Optimized Mix Design Proportioning Procedures COMPASS

Transcription

1 Virginia Concrete Conference 2008 FHWA CPTP Task 64 Develop computer-based guidelines for job-specific optimization of paving concrete Optimized Mix Design Proportioning Procedures Presented by: George Chang, PhD, P.E. The Transtec Group Considerations: Used by concrete pavement engineers, materials engineers, and paving concrete suppliers Balance practical and reliable For JPCP, CRCP, and patch/repair mixtures Conventional concrete-making materials 2 Job-specific optimization Why optimize? Job specific Structural design Environment Construction Early-age Long-term Yesterday Rock Sand Cement Water 28-Day 28-Day strength strength spec spec Available Materials Optimum materials and proportions Mix design criteria Today Multiple rock sources Fly-ashes, GGBFS, S. fume, nat. pozz. Multiple sands Water Multiple cements Chem. admixtures PRS PRS Warranty Early traffic opening Time Competition 3 4 How to Optimize? All Available Materials Multiple Rock Sources Multiple Sands Multiple Cements Knowledge Base Subset of Materials 2 Rock Sources 1 Sand 2 Cements Fly-Ashes, GGBFS, S. Fume, Nat. Pozz. Chem. Admixtures 1 Class F Fly-Ash 1 GGBFS 2 Chem. Admix. Computerized Guidelines Rock Optimum Materials Sand Optimum Proportions Cement Class F Fly-Ash Chem. Admix

2 Site Specific Conditions Project Type Design & Construction Info Climate Exposure Mix Expert Mix Expert Mix Design Criteria Recommendations Integrated Materials and Construction Practices for PCCP (IMCP Manual) Design Criteria Recommendations Important PCC Properties Recommended Test Methods Recommended Materials 7 8 Materials Recommendations Mix Expert 9 10 Aggregate Gradation Purpose: to determine optimal proportioning of available aggregates to Improve durability Maximize strength potential Achieve workability requirements for paving applications Minimize cost Aggregate Packing Packing Models Identified Dewar - Theory of Particle Mixtures De Larrard - Compressible Packing Model Toufar (SHRP) Characteristics Volumetric models Minimize voids in aggregate structure All validated with actual mix data Gap-Graded Mixture Well-Graded Mixture

3 Aggregate Packing Aggregate Packing Reality Checks Application of practical principles learned from construction practice Aggregate gradation for PCC mixtures Coarseness Factor Chart 0.45 Power Chart (Asphalt Industry) Percent Retained (8-18 Chart) Water Content 15 (taken from Hover 2001) Mixture Proportioning 18 3

4 How to optimize? Mixture optimization Design of Experiments Statistical Criteria Materials Cost Coarse aggregate Permeability Strength Stiffness Models Fine aggregate Cement Water SCMs 4-D Space Plastic shrink. Performance Optimum PCC mixture Chemical admixtures Fibers 19 2-factor model 3-factror model 4-factor model Number of trial batches increases based on number of factors 20 Response Models In general, a response is a property of interest that can be expressed in terms of one or more factors Statistical Response Surface Methodology (RSM) Contour Plot Response (e.g. Strength) For concrete, response models relate the materials proportions to concrete properties (mix design criteria) w/c Optimum FA% w/c FA% 21 Response Surface 22 Desirability functions Desirability functions are used to optimize for several mix design criteria simultaneously. Desirability (d i ) 1 Optimum Desirability varies from 1 at the optimum value to 0 at non acceptable values. Desirability functions Desirability (d 1 ) Desirability ( d 2 ) Desirability ( d 3 ) For each mix design criterion an individual desirability function d i is defined: Desirability ( d 4 ) Cost ($) Strength Plastic shrinkage 0 23 Overall desirability is a combination of the individual desirabilities reflecting the relative importance of each criterion 24 4

5 Overall desirability w 1 =10% Strength process 1. Define mix design criteria 2. Define initial set of trial mixtures w 3 = 30% Plastic shrinkage $ w 2 = 40% 3. Develop response models 4. Predict properties from response models for extended set of mixtures w 4 = 20% Identify optimal mix using desirability functions 26 Material Factors Statistical Trial Batches Statistical Mix Design Criteria Statistical Batching Results

6 Statistical Desirability Statistical Top Ten Optimum Mixes Mix Statistical Model Fit