By Greg Wilkinson, Graniterock Research and Technical Services
PART 1: THE MOST IMPORTANT STEP
Representative sampling is defined as the taking of all materials in the same proportion as they exist, or will be used. The best laboratory and technicians cannot compensate for a bad sample, and all of the time and effort spent testing construction materials is a waste if the sample does not represent the material being evaluated.
Whether we are producers, contractors or owners, we must be confident that the results that we generate are reliable. Reducing the possibility of sampling error is the most important step we can take to achieve this.
Sources of Bias
Bias is defined as systematic favoritism in the sampling process, and along with poor sampling procedures, is a significant cause of non-representative sampling. Sources of bias include:
Non-random (predictable) sampling time
Improper or non-standard sampling and splitting
The use of a standard sampling procedure, or adherence to a standardized method such as ASTM D75 Standard Practice for Sampling Aggregate or CT 125 Method of Test for Sampling Highway Materials and Products Used in the Roadway Structural Sections will significantly reduce the effect of bias, and should be used without exception.
Material shipping and handling usually introduces segregation (a source of bias) into the material which is pernicious and very difficult to remove during product sampling. Segregation is also present in stockpiles, where vibration and conveyor systems cause coarser particles to fall to the edge of the pile. For this reason, samples taken from a conveyor belt are generally more representative than samples taken from other locations, since the factors that lead to segregation have yet to affect the material.
A transverse removal of all material from a conveyor belt (belt cut) at a point in production that precedes stockpiling will virtually eliminate segregation, and drastically improve sample quality. The advantages and disadvantages of this location as well as others are described in the chart below
Sampling from Stockpiles
For reasons of convenience, samples are most often taken from stockpiles. Depending on the type and condition of material, the quality of stockpile samples can vary. Well-graded materials such as aggregate base, as well as overly-dry or non-homogenous materials are particularly problematic, and standard methods should be adhered to which will reduce the effect of segregation and other factors that may compromise sample quality.
The use of heavy equipment to blend segregated material back into its original condition is recommended for large stockpiles or piles that segregate easily, and the use of sample tubes that penetrate into sand stockpiles can remove surface aberrations and improve access to more locations within the pile.
In all circumstances, an understanding of the need to represent all of the material present in the sample and the importance of representability will serve as the basis for effective sampling. The improved efficiency and accuracy of the data obtained through testing only the highest quality product samples will justify the efforts required.
|Belt Cut||Stockpile||Field Sample|
|Quality||Excellent||Good, if done carefully||Fair/Poor|
|Represents||Material produced or loaded. Product blend||Final.intermediate state of material used.||Final state of material to be used.|
|Advantages||Low potential for segregation. Larger samples yield more accuracy.||Easy to obtain. Good for chip seal aggregates (single-sized).||Best representation of material to be used in its final state. May reflect segregation or degradation from shipping.|
|Disadvantages||Larger samples needed, less conveient, difficult to obtain.||Difficult to get representative sample for well-graded products.||Difficult to get representative sample for well-graded products.|