the tip and moves reliable to predict compression failure of the structure being printed.
into the sample at constant speed. The shear surface is therefore ex­
pected to move into the sample with the penetration tip during the
5.3. Measuring recommendations
measurement (see Fig. 1.c). This contrasts with other methods such as
compression or vane tests, in which the shear surface is static. As a
5.3.1. Measuring conditions
consequence, the material that is tested at this shear surface at a given
For measurements in conjunction with a building process, either for
point in time has been at rest since the beginning of the test and can
development or process control, the sample is required to be measured
therefore be considered as untouched. Thus, conical penetration tips
off-line. Initial yield stress and hydration kinetics should be equivalent
appear to be a good compromise for continuous penetration tests,
to the material from the process, meaning that the conditioning should
measuring yield stress. However, the size of the shear zone should in­
be as close as possible to the process conditions (temperature, resting
crease with penetration depth (Eq. (4)) as it should extend into the
times, shear rate history, etc.). Plastic mortars are sensitive to me­
material above the conical sheared zone, something that fast penetration
chanical shock and compression samples can be damaged when
tests (Fig. 6) corroborate up to a penetration depth of 20 mm.
demoulded. Also, the time of sample casting can strongly affect mea­
From the vane experiments (Cf. Fig. 8.b), it is evident that a partially
surements. Thus, material samples should be placed into the measure­
destructive deformation is applied before the yield stress is reached (two
ment container at a similar time as material deposition in the building
separate linear regimes). Thus, some deformation until yield stress is
process.
measured occurs, with a decrease of stiffness that captures partial Depending on composition, the measurement range, in which slow
structural breakdown. This characteristic strain dependency should also
penetration scales with compressive strength can vary. This is caused by
exist for penetration tests [29]. Owing to the existence of these different an internal friction angle increase as hydration occurs (compare
damage states measured by vane rheometry, it appears that assuming
Table 2). Comparative compressive strength measurements, validating
yield stress as the only characteristic stress may limit the validity of the measurement range, are thus essential for compositions, without
analytical approaches to penetration tests. In such experiments, the
previous experimental data.
shape of the sheared surface is influenced by the material in­
homogeneity introduced by the lower shear strength of previously
5.3.2. Size of sample container and penetration tip
deformed material. Thus the determination of the exact shape of the The outermost characteristic defines the minimum container diam­
shear surface is not possible.
eter. For penetration tips with apex angle smaller than 90◦ , the outer­
most characteristic has a radius equivalent to the penetration tip height
5.2. Slow penetration test for yield stress characterization H. The minimum container diameter for the outermost characteristic to
form regularly is thus 2H. This minimum diameter is also the minimum
The slow penetration test is an easy-to-use experimental setup that size for a container with a rough surface. However, a container diameter
measures structural build-up in pastes over time on a single sample. As of at least 3H is warranted to reduce the risk of wall slip on the container
such, this method can provide real-time measurements of yield stress on walls. The upper limit for the sample size is the time necessary to take a
a material at rest as is the case in digital fabrication applications. homogenous sample, which can be a limiting factor in rapidly evolving
Especially in layered extrusion and slip-forming, where concrete set-accelerated mortar used for 3D-printing. Moreover, the penetration
strength build-up at rest is used technologically to support the structure tip diameter should be at least 5× the size of largest aggregate and have
as it is produced [2]. Its measuring range of 1–200 kPa is ideal for an apex angle between 40◦ and 70◦ (constant Nc).
capturing the yield stress range that is required for such self-support.
This measuring method can predict failure, when used during fabrica­ 5.3.3. Yield stress calculation
tion, acting as a quality control method for the placed concrete, or it can Yield stress can be calculated from penetration force measurements
inform the process, giving feedback on how fast the process can build with Eq. (3), using the bearing capacity factor Nc for a conical tip and
vertically, or how much set accelerator is needed for a target vertical dividing the found yield stress by the correction factor from Eq. (4). This
building rate. To implement such a test in a practical way brings up a gives:
paradox, one of avoiding failure through prediction, but needing to
actually measure material failure in order to predict it. Thus the best τ0 =
Fpen
(10)
way the authors have found is to take a sample just before a printing Nc ⋅αd ⋅Apen
process, right before printing the first, most strength-failure sensitive
layer. Monitoring the strength development of this material closely 6. Conclusion
mirrors what is happening in the first layer and can dictate printing
speed for day-tp-day variation (for example, from temperature varia­ The studied slow penetration test is an easy-to-use experimental
tion). The major strength of this method is that rather than giving a setup that measures the structural build-up of cement pastes and mortars
single point snapshot, it gives the continuous strength development over over time on a single sample. The penetration force and yield stress are
time. The method implemented as such, however, is not a true “in-situ” experimentally found to scale linearly in the range of 1–200 kPa in
method, thus it does not give information about subsequent layers, cement pastes, the range of greatest interest for structural build-up of
which could potentially be more failure sensitive due to in-situ process concrete in digital fabrication during production.
variation or geometry. To overcome this limitation, one could correlate By modelling penetration as a geo-mechanical soil stability problem,
the method to another method better suited for in-situ process it is possible to predict the relationship of penetration force to yield

10
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

stress for plastic materials. Penetration tests in this approach were [3] N. Roussel, Rheological requirements for printable concretes, Cem. Conc. Res. 112
(2018) 76–85, https://doi.org/10.1016/j.cemconres.2018.04.005.
shown to provide cohesion measurements from a yield stress of 1 kPa
[4] G. De Schutter, K. Lesage, V. Mechtcherine, V.N. Nerella, G. Habert, I. Agusti-Juan,
upwards or from 0.1 kPa at shallow depth, but are limited to early age Vision of 3D printing with concrete — technical, economic and environmental
strength range with plastic behaviour. potentials, SI Digit. Concr. 112 (2018) 25–36, https://doi.org/10.1016/j.
Penetration tests with a conical tool moving at slow rate are adequate cemconres.2018.06.001.
[5] V. Mechtcherine, et al., Extrusion-based additive manufacturing with cement-based
to measure the structural build-up of both cement pastes and mortars materials – production steps, processes, and their underlying physics: a review,
over time on a single sample. The shear surface on which yield stress is Cem. Concr. Res. 132 (Jun. 2020), 106037, https://doi.org/10.1016/j.
measured, appears to have a homogenous microstructure and corre­ cemconres.2020.106037.
[6] E. Lloret-Fritschi, et al., From smart dynamic casting to a growing family of digital
sponds to a surface that is constantly moving with the penetration tip. casting systems, Cem. Concr. Res. 134 (Aug. 2020), 106071, https://doi.org/
This theoretical finding is substantiated by the good agreement of all 10.1016/j.cemconres.2020.106071.
characterization methods with the slow penetration measurements. [7] E. Lloret, et al., Complex concrete structures: Merging existing casting techniques
with digital fabrication, Mater. Ecol. 60 (Supplement C) (Mar. 2015) 40–49,
Measurements on cement pastes also explain the effects of setup pa­ https://doi.org/10.1016/j.cad.2014.02.011.
rameters and give a basis for recommendations for testing. [8] A. Szabo, L. Reiter, E. Lloret-Fritschi, F. Gramazio, M. Kohler, R.J. Flatt, Mastering
With an increasing internal friction angle, the ratio of penetration yield stress evolution and formwork friction for smart dynamic casting, Materials
13 (9) (2020), https://doi.org/10.3390/ma13092084.
force to compressive strength and yield stress increases. As internal [9] R.J.M. Wolfs, F.P. Bos, T.A.M. Salet, Early age mechanical behaviour of 3D printed
friction angle cannot be directly determined in a penetration test, yield concrete: numerical modelling and experimental testing, Cem. Concr. Res. 106
stress quantifications are limited to plastic materials. Measurements on (Apr. 2018) 103–116, https://doi.org/10.1016/j.cemconres.2018.02.001.
[10] N. Roussel, P. Coussot, ‘Fifty-cent rheometer’ for yield stress measurements: from
broadly different mortar compositions indicate some need for verifica­
slump to spreading flow, J. Rheol. 49 (3) (2005) 705–718, https://doi.org/
tion of the bearing capacity factor for each type of mortar used. 10.1122/1.1879041.
Finally, some practical recommendations for slow penetration [11] A. Pierre, C. Lanos, P. Estellé, Extension of spread-slump formulae for yield stress
testing of cement-based mortars were proposed, in particular with evaluation, Appl. Rheol. 23 (6) (Nov. 2013) 63849.
[12] J. Assaad, K.H. Khayat, H. Mesbah, Assessment of thixotropy of flowable and self-
respect to implementation for process monitoring. The method seems consolidating concrete, Mater. J. 100 (2) (2003) 99–107.
well-suited for prediction of process success with respect to strength- [13] J.J. Assaad, K.H. Khayat, Effect of viscosity-enhancing admixtures on formwork
based failure if implemented just before printing, and allows for pressure and thixotropy of self-consolidating concrete, Mater. J. 103 (4) (2006)
280–287.
adjustment based on that initial material behaviour. To truly monitor [14] A. Perrot, D. Rangeard, A. Pierre, Structural built-up of cement-based materials
the process in-situ, however, it should be well connected to a true in-situ used for 3D-printing extrusion techniques, Mater. Struct. (Feb. 2015) 1–8, https://
process monitoring method in a physically meaningful way, an inter­ doi.org/10.1617/s11527-015-0571-0.
[15] P. Billberg, Form Pressure Generated by Self-compacting Concrete: Influence of
esting challenge for future research. Thixotropy And Structural Behaviour at Rest, KTH, Civil and Architectural
Engineering, 2006.
[16] H.A. Barnes, Q.D. Nguyen, Rotating vane rheometry — a review, J. Non-Newton.
CRediT authorship contribution statement
Fluid Mech. 98 (1) (Mar. 2001) 1–14, https://doi.org/10.1016/S0377-0257(01)
00095-7.
Lex Reiter: Conceptualization, Methodology, Investigation, Valida­ [17] A.F. Omran, S. Naji, K.H. Khayat, Portable vane test to assess structural buildup at
tion, Formal Analysis, Data Curation and Visualization, Writing Original rest of self-consolidating concrete, ACI Mater. J. 108 (6) (2011).
[18] D. Lootens, P. Jousset, L. Martinie, N. Roussel, R.J. Flatt, Yield stress during setting
Draft. of cement pastes from penetration tests, Cem. Concr. Res. 39 (5) (May 2009)
Timothy Wangler: Supervision, Review and Editing of Original and 401–408, https://doi.org/10.1016/j.cemconres.2009.01.012.
subsequent drafts. [19] A. Pierre, A. Perrot, A. Histace, S. Gharsalli, E.-H. Kadri, A study on the limitations
of a vane rheometer for mineral suspensions using image processing, Rheol. Acta
Nicolas Roussel: Supervision, Review and Editing of Original and 56 (4) (Apr. 2017) 351–367, https://doi.org/10.1007/s00397-017-0993-4.
subsequent drafts. [20] N. Roussel, G. Ovarlez, S. Garrault, C. Brumaud, The origins of thixotropy of fresh
Robert J. Flatt: Main Supervision, Writing – Review and Editing of cement pastes, Cem. Concr. Res. 42 (1) (Jan. 2012) 148–157, https://doi.org/
10.1016/j.cemconres.2011.09.004.
Original and subsequent drafts, Project administration, Funding [21] S.P. Jiang, J.C. Mutin, A. Nonat, Studies on mechanism and physico-chemical
acquisition. parameters at the origin of the cement setting. I. The fundamental processes
involved during the cement setting, Cem. Concr. Res. 25 (4) (May 1995) 779–789,
https://doi.org/10.1016/0008-8846(95)00068-N.
[22] S. Mantellato, Flow Loss in Superplasticized Cement Pastes, ETH Zürich, 2017.
Declaration of competing interest [23] R.J.M. Wolfs, F.P. Bos, T.A.M. Salet, Correlation between destructive compression
tests and non-destructive ultrasonic measurements on early age 3D printed
The authors have no affiliation with any organization with a direct or concrete, Constr. Build. Mater. 181 (Aug. 2018) 447–454, https://doi.org/
10.1016/j.conbuildmat.2018.06.060.
indirect financial interest in the subject matter discussed in the
[24] D. Lootens, R.J. Flatt, P. Juilland, Pragmatic (and) scientific characterization of the
manuscript. early ages properties of cementitious materials, in: Creep, Shrinkage and Durability
Mechanics of Concrete and Concrete Structures, Two Volume Set: Proceedings of
the CONCREEP 8 conference held in Ise-Shima, Japan, 30 September-2 October
Acknowledgements 2008 1, 2008, p. 305.
[25] L. Nachbaur, J.C. Mutin, A. Nonat, L. Choplin, Dynamic mode rheology of cement
This work was developed within the National Centre for Competence and tricalcium silicate pastes from mixing to setting, Cem. Concr. Res. 31 (2) (Feb.
2001) 183–192, https://doi.org/10.1016/S0008-8846(00)00464-6.
in Research (NCCR) Digital Fabrication, Agreement #51NF40-141853. [26] L. Reiter, T. Wangler, N. Roussel, R.J. Flatt, Distinguishing Flocculation And
Hydration Effects on the Thixotropy of Cement Pastes, 2016.
[27] M.A. Schultz, L.J. Struble, Use of oscillatory shear to study flow behavior of fresh
Appendix A. Supplementary data
cement paste, Cem. Concr. Res. 23 (2) (Mar. 1993) 273–282, https://doi.org/
10.1016/0008-8846(93)90092-N.
Supplementary data to this article can be found online at https://doi. [28] S. Amziane, A. Perrot, T. Lecompte, A novel settling and structural build-up
org/10.1016/j.cemconres.2022.106802. measurement method, Meas. Sci. Technol. 19 (10) (Aug. 2008), 105702, https://
doi.org/10.1088/0957-0233/19/10/105702.
[29] L.K. Mettler, F.K. Wittel, R.J. Flatt, H.J. Herrmann, Evolution of strength and
References failure of SCC during early hydration, Cem. Concr. Res. 89 (Nov. 2016) 288–296,
https://doi.org/10.1016/j.cemconres.2016.09.004.
[30] H.A. Barnes, A review of the slip (wall depletion) of polymer solutions, emulsions
[1] R.A. Buswell, W.R.L. de Silva, S.Z. Jones, J. Dirrenberger, 3D printing using
and particle suspensions in viscometers: its cause, character, and cure, J. Non-
concrete extrusion: a roadmap for research, Cem. Concr. Res. 112 (2018) 37–49,
Newton. Fluid Mech. 56 (3) (Mar. 1995) 221–251, https://doi.org/10.1016/0377-
https://doi.org/10.1016/j.cemconres.2018.05.006.
0257(94)01282-M.
[2] L. Reiter, T. Wangler, N. Roussel, R.J. Flatt, The role of early age structural build-
[31] T. Wangler, et al., Digital concrete: opportunities and challenges, RILEM Tech. Lett.
up in digital fabrication with concrete, Cem. Concr. Res. 112 (2018) 86–95,
1 (2016) 2016, https://doi.org/10.21809/rilemtechlett.2016.16.
https://doi.org/10.1016/j.cemconres.2018.05.011.

11
L. Reiter et al. Cement and Concrete Research 157 (2022) 106802

[32] G. Ovarlez, 2 - introduction to the rheometry of complex suspensions, in: [46] H.-Y. Fang, Foundation Engineering Handbook, Springer Science & Business
N. Roussel (Ed.), Understanding the Rheology of Concrete, Woodhead Publishing, Media, 2013.
2012, pp. 23–62, https://doi.org/10.1533/9780857095282.1.23. [47] G.T. Houlsby, C.M. Martin, Undrained bearing capacity factors for conical footings
[33] D. Taylor, Fundamentals of Soil Mechanics, Chapman And Hall, Limited, New on clay, Géotechnique 53 (5) (2003).
York, 1948. [48] T. Koumoto G. T. Houlsby, “Theory and practice of the fall cone test,” p. 12.
[34] K. Terzaghi, Theoretical Soil Mechanics, Chapman And Hall, Limited, London, [49] G.T. Houlsby, Theoretical analysis of the fall cone test, Géotechnique 32 (2) (1982)
1951. 111–118, https://doi.org/10.1680/geot.1982.32.2.111.
[35] J. Yammine, M. Chaouche, M. Guerinet, M. Moranville, N. Roussel, From ordinary [50] S. Hansbo, A New Approach to the Determination of the Shear Strength of Clay by
rhelogy concrete to self compacting concrete: a transition between frictional and the Fall-cone Test, Royal Swedish Geotechnical Institute, 1957.
hydrodynamic interactions, Cem. Concr. Res. 38 (7) (Jul. 2008) 890–896, https:// [51] Juang C. Hsein, Yuan Haiming, Lee Der-Her, Lin Ping-Sien, Simplified cone
doi.org/10.1016/j.cemconres.2008.03.011. penetration test-based method for evaluating liquefaction resistance of soils,
[36] D. Feys, 13 - understanding the pumping of conventional vibrated and self- J. Geotech. Geoenvironmental Eng. 129 (1) (Jan. 2003) 66–80, https://doi.org/
compacting concrete, in: N. Roussel (Ed.), Understanding the Rheology of 10.1061/(ASCE)1090-0241(2003)129:1(66).
Concrete, Woodhead Publishing, 2012, pp. 331–353, https://doi.org/10.1533/ [52] P.K. Robertson, Soil classification using the cone penetration test, Can. Geotech. J.
9780857095282.3.331. 27 (1) (Feb. 1990) 151–158, https://doi.org/10.1139/t90-014.
[37] P. Coussot, C. Ancey, Rheophysical classification of concentrated suspensions and [53] O. Wuensch, Experimentelle Bestimmung Binghamscher Stoffparameter, Rheol.
granular pastes, Phys. Rev. E 59 (4) (Apr. 1999) 4445–4457, https://doi.org/ Acta 29 (2) (1990) 163–169.
10.1103/PhysRevE.59.4445. [54] U. Pott, D. Stephan, Penetration test as a fast method to determine yield stress and
[38] P. Coussot, Introduction to the rheology of complex fluids, in: Understanding the structural build-up for 3D printing of cementitious materials, Cem. Concr. Compos.
Rheology of Concrete, Woodhead Publishing, 2012 [Online]. Available: https:// 121 (Aug. 2021), 104066, https://doi.org/10.1016/j.cemconcomp.2021.104066.
books.google.ch/books?id=1AbKygAACAAJ. [55] A.S. Vesic, Bearing capacity of shallow foundations, in: Foundation Engineering
[39] A. Yahia, S. Mantellato, R.J. Flatt, 7 - concrete rheology: a basis for understanding Handbook, 1975. Ed. By.
chemical admixtures, in: Science And Technology of Concrete Admixtures, [56] A.W. Skempton, The bearing capacity of clays, Proc Build. Res. Congr. 1 (1951)
Woodhead Publishing, 2016, pp. 97–127, https://doi.org/10.1016/B978-0-08- 180–189.
100693-1.00007-2. [57] L. Reiter, T. Wangler, A. Anton, R.J. Flatt, Setting on demand for digital concrete –
[40] A. Atterberg, Die Plastizität der Tone (The plasticity of the clays), Int. Mitt. principles, measurements, chemistry, validation, Cem. Concr. Res. 132 (2020),
Bodenkd. 1 (1911) 10–43. 106047, https://doi.org/10.1016/j.cemconres.2020.106047.
[41] T. Lecompte, A. Perrot, V. Picandet, H. Bellegou, S. Amziane, Cement-based mixes: [58] A. Perrot, D. Rangeard, A. Pierre, Structural built-up of cement-based materials
shearing properties and pore pressure, Cem. Concr. Res. 42 (1) (Jan. 2012) used for 3D-printing extrusion techniques, Mater. Struct. (Feb. 2015) 1–8, https://
139–147, https://doi.org/10.1016/j.cemconres.2011.09.007. doi.org/10.1617/s11527-015-0571-0.
[42] K.T. Fosså, Slipforming of Vertical Concrete Structures. Friction Between Concrete [59] L. Reiter, Structural Build-up for Digital Fabrication With Concrete - Materials,
And Slipform Panel, 2001. Methods And Processes, ETH Zurich, 2019, https://doi.org/10.3929/ethz-b-
[43] T. Craipeau, T. Lecompte, F. Toussaint, A. Perrot, in: Evolution of Concrete/ 000456199.
Formwork Interface in Slipforming Process, 2019, pp. 12–23. Cham. [60] A. Szabo, L. Reiter, E. Lloret-Fritschi, F. Gramazio, M. Kohler, R.J. Flatt, Processing
[44] L. Prandtl, Über die härte plastischer körper, Nachrichten Von Ges. Wiss. Zu Gött. of set on demand solutions for digital fabrication in architecture, in: RheoCon2,
Math.-Phys. Kl. 1920 (1920) 74–85. Dresden vol. Rheology and Processing of Construction Materials (RheoCon2),
[45] J.B. Hansen, A Revised And Extended Formula for Bearing Capacity, 1970. 2019.

12