Grain size distributions

[kcollins6778]

I am not an engineer, but I have been asked by one to help with creating a grain size distribution chart and I honestly have no idea where to start. Can someone send me an example of the dataset that is required to produce one of these charts (preferrably in Excel)? I would really appreciate it!!

[DPlotAdmin]

Karen,
There's nothing really magic about a grain size distribution plot; it is just grain size in MM on the X axis and % finer by weight on the Y axis. For example, here are the values for the "gravelly clay" curve in the EX07 example plot:

26.0001,100
19.4553,98.5911
13.0582,96.8746
9.71421,95.329
6.52354,92.8169
4.82443,91.032
3.48282,89.0852
2.04661,86.876
1.23944,84.9089
0.867747,83.5154
0.607458,82.281
0.440892,81.369
0.308626,80.2142
0.218626,79.3789
0.154921,78.0664
0.108451,76.832
0.0755457,73.9264
0.0711239,73.5217
0.0604565,71.9119
0.050485,69.6636
0.0404271,66.6946
0.0301384,61.967
0.0203072,54.4435
0.0100982,39.1682
0.00906507,37.4852
0.00818797,35.6437
0.00704737,33.4785
0.00602838,31.4716
0.00497372,29.1424
0.00402923,26.9702
0.00301835,24.073
0.00202739,21.2428
0.00192012,20.9978

You can select those numbers with your mouse, right-click and select "Copy". Then click the Paste button in DPlot or press Ctrl+V. You'll initially get a linear X, linear Y plot. Right-click within the plot and select "Grain Size Distribution". That's it. You might also want to force the extents to values different than what DPlot selects by default. To do so select Options>Extents/Tick Marks/Size and check the "Specify extents" box.

If your grain sizes are in some units other than millimeters, right-click on the plot and select "Units".

If your Y values are % coarser rather than % finer, select Edit>Operate on Y and enter "100-Y" for the formula.

Let me know if this doesn't answer all your questions.

[kcollins6778]

Ok, I understand how your example works and I was able to create a chart. I guess now I just have to figure out how to get all those data values. The text file that my engineer gave me contains the Sieves and the percent finer; not the actual grain sizes. As I said, I am not an engineer so I'm not sure how the entire process works to get the data, but is there a way to convert the Sieve labels to grain size or does it not work that way?

Thanks so much for your help!!

[DPlotAdmin]

Karen,
DPlot won't take sieve numbers as input because, unfortunately, there isn't a linear relationship between opening size and sieve number. You'll need to swap your sieve sizes for the grain size before importing. Below are the sieve sizes built in to DPlot. If these don't cover your needs you can Google "ASTM sieve" to find several references that might include additional sizes.

Code: Select all

Sieve size    Opening in mm
6"            152.4
4"            101.6
3"            76.2
2"            50.8
1-1/2"        38.1
1"            25.4
3/4"          19.05
1/2"          12.7
3/8"          9.525
#3            6.35
#4            4.75
#6            3.35
#8            2.38
#10           2.00
#12           1.68
#16           1.18
#20           0.850
#30           0.600
#40           0.425
#50           0.300
#60           0.250
#70           0.212
#100          0.150
#140          0.106
#200          0.075
#270          0.053
#400          0.037

[Probable1]

Karen,

I happen to be in the business of "creating grain size distributions". My company designs builds and sells pulverizing and classifying systems for mostly non-metallic minerals. I work with distribution curves all the time. First, the grain size is based on the assumption of a spherical partical which will pass through a given mesh opening. The original opening series was to be based on the "root of 2" series, or a sequence either multiplied or divided by 1.414. The secondary series when more precision was needed was the quarter root of two series or as before but a number of 1.1892. ASTM is a good place to go, but the two major suppliers of screens for determining a grain size distribution from a sample, is Tyler Screens (Might be owned by someone else, we used to own them) and U.S. Screens. Plus, what you will find is that they will differ in both the "Mesh" designation in a few places, but they will differ on the openings as well. The term "Mesh" means the number of screen openings in a line per inch. It will be the same in either direction. So, a 12 mesh will have 12 openings per inch, with 144 openings per square inch. The differences between companies, and in industrial size screens is the diameter of the wire. This affects the actual size of the opening. The heavier duty the screen, the thicker the wire, the more likely the opening being smaller than the same mesh screen such as a test screen size, 9" dia. round if I remember correctly, and about 3" deep. They can be stacked up to 7 high, the sample placed on the top screen, the stack placed in a "shaker" for 15 minutes or so. Then each screen, top to bottom, has the residue left on it weighed. If you are doing a cummulative percent passing, you leave each screens residue on the scale pan, and add the next screens residue to it. From the weights at each point, depending on the weight you started with, you can calculate the %passing for each Mesh, size opening In., mm, or microns. You will have one shaped curve if you use a linear Y axis and a Log X axis using mm or microns. I have communicated with DPlotAdmin for a method of having the percent passing be the Y axis and be probability, and the screen openings be linear with the root of 2 openings be the major lines and the quarter root of two openings be the mid lines. On Naturally occuring materials, such as limestone and similar materials, it produces a straight line about half way then there is usually one but sometimes two breaks (Bi-modal). Each section still plots as straight lines. It enables you to see right away if the data is bad, from sampling, or screening method or if there was blinding of a screen during testing or even if one had a small hole in it.

Sorry to run on like this, but I have 35 years of experience. I will post my work e-mail to the DplotAdmin. You can e-mail him to get it and contact me Monday if you want. You can describe your problem or needs more clearly and I will help you.

Probable1

[kcollins6778]

I just wanted to say thank you very much for the fast and thorough replies to my questions. This is by far the best customer support forum I have ever participated in and in my line of work, I use quite a few of them. I am very impressed by the service and from what I see so far of the product.

Thank you again!

[DPlotAdmin]

Karen,
Thanks very much. This is a good way to start a Monday morning