It is now time to start discussing the “how to” of selecting aggregates for a concrete mix design. You will find there is no single “best method” for proportioning aggregates. Aggregate design methods basically fall into 2 different categories: particle packing methods and combined grading methods. Each method has multiple variations and each variation has its own strengths and weaknesses. Particle packing techniques can produce initial mixes that are very close to the optimum combination of materials, but these techniques are often difficult to use and require either expensive computer simulators or performing a lot of combined aggregate unit weight tests. On the other hand, combined aggregate grading techniques can often get you 90% of the way to a “best mix” using standardly available grading data and simple spreadsheets. The right technique for you will depend on the data available and your exact situation.
Particle packing theories assume that the way to minimize water demand in concrete is to fill as many spaces between aggregate particles with other aggregate particles. Then a minimal amount of paste and water will be required to fill the remaining spaces and produce a workable concrete. This concept has been applied to the coal industry, where they use to mix crushed coal with water to form a slurry that could be pumped through pipes, and to the asphalt industry. In the asphalt industry if they have a bad combined grading they must add expensive oil. In the concrete industry if we have a bad grading we add cheap water. Guess who is more concerned with combined grading?
All serious concrete technologists today recognize the importance of particle packing, even down to the cement size particles. Work by Francois de Larrard, Lafarge, Gunar Idorn and by iCrete has shown that particle packing can effectively be used to proportion concrete mixes. The ternery packing diagram is a common tool used to optimize particle packing. An opposite perspective on particle packing would be to look at the voids in the combined aggregate. Of course, maximum aggregate density would imply minimum voids. Work by Joe Dewar, http://www.mixsim.net/, and Eric Koehler and Dr. David Fowler, http://www.icar.utexas.edu/publications/108/ICAR%20108-2F%20(Final%20Report).pdf, examine either voids in the combined aggregate or paste demand in concrete mixes to determine the optimum combination of materials. The b/bo procedure used in ACI 211 is another variation on the percent voids method.
The biggest problem with optimized particle packing is that it can best be determined by physically combining the aggregate particles, then weighing them. Aggregate unit weight is not a linear function of the percentage of each aggregate. Combining two aggregates can result in a combined material that is heavier than either material individually. Determining the optimum combination either requires weighing a significant number of combined aggregate samples or using expensive computer models to simulate the combined material.
Another problem is that maximum density concrete mixes don’t always produce the best concrete. Back in 1918 in Lewis Institute Bulletin 1 Duff Abrams said that a mix designed by maximum aggregate density or maximum concrete density did not produce the best concrete. He chose a method based on aggregate grading – the combined fineness modulus. Studies done during the SHRP program of the 1990s appeared to look good on paper (http://onlinepubs.trb.org/onlinepubs/shrp/shrp-c-624.pdf), but word was that once these mixes started to be used in the field they were too harsh to place. When an aggregate blend is at maximum density, it is also at a point where there is a maximum amount of aggregate-to-aggregate contact which must be overcome in order to obtain a workable mix.
It is now evident that the target in particle packing should not be to achieve maximum density, but something less than maximum. This is a common practice in the asphalt industry. Duff Abrams said that more coarse aggregate should be incorporated into a concrete mix than a maximum density curve would indicate, but the result of that would be the increase the demand for paste to fill voids that were unfilled by sand. There may be better techniques, such as adding a slight excess of mortar, or by slightly decreasing intermediate aggregate, that would result in a more workable concrete.
Combined Aggregate Grading
The nice thing about combined aggregate grading techniques is that combined aggregate grading is easy to determine mathematically. Simply multiply each individual aggregate grading by the percent of that material by volume, and the answer magically appears. (I know the argument about weight % vs. volume %, but I support the volume % for reasons that will be explained in a later post.) Just about any high school math geek can write a spreadsheet to calculate combined aggregate grading. Once you have a little understanding about the importance of various aggregate particle sizes, it is easy to recognize what makes a good concrete mix and what makes a bad mix. (Of course you can also read my previous and future blog posts about combined aggregate grading to learn these “secrets”.)
There are many combined aggregate grading approaches, including the 0.45 power chart, based on work by Fuller & Thompson; the 8-18 spec for an individual percent retained chart; and various “banana” curves developed by pumpers, Departments of Transportation, and others. The Bolomey curve is based on combined aggregate grading plus cementitious content. There are also hybrids, such as Ken Day’s Mix Suitability Factor which is based on a modified specific surface calculated from the grading, or Duff Abrams combined fineness modulus. The Coarseness Factor Chart, developed by my father, Jim Shilstone, uses only the 3/8” (9.5mm) sieve and the #8 (2.36mm) sieve to divide the aggregate into three particle size classifications, then determines a broad range of combinations that will produce a good combination of aggregates.
One problem with all combined grading approaches is that none of them provides a mathematical component to compensate for particle shape and texture. Most of the techniques can be modified based on guidelines relating to shape and texture, but it is hard to program a guideline into a computer.
Over the next few months I will be discussing each of these techniques individually. However, the biggest thing to remember is that it doesn’t matter how concrete looks on paper. What matters is how concrete performs in the field. Also, aggregate is a product of nature and is inherently variable. You can spend months optimizing aggregate combinations in the lab, but each shovel full of aggregate is different, so your combination of aggregates must allow for variations in each material and still provide a good quality concrete blend.
What method(s) do you use for proportioning your aggregates? Just leave a comment on your preferences.
Until next time,