Good morning all! I created this topic to continue a discussion that @Rogerio_Penna brought up about calculating quantities on a complex stockpile. Below is the offline process that we used to test and verify the quantities and DroneDeploy’s system, as well as, my standard method for performing take-offs.

If you just happened upon this post you can visit the origin post to catch up.

So the mission started off by Rogerio inquiring about how you can possibly know that you have taken off good quantities with a complex stockpile or in his case, one that is on the side of a hill. Here’s his pile.

So here we have a stockpile on multiple slopes, so the first thing I am hearing from allot of people is that they are trying to take-off “the pile” when you should actually be thinking in terms of taking of sections. In this case there are 3 seperate planes and another pile all together so we used 4 calculations.

After doing this takeoff and feeling pretty good about the numbers we then took it to the next level and exported the XYZ point cloud and brought it into Carlson Precision 3D.

Once you get the point cloud into Carlson P3D you can do just about any filter techniques, collaboration with other CAD files and analysis that you might want to do. It very powerful and very fast even when using point clouds of 10 million + points. Here’s the point cloud in Carlson P3D. As you can see it already has the pixel colors as derived by DroneDeploy and it is on the correct projection which is embedded in the file from your choices on the DroneDeploy site. This one was exported at 5 million points which is my default for accuracy vs performance. We’ve got to get these results out quick!

From there I drafted polylines to match (as close as possible) what had been digitized in DroneDeploy. Nice thing about P3D and maybe DroneDeploy some day is 3D take-off. You can even see spot elevations if you really want to dial it in. @Jamespipe@Anya

Beyond this, I have heard from others that taking the pile of 5 different times produces 5 different results. No doubt! Try to survey the same stockpile traditionally and see what you get. The idea is to create a proven system and stick to it. Here’s my best effort.

Plan your takeoff. North to South, East to West… whatever suits you.

Use standard colors and naming conventions. Blue = Rock, Green = Topsoil and so on. By using standard naming you will make your life allot easier down the road when you want to export the data. I even go as far to label rock piles with the range of sizes, I.E. “Limestone 12-36” (inch).

Now for the takeoff. I typically start by quickly drawing the simplest shape over the pile to get a good idea of the quantity, but also to filter out smaller piles as I really don’t care about anything less than 50cy. As you can see with this one my initial number is 5,061cy.

Next I will turn on the Elevation map mode and fine tune the points. As you can see it is much easier to spot the grade breaks. This one doesn’t have much variation because it is on a 2-mile site with 120’ of variation across the site.

Dialing in the points I then use the Distance (and section) annotation to prove up and spots I am not sure of. As you can see the vertex is verified to be in the flat spot.

After dialing in the pile I get 4,774cy. Now I am comfortable with these numbers and this process because it works. Not only did I do a traditional topo, but we also tracked the trucks, We were within 2% or 100cy, or 5 trucks. The traditional survey was 8% off the truck quantity and I took more time than I would expect any of our crews to take. If you need better accuracy than that go buy a laser scanner and have fun getting to the top of the pile.

Hi all, To be more certain with the volumes of this site you need to add two components to be more confident with the result.

1.) The elevation at the base of the gravel stockpile needs to be modeled in the best way possible. The best would be a DTM prior to this gravel pile being built. Perhaps some land survey has surveyed the site in the past when the pile was not thier. Another method would be to build a dtm in Whitebox or other GIS software utilizing the area outside the pile, and interpolate a few points that reflect likely ground breaklines under the pile. Then you can simply run an analysis between the two dtms.
2.) The other item that will improve your accuracy would be to add GCPs. This could bring RMS error down from 1-2meters to 3-4cms. On a small stockpile like this that could be a significant percentage.

If this is an ongoing project, make sure to look at haul records as this pile is removed to verify your estimate. Then Whenever there is significant “base” exposed make sure to get in a flight so you will know exactly what is under the pile.

If the gravel operator is only interested in knowing haw much inventory that he has relatively close is good. But if he is trying to verify sales volume against trucking receipts, he may want to have more precision.

@Andrew_Price, thank you for sharing your experience!

The whole purpose of the original post and this share was to verify the quantities of DroneDeploy. The bottom surface was interpolated from the surrounding terrain which is how you get the volumes in the first place. The exercise was to show that by modeling in this manner you would get the same results that you get in DroneDeploy and that checked out. It is rarely feasible to expect to have a topographic survey of exactly where the pile was placed.

We have done comparisons between robotic instruments, GPS survey equipment and the drone on several different stockpiles including verification with Trucking quantities and gcps do not help stockpiles enough to substantiate the effort required to collect the data and extra processing time, not counting the fact that you have to pay for them.

I can see them helping if you have a permanent mining operation and you’re talking about hundreds of thousands of cubic yards, but in construction the stockpiles move fast and are rarely larger than 100,000cy. The RMSE values also do not apply to our stockpiles because it is all locally relative and doesn’t matter on a global scale. With Delta’s of less than 5% we are doing good.

First thing I did… circulated entire pile volume and then selected LOWEST POINT.

Got a volume of 58096 cubic meters

I also drew two lines, to get top height and lower height along the slope. Also, to know the lines lenght.

So I took a screenshot and took it to Sketchup. I imported the screenshot and using the lines as base, I re-scaled the image until the lines were the same lenght as in DroneDeploy.

Now I circulated the exact same path around the pile as in DroneDeploy. Saved the shape for later.

Remade a simple shape terrain with the heights I got from the sloped lines at Drone Deploy.

The shape I saved later… put it above the 3D simple shape of the terrain. Draped it, Deleted the borders.

Ended up with a volume that is quite similar to the extra volume I got in Drone Deploy when using the lowest point.

USed Tig’s Volume Calculator plugin, for irregular 3D volumes.

22.382 cubic meters. Subtracted it from the 58096 cubic meters.

important to note, that using only DroneDeploy, and separating the pile in 4 parts (top, 2 bottom and middle), and using lowest point for the 2 at the bottom and top part, I got 33153,4 m³