Creep behaviour and microstructure of hardened cement
pastes
Citation for published version (APA):
Willems, H. H. (1985). Creep behaviour and microstructure of hardened cement pastes. Technische Hogeschool
Eindhoven. https://doi.org/10.6100/IR240257
DOI:
10.6100/IR240257
Document status and date:
Published: 01/01/1985
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.
CREEP BEHAVIOUR AND MICROSTRUCTIJRE
OF HARDENED CEMENT PASTES
Proefschrift
ter verkrljglng van de graad van doctor in de technische wetenschappen aan de Technische Hogeschool Eindhoven, op gezag van de rector magnificus, Prof. Dr. F.N. Hooge, voor sen commissle aangewezen door hat College van Dekanen in het openbaar te verdedlgen op dinsdog 10 December 1985 te 16.00
uur
door
HUBERTUS HENRICUS WILLEMS
Dit proefschrift is goedgekeurd door de
promotoren:
Prof.
Ir. P.e.
Kreijger
en
Prof. Dr. JA Poulis
CONTENTS
1. SuMKARY
2. lNTRODUCTION
3. EXPERIMENtAL PROCEDURES AND APPARATUS
3.1. specimens
3.1.1. Shap~ of the specimens
3.1.2. Cemflnt 3.1.3. Water-cement ratio 3.1.4. Specimen moulds 3.1.5. specimen preparation 3 9 ;I 9 10 10 10 14 3.1.6. compressive strength of th~ specime~s 15
3,2. Sh~l~kQge and creep experiments 15
3.2.1. Apparatus 15
3.2.2. Experimental procedures 20
3.3. Water vapour sorption mea$urewents 20
3.3.1. Apparatus 20
3.3.2. EIpe~ime~tal Procedures 23
),4. silicate polymerisation Analysis 24
3.4.l. Silylation Technique 25
3.4.2. GaS-liquid chromatography 25
3.4.3. Non-volatile fraction of TKS-derivatives 26
4. EXPERIMENTAL RESULTS
4.1. Compressive strength of the sp~cimen$
4.2. Shrinkage and c~eep experiments
4.3. Water vapour sorption experiments 4.4. Silicate polymerisation analysis
27 27 28 31 36 1
5. ANALYSIS OF EXPERIMENTAL RESULTS
~.l. Compressive strength of the specimens
5.2. Shrinkage and creep experiments 5.3. Water vapour sorption experiments
5.3.1. Interpretation of isotherms 5.3.2. Di~cus~ion of experimenla1 resulls 5.4. silicate PDlyme~isation Analysis
6. CONCLUSIONS AND FINAL REMARKS
REFERENCKS SAMENVATTING ACKNOWLEDGEMENT CURRICULUM VITAE TABLES EXPERIMENTAL OUTLINE 2 39 39 40 44 44 49 56 59 62 65 66 67 66 115
chapte", 1
1. Summary.
Creep is a mechanical property, that has to be taken into
account in the design of concrete constructions. Earlier stu-dies showed, tohat cl"'eep of concl"'ete, among other- thln&s, de-pends on the relative humidity of the environment and on the type of cement, that is used. The origin of creep lies in the hydrated cement paste, which is present between the aggregates. In this thesis a study is des(:cibed. dealing with the relation between the creep behaviour and the microstructur-e of hardened
portland and portland-bla~tfurnace cement paste~, At several
relati ve hwrddi ties shl"'inkage and cl"'eep expal"'iments hava been per-formed with t.hin-waned hollow cylindrical specimens. Next. the pore structure and the silicate structure of the specimens has been studied. The pore str-uctUl"'e was determined by means of physical sorption of water with equipment, specially developed for this purpose. The silicate st.ructur-e has been studied by means of tr-lmethylsilyl-derivatisation of t.he silicates pr-esent in the hydrated cement pastes, followad by gas-liquid chromato-graphy.
Clear differ-ences exist between both the pore and the silicate st.ructure of hydr-ated portland and por-tland-blastfurnace cement pestes. The influence of the relative hWllidity and the pre-traa.tment of tile spaeimans (shl"'int.;ag<i! Or creep) On tile
PllCrO-structure of the hydrated cement pastes, app~ars to be var-y
small, No univoeal relation could be found between the ereep behaviour and the mierostructure of the specimens investigated.
The intera~tion between the hydrated cement paste and the water
present inside of it, needs fur-ther study.
Chapter <1
Z, Introduction.
Conc~ete is the moat widely used buildins m~teri~l in the
wOl:"ld. !lence the~e is a s~eat interest in conl:l'olling and
im-p~ovin& il:. mechanical properties. Shrinka~e and cl'eep are
vital material propertie~. that have to be taken into account
in the desisn of I'einfol'ced, prest~essed and post-tensioned
concrete conBtructions.
C .. eep is the non~clastic time-dependent defol'mation of a body
due to ~n applied ~\:I'ess. I t may be subdivided into basic and
dryin~ Cl'eep. Basic c~e@p occurs. when ~ body is in hygrical
equilib~ium with its environment. Dl'ying creep is the
additio-nal creep, that occurs when R specimen lOGes water,
Shdnk~~e is the diminution of specimen dilllensions without the
pl'esence of an extern~l load.
If ~ loaded specimen simultaneously loses wnte .. , both shrinkage
~nd creep will OCCUI'. Usually creep is re~~rded to be
indepen-dent of sb~lnkage. In that eag", the dlffel'ence between the
total deformation of a lo~ded specimen and the shrlnkae;e of an
unloaded specimen, is assigned to cl"eep, Actually creep is not. entirely independent of shrinkage.
When a specimen is loaded, an inunediate elastic deformation
will OCCUI', followed by a slow time-dependent defo~mation. When
after some time the lo~d i. removed, an immediate elastic
reco-ve .. y will take place, followed by a slow non-elastic I'ecove~y.
usually the elastic recovery of concrete i~ smaller than the
elasti~ deformation at the sta~t of an experiment. This is
cau-sed by an inc.ease in the modulus of eluticHy with time, due to the hydl'ation l'eacHons. Usually part of the creep appears
to be il'recovel'able. This ~esults in a permanent deformation of
a specimen after unloadins.
the val'ious defo.mations mentioned above are illustrated in
Fi S, <" J.. FOI' I:he ease of presentation all deformations are
time _ time _ a. b. basic creep deformation basic creep shrinkar;e defol"mation recovery non-elastic recovery pel."l1lanent deformation t u time ~ Deformation of an unloaded, dryin& specimen. Deformations of a loaded specimen in hyr;rical equilibrium with its environment"
Deformations of a loaded, dryinr; specimen. Deformations of a specimen unloaded at t=t . u
Deformations in shrin~age and creep processes.
Shrinkage may occur due to carbonat.ion, cooling Or" loss of
water due to hydration andlor evaporation. 1t results in
internal stresses, which may lead to the formation of cracks.
thl!se are detrimental to thl! stabili ty, lifetime l). .... d a".th"Ucal vQlue of concrete co .... at.:-ud:ions. Internal str0SSel>
ffiay be reduced partly Ot" corn~let"ly due to creep.
C.:-eep ffiay Qlso Qdversely affect concrete constructions:
In 1ar1\e-·;;"an constructions inadmi 5S i ble deflections may
occur due to creep, resulti ng in crack fo!:"mat io... In the case of roof constructions thi$ ml).Y .:-esult in leakages, which on their turn may cause inconve .... iences Qnd/or damages.
In lQrge· scale construction~ ~nt".:-nQl stresses originate
from the heat evolution caused by the hydration reaction',"
~$ at!:"eady ffientioned they may (partly) be reduced by creep.
When, however, in the course of time the temperature inside the constr;-Ilction fQlls, tensile stresses will occur- du" to thermQl shrinkage. Since the ten$ i Ie strength of concrete 1 s smaller than its compressio .... st.:-engt.h, crack form!!.tiol) may occur, which unfavourably affects the construction·s durability.
In prestressed coneret.e, creep is always an undesirllble
phenomenon- Due to creep the tension in the steel ba~G will
decrease, r-esultinl:; i ... mQller beQring capacities of the constt"uctions.
The safety factor in buckling ·of excentdcQlly compressed
columns ~s ~educed considerably by creep. Especially in the
design of curved str"ctures under cQmp.:-ession this has to be considered.
Although the creep behaviour of concrete is influenced by the
aggregates, its origine lies i .... t.he hydrated cement pa~te,
which is present between the !!.t,t,t"el'illtes. Hydrated cement pl).>:lte
is a highly porou~ and heterogeneous mat.erilll. It develops as a
result of the chemicd reactions, thQI: take place when cement is mixed with water" Roughly speaking it consists of a skeleton
of hyd~QI: ion produc t s wi th vo ids i I) between. In [1] Ii
sur-vey of the various chemical reactio .... s occurring during the
,he IIIILin hydr;-ation pr-oduc:ts ar-e the calc iUID ~il i<;ILte hydrILt;e~, They usually appear in the form nf laminated fnils. This str:-uc-ture is similar to the one found in the mineral tnbermnrite and in sOllie cllL¥ minerab. BecaUlie of the emIL1I dimensions of the hyd!."atinn p!."nducts the term "CSH-gel" is oHen used (N. B. In cement chemistr:-y the initials C, S and II stand for- cao, Sia
Z ILnd H
20!).
Dep"ndint on the r-elal:ive hWllidity, "ar;-yin~ amOunts of water
may be present between the layers in the gel particles (inter-layer water). Further, wate!." may be present on the sur-face of
the ~el pILrticles (physicILlly bound water), in the spaces
between the gel par:-ticles (gel water) and jn the voids between cr:-ystal structur:-es (capillar:-y water).
1n the course of time nUlDerOUB theories cnnce!."ning the creep
behaviour:- of concr-ete have been developed. In [2] anti [3)
a survey of the various theor-ies is pr-e$ented, In some of the theories the presence and/nr the displacement of water in the
pore system plays an impor-tant par-I;. So the relat i ve h\lffiidi ty
of the envir-onment will be an impor-tant par~eter for the creep
behaviour-. The r:-igidil:y of the hydr-ated <;emen!: paste skeleton is another deter:-mining factor- fol." the creep behaviour. In this connection the type of cement used, will be an important
factor:-. The r-ellults obtained by Cornelissen [4] also point
to this. He found that the creep of ~oncr-ete made with
portland-blastfurnace cement was small~~ than the cr-eep of
concrete m~de with portllLnd cement.
The -woJ:"k described in this thesis, deals with a stUdy o( the rellLtion between the creep behaviour:- and the microatr-uctuce of
hydrated portland and portland-blastfur:-nace cement pastes.
Herewith the approach of Mindess Elt d. [~.6,7.8,9] h~s been
partly adopted.
Jl"ollowin& to shrinka!;e and creep experiments at sever-al r:-ela-tive humidIties, the por-e str-uctur;-e of thin-walled, hollow cy-lintlriclLl specimens of hydrated por:-tland and portland-blast-fur-nace cement pa£'tes has been studied by means of phys ical sorption of water vapour. These eIperiments have been carried ou!: in a pure water vapour atmosphere with equipment, specIally
developed for this purpose. The degree of polymerisation of the
silieates, a measu~e fo~ the ~i~idity of the cement paste ske·
leton, has teon examined l;Iy me'l.n$ Qf
tdmethyhilyl-deriVII.ti-sation a~cot"din& 1:0 Tamils. Sa~kat" and Roy [10,11]. followed
by ~as-liquid cht"omatog~aphy of the de~ivatlves.
On the pages 115, 116 and 117 the ellpe~imen~al oulline and a
suney of the "adous specimens and thei!." testin& conditions
at"E! presented.
Chapter 3.
Experimental Procedures and Apparatus
.
3
.
1. Specimens.
3.1.1. Shape of the specimens.
The shrinkage and creep experiments were performed with
thin-walled hollow cylindrical specimens of hardened cement paste.
The specimen dimensions were: length 100
mm,
outer diameter
20
mm
and wall thickness 0.5
mm.
(see Fig
.
3.1).
This type of specimen has been chosen in order to achieve rapid
hygrical equilibrium between the specimen and the environment
to which it was exposed.
This is important in studies of
shrinkage and creep phenomena, since a non-uniform distribution
of water over the cross-section of a specimen may cause
devia-tions of the equilibrium behaviour to arise because of internal
stresses.
The same type of specimens had been used before by [5] and
[12].
Fil5. 3.1.
Thin-wall cylindrical specimen for shrinkage and
creep experiments
.
3.1. 2. c~m~nt. In thi s po~tland
(CEHIJ)
inv~stlgation two types of cement have been used: a
cement (ENCI) and a po~tland-blastfu~nace cement
with p~acticQlly identical Blaine su~face a~eQS and
g~anulomet~y (see Table 3.11.
This was achieved by choosilll';: a somewhat coa~se B-type
port-land-blastfut"nace cement and by blendin~ B- and C-type po~tland
cements, until the Blaine sut"face a~eas of the two cements were
approximately the SMl0. He~ewi th it was achieved, that both
cements we~e practically identical with tefe("e[II~'" to the
volu-met~ic composition. The cements Wete stated in wax-sealed tins.
The content of one tin (1 kg) WaS enough for a complete series
of meai;uI'ements at th~ee diffe~ent relative humidities,
inclu-ding the mechanical st~ength measu~ements of the cement.
The mechanical st~ength of the cements afte~ 28 days was tested
acco~dinl\ to NEt! 3072 and could be considel"ed as no~mal (see
Table 3 .• ).
The po["tland. cement will be indicated fUI'theI' on as PC and the
pOI'tland-bla~tfurnace cement as PBC.
3.1"3. Wate["-cement ratio.
Tn view of the ~equi~ements which had to be set for the
work-ability of the cement pastes. a water/cement ~atio of 0.3:;0
(m/m) was chosen fo~ ths pOl"tland cement pastes. Because the
density of the portland-blastfut"nace cement was about 5~ lower
than that of the po~tland cement (see Table 3.1) a w./c. ~atio
of 0.369 (m/m) was used in the case of the porUand~blast furnace cement pastes. In this way the w./e. rati05 (vol./vol.)
and so the initially introd~ced porevol~me5, we["e the same for
both cement paates.
3.1.4. Specimen moulds.
Startin~ f~om [121. two mo~lds were me-de of which the oute~
pa~ts cons isted of two half shells. wi th a n\lll\ber of pinG and
an .elastic clip. both halves were pl"essed together.
since the spec imens often c~acked when the oute~ parts of the
mould weI'e sepatated, the mould design was imp~oved.
UHima.t;ely this lead 1;(1 the design of a mould a diagram ol' which i~ p,esented in Fig. 3.2.
r
nut (stainle~~steel)
c
top cove~ (~tainleS~steel)
K I:'ubbel:' O-ring
A outer jacket (PVC)
B
innerjacket
(tel'loo)J specimen
F inner mould
(El'ts.lyte)
E guiding l'od (stain~
less steel)
D bottom cover
(stain-lea!! steel)
H outlets
~i&.3.2. Hould fol' p~epa~ins
specimens.
The various parts comprisint; the mould ~",e ,,-Iso shown in the
Fit;s. 3.3 ~nd 3.4. Most parts of the mould were m~de on a lathe.
The mOllld consists of a lS'; rom lont; outer j~cket (A) made of
P.V.C. (Trovidur), its outer di'l-\1leter being 50 rom and its inner
dlamet.er 23.0 rom. Into the outer jacket. is pressed a closely
fitting teflon cylinder (0) with a wall thickness of 1.5 mm. At
one place. this teflon ~ylinder is cut throut;h lengthwise (see
Fit;. 3.4). The mould is enclosed with stainless steel covers
(C,D). In t.he centre of the bottom cover, a stainless steel rod
(E) (06.0 mm) is mounted; its purpose is to guide and
"el1--tre the inne~ mould (F), when this is brought into position.
The i "net:" mould (0 19.0 x 0 8.0 mm) is made of Ertalyte
(P.E.T.P.), which is a synthetic m~tedd with a great stiff·
ness, rigidity, almost no moistllre absorption and a low thermal expansion coefficient. As a result of these properties, it has turned out to be extremely suitable for this purpose.
In the bottom cover a small chamber (C) is also present _ This
ohamb"" .erves to collect the surplus cement paste, that is
spread upon thc inside of the inner;' jacket and to carry it off
via the outlets (H). when the inner mould is pushed downward
ove. the center rod dudll& pre.;>aratlon of the specimens. Imme-· diately after closure of the mOUld, the cement paste could be
rinsed out of the chambe~ by inje~ting Gome water into the
out-lets.
The varIous p~rt5 of the mould could be pressed firmly toðer
by tithtening of the nut (II. To ensure {I, ~omplete isolation of
the ap.:.cimen (.l), all sliding p~rts of the mould that did not
come into cont~ot with the cement paste were greased. TO
faci-litate the removal of the top and bottom ~Ovet:"s when the mould
was f111ed with cement paste. the rubber o-rings (K) were fit-ted; for this purpose, two aeration holes were also made in the top cover.
Fig. 3.3.
Various parts of
specimen mould.
Fig. 3.4.
PVC outer jacket
with teflon inner
jacket (cut throush
lensthwise) .
3rl+~. gp~cimen preparation.
For the vreparation of the specimens boiled-out,
double-distil-led water was used. The w./c. r~tio was O.3~O for the portl~nd
cement and 0.369 for the po~tland-blastfu~nace cement (see
section 3.1.3.).
Firstly, the cement and the watel:' we.;-o:> mixed by hand in a beaker for one minute. The paste was then allowed to stand for
another minute. Next, the paste was mixed a~ain for one m~nute.
For homogeneity pu~poses. always an exce$S of cement pa~te was
used to fill the moulds.
Wi th a spatl,lha. the cement vaste was sp~ead on the ins ide of
the teflon jacko:>t, The outer part of the mould was then placed in the bottom cover and the inner mould carefully slid down
OVM' the guiding rod. Sy covering the outlets in the bottom
unt i 1 th" cement paste squeezed out of the top of the mould, the ait' could be t'emoved completely from inside the mould. In
this way, it was poss~bte to produce Bpecimens without air
bubbles or holes.
The top CoVer was then placed into position and the nut (I) was fastened; finally. the chamber in the bottom (lover was rin,ed
wi th water: and the mouldS Wete vibrated for 30 seconds on a
small vibt'ating table.
Then the mould. were totated fot 24 hours in a vertical plalle around an axis perpendicular to their length-axis, in order to maintain a IIniform water ,!jl1tdblltion over the specimens. To
pt'event carbonation, the l:'otation took place in It plastic ba~
filled with nitrogen. The temperature was kept at 25°C.
Aner 24 hout's, the specimens were removed from the moulds.
80th covet's were removed. "nd 300 ",1 of methanol, cooled wi th
dry ice <initial temper:atul:'e -60"G). was poured five times
through the openi n~ in the centro:> of the inner mould. (Wi th
thermocouples, previously it Was found, that the temperatut'e of
the specimen did not fall below 4"C then). with a special
devi~e, the whole of inner mould, specimen "nd teflon jacket
could be pulled out of the outer ja¢ket. The teflon jacket could then be stripped off easily. Finally, the inlier mQuld
was p~lled out of the specimen. The specimens we~e then sto~ed
sepa~ately at 25"C in closed test tubes filled with lime water.
AI: the age of 25 days, the specimens were trimmed wi th a dis.-mond saw to a length of 100 rom and put back again into the test
t~bes. ~inally, at the age of 28 days, the specimens were
placed in the shrinkage and c~eep cells. preliminary
investiga-tions [1] had shown that the degree of hydration at this
time was about 0.7. Since the hyd~ation rate was very low at
the time of the expe~ iments, the inc~ease in the degree of
hyd~atiQn wa~ negligible.
3.1.6. Comp~essive st~ength of the specimens.
The comp~essive strength after 28 days waS dete~mined afte~ the
specimens had been d~ied in the cells fo~ 24 hours at a R.K. of
$:)1.. 1n thi sway, the spec imens we~e cons ide~ed to have an
intermediate state of dryness, with regard tQ the future
shrinkage and creep experiments.
The results of the compressive st~ength measu~ements were used
tQ dete~mine the magnitude Qf the load, that was applied to the
specimens during the c~eep expe~iments.
The expe~iments we~e pe~fQrmed with a. Monsanto Tensomete~.
type
w,
equipped with a p~ecision compression cage, adapted fo~this pU~PQse. The specimens were fitted between two di scs
1n-~ide the cage. The fQ~ce was a.pplied to the specimens via balls
centrally positioned outside the di.scs. In this way the fQ~ce
wa.s exerted proecisely along the length-axis of the £Ipecimen.
The applied st~e5S and. the cor~espond1ng length changes of the
~pecimens we~e measu~ed wit.h LVDT's and ~eco~ded with an It-y
roecorder.
3.2. Sh~inkage and croeep ell:pe~iments.
3.2.1. Apparoatus.
The $ht"inkage and c~eep experiment.s weroe perfo~m@d wi th lohe
apparatus shQWO In Fig. 3.5.
It consisted Qf a clQsed syst.em incorporat.ing two shrinkage and
creep cells. Nit~Qgen was used as t.est at.mosphere. The flow of
tas th~ou~h the cells was 50 I/h. In this way. the total volume of tas incorporated in the system was cil"culated every minute and the linear gas velodty in the cells was approximately 4 cm/s.
Fir;. 3.5.
saturated
sat! solution '---r--~---' 25.
o· (
Diagram of the shrinkage and creep appal"atus.
the relative humidity of the test atmosphere wag established
and kept constant by meanS of saturated salt ~olutions, thermo~
stated at 25.0~C. Accordint to [13J. the followin~ salt~
were ... sed,
CH
3COOK (R.H.22~), H~(N03)2.6H2o (R.H.S3~) and NaCl (R.H.75~).
A r-elative humidity <;>f 100~ was established by passing the
nitroten thr-ou~h distilled water-.
To determine the R.H. and the temperature inside the system, a
hai r-hygromete!' and a Pl-IOO res i stance thermometel" were also
inclUded in the system.
The equipment was housed in a climatic ,.oom in which the
temperature was kept constant at 2S·C.
•
~~~----~ Specimen
Atmosphere out
~...J',4~~- L.V.D.T.
AtmQ£'pbere in
Fig. 3.7.
18
Top view of the shrinkage and creep apparatus.
Fig. 3.8.
Creep cell.
In Fig. 3.6, a schematic diagl:'$J\\ of a cl"eep cell is shown. The cell consists of two plexiglass tubes (030 x 040 mm)
that are sep&r&ted hy &n t!.luminum par-t i !:ion" The lower compart·· ment is closed by means of an aluminum cover. In the centre of
the separating wall a LVDT is mounted (sangamo DC 2.5 mm;
l~near~ty better than O.l~).
The specimen is placed i~ a stainless steel disc in the
sepal:'a-ting wall. An &luminum cap, also equipped with a stainless steel disc. is placed on top of the specimen.
In the centl:'e of the c 8.1' i $ mounted a threaded rod wi th a
st&inless stflel plug, which I'es\:s on the top of the LVDT. In this way, the changes in lenght of a specimen could be measured
with an accuracy of ± 1 ~.
The rim of the cap (made from a tube) is placed in a circular l"esel"voir filled with oil (see Fig. 3.8)" Thi$ enables vertical movements of the cap. while the cells remain isolated fl:'om tile outsid'L All Qthel:' pal"ts of the cell were eithel' sealed wi til silicone gum Ot' with vacuum grease.
The gas stream was lead fl:'Om the lower comp&rtment via ducts in the pal"tition and in the cap, to the oubdde of the specimen,
befo~e carrying it away from the cells. It the gas flow was In
the o?po!>i te dil"ection, the specimens appear:ed to Cl"ElCK during
the first dry~ng_
The creep specimens wet'e loaded with a stress/strength ratio of
abo\lt 0.1 (see section 5.1). The load (1116 N) was applied by
means of a level" and 8. weight of approllmately 4.8 kg (see Fig.
3.7). In the de~ign of the apparatus the Weight
0'
the cap andthe level:' have aho been t&Ken into accounL By mea.ns of the
~wo balls at the top and the bottom Of the cell. the load was
applied a]l:i.a11y to the specimen. ACtordin& to [14] and
[IS], bucklint of the specimen would nol: occur due to the
applied stress.
The Shl"inka&e cell is almost Identical to the cl"eep cell; only
the bottom ~oVer: is attached dirQctly to the bue plate and,
instead of a ball, there is II. pin in I:lIe UP?el" cap (see Fig.
3"7)" The pin is locked in a guide b\lSh in order to prevent the ca.p tilt ing. The load exerted on the shdnkage specimen by
the weitht of the c~p (1.5 N) can be nel!Olected in comparison
with the st~ess applied to the creep specimen (196 Nl.
J.2.2. Experimental PrQced~res.
At an a~e of 28 days, the specimens were placed into the
;;hrink~~() and creep cells. previously. these cells had been
washed with nitro~en for lS min~teB. After pl~cint the top caps
into position, the initi~l len~th of the specimens WaS me~su·
red. When ~ll the specimens h~d been placed into the cells, the
data recording system WaS st~rted and the cells W'H"e washed
ag~in wi th ni tt"oeen for some 15 min~tes. Then the system wB
closed end the pump started.
Stat·tint from b 0, the 1enl!Oth chan~es were recorded every 5
hou~s with ~ Oatron deta lOl!Ol!Oing system (1200 se~ies).
Together with the specimens also some free w~ter was introduced
into the cells. so a meas~rable chanEc in 1enr;th only occurred
afte~ a certain time, which depended on the R.H. of the test
etmosphe["e.
After 8 days, the creep specimens were ~nloeded and their reco"
very was measured for 6 more days. Prel~minary experiments h~d
shown, that ell:tendint, the measurin!,; time did not give m~ch
additional infQr;matiol'l, either (or the c~eep or fo~ the
reco-very measurements.
3.3. w~ter vapour sorption me~surements.
After the shrinka!,;e and cree(> experiments hed been performed,
the pore structure of the hydrated cement pe~tes were stUdied
through adsorption end de~orplion of water et 25 .O·C. The
iso-therms were dete~mined in a pure water vapour atmosphere.
~.3.1. Apparatus.
The amount of water va~our adsorbed or de50rbed was determined
gravimetri cally. Fi 1>. 3.9 shows a 5chemet ic d lar;ram of tl)e
set-up, that wes developed for this p~rpo~e.
b:liancc.oe- cant.rc.i unit I, mic:rob.al.l..llnc:oC 2, vacuum bottle ), t>J:'I!I':S~l,Iri! 1;rAr,.s..:h)'~liIr 4. thc:rlIlOCQUp 1..:: 5. cool~r ( .... ,ater) b. ho..:llco.:r (JtllnclHll "'lr...:) 7. tempeI'.tIcutl:! 5tmaOr fl. i:lilolar::cd cllblnt:t 9. rit-!l.OL v~"::uvrn *~""ij.~ 10. m.aln vlllvc
Il. ~l~c:trom.:J.;g~tit valve
I~. v.a.ri.abl~ l~.&k V4l\lif!
I:L bulb for prc.liIliIure: control
,-4.
p):.'ul:l;'t'.tIJWo'I"bl":!- Chetm¢!j:tp,[Fig. 3.9. Schematic diagram of the sorption AppArAtus.
A Cahn 2000 microbalance was mounted in a glass ~acuum bottle. The inside of the vacuum bottle and the hangdown tubes of
s3mple and t8.~e p8.ns were c03ted with tin oxide. This
trQnsPQ-~ent end conductive coatin& was connected electrically to the
earth of the balQnce, in order to eliminate ele~tro$tati~
chaq;es.
The wQ\:;,r vapour pressure inside the vacuum ~y~tem was
est8.-blished by means of a smQll bulb, partiQlly filled with
double-di~tilled and de&a~sed water. The temperature of the
water in the bulb was varied between -24.8 and +24. FC, by
means
Qf
a prQ&_ammable thermostat and a close~ system in whichelhanol i~ ¢irculated, In this way, vapour pressures between 2
and 9$'1. of the saturat ion vapour pressure of water at 25. O'C
(3167 Pal could be achieved (see Tables 3.3 and 3.4)_
Since no other vapours or gases were present, the totQl
pres-sure in the vacuum system was identical to the saturation
vapour pressure of the water in the bulb. It was measured with
a Druck PDCR 110 W ~ie~o-~eBi~tlve p~e~sure transducer, mounte~
in the lid of the vacuum boUle. Thi s trQnsducer is able to
wea~ure pres~ures
up to 2 x 104 Pa with an accuracy of±
2 Pa.The temperature of the water in the bulb and th~ t~m~~t'"Qtut'"a
nea~ the sample were measured with ~hermocouples
(iron-con-stQntan) sealed in stainless steel housin&s.
The vacuum bottl~ and pa~t of the va~uum system were housed in
an isolated cabinet, which was th~rmostQtlcally controll~d at
25.0·C by means of Qn air-hQndling system fitted to the top of
the cabinet. H. was posslblOl to achiev~ high l""~lative vapour
press\,lI:es (up to 0.99) without any condensation, due to the
&ood tem~e~ature stability and homogeneity inside the cabinet.
To reduce the influence of buildin& vib~ations, the cabinet was
fixed to a heQvy stone tQble (see Fi&. 3.10).
By means of a flexible metal tube, the vacuum system was
con-necte~ to both a.n oil··diffus ion pump a.nd a rotary·-vane vacuum
pump. A set of va.lves was pla.ced between the pumps and the
system to control the pumping rate. For getting a rough
indication of the vacuum in the "'yat~m, a Pirani VQcuum tQuge
was us~d. Th~ pert of the system outside th~ cQbin~t was
Fig. 3.10.
General view of equipment for water vapour
sorp-tion experiments.
enclosed in a plastic bag, thermostatically controlled at 27°C.
The whole experimental equipment was installed in a room,
con-trolled at 22°C.
The experimental data like sample weight,
water
vapour
pressure
and
the
various
temperatures
were
recorded with a data logging system and a multi-pen recorder.
3.3.2. Experimental Procedures.
Next to the shrinkage and creep experiments, water vapour
sorp-tion experiments were performed with pieces taken from the
hydrated cement paste cylinders. Firstly, this was done with
the creep- specimens.
During the sorption measurements on a creep specimen, the
shrinkage specimen was maintained at a temperature of 2.5°C, in
a test tube containing ni trogen and the same saturated salt
solution as had been used wi th the preceeding shrinkage and
creep experiments.
At the starting point of an isotherm, the system was evacuated
wi th all the valves opened. After about 1 minute, the main
valve
was
closed.
Evacuation was
then
continued via the
electl:omlLgnetic and tile v<td able leak valve. In this way, the
~y~tem wat slowly evacuated further <tnd B imultoaneously w<tshed
with water v<t~our. Because of the very low pumping rate
(10- 2 lis). th", temperlLture of the watel: in the bulb did not vary from the temperatur", of the ethanol in the closed $yetem. Tile samples were dded by immersinr, the hulb in a Dewal: flask
filled with dry ice Rnd ethRnol (-79·c). In t.his case the
saturatoion vapoul: pressure of water wa~ only 0.07 Pa. Acco~ding
to [2. pr,.271J and [16), all evaporable (not
chemically combined) water was thus r:emov",d fr:om the hydr:at",d cement paate. This procedure is also known as D-dtying.
Satur:atolon of a sample occurred at plpo = 0.98, by keepin&
the temperature of the water in the bulb at 24.7·c.
When the weight of the sample became constant. the
eh,ctro-ma&netic valve WILS closed and the adsol:ption or: desorption
experiments were started. The water vapour prcssuI:e inside the
system WI'S varied step by step. The toime needed to reach
equi-librium at ~fich stlL&e. varied from 1 to ~ day~. It took 3 weeks
to detoermine one ad~o~plior. or desorption brar.eh, when
consi,-t in!'; of 12 steps. Ir. oc-der to prevent di stourbances of the
ad-sorption or de~orpHor. processes, the vacuum pump WI" only run
a.t the first and lhe final step of ar. isotherm, as de«cc-ib.:.d
previously. Thi s could be done. s ir.ce the leakage ir.t.o the
vacuum system WaS very sml'll. During one experiment, the
incr~ase in pressure due to leaka~e. Was less than 10 Pa.
~.4. silicate Polymerisatoion Analvsis.
The silicate structure of the hydrated ¢e~ent paste cylinders,
after the shdnkage and creep experhller.ts, WaS studied by the
tt"imethyh HylRtion (TKS) techn~que d.:.veloped by TamAs, Sarkll.r
1'011 I;l:oy [10].
This method was considered to be the most r:.:.liable for
stu-dying the silicate structures occurring in hydrated cement
pastes [11,17].
In the (:ouc-se of this study. however, an improved method of det"ivatisatiol'l, based on thl't of T_As, Sarkal: and Roy, WaS
pur-poses, howeve~ (compatison of ~esult~). the o~i&inal method was used in all ou~ expe~iments,
3.4.1. Silylatioo Technique.
Two mixtut"es of 20 ml dimethyHormamide (DH~·). 10 ml
hexa-methyld~$ilOlCl).ne (HMO) and 10 m1 tt"imethylchlorosilane (TMCS)
were stirred l<Iell at ("oom tempe~atu~e. All chemicals had been ft"e$hly distilled pt"iot" to the ~ilylation experiments. After 15 minutes, 0.400 g of a Sl'ound $Mlple, taken respectively f~om
the shrinkage and the creep spe~imen, l<Ia$ added to each of the mixtures and stirdng was continued fOt" I). j;'uttller 30 minutes.
Then. the reaction mixtures were t.("ansfer("ed into separatory (1)nneh and the uppet" laye~s we~e washed several times with water. Next. the upper layers were dried Over anhydro1)s calci1)m chloride fo~ one hour and treated overniKht. with 4 g of Amber-lyst-15 ion eXChange resin (aohm atld. Haa'l' Benelux N. V.) in ordet" to complete the det"ivatisation.
The nexl: mOl:n1ng. I:he liquid wa~ filtered off into volumet.dc flasks and the f~l\:.et". containin& the resin were washed several
t~mea with HHD. Furthe~ 10 ml of an internal standard solution
(O.l" vol.lvol. n-tetradeeane in HMD) was added to each
fil-t~ate. which was then made up to 8. volume of 100 m1 by adding
mO~e HMD. Finally the resultant mixtures were analysed by
gas-liquid chromatoKraphy without furthe~ treatment.
3.4.2. GaB-liquid cbromatog~aphy.
To analyse the TMS-de~ivatives of the silicates, a Perkin-KIrner Sigma 2B cbroma\:.ograph, equipped with a flame ionisation detector, was used. The detector tempe~atu~e was 300°C.
The chromatograph was fiHed l<Iith a 25m fused silica capill&ry column (I.D. 0.22 rom). Cp-sil 5 (a ?olydimethylsilicone) with a film thickness of 0.12 ]llII was used u the atationat"y phase and helium was used 80S the carrier gas. The inje<:l:or
tem-perature was 280"C a.nd the split injection technique
wu
used (splittint ratio 1:20).The column temperatul'e Wa$ raised linearly from 80 to 280°C at 10°C/mln. ~inally. the column was ope~ated isotherma.lly for
further 20 minutes. The <;hromatogratlls recorder and analysed with a Spectra system.
W0re plotted on a x-t
Physics SP 4000 data
~_4_3" Non-volati10 fraction of IMS-derivatives.
The non-volatile fraction of the TMS-de"dvatives has been
de-tel"mi ned by heati ng pal"t of the react ion mixture during 24
hours in an oven at lSO·C. After cooling in a dessicator the residue was weighed with an analyHcal bal&nce. All silical:" i5ll'ucl:.ures consisting of mo .... e than six silicate gt"oups a .... e
sup--posed to be included in the nonvolatile fl'action of the
1HS-deri"a!:;v",; as det.el"mined in this way [6].
Chtlptel' 4
4. Experimenttll Results.
4.1. compl'essive stl'ength of the specimens.
As all'ea.dy mentioned in section 3.1.6.. the compressive
strength of some 28 dtlys' old spec imens was determined
afte.:-they had been dried fol' 24 hou~$ at a R,ij. of S3~- The obtained
stress-strain diagl'ams showed nearly straight lines and all specimens failed due to brittle fracture (see Fig. 4.1).
,
E .!!:s
~ ~ w ~ 100 50 Fit.. 4.1. Typical stress-strain diagtam.The I'esults Of the mea"urements are shown in Tables 4.1 and 4.2. The mean compress! ve strength of the PC spec Imens WtlS 70
N/mm2 (0 20 N/mm2). The metln valU4i! of tlJe elutic
modules was 14.3 x 103 N/mm2 with tl standtl.:-d deviation of
0.7 x 103 N/mm'1.
For the PBC specimens the mean compl'esslve str4i!n&th m4i!asured
60 N/mm l (0
was 20 N/IIIIl ) and the mean elastic '1 modlllu~
was 14.1 1 103 N/wm2 (0" =
O,S
x 10 N/IlUII). 3 '1H is stdkinS. that the sta.ndard deviation of the elastic
modulus is significantly smallel' compal'ed to the standal'd
deviation of the compressive stl'ength values. Allpa .. ~ntly. th~
comp"elisive st"ength of & specimen was mor-e s~n!iiti>le to th~
e~pe~imeqtal conditions than the elastic pl'oper-ties.
4.2. Shrlnk~&e anQ G~eep Expe~imentB.
The result~ of the gh~inkage and creep experiments are
presen-ted i n Tab1e~ 4.3. a t.o 4.10. b. The r'Hylt~ are prnel)ted in a
conden~e<;l form in Fi&~L 4.:;> and 4.3. l'igut:'e 4.2 shows the re··
suIts of the ~hrinkage and creep experiment.s with PC specimens
al 100. 75. 53 and n'l'. r",lal: i VE! humidity t:'especti ve1y. Figure
4.3 shows the t:'esu1ts of the experiments with PSC specimens. At. a t:'clativc humidity of 53'1'. the defor;-mal:ion cut:'ves of both
the loaded and t.he unloaded PC and the PBC spe¢imel)~
pl'acti-cally coincided. At the ot.her relative hYmidities, these
defOI'-malions were alway~ smaller for the PBC specimens.
sao
400
en300
0 ..-X 0 ."200
:
:
:::
,
100
50
100
150
200
250
300
500
400
f
'"
~300
x..3
-.
<I200
I100
50
100
150
200
250
300
time(h)---Iri!;. 4.3. Deformation curves pac (time of unloa~in~ 192h).
RH(%)
22
5315
100
shrinkage+ct<!ep
• •
..
•
shrinkage 0 {]
6.
0The ~h~inkase and total defo~mation cu~ves (sh~inka&e + c~~~p)
of PBC at a ~elativ~ hum~dil:y of 221. show an i~regula~ path
after about one day _ With all the other nperiments, thi"
phe-nomenon was not obse~ved. III secUon :;. 'l- I:his behaviou~ will
be discussed furthe~.
In the tot.al deformation cur1fe~, the elnstic deformations due
to application of the load at t= 0 and the elastic ~ecove~y due
to ~~moval o{ the load at t= 192 h. have not been inco~porated.
The values of the elastic defo~mations a['e (Siven in Tables
.... 3.c to "'.S.c and in Tabl~s 4.9.b and 4.10.b. They a~e shown
in Fi~ •. 4.4 and 4.5. As it can be seen. ha~dly Qny differences
are obse~ved in the ~lastic defo~mations of th~ wet PC and PSC
specimens due to application of the load.
ln d~y conditions. the ~lastic ~ecovery of the PC specimens is
approximately 10'1, less than the recovery of the PBC specimens.
Althouth this effect. could always be measured, it was not
!j-ignlflcant, because of the larse ~tanda~d deviation of the
1'I'IeaSur-ernents.
!
60'"
~o
ItO~
20o
•
•
•
-(l_--~et---·~ 50 100 R.H (%l-Flg;. 4.4. Ela~ti~ deformations pc.•
t" 0 l= 192 h ;\0t
60'"
~..
40•
.::;f
zQo
Fi~. 4.5.•
o 50•
Q 100 RH.I%l---Elastic deformations PBC.•
t= 0o
t=; 192 h4.3. Water vapour ~o~ption experiments.
Following the shrinkage and creep experiments, the po~e
struc-tu~e of the specimens was .tudied by means of physical sorption
of water vapou~ in a pure wate~ vapour atmosphere (see section
3.3.) .
The equipment needed for these experiments was developed,
con-structed and perfect.ed, durill~ the course of the experiments.
Also, some problems concerning the operation of the mi~ro
balance at high relative water vapour pressures had to be
solved. The~efo~e. it was not possible to obtaill as many
~esults as originally aimed at.
At first, only adsorption isotherms were measured to a relative
vapou~ pressure of about 0.88. After ~nstallion of an improved
temperature control uni t, r",lati ve vapour pressures of up to
0.99 could be establiShed. From that time, desorption isotherms have also been measured, sometimes followed by adsorption iso-therms.
The results of the water vapou(' sorption experiments are
pre-sented in TablES 4.11 to 4.23 and are shown graphically ill
Figs. 4_6 to 4.13.
AS it ~an be seen, the adsorption branches belong to type II of
the BET-cla~~ification [19J- !lowever, the complet(:
iso-therms show ~emarkably large hysteresis loops. that only close
at pip" '" O. This phenomenon Wa$ also reported by Krasilnikov [~O],
At first glance, differences in the pore structure of the variolls specimens of one kind of cement were hardly visible. Small differences could o1l1y be found in the cours"" of the various cllrves. with respect to pretreatment (shrinkagE, creep
and/or
R.H.l.
!lowever, clear differences exist between theiso-therms of the PC and the
pac
spec imens. The latter show alar&er total water uptake at hi&h values of plpo and a
smaller loss of wate~ at low plI'o values during desorption
0.10
'"
on ';; 0.15 010 M5 S8ETd.'.~ 242ml/9 5BETad~.~ 123m21g~_li.: ... .L.h WAt .. iu' 'o'u.?our liorptiljn ~t=I)~hh',"m or h~dt'o.tQd E-'oC
pute o.ftC'[" I;r~ofjF .1.~tiI"l!'ill:'nt.~ .... t '~'1. R. H I
'"
'"
'";; 0.15
0.10
~p .. <;iIllIU> ~3d. (:~tol.['t _ d-o(:.(l~p.,i.Q=);
SiI'""," 'am'!'D 13d. Q ("~.,,"l - u..jIlOi'pt lorJ).
[J
G.S
~ WOoter ".p¢ut" ~ol!'ptiol'l iD,othl;l,m ~(' byd"u.l.::d PC
put., ""F~~ ... :8h.~!nk.Q.B.Q OKporim"nt¥ H 7S.~ R.H.
) 0.20
'"
on -;:; 0.15 0.10 seE" I des. "Z~2m2/g. SSET ~~;_126 m2, g, 0.5 I I I I J!:.i.!..:... Water' YII.PO\l[' soO!';ptlol'l iIillOth".n. <;>f hy<J. •• t.':l<i FC
puto .tftlllt-~r •• p .'l'ip.dmontll' o.t ;:J.';' ~.H.
'"
<II 0.20 " 0.15 0.10 0,053p.ciril"" ~~4 • i."".t .. dolt'Ol'.P'tkonl ~
;p ... 'im~n llll.> 0 (Qtd.~t .. tl.dlll"''':p-lign) .. 0,5 I J I I I
!!..I.:...:i ill .. ,"". ... .. PQ'Iu.. .ol'ttion i!loth-'!l."iL1. flf :t,.y4 ... t.<;t poe; putll .rt'llll." III'It"idltoll!;t .IIiP.,.:dm.nta 'It ~J1. R.H.
s~'" i.mun '21", • (start .. d":IICI:'p~!.~~) i
SplllC lman 111.1. 0 tt:h('~ .... d1!o['Ptlon} ..
1
0.20 c;, ;; -;;: 0.15 0.10 'BETdes.= 21B mZ,g. SBET ads.= 109 mZ'9 0.5 I I I.!::.l.!...:... WIi. L", I.' VIl~OI.L" ~ot'ptlon l!;'O'thoo:.>t"fT1 <;It: \'ydt'"lIt.l"! PC
1
0.20'"
'"
>< 0.15
0.10
p ... lOt.c .o.tt~r' C['(!~II 0~p{ldl'T\tlntl;l Kl i!2"1.. R . .11.
~p..,¢i'T'''''' ~~d • \:Hlll:'t • dQ:IOo['Ption};
SI?-Cl-.:dm",n lSb I) (~tQ,t"t • • ;u;i'l' 'l'<: fI't: lur'l.). SET =Z18mZlg.
B d",.
-
,
'SETad •. -11Zm 19·
0.5
~...,.:L..ll.:.. W.IIt"" VIlP'OIJt" 1F¢,,-.;pUon hotnur~ {Joe hydr.ul;,,,o;t p¢
!?'utc, (littl[' u:~~11'11:.1L1',.':: cJl.po;rim ... ni;:jI: at. 22", R.11,
O.LO ~ ';; 0.15 0.10
o.os
SBETde,~ 197mL/g, S E=
116m2/g. B T M,. I 05 I { / I I~ "... .. t:. ... r- """"po.lil.' :ao:t'ptlon il;Qthvt"m ... f hy'jl."..!;ad f'HC
P."~" *CL"r f.t'IlI(JI;! C'X~0I"lJnont[J 1.1:. 7~1. g,lt,
S~ec lmen. l~b • (lFhlj"l;. .II d.¥I~II:p!:l':l<">:
:!;pllldlHiln. lId 0 ts;tart .. I.dIlQt'"pti';m>,
1
0,20 ~""
;; 0.15 0.10 0,5~ Watlt[' ", .. pr;1,lI1L IIO['ptllUl:r.l i":ot.htt'[';11 of hyd['".t.od. 1"0-;;
pu~. d"l'llll." Jl:hl."lnk&t,41 Ilolt:PtiJ~in'ujl.r'lt;. .. d 1~'I.. ~.H_
:Splllclman 16.1 . • i.ltart .. dQ:5lo['J;ltlC1n~.
4.4. Silicate polyme~isation analysis.
Following the shrinkage and cr-ee:p expedmenlos. tmQll sHll'lples
wer-e der-ivQted in the way pr-e:viously deser-ibed in section
3.3.1 .• in or-der- to study the ~ilicate str-uctur-e of the various specimens. The resultant THS-derivatives were analysed by means
of gas-·liquld chl'omatogl'aphy (1;.1.10.). 1'il:;_ 4.14 shows a
typical
chromatogram-NO pu .... e dedvatlves we .... e aVQilable. However, the eh~omatol:;~ams
obtained all showed ch8.r-8.cter-istlc pr-oflles, agreeing very well to the ones shown in Hte .... ature. Th.:.refo .... e. identification of
the peaks
in
the ehr-omatograms was carl'ied out Onthe
basis ofdiagr-ams shown 1n [18] and [21J. TQble 4.24 shows a list
of the silicate anions. assigned to the various peaks in the way mentioned above.
1n Tables 4. 2S to 4.3& the peak areas of the various compo·
nents, with r-egar-d to the inter-nal standard peak ar-ea are
pl'e-sented_ In 1'i&5. 4.15 and .... If> these results are shown in a
compressed way. From these di ag .... ams follows that.
th~ mono· and the di·silicate (peaks land 5).
apa .... t from only minor
amounts of the highe~ silicates coulu be observed in the
chro-malogt"a.rns.
~'or- the PEC specimens the peak Qreas of the disilicate wer-e
a.lways la.r-ger- t.han the peak arells of the monosilicate; for the
PC specimens the peak !lreas of the dis i licate were slightly
smaller- than the peak. a["eaS of the mOnO. iIi cate. Howevet", the
absolute vQlues of both the mono·· and disilicate peak Qreas
we .... ~ always lllt"ger fo~ the FC specimens.
The results presenteu in Tables 4.Z5 to 4.36 show lal'~e
standa["d deviations, mainly caused by deviations in the .... esults
obtained with identical sa.rnples and by the different lime
i nterval:5 between der- i vat i sll.Uon Qnd analys i s of the various
samples.
In Table 4.37 the non-volatile quantities of TKS-der-ivatives in the var-ious .... eaction mi](tut"eS are pt"esented. For- the PBC
sam-ples the amounts appear-ed to be appt"oximately 5070 larger thlln
fo~ the PBC samples.
\
1
j
Fi&. 4.14. ,-10 /6 98~
7 ____ _ II 15----..Typical chromatogrQffi of TKS-derivatives of hardened cement: paste after shrinkage and " .... eep experiments.
fOr
explanation of the
peaknumbers, see Table 4.24.g~ ]S 'Yo> ~ep';l~% :Vlrlr"'-~gl1' ?s"%
:1
c:r1l'1'.p1Y%~
:
'
rhR
D d:J"
..
"
>, ,;"
'5Jkno--- p~9;no - -p;;;;;kno~
p'i!akno.--:j
S! )~% ' IQ
5,..,..ink",g .. S3% [rl!.'!PO S-l % 3-~~
'l
1-' ~I
O~ 15"
15 :01"5-"
IS~k n~-- ceakno- pea~ ... ~-
pE:;;:'nn.-I
,J
~e ~ ~ 0;'Ii
( .. @@p1~% ~h":~~9t :'::1"";" u;I'epn--t,.~
3 ~r:
In
,
~
I'1
I
:l
Chapter ~
S.
Analysis of e~perimental resul~B.~.1. compre~~ive stren~th of the specimens.
According to Neville [31 creep is directly proportional to the stress applied up to about half the ultimate strength of a "peeimen. To be ,,"ut"e. that dUring the creep e"pedments the stru5 applied would be of the right magnitude. the ultimate ¢ompt"essive strength of a number of specimens was 4etet"mined in
a way. described ~n section 3.1.6.
The t"esults of these experiments. pt"esented in Tables
".1
andthat
4.2, show
70 N/I1lI/ and for
tions of 20 NII!rI/
for
can
PC the mean compt"essive strength was
60 N/mm2. The large standard
devia-be explained by cons idedng, that the
ul tlmate stt"ength of a Ii1pecimen, beyond ma\:;eri al propet"ti flS,
also depends on many other mOre or leB$ coincidental factors
cohet"ent \:;0 the performance of the experiments (transmission of
\:;he load to a specimen) and to the quaU ty of the spec imen
(pt"esenee and amount of' cl'a~l(s or airholes inside the speci.
men). The standard deviation could haVe been reduced by per·· forming more experiments. However, the t"esults obl:"'-ined wero:.
considered ~o be se.tisfa.:li:ot"y, since the purpose of these pre··
liminary experiments was only to get an impt"ession of the
magnitude of the ultimate strength of the type of specimen used. The modulus of elasticity is a pure matedal propet"ty and, therefol'e, less sensitive to th", coincidental factor;s formerly mentioned.
starting from these results and considering some practical data lilr.e dimensionS of creep appnatus, weight size and length of
level', it was decided to apply a load of 196 N to the specimens
dudng the Cl,'eep experiments. This was t"eu.lised by means of a weight of approxime.tely 47 N on a level' (distance to specimenl
distance to weight. 1:4). (The weight of the upper cap and the
lev9r was abo taken into account in the calculation of the wei.ght size).
S~nce th~ the stress 6.4 Nlmn? cross-sectional area of applied to creep which is approximately
stren~th (see Tables 4.1 and 4.2).
;"2" Shrinkage and creep ~xp~riments.
specimen 0.1 of was WllS the 30.6 mm • 2 equal to ultimate
The results of the shrinkal'.e and creep ell:p~dments on PC and
PBC pastes, pcesented in Fi!,>s. 4.2 and 4.3, showed that the
d~formation5 du~ to shrinkal',e and shrinkage plus cre~p were
!,>enerally smaller in the case of
pee.
Only the $hrinkage curvesof PC and PBC pastes at S3~ R.H. coincided almost completely.
In Fi!5. 5.1 the shdnka~e and total defoClflations at t=
19::-!>ours (the time or unloading the creep "pecill'l~ns) are plotted
against the relative humidity of the test atmosphere" From this
di atram i t follows. thaI: in the R. H. J;'Qnte from 50 to lOO'E.,
only mino[" d i rfeJ;'ences in the sht"i Ilkll!5e and total deforIM.tion
b~h8viour of PC and PSC spec~men5 exisl:.. Below II R.H. of
al?I?roximate1y :;0"1. "lear differences in the deformat~on
beha-v~Ollr ~Bn be noticed" 500
~
400~
"'~ 300.
J ::::..,
200-=pe
tot al
d~t o=PC shrinkag~ 100 .~PBC lotal d2i D"PBC sh,inkag~ 0.2 0.4 0.6o.a
10 R.H.Fi&. S .l. Shcinkage and total defo["m8.tioll~ of specimen~
after 19ih versus R.H. 40
100
80
t
~....
60
)<"
-:;
40
r20
Fig. 5.2.100
80
t
.... 060
)< o"?
40
20
Fig. ~.3.time(hl---Creep curves of PBC spe~imen$.
50
100
150
200
250
300
time(hl-c~eBp ~urve~ of PC specimens.
Cornelissen (4) earlier reported, that for lO~10xSO em
con-crete specimens no sir,nH'icant; difference in ~hdnkar,e
beha-"iour could be found at SO~ R_ H. However, for PBC ~pec imens a
smaller specific creep wa~ found tban for PC specimens.
When in Fir,s. 4_2 and 4.3 the deformations due to shrink&r,e are
subtracted from tbe total deforlllations, the creep defo .... mations
of tbe loaded spec illl",ns ace obtained. Fir,~. 5.2 and 5 _ 3 show
the obtained creep curves. Tbe influence of the relative humi-dity on creep clearly follows hom tbese diar,rams. "or both
types of c€ment the creep ince€ases with decr-easinr, l"elative
humidity of the test atmosphBl"e.
In fil- 5.4 the creep at t= 192 boues is plotted as a function
of the eeh.tive humidity at wbich the €xper;illlents were perfor·
med. h"om this diar,rRIII the int.'i.uenM of the R.ll. OJ) the creep
of hardened PC and PBe pastes can be seen. Under normal
circum-slance' (R.H. ;>$O'l'.) the cr;-eep of PBC specimens appeared to be
smaller than tbe creep of PC specimen.. bu~ at \L H. <: 351. the
creep of PC was smaller tban that of PBC. Apparently the cr;-eep
of
rac
,pecim€ns is influenced si:.t"onr,er:- by" tbe rela.tivehumi-dity than the ~reeP of PC specimens.
100
80
cr. 0 ..---PC
x60
0 ... ~..2.-
40
0-Q) OJ <.... u20
0
20
40
60
80
100
RH.(%l-fi/;- S. 4. Cr:-eep defot"mations of specimens after 192h versus
Most of the shrinkage and cr~ep experim~nt$ report~d in litera-ture had been perform~d with relatively large ijpetimens in relation to the specimens, that were used in this work. Accor-ding to [22] the specimen size will have consequences fOr the experimental results. Thin specimens will rapidly loose capilla.ry watel;" d.uring dryint. This proce.!lS is accompanied by shrinka.!le due to ca.pillary forces, occurring during recession of the menisci in the p01'es. When cl:'~ep simultaneously occurs with shrinkage, the creep strain is strongly enhanced; the so-called Picket.t effect [23J.
Due to the ~elati vely large surfa¢e/volum~ I:'at io the ini l;ial evaporation rate and consequently the shrinkage and simul-taneous cre~p will be large for thin specimens. When a specimen hM rea.ched the equ 11 i bri um capillary wate:r content, with respect to the relat ive humid.i ty of the test atmoaphere, the rate of moisture loss will ~tron!lly decrease. 1:his effect can be observed clearly in Figs. 4.2' and 4.3. Uter about one day at
;on
R.H. , a rdatively sha.rp bend is visible in the shl:'inkate and tbe total deformation curves of the experiments. When I:'elatively lal:'te specimens a.re used for shrinkage snd creep e~periments, initially only the outer parts of the speci,men~ will reach hygrical e<:\\lilibrium with the test atmosphel:'e.
As a cOnsequence, the messured !>hrinkage will primarily be an induced stl:'ai n of th", ... el;. intedol:' of the spec imen. caused by sbl:'il'lkage of the outer part. Through diffusion the water from inside the specimen is tra.nsport:.ed to the sUl:'fl).ce. Since the diffusion process is slower than the evaporation rate frOm the s\lrhce [22]. the I:'elative moiat\lre loss will be smaller for larger specimens than fOr small (thin) specl.mens. Over longer periods of time also additional hydration may occur due to the high w&ter content of the core. Therefore, the shrinkage a.nd. cl:'eep curves of l&rge specimens will $how a. mot'e gradual path than the corresponding curves of thin-walled specimens.
probably due to the: described effect, the results of ma.ny eXperiments performed with lsrge $pecimens, cather reflect shrinkage and creep behaviour corl:'eaponding with a hi!lher I:'ela-tive humidity, than used durin!; the experimenh. This effect
will be more pronoul1l;'ed, when the mean moist.ure I;'ontent of a
specimen differs more f~om che equilibrium moisture content.
No explanat.ion could be found for the peak in the shrinkage and
tot.al deformat~on curves of PBG, after about one day at
R.H. , 22~, This behaviour could be explained by an increase in lernperacure of a specimen, caused by a decrease of the
evapora-tion rat.e. At 25D
C, che wet bulb temperat.ure correspondin& to a
R.H. of 22~ is 13·G. SO a temperature rise of lZ'G (at. the
ut-most.) ~ould occur when a ~pecimen becomes dry. With a value of
14 x lo6,DC for the thermal expansion coefficient (2) the measured expansion might occur,
Anothor poss i bi l i ty mi!;ht be the phenomenon called
"hy!;ro-thermal shdnka!>e" described by f>ower~ [24] (shrinka!;e 01'
swell in!; due to displacement of water (rom the capillary pores
into the &el pores and rever~e, caused by temperat.ure change~).
Howeve~. th~ measured effect appears only in the PBC cu~ves at
a R.H. of 22'-.
Yet i t is believed, that the effect is somehow related to the
~hrl nkal\e of the type o( cement used a.nd esped ally to the
experimental conuitions.
~.3. ~ater vapOur sorption experiments.
~.3.l. Interp~etation of isotherms.
For the interpretation of the isotherms the theory developed by
Br;-u~auer-, Emmet t and Teller was used [25]. If the number of
adsorbed la.ye~s i$ assumed to be infinite at p/po" 1, the
BET-equat.ion is !;iven by:
where
~1_ + C-l
x.C x.C
m !II
p" vapour pressure of adsorbate.
po· saturation vapour p~e5sure,
(5.1)
x. amount of water adsorbed per!; dry cement paste.
C eXp.(E1-EI)/RT, whe~e Kl is the adsorption in the ~ir~t layer and El is
con- densation of the adsorbate.
heat of the heILt of
From Eq. (5.11 it follows, tha-t when P/X(Po-p) is plotted
aga~Mt p/po' a straight line is obtained with slope
s " (C-l)/x m • G and intercept i = lIx .G • This straight line
m is called the BET-plot:.
calculated a~ follows:
_l~ i .. s
From the BET-plot x ana
m C can be
(5.2)
(S.3) "<"he specific su~face area S is calculated from the monolayer capacity as follows,
(5.4) whltI"e
" molec~lar weight of adsorbate (18,0 ~/mole).
N Avogadro's
con~tant
(6.02 x l023/mole).A = molecular cross-sectional area of adsorbate in m2• m
Am hu been calculated from the 1 iquid density of wateI" at
2$.0·C
(Pwater,2S.0~C-
0.997 stcml),assum~n~
that thea~rangement of adso~bed water molecules on the surface is the
same as in the bulk of the liquid:
A = f
m wheI"e
iii UJ
p.H )
f = packing factor (1,09 fOI" a l~quid. see [2]).
Although the B.lisumptlon made is not correct, this procedure is commonly used to determine the value of A •
rn In this way a