ICT and statistics as useful tools for the optimisation of the production process of concrete Peter Minne Robby Caspeele Concrete Innovation Forum - February 14 th, 2010
Statistics can be difficult
Remember this feeling?
Objection 1 However some objections Objection 2 Objection 3 Do not worry about your difficulties in Mathematics. I can assure you mine are still greater. Albert Einstein Robby Caspeele 4
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 5
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 6
Process steering approach effect input 7
Process steering approach 8
Examples of statistical tools for quality control
Probabilistic design and evaluation of conformity criteria Unsafe region AOQL = 5% EN 206-1 Numerical analyses Monte Carlo simulations 10
Bayesian statistics using all available information Updating strength distributions based on test t results Bayesian non-linear regression for updating strength prediction models Assessment of in-situ characteristic concrete strength Influence of conformity control on concrete properties p Influence of conformity control on the safety level of concrete structures 11
Design of concrete strength prediction models Co ompressive e strength f c [MPa] non-linear regression W/C 12
Updating concrete strength prediction models
Updating concrete strength prediction models 14
Quality control charts C i i x j j 1 0 15
Quality control charts Concrete production at a concrete plant 16
Quality control charts Actions should be taken in order to avoid nonconformities 17
Quality control charts NON-CONFORMITY! 18
Monte Carlo simulations Random numbers: realizations of U(0,1) pseudo-random numbers r i Monte Carlo simulations How to calculate realizations x from X with F X (x)? x i 1 F r or r F x X i i X i 19
variable 1 variable n Monte Carlo simulations Monte Carlo 140 120 simulations 100 MODEL RESPONSE 80 60 40 20 0 136.2 Frequency 141 147 152 158 164 169 175 180 186 191.8.4.9.5.1.6.2.8.3.9 Water demand 20
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 21
Water demand and consistency prediction models Modelling of solids and voids: 1 grain 3D model: cubes Volume solids: Volume voids: Voids ratio: 3 D X 3 U X D 3 3 22
Water demand and consistency prediction models Modelling of solids and voids: 2 grains m, z spatial parameters D0 md1 D D0+mD1 X" D0+mD1 X0" (1+z)D0 Volume of voids of fine grains increases due to the introduction of course Volume of voids of course grains increases (wall effect) due to the introduction of fine grains " (loosening effect) U1 U1 " U0 U 0 23
Water demand and consistency prediction models Vo oids ratio U 1.0 0.9 0.8 0.7 Power s diagram U 1 - In the range U 0 M : U 1 U n U 0 n( ( 1 U0 - In the range MU 1 : 0.6 U 0 0 U n nu1 0.5 U0 n M 0.4 (1 U 0 U ) ) ( 1 0.3 02 0.2 0.1 O 0 0.1 M X 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Fine fraction n n X X U M U0 (1 U ( 0 ) U0U1 (1 U U ( 0 U1 ) 24
Water demand and consistency prediction models Power s diagram Dewar s real mix + U 0 U 1 U n U 0 U 0 U 1 U 1 25
Water demand and consistency prediction models Power s diagram Dewar s real mix 1.2 1.1 " U 0 1.0 size ratio r=0.080 U " U 1 + 0.9 0.8 U 1 + F 0.7 U Voids ratio 06 0.6 U 0 0.5 A 0.4 0.3 B C D E = 0.2 0.1 0 0.1 M X 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Fine fraction n 26
Water demand and consistency prediction models 0.3 D 1.75 D1 0 3 D1 7.5 D1 27
Water demand and consistency prediction models 1.0 Influence of ratio of average grain size r 0.9 0.8 0.7 U0 = 0.80 r=1 A size ratio E F U 1 = 0.80 Vo oids ratio U 0.6 0.5 0.4 0.3 B r=0.2 C r=0.04 r=0.01 D r r av. grain size of fine grains av. grain size of course grains 0.2 r=0 0.1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Fine fraction n 28
Water demand and consistency prediction models cement 1+2 sand 1+2 gravel 1+2 + cement 3 + sand 3 + gravel 3 cement + sand voids skeleton 29
Water demand and consistency prediction models Modelling the consistency Reference slump = 50 mm The difference in water demand for other slump values independent of raw material and concrete parameters Empirical function for the difference in water demand F s F s 1 SL 50 6 SL 50 30
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 31
Mix design, mix proportioning: a multifunctional ICT tool 32
Mix design, mix proportioning: a multifunctional ICT tool Management of raw materials Grondstoffen d t Cement Cement Toevoegsel Toevoegsel Fijn Granulaat Fijn Granulaat Grof Granulaat Grof Granulaat Hulpstof Hulpstof Parameters Korrelkromme Abosolute volumieke massa Waterbehoefte bij standaard consistentie Blaine waarde Beta-p-waarde Cementsterkte (1, 2, 7, 28 dagen) Natrium-equivalent Chloride gehalte k-waarde Berekende ee e degoo grootheden Gemiddelde korrelafmeting Holle Ruimten Ratio Cementtypes Portlandtypes Portlandcomposietcement Hoogovencement Samengesteld cement Toevoegsels Vliegas Kalksteen Parameters Korrelkromme Korrelvolumieke massa Schijnbare volumieke massa Waterabsorptie Natrium-equivalent Chloride gehalte Deeltjes < 63 µm Berekende grootheden Gemiddelde korrelafmeting Holle Ruimten Ratio Specifiek oppervlak Specifiek oppervlak Day Fijnheidsmodulus Parameters Volumieke massa Droge materie Chloride gehalte Natrium-equivalent Types Plastificeerder Superplastificeerder Luchtbelvormer 33
Mix design, mix proportioning: a multifunctional ICT tool Concrete specifications according to NBN EN 206-1 and NBN B 15-001 34
Mix design, mix proportioning: a multifunctional ICT tool Processing of raw materials Verwerking V granulaten l ten - Last analysis - Random analysis - Mean of analyses - Mean over a time period - Previous calculated mean Descriptive statistics of variables 35
Mix design, mix proportioning: a multifunctional ICT tool Granulatensamenstelling Design of inert skeleton Design of grain size distribution - target curves - Power s diagram and Dewar real mixes 36
Mix design, mix proportioning: a multifunctional ICT tool Concrete strength th prediction models - Feret - Bolomey - Abrams -Dutron - Dewar -Buist - Hanke 37
Mix design, mix proportioning: a multifunctional ICT tool Mix Mix Design Ontwerp betonmengsel 38
Mix design, mix proportioning: a multifunctional ICT tool Mix Mix Proportioning Production of actual concrete mixes Variatie van grondstoffen Variatie in de productie Vochtgehalte grondstoffen Mix Proportioning Aanpassing recuperatiewater Beoordeling ten opzichte van bestaande Mix Design 39
Mix design, mix proportioning: a multifunctional ICT tool Updating concrete strength prediction models 40
Mix design, mix proportioning: a multifunctional ICT tool Simulation of water demand and concrete strength 41
Mix design, mix proportioning: a multifunctional ICT tool Cusum Quality control of concrete production 42
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 43
Some properties of raw materials and their variation - The raw materials which are used for concrete have intrisic properties and fabrication characteristics - All these properties and characteristics are subjected to large variations - Only limited attention is paid to these important variations -Most often only the econcrete ceestrength is systematically y predicted pedcedand monitored in time - The water demand and consistency are most often not predicted, although these are significant variables influencing many concrete properties 44
Some properties of raw materials and their variation 100 80 60 sand (fine sand 0/2) - grain size distributions analyses in time 40 100 20 80 0 60 0.1 1 10 d [ mm] Y [%] 40 20 0 0.1 1 10 d[mm] 45
analyses in time of passing through 100 Some properties of raw materials and their variation DOORVAL fijn zand 0/2 fijn zand_0/2_fractie 0.125-0.250 fijn zand_0/2_fractie 0.250-0.500 fijn zand_0/2_fractie 0.125-0.500 80 60 Y [%] 40 20 0 11-01-08 20-04-08 29-07-08 6-11-08 14-02-09 25-05-09 2-09-09 11-12-09 21-03-10 46 tijd
Some properties of raw materials and their variation sand (fine sand 0/2) - grain size distributions analyses of grain size distribution Histogram fractie 0.125-0.250mm Histogram fractie 0.250-0.500mm 30 25 25 20 Frequentie 20 15 10 Frequentie 15 10 5 5 0 9.2 13.025 16.85 20.675 24.5 28.325 Doorval [%] 32.15 35.975 Meer 0 54.7 58.0375 5 61.375 64.7125 68.05 71.3875 Doorval [%] 74.725 78.0625 r Meer 47
Some properties of raw materials and their variation sand (fine sand 0/2) - grain size distributions Log (mean size of the size fraction)=0.5(log(upper size)+log(lower size)) analyses of derived properties: mean size vol.propn x log(mean size of the size fraction Log(mean size) ) 100 80 60 mean size = 0.3145 mm 40 20 0 0.1 1 10 d [ mm] 48
Some properties of raw materials and their variation sand (fine sand 0/2) - grain size distributions Distribution of mean size Monte Carlo simulations of mean size 25 20 [mm] gemiddelde 0.3099 st. Dev 0.0187 Freque entie 15 10 5 0 0.2 794 0.29 9165 0.3 3039 0.31 615 0.3 3284 0.34 4065 0.3 3529 0.36 6515 Meer Mean size [mm] 49
Some properties of raw materials and their variation sand (fine sand 0/2) - grain size distributions analyses of derived properties: volumetric mass and voids ratio grain density (pyknometer) bulk density (recipient) voids ratio 50
Some properties of raw materials and their variation sand (fine sand 0/2) - grain size distributions Distribution of voids ratio Monte Carlo simulations of voids ratio Freque entie 20 18 16 14 12 10 8 6 4 2 0 [-] gemiddelde 1.1201 st. Dev 0.0801 0.9 9042 0.9518 8375 0.999 9475 1.0471 125 1.09 9475 1.1423 3875 Voids ratio [-] 1.190 0025 1.2376 6625 Meer 51
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 52
Influence of variation of concrete properties on the water demand variable 1 Monte Carlo simulations water demand MODEL RESPONSE 140 120 100 80 60 40 variable n 20 0.2.8.4.9.5.1.6.2.8.3.9 Frequency 136 141 147 152 158 164 169 175 180 186 191 Water demand 53
Influence of variation of concrete properties on the water demand Concrete recipe Mix Design [kg/m³] CEM III/A 42.5 LA 260 Fly ash 15 Water 170 Sand 0/2 504 Sand 0/4 393 Coarse Aggregates 6/20 954 Water reducer (Sky) 1.62 54
Influence of variation of concrete properties on the water demand Statistical characteristics of raw materials (based on an observation period of 1 year) Constituents Mean size Voids ratio mean [mm] (st.dev.) mean [-] (st.dev.) CEM III/A 42.5 LA 0.01202 (0.00068) 0.8643 (0.0126) Fly ash 0.0194901949 (0.00105) 00105) 0.6787 (0.0247) 0247) Sand 0/2 0.3114 (0.0205) 1.1463 (0.0822) Sand 0/4 1.0259 (0.1275) 0.7794 (0.1114) Coarse Aggregates 14.0771 (1.3259) 0.9183 (0.0605) 55
Influence of variation of concrete properties on the water demand Random simulation of the water demand 140 120 Simulation Water demand mean [kg] (st.dev.) Properties of all constituents are varying 169.8 (9.29) Only properties of the binder are varying 170.4 (1.18) Only properties of sand are varying 173.9(581) (5.81) Only properties of coarse aggregates are varying 171.1 (7.29) Frequency 100 80 60 0 Concrete strength results estimation of the actual variation in water demand 40 Water demand mean [kg] (st. dev.) 20 Actual variation of the water 173.5 (11.52) 136.2 141.8 147.4 152.9 158.5 164.1 169.6 175.2 180.8 186.3 191.9 56 Water demand
Influence of variation of concrete properties on the water demand 100 80 Random simulation of grain size distribution 60 40 grof_zand_onder grof_zand_boven grof_tussen 20 0 0.01 0.1 1 10 100 80 60 40 grof_zand_onder grof_zand_boven f d b grof_tussen 20 0 0.01 0.1 1 10 57
Influence of variation of concrete properties on the water demand 100 80 cumulativ ve passing (%) 60 40 20 0 0.01 0.1 1 10 diameter (mm) Simulation good coarse_sand_u good coarse_zand_a a good coarse_sand_b sand EN12620 sand NBN B11-011_u sand NBN B11-011_a sand NBN B11-011_b Water demand mean [kg] (st.dev.) Sand according to NBN 163.69 (8.78) Sand according to EN 12620 184.9 (10.97) Good concrete sand 166.5 (4.23)
Content 1. Introduction using statistics in concrete production 2. Water demand and consistency prediction models the building stones for computational concrete design 3. The Mix design, Mix proportioning software a multifunctional ICT tool 4. Some properties of raw material and their variation input for statistical simulation models 5. Case study influence of variation of concrete properties on the water demand 6. Conclusions 59
Conclusions The simulated water demand are comparable with the water demand obtained by the strength results diagrams of Powers and the theory of the particle mixtures of Dewar provide a useful method for estimating the water demand The magnitude of the variation in water demand is of the same magnitude as the variation in the actual water content the variability of the properties of the raw materials is the main origin of the variability in water demand More stringent specifications are required for the acceptable boundaries of the grain size distribution according to the standard The models can be used to predict the water demand for new mix designs and to predict the water demand when the raw materials parameters are changed 60
Thank you for your attention! Ing. Peter Minne Gebroeders Desmetstraat 1 9000 Gent Peter.Minne@kahosl.be Dr. ir. Robby Caspeele Technologiepark-Zwijnaarde 904 9052 Zwijnaarde Robby.Caspeele@UGent.be