Isothermal calorimetry

Strength prediction

Lasse Frølich, Concrete Technologist

Lund, June 2016

2Background

Intention:

Replace daily cement mill 1 day mortar strengths with isothermal

calorimetry predicted strengths

From this… …to this

3Background

• Isothermal procedure determined

1. Paste with 50 g cement and w/c 0,50

2. Water preconditioned inside calorimeter

3. Handmixed externally

4. 24 hour measurements (first 2 hours exclueded)

5. Establishment of strenght/heat correlation for

each cement type

6. New correlation must be established if cement

variations exceed specific limits

7. Reference cement measured weekly as control

4Background

Preliminary studies

- Good strength/heat correlation for factors influencing C3S

- Poor strength/heat correlation for factors influencing SO3/C3A

Changes in C3S Changes in SO3/C3A

C3S content SO3/C3A balance

C3S reactivity SO3 content

Fineness SO3 solubility

Filler C3A content

Alkali and Chloride C3A reactivity

5Background

Assessment of results

Coefficient of variation (CV) of residuals compared to combined

uncertainty of the two measurement methods (acceptable uncertainty)

Acceptable uncertainty:CV = 4,9 %

Calculations:Mean strength 23,9 MPa

Std. of residuals 1,1 Mpa

CV of residuals �,�

��,�∗ 100 = 4,6 %

Residuals

4,6 < 4,9Measurements:CV = 4,6 %

Example

6Background

• Implementation – 1 day strength prediction

– 3 month implementation period

– CV below 4,9 % for predicted strengths for 2 different cements

– Accuracy and precision acceptable

– Stop measurements of mortar strengths

7Predicting strengths - during a long period of time

Results

Residuals mean 0,0 MPa

Residuals std. dev. 0,5 MPa

Residuals CV 4,1 %

Initial period

Control testing

Results

Residuals mean -0,2 Mpa

Residuals std. dev. 1,0 Mpa

Residuals CV 10,0 %

8Predicting strengths - during a long period of time

• Change in correlation over time

• Only control measurements are shown

Time [Years]

0,0 0,5 1,0 1,5 2,0 2,5

9What's the problem?

• Changes in chemistry?

– No significant correlation to results of XRF analysis

• Changes in physical parameters?

– No significant correlation to changes in strength level

– No correlation to fineness

– No correlation to setting time

• Lets look at the curves…

10Isothermal calorimetry heat flow curves

Time [Hours]

Heat

flo

w [

W/g

cem

en

t]

Confused?

…better to look at them one at a time

11Shapes

• Division into defined shapes

• Focus on alite and sulfate depletion peaks

• Shapes defined by sulfate depletion peak compared to alite peak

Alite Sulfate

depletion

Late sulfate depletion

Alite Sulfate

depletion

Early sulfate depletion

12Shapes

5 distinguished shaped defined

13Strength prediction and curve shape

3

2

1

0

-1

-2

PCSE1D-PIKAL

1

2

3

4

5

6

ShapeM

ort

ar

–is

oth

erm

al [M

Pa]

Start date

14Strength prediction and distance between peaks

Mort

ar

–is

oth

erm

al [M

Pa]

Minutes

15Strength prediction and sulfate balance

16Sulfate depletion

• Why is the sulfate depletion of interest in general?

– To achive the highest possible strengths

”As a rule of thumb the sulfate depletion peak

for a well-balanced cement should appear

several hours after the maximum of the main

hydration peak”

Reference: A practical guide to microstructural

analysis of cementitious materials

17Sulfate depletion

Is the sulfate depletion “peak” actually interesting?

18

2. C3S + 2.3H2 → C1.7SH + 1.3CH

Sulfate depletion

Heat flow

Time

C3A

C3S

19

2. C3S + 2.3H2 → C1.7SH + 1.3CH

Sulfate depletion

Heat flow

Time

C3A

C3S

C3A + C3S

20

2. C3S + 2.3H2 → C1.7SH + 1.3CH

Sulfate depletion

Heat flow

Time

C3A

C3S

Gypsum used up hereC3A + C3S

21

2. C3S + 2.3H2 → C1.7SH + 1.3CH

Sulfate depletion

Heat flow

Time

C3A

C3S

Gypsum used up hereC3A + C3S

22

2. C3S + 2.3H2 → C1.7SH + 1.3CH

Sulfate depletion

Heat flow

Time

C3A

C3S

C3A + C3S

C3A + C3S + sulfate depletion

Gypsum used up here

23Conclusions

• Strength prediction with isothermal calorimetry not sufficiently accurate for normal cement variations over time

• Distance between alite peak and sulfate depletion peak influence strength prediction

• Correction of sulfate depletion might provide accurate strength predictions

• Very time consuming to do manual analysis of curves

– Automated software needed

24Example

• Cement mill influence on cement reactivity

– Same clinker

– Grinding in two different cement mills

– The two mills produce same 1 day strengths

– Mill A without separator → High temperature

– Mill B mill with separator → Low temperature

– Mill A: 50/50 gypsum/anhydrite

– Mill B: 100% gypsum

25Influence of grinding to cement reactivity

Mill A

Mill B

Mill A:

No separator

50/50 gypsum/anhydrite

Temp. 130 C

Mill B:

Separator

100% gypsum

Temp. 110 C

Mill A sample 1 -

Mill A sample 2 -

Mill A sample 3 -

Mill A sample 4 -

Mill B sample 1 -

Mill B sample 2 -

Mill B sample 3 -

Mill B sample 4 -