1

Quality Control and traceability; the guarantee for reliable product in Cement Manufacturing

Maggi Loubser

Test and Measurement 2013 Conference Misty Hills Conference Hotel, Muldersdrift 6-9 October 2013

2

• 3000 BC– Egyptians used mud mixed with straw to

bind bricks. Also furthered the discovery of lime and gypsum mortar as a binding agent for the Pyramids

A bit of history

3

A bit of history

• 300 BC• Romans used slaked lime and volcanic ash called poz zuolana,

found near Pozzouli at the bay of Naples. They used lime as a cementitious material. Pliny reported a mortar mixt ure of 1 part lime to 4 parts sand. Vitruvius reported a 2 parts pozzolana to 1 part lime. Animal fat, milk, and blood were used as admixtures

4

Not just grey powder anymore

5

Sophistication in design and engineeringby construction industry

• Today complex mix designs are used to prepare a concrete for a specific application and the client demands consistency and strength performance from the cement.

• In the production of cement it is the manufacturer's task to ensure that the properties of cement are kept at a certain level, with variations as small as possible to meet the standard specifications and to comply with the demands and needs of the market.

• This implies that variability in material composition and processing throughout the manufacturing process must be minimised.

6

Quality Control

• At the beginning of the 20th century, quality control in cement production essentially consisted of CaCO3 titration, litre weight determination on clinker and Blaine fineness determination of cement to control the product quality.

7

Demands for consistent product quality for both the finished product and the in-process materials

• The control parameters include chemical composition, chemical ratios, particle size distribution, colour, mineralogical composition and free lime content.

• Quality assurance systems must be developed to control the composition of all inputs to the manufacturing process and all process conditions in the manufacturing process.

8

Backup by GLS

• Classical Wet methods• Instrumental Methods

• Colorimetric• Atomic

Absorption/Emission• X-Ray Fluorescence• X-Ray Diffraction• Microscopy• Inductively Coupled

Plasma• Ion Chromatography• UV-VIS

Spectrophotometry

9

PPC is striving not just for conformity but operational excellence

• Cement is manufactured to the specifications of the European standard, EN 197-1 for Common cements (SANS 50197-1, SANS 50197-2).

• The standard covers cement types based on composition and strength classes, with specifications for performance and conformity criteria (South African Bureau of Standards, 2000; 2004; 2011).

• Regarding management systems, all PPC factories are also ISO9001:2008 and OSHAS 18000 and ISO 14001:2004 certified.

• Group Laboratory Services are also ISO17025:2005 accredited

10

Quality Control – continually improving performance

• This is where real value add comes in:• Confidence in analytical results

• Control systems in place• Use control systems to trouble shoot problems in laboratory and plant

• Data mining to optimise processes and reduce standard deviations on controls

11

Quality control in the XRF labs

• Best practises used throughout the group• Daily (once a shift) check samples and drift monito r samples• “Sample 99” prepared on each shift to check sample preparation

SiO2 Quality control chartSiO2 Quality control chartSiO2 Quality control chartSiO2 Quality control chart

11.00

11.50

12.00

12.50

13.00

13.50

14.00

14.50

15.00

19/0

2/20

09

26/0

2/20

09

05/0

3/20

09

12/0

3/20

09

19/0

3/20

09

26/0

3/20

09

02/0

4/20

09

09/0

4/20

09

16/0

4/20

09

23/0

4/20

09

30/0

4/20

09

07/0

5/20

09

14/0

5/20

09

SiO2 Average sd 2sd 3sd -2sd -sd -3sd

12

Setting up of a control system

• A stable material with relative high concentration of all relevant analyte elements is needed

• The material does not have to be in the same matrix as analyteelements – a solid glass or metal works well

• This check sample is subsequently analysed twenty times over at least ten days and the averages and standard deviations calculated from where control (3sd) and warning limits (2sd) are set

• This data is then plotted on a control chart • Subsequent daily data is plotted on the same chart and when

control limits are exceeded corrective action needs to be taken.

13

Control Chart

14

“Sample 99” prepared on every shift and analysed in combination with check sample

CaO by Analyst

41.40

41.41

41.42

41.43

41.44

41.45

41.46

41.47

Wee

k 1

Wee

k 3

Wee

k 5

Wee

k 7

Wee

k 9

Wee

k 11

Wee

k 13

Wee

k 15

Wee

k 17

Wee

k 19

Wee

k 21

Wee

k 23

Wee

k 25

Wee

k 27

Wee

k 29

Wee

k 31

Wee

k 33

Wee

k 35

Wee

k 37

CaO by Analyst +2SD -2SD

CaO by Testers

40.4

40.9

41.4

41.9

42.4

42.9W

eek

6

Wee

k 6

Wee

k 7

Wee

k 7

Wee

k 8

Wee

k 8

Wee

k 9

Wee

k 9

Wee

k 10

Wee

k 11

Wee

k 11

Wee

k 12

Wee

k 12

Wee

k 13

Wee

k 14

Wee

k 14

Wee

k 15

CaO +2SD -2SD

Intervention:re-training

15

Round Robins

FINAL DATE FOR

SUBMITTING RESULT

SiO2 Al2O3 Fe2O3 Mn2O3 TiO2 CaO MgO P2O5 SO3 K2O Na2O LOI Total

22-Feb-10 DEH 19.73 5.63 3.61 0.42 0.41 62.53 2.89 0.00 2.25 0.46 0.19 1.58 99.7DWB 21.17 4.58 2.39 0.41 0.55 62.84 2.30 0.81 2.16 0.23 0.12 1.51 99.07GLS 20.72 5.26 4.01 0.41 0.46 62.21 3.00 0.08 2.46 0.35 0.11 1.51 100.6

HERC 20.49 5.18 3.90 0.41 0.46 61.52 2.87 0.06 2.59 0.24 0.10 1.76 99.6PE 21.8 6.2 3.8 0.4 0.5 61.7 2.9 0.08 2.58 0.31 0.25 1.67 100.5

RBK 20.62 5.37 3.74 0.62 0.57 60.46 3.42 0.07 2.46 0.32 0.00 1.75 99.40JUP 21.4 5 3.8 0.4 0.5 62.0 2.7 0.1 2.7 0.4 -0.1 1.92 100.82

JUPfb 20.6 5.5 3.8 0.4 0.5 62.0 2.8 0.1 2.4 0.3 0.1 1.92 98.5SLR 20.11 5.00 3.30 0.43 0.47 60.19 2.50 0.06 2.51 0.33 0.11 1.51 96.51

08-Mar-10 DEH 19.73 5.63 3.61 0.42 0.41 62.53 2.89 0.00 2.25 0.46 0.19 1.68 99.8DWB 19.78 5.87 3.69 0.43 0.42 61.35 2.96 0.00 2.23 0.46 0.18 1.55 99.11GLS 21.65 5.29 3.88 0.38 0.46 62.47 2.80 0.08 2.53 0.35 0.11 1.54 101.5

HERC 20.31 5.20 3.84 0.41 0.46 61.35 2.86 0.06 2.79 0.25 0.10 1.71 99.4PE 22.0 6.4 3.8 0.4 0.5 61.9 3.0 0.08 2.66 0.30 0.26 1.85 101.2

RBK 20.60 5.36 3.76 0.62 0.58 60.50 3.52 0.07 2.43 0.32 0.02 1.60 99.38JUP 22 5.2 3.8 0.4 0.5 62.4 2.9 0.1 2.5 0.4 0.1 1.52 101.8

JUPfb 21.1 5.6 3.9 0.4 0.4 62.8 2.9 0.1 2.4 0.1 0.0 1.52 99.6SLR 20.15 5.07 3.36 0.43 0.47 59.90 2.53 0.06 2.43 0.33 0.10 2.00 96.82

22-Mar-10 DEH 19.32 5.68 3.66 0.42 0.41 62.51 2.89 0.00 2.23 0.46 0.18 1.66 99.4DWB 21.18 4.59 2.40 0.41 0.55 62.83 2.30 0.80 2.12 0.23 0.12 1.91 99.46GLS 21.57 5.28 3.8 0.38 0.46 62.24 2.79 0.07 2.46 0.35 0.11 1.59 101.1

HERC 20.56 5.20 3.86 0.40 0.44 61.24 2.88 0.06 2.56 0.24 0.10 1.75 97.9PE 22.0 6.3 3.8 0.4 0.5 61.4 2.9 0.08 2.68 0.30 0.28 1.65 100.5

RBK 20.52 5.32 4.06 0.63 0.58 60.18 3.62 0.07 2.40 0.32 0.02 1.63 99.35JUP 22.00 5.10 3.90 0.40 0.50 62.50 2.90 0.10 2.30 0.30 0.10 1.63 101.73SLR 20.62 5.06 3.44 0.442 0.474 60.21 2.58 0.06 2.34 0.33 0.11 1.96 97.61

16

Weekly reports

17

Weekly reports

18

Case study

• Check samples outside control limits – light elements correct after manual intervention (as drift exceeded 10%)

1. too high for some time

3. repair spectrometer

5. within spec

2. disaster

4. drift correct

19

Case study

• Check samples outside control limits heavier elements (SC detector)cannot be drift corrected

20

On closer inspection

21

Case study II: two factories use the same clinker – analysis differ from weekly reports

• First action is to check control charts and “sample 99”• Additional a reference material is analysed by both laboratories

• On face value differences do not look significant

Sample Reference SiO 2 Al2O3 Fe2O3 Mn2O3 TiO2 CaO MgO P2O5 SO3 K2O Na2O LOI Total% % % % % % % % % % % % %

Clinker 17/11/2011 CERT 21.135 4.362 3.688 0.59 0.379 66.426 2.92 0.054 0.288 0.148 0.086 0.37 100.45Factory 1 results 21.16 4.61 3.73 0.61 0.4 66.19 3.07 0.05 0.31 0.24 0.06 0.37 100.80Factory 2 results 21.04 4.39 3.68 0.59 0.39 67.03 2.96 0.05 0.27 0.17 -0.01 0.3 101.08

LSF =CaO - 0.7*SO3

2.8*SiO2 + 1.2*Al2O3 + 0.65*Fe2O3

LSF

97.8

96.8

99.0

22

Control charts Ca

Something happened and since drift correction trying to correct

Ca high

23

Control charts Si

No automatedcorrection done as drift was larger than 10%

24

After correction

Sample Reference SiO 2 Al2O3 Fe2O3 Mn2O3 TiO2 CaO MgO P2O5 SO3 K2O Na2O LOI Total LSF% % % % % % % % % % % % %

Clinker CERT 21.135 4.362 3.688 0.59 0.379 66.426 2.92 0.054 0.288 0.148 0.086 0.37 100.45 97.8

Before drift corr 21.04 4.39 3.68 0.59 0.39 67.03 2.96 0.05 0.27 0.17 -0.01 0.3 101.08 99.0

After drift corr 21.24 4.39 3.65 0.59 0.39 66.94 3 0.05 0.27 0.17 -0.01 0.3 100.98 98.1

• Difference is not is statistically significant on reference material

• Problems not totally eliminated on real material

• Representative sampling and material change with transport is significant problems that cannot be solved by fine tuning a analytical spectrometer

25

Automation – Hercules factory

• Storage• Mill and press• Sample receiving• Particle size

analyser• Conveyor belt

26

Automation – Hercules factory

Mill and pressXRD

XRF

27

But not a guarantee of a quality sure life!

• More and more industries move towards automation to improve quality, but more expensive sample preparation equipment does not necessarily guarantee smaller analytical uncertainties either.

• Pressed powder briquettes prepared on an automated mill and press system.

• Alarm bells went off when results for a batch of samples were consistently low, but the instrument check samples showed no deviations.

28

Other side of the coin –Expansion into Africa

29

Other side of the coin –Expansion into Africa

30

How do you create a culture of quality

• How do you export best practises and corporate culture?

• You don’t…….• You adapt your quality practises… fit for purpose• You convince over time and with lots of patience by

exactly the kinds of exercises I showed you• When profit increases, and the job becomes easier, you

will get they buy in you need to implement a worldclassquality system.

31

Results

• This is when Quality Control is not just conformity• This is when it becomes part of who we are and what we do

• Buy in only comes through positive reinforcement

32

By making sure of the product leaving the gates