OPTIMIZATION OF STOCKPILE BLENDING WITH THE ABB SPECTRAFLOW ONLINE ANALYZER AND RMP SOFTWARE

AT FAUJI CEMENT COMPANY, PAKISTAN

Rais Ahmed, General Manager (Plant) Fauji Cement Company Ltd.

Islamabad, Pakistan

Christian Potocan, Product Manager SpectraFlow ABB Switzerland Ltd.

Baden-Daettwil, Switzerland

Summary For the expansion of the production Fauji Cement Company built a new second kiln-line. At this occasion Fauji Cement Company also invested in a SpectraFlow Online Analyzer and a PreBlending Software to optimize its Stockpile blending as in the expanded operation both lines are fed with material from the existing stockpile and therefore the turnover at the pile will be significantly higher. This optimization is based on the Raw Material analyses of the SpectraFlow Online Analyzer, which are used in the Preblending Software to evaluate the best setpoints of the Clay and Limestone feeders into the crusher. This paper will explain the setup of Fauji Cement’s operation and the results of the Stockpile blending optimization with a special focus on the ABB SpectraFlow Online Analyzer optimization and performance.

Fig. 2 The stacking overview of the PreBlending Software at the crusher. The average LSF over the last stacking vs. the last 500 tons are shown

Fig. 1 Overview of the Control Circuit at Fauji Cement

Operational Setup at Fauji Cement Company

Fauji Cement Company is using two feeder bins before the crusher. These bins are filled with Clay resp. low LSF material and limestone. The LSF target is always between the chemical composition of the bins. By adjusting the speed of the feeder out of the bins, the mixing ratio of low LSF material versus high LSF material can be adjusted. The material is forwarded via the ABB SpectraFlow Online Analyzer to a circular stockpile with a capacity of 55’000 t. The pile is feeding Line 1 with a capacity of 250 t/h and the new Line 2 with a capacity of 570 t/h. Before each Raw Mill 3 additive bins with high LSF material (High Grade Limestone), low LSF material (Clay) and Iron Corrective (Laterite) are available and as corrections in both ways are possible the pile can be adjusted right on the target of the Raw Mill LSF setpoint. To control the Raw Mill LSF setpoint ABB installed 2 sets of online control software: The PreBlending Software controls the apron feeders out of the crusher bins on the basis of analysis of the ABB SpectraFlow Analyzer and the Raw Mix Proportioning Software controls the additive input at the Raw Mill on the basis of the XRF hourly samples.

The ABB SpectraFlow Online Analyzer By introducing the SpectraFlow Online Analyzer ABB made its well-known Near-Infrared technology available for the minerals industry. Advantages like the absence of a radioactive source as well as low maintenance and investment costs while achieving very accurate measurement results makes the SpectraFlow Analyzer very attractive especially for the cement industry. The SpectraFlow Online Analyzer can not only be used for the measurement of raw material on a conveyor belt but also - unique on the market - for the online measurement of Raw Meal, Kiln Feed or Finished Cement in the Airslide. Combined with the well-known optimization software Expert Optimizer the SpectraFlow Online Analyzer is a powerful device for the cement industry to achieve a better quality performance and to reduce costs.

Fig.3 The ABB SpectraFlow Online Analyzer in operation

Installation and Optimization of the ABB SpectraFlow Online Analyzer The ABB SpectraFlow Analyzer was commissioned in October 2011 and is until today running continuously. The optimization is based on a back-calculation of the XRF Raw Mill results as this is the only possibility to evaluate the performance of the SpectraFlow Online Analyzer. A direct spot sampling from the conveyor belt would only deliver random results as the SpectraFlow Analyzer is scanning the material every 7 seconds and is averaging 9 scans to an one-minute average value. At Fauji Cement Company the throughput at the crusher is about 2000 t/h and therefore each one minute value of the SpectraFlow Analyzer represents around 30 tons of Raw Material. It is obvious that it is not possible to compare 30 tons with a spot sample of some kilos, especially if there are high variations in the material. The back-calculation compares the average SpectraFlow results with the average XRF results of each constituent over one complete pile, which is at Fauji Cement Company 55’000 t. To be able to compare the results all influences on the material stream from the SpectraFlow Analyzer to the Raw Mill Sampling Station have to be considered. First the additives have to be subtracted according the hourly feeder setpoints at the additive silos. To be able to do this the additive average chemical composition has to be evaluated. This has to be combined with the hourly mill throughput. The result is the chemical composition of the raw material from the stockpile and all these results have to be averaged over the complete pile. The SpectraFlow one minute results also have to be tagged with the one minute belt weigher values. These values have to be moisture corrected as the XRF Raw Mill results are also dry values. The moisture measurement is made by the ABB SpectraFlow Analyzer itself and no additional moisture measurement device is needed. Afterwards all the SpectraFlow results have to be averaged over the complete pile. Now the averaged back-calculated XRF value for each constituent can be compared with the averaged SpectraFlow value for each constituent. The difference of these values is the offset.

Since the analyzer has been commissioned two Optimizations were performed to firstly evaluate the rough offsets and secondly to adjust the offsets right to perfection. These two phases were necessary as before the analyzer was commissioned the variations on the stockpile were very high. This required additional emergency corrections of the pile had to be done via a separate feeder, which was not covered by the SpectraFlow Analyzer. Therefore a huge amount of unknown material was forwarded onto the pile which explains why the optimization had to be done in two steps.

stockpileoftonstotal

tonsCaOtonsCaOtonsCaOOffset Stockpilecomplete Stockpilecomplete

tweigherCrusherBelutewSpectraFloStockpilecomplete

nvehourAdditineAvgAdditivlhourRawMeaXRF

CaO ___

)***(_ _

_min_

...1...1

)(

∑ ∑∑ −−

=

Performance of the ABB SpectraFlow Online Analyzer and the ABB Control Software The first raw material from the stockpile after the first Offset Correction was reclaimed on 27.11.2011. The impact of the ABB SpectraFlow Analyzer and ABB control software was significant, as the LSF Standard Deviation of the hourly Raw Mill XRF values was immediately reduced from 4.385 with a Limestone/Clay additive consumption of 11.778 % to 2.695 with a Limestone/Clay additive consumption of 3.267 % with the SpectraFlow Analyzer. Therefore no more additional corrective material via an emergency feeder had to be forwarded to the pile and the second optimization could be started. In the beginning of January 2012 the first raw material after the second offset correction was reclaimed and from then on the LSF Standard Deviation of the hourly Raw Mill values was decreased to 1.503 with an average Limestone/Clay additive consumption of 1.248 %. Laterite as an Iron corrective is added about 5 t/h generally to the mix as iron is very low in all quarry raw materials. To summarize after the final optimization the combination of the ABB PreBlending Software with the analytical results of the ABB SpectraFlow Analyzer ascertains that by taking more than 96% of the raw material from the stockpile the Raw Meal has already Kiln Feed quality.

Fig. 4 Overview of the ABB SpectraFlow Online Analyzer performance after the 1st and final offset correction. The LSF Standard Deviation at the Raw Mill vs the Additive Consumption

STDEV4) SpectraFlow XRF Deviation Consumption  of  PILE  1 RM  22) 1.32 133.6953 133.1676 0.5277 PILE  1

RM  11) 1.61 MIX3) CLAY LATERITE TOTAL

RM  22) 0.97 53083.61 2901.29 1449.60 57434.50 t

PILE  1 RM  22) 0.20 45.7054 45.7451 -­‐0.0396 92.42 5.05 2.53 100.00 %

RM  11) 0.21

RM  22) 0.17 PILE  2

PILE  1 RM  22) 0.17 10.6967 10.7933 -­‐0.0966 MIX3) CLAY LATERITE TOTAL

RM  11) 0.20 48509.18 3430.65 1003.93 52943.76 t

RM  22) 0.11 91.01 7.00 1.99 100.00 %

PILE  1 RM  22) 0.07 2.8277 2.7754 0.0523 RAW  MILL  11)  /  LINE  1

RM  11) 0.09 23233.83 2108.65 478.53 25821.01 t

RM  22) 0.06 89.94 8.20 1.86 100.00 %

PILE  1 RM  22) 0.11 1.2957 1.2306 0.0651 RAW  MILL  22)  /  LINE  2

RM  11) 0.08 25275.35 1322.00 605.40 27202.75 t

RM  22) 0.05 92.91 4.85 2.23 100.00 %

PILE  1 RM  22) 0.04 1.2750 1.2805 -­‐0.0055

RM  11) 0.03

RM  22) 0.03 1)  RM  1  is  without  ABB  add.  feeder  control

PILE  1 RM  22) 0.01 0.3903 0.3841 0.0062 2)  RM  2  is  with  ABB  additive  feeder  control

RM  11) 0.01 3)  Raw  Material  from  Stockpile  

RM  22) 0.01 4)  STDEV  from  hourly  XRF  Raw  Mill  values

1.3009 1.3158 -­‐0.0149

0.3777 0.3819 -­‐0.0042

0.0931

2.6781 2.7367 -­‐0.0585

1.2271 1.3660 -­‐0.1389

PILE  2

PILE  2

137.0525 137.2077 -­‐0.1552

45.9125 45.8271 0.0854

10.5316 10.4385

CaO

PILE  2

PILE  2

PILE  2

PILE  2

PILE  2

LSF

SiO2

Al2O3

Fe2O3

MgO

K2O

Fig. 5 Overview of the Raw Mill performance after the final optimization of the ABB Control Package

Performance Validation of the ABB SpectraFlow Online Analyzer

After implementing the final offsets and some months of successful operation another comparison of the ABB SpectraFlow Analyzer values with the back-calculated XRF Raw Mill values was made to validate the accuracy of the SpectraFlow measurement. 2 complete piles from the 06.03.2012 to the 20.03.2012 were stacked and reclaimed with a total amount of more than 110’000 tons. The accuracy of the SpectraFlow Online Analyzer measurement was very high, as there was actually no difference with the XRF values. The calculation itself can be considered as very good, as more than 91 % of the total mix was material from the stockpile, while around 2 % was Laterite, which is a very stable additive and does not add uncertainties. With the highly accurate SpectraFlow Online Analyzer measurements the PreBlending software could adjust the feeder setpoints at the crusher bins perfectly, which resulted in a very stable stockpile on the target LSF value. The stability and homogeneity of the stockpile is also visible in the very low LSF standard deviation of the hourly XRF values. During the reclaimation of pile 1 only Raw Mill 2 was in operation and with the Raw Mix Proportioning software controlling the additive feeders the LSF standard deviation of the hourly XRF raw meal values was at 1.32. The LSF standard deviation of the hourly XRF values after the Raw Mill 2 of pile 2 was even better at 0.97. Raw Mill 1 is run without ABB’s Raw Mix Proportioning software, which is automatically adjusting the additive feeder setpoints according the hourly XRF analysis of the Raw Meal and therefore the LSF standard deviation was at 1.61. Conclusion The combination of the Preblending Software, the Raw Mix Proportioning Software and the ABB SpectraFlow Online Analyzer resulted in the direct tangible benefit for the customer that the amount of additives was significantly reduced. Minimum 90% and up to 98% of the Raw Meal comes now from the stockpile. This does not only apply to the new line but for the existing line as well because the raw mills of both lines are fed from the same stable stockpile. The less tangible effect is that the LSF standard deviation of the hourly raw meal feed is significantly reduced, which means less energy consumption in the homogenization silo and a smoother kiln operation.