1. COMPARISON OF STATISTICAL MIX-DESIGN PROPORTIONS OF HIGH STRENGTH SELF-COMPACTING CONCRETE Özlem AKALIN, Plustechno Bahar SENNAROĞLU, Marmara University

2. Outline Objectives Need for optimization of HS-SCC Statistical mixture experimental design Comparison of SMD method results with Okamura’s Rule Conclusion

3. Objectives Need for HS-SCC mixture proportioning Statistical Mixture Experimental Design Method Optimum proportions of HS-SCC (C100/115) concrete class using SMD method Comparing the results obtained from SMD method with Okamura’s Rule

4. HSC demand is increasing due to its technical and economical benefits Concrete or composite column is more economical than building with a pure steel Taking full advantage of increased compressive strength : reducing amount of steel, reducing column size to increase usable floor space or allowing additional stories without detracting from lower floors

5. The development of SCC was started in 1983 to find a solution for more durable concrete structures in Japan. Self Compacting Concrete

6. Prof.Dr.Hajime Okamura Kochi University of Technology

7. SCC is a special type of concrete that has a high resistance to segregation Adequate compaction to pour concrete Better concrete quality Shorter construction period

8. Concrete design is an optimization of mixture Concrete Classes (TS EN 206-1) C8/10 C12/15 C16/20 C20/25 C30/37 C35/45 C40/50 C45/55 C50/60 C55/67 C60/75 C70/85 C80/95 C90/105 C100/115 HSC

9. HSC mixture proportioning * HSC mixture proportioning is a more critical process than the normal strength concrete. Many trial batches are required to generate data that enables the researcher to identify optimum mixture proportions. *ACI Manual of Concrete Practice,1997.

10. Mixture Experiments The measured response is assumed to depend only on the proportions of ingredients present in the mixture and not on the amount of mixture Experiment and you’ll see! (Cole Porter)

11. Mixture Experiments A q-components mixture in which represents the proportion of the i th component present in mixture, The composition space of the q components takes the form of a regular dimensional simplex.

12. Physical, theoretical, or economic considerations often impose additional constraints on individual components * Quenouille, M.H

13. Mixture Experiments The purpose of mixture experiments is to build an appropriate model relating the response(s) to components. Most commonly used mixture model forms in fitting data are the second-degree polynomials introduced by (Scheffé, 1958) of the form

14. D-optimal Design for HS-SCC was used to mathematically model the influence of eight mixture parameters and their 2-way interactions on responses 8 mixture parameters Cement (c), Silica fume(sf), fly ash(pfa), water(w), natural sand(n-s), crushed sand(c-s), aggregate(agg), chemical admixture(adm) Responses T50 slump flow time, Slump Flow, Compressive Strength, Appearance, RCP

15. Constraints on Mixture Components (L/m 3 ) ComponentIDMimimumMaximum Cementc109.65172.28 Silica fumesfsf7.9527.58 Fly ashpfa18.0685.31 Natural sandn-s108.08175.63 Crushed sandc-s137.33206.06 Aggregateagg338.68414.77 Admixtureadm610 Waterw139.99160.02

16. D-optimal Design Mixtures Proportions (L/m 3 ) Runcwsfpfan-sc-saggadm 1126.37139.9927.5885.31108.08137.33347.0410.0 2121.61139.997.9518.06131.99137.33414.7710.0 3131.16160.0218.2918.06108.08201.41338.686.0 4131.16160.027.9518.06175.63138.29340.5910.0 5116.82147.177.9578.38108.08178.61338.686.0 6156.46140.197.9518.27108.08206.06338.686.0 7131.16160.0227.5851.92108.08137.33355.6110.0 8172.28158.137.9553.25108.08137.33338.686.0 9156.24139.9927.5818.06123.33167.82338.6810.0 …….46 mixtures

17. Experimental Test Results Run Flow (cm) T50 (S) U.wt. (kg/m 3 ) App.By Sight 1 day (MPa) 7 days (MPa) 28 days (MPa) RCP (C) Cost ($/m 3 ) 1709.52.425527.598.0129.211.9122.6 25711.42.46324.357.380.040.375.2 3674.62.456417.471.994.130.294.2 4715.52.45322.345.864.961.277.2 5708.72.438518.875.0103.040.272.4 66513.72.469531.997.3123.840.577.5 ……46

18. Analysis of mixture experiment requires 1)Developing regression model relating response variable to components 2)Use of model for prediction and optimization Second degree Scheffé Polynomials are considered since observations indicate that the interaction terms are important

19. Statistical Analyses Results ResponsesSPRESSR-Sq (%) R-Sq (adj) (%) MSEModel P-value Lack-of- fit P-value Flow7.02224839248.295.4079.3249.320.0030.076 T504.6521922221.495.3779.1621.640.0030.059 1 day3.543366958.297.3888.2312.550.0000.280 7 day3.4232414290.898.9295.1311.710.0000.017 28 day4.1524512791.398.9095.0317.240.0000.068 RCP7.3149853385.092.9768.3853.510.0150.405 Appearance0.834594157.02691.5561.980.6960.0310.914

20. Desirability Objective Function where n is the number of responses in the measure. The numerical optimization finds a point maximizes desirability function. In this study desired response parameters were defined as target, maximum or minimum by giving importance degree and response optimization suggested input variables by predicting responses and desirability are tabulated in table.

21. Optimization Targets ParametersGoalLower (L) Target (T) Upper (U) Weight (W) Importance (r ) Flow in. (cm) Target25.59 (65) 27.56 (70) 29.53 (75) 15 T50 (s) Target45615 C. Strength psi (MPa) Target15950 (110) 16675 (115) 18125 (125) 15 AppearanceMaximum 45514 Cost, $/yd³ ($/m³) Minimum 53.5 70 53.5 70 68.8 90 14

22. Optimization Solutions Components(L/m³)(kg/m 3 )(lb/yd 3 ) Cement138.5431149.5 Water150.715152.4 Silica fume14.23110.8 Fly ash20.84515.6 Natural sand147.0385133.6 Crushed sand165.8441153 aggregate338.7914317 admixture6.06.42.2

23. Comparison of results Predicted Results From analysis Confirmation test results Trial&Error results Slump Flow71.0 cm69 cm65-72 cm T50 Flow time5.00 s8.00 s4.8-9.6 s Compressive Strength at (28 days) 116.0 MPa106 MPa61-121 MPa Appearance4.885NA

24. Comparison of Results (L/m 3 )Trial & ErrorStatistical Mixture Cement  144 138.5 Silica fume  16 14.2 Fly ash  33 20.8 Natural sand111-146147.0 Crushed sand138-170165.8 Aggregate312-370338.7 admixture  9 6 Cost ($/m 3 )97.687.6

25. Okamura’s Rules for SCC 1) The volume of cement and fine powder: 170 <V c +V f = 187 < 200 2) Water/(cement+fine powder) by volume: 0.85 <V w /(V c +V f ) = 0.96 < 1.20 3) Volume of coarse aggregate : V G ≤ 340 L/m 3 4) Maximum size of coarse aggregate: D max ≤ 20

26. Okamura’s RulesSMD’s Optimum Results 0.85 <V w /(V c +V f+ V sf ) < 1.200.85 <151/(138.6+14.1+21.0) = 0.87< 1.20 170 <V c +V f +V sf < 200 (L/m 3 )170 <173.7< 200 V G ≤ 340 (L/m 3 )V G =338.5< 340 D max < 25 (mm)D max = 12 < 25 Comparison of SMD results with Okamura’s Rules

27. Conclusion  Statistical experimental design provides systematic approach for concrete design,  Mixture experiments give advantage to reach optimum proportions of concrete mixture components at a minimum cost,  Results of SMD for HS-SCC (C100/115) are confirmed with Okamura’s Rules for SCC.

28. THANK YOU