Two reports come to rest on the same desk. They are both professional and have used adequate equipment. One assigns to the pile a volume that is a few thousand cubic metres more than the other. Which one is correct determines what the payment certificate, the inventory figure, or the invoice will say.
The simple answer is that it doesn't follow that either of the surveyors has made an error; the majority of disputes over stockpiles begin with questions regarding definitions rather than with measurements—specifically, what constitutes the bottom of the pile, where the pile ends, and what figure is multiplied to arrive at the number of tonnes.
The article we wrote about breaklines explained how a smooth surface can cause the lines to be lost at places where the shape of the ground changes, such as at the toe of a stockpile. This next point takes it a stage further by asking how you are to find out the reason when two surveys give different answers, and also how you can prevent this from happening once more.
A Volume Is the Space Between Two Surfaces
A stockpile volume is not a measure of the pile but a calculation based on two surfaces, namely the top of the pile and the surface of the ground on which it stands, this being restricted by a boundary surrounding the pile; if any of the three elements or the way in which they are combined is altered, the figure will change; and when the result is given in tonnes, a density assumption is added on.
All disagreements arise from one of six sources: the base surface, the boundary, the top surface, the calculation method, the survey date, or the conversion to tonnes.
1. The Base Surface
This is the least obvious and at the same time often the largest source of variation. According to DroneDeploy's mining guidance, the selection of the reference surface is often what causes the greatest discrepancy between surveys, which is the reason why it is more important to be consistent than to rely on any one method.
The common options give different answers:
- a flat plane at a chosen elevation, which only works on level, prepared ground
- a surface interpolated from the ground around the pile
- a survey of the original ground taken before the pile was formed
Birdi's reconciliation guide stresses that the base must show the original ground level prior to the pile having been constructed, and the issue can be eliminated by carrying out a verified base survey before a new pile is built. This relates to the DTM aspect of the distinction between DSM and DTM that we discussed in our earlier post on engineering-ready deliverables. The top of the pile is a surface model, while the bottom is a terrain model and a decision has to be made as to what it is.
2. The Boundary (the Toe Line)
It's clear where the pile ends and the ground begins on a clean conical pile; on an actual one, though, with spillage and a mixture of materials and vegetation at the edges, it isn't.
Propeller carried out direct tests on this issue. When it tried to measure the same stockpile 20 times, it found that tight outlines yielded a volume of about 2,900 cubic metres whereas generous outlines gave about 3,000 cubic metres; this represents roughly a 3% difference based just on the outline using the same data. A research paper concerned with footprint delineation reaches the same result from the other side of the question: that footprint delineation is still one of the main causes of error when it comes to stockpile volumes, and the volumes calculated from an automatically generated concave hull are closer to those obtained from a manually drawn outline than the volumes derived from a convex hull.
The toe serves as a breakline; if two surveyors draw it differently, then they are dealing with two different piles.
3. The Upper Surface: Method and Coverage
It is also important how the surface is captured. A total station records separate points along the base and crest, while a drone or laser scanner records a dense surface. In one of the DroneDeploy case studies, a survey team used a robotic total station to gather 65 points, 33 of which were taken around the base and 32 around the crest. The area between those points is then interpolated. Photogrammetry too has its own shortcomings, with Propeller listing steep faces, overhangs and vegetation creep as the features which make a pile harder to model.
It's useful to have an idea of what normal is. Vendor figures give the accuracy of drone stockpile measurements as one to two per cent when good control is exercised, and these should be treated as claims. Peer-reviewed comparisons involving a single test pile show a few per cent for both methods; Arango and Morales found the total station volume to be 2.88% off the actual volume and the UAV volume to be 0.67% off in the other direction. A study of a twin gravel stockpile showed 2.94% for the total station and 2.31% for the UAV. Both of these are small-scale tests, and neither establishes a general rule. Yet they do indicate that two skilled surveyors can differ by a few per cent simply because of the method used.
4. The Calculation Method
The volumes obtained from the same points will vary according to the way the surface is constructed and the method used to calculate the volume. As noted in the footprint paper, TIN surfaces provide more accurate volumes than grid models. When the three common methods were compared, the contour method proved to be less precise than both the grid and TIN methods. The size of the grid cells, the software defaults, and surface smoothing all have an effect. Another area to investigate would be if the two surveyors had used different software settings.
5. The Survey Date
The stockpiles are not fixed; material is removed, added, compressed, and worn away, as well as altered by weather conditions. The amount of material recorded when a survey is carried out on the first of the month will be different from the amount recorded when a survey is carried out on the tenth unless you take into account the material that moved during that time. Birdi includes the gaps between survey dates and the actual movement of material among the four most common causes of reconciliation discrepancies.
6. Conversion from Cubic Metres to Tonnes
The amount recorded in many reports is given in tonnes, which requires multiplication by a bulk density, and this density is based on an assumption. To illustrate, if on a pile having a volume of 50,000 cubic metres the assumed density is changed from 1.60 to 1.55 tonnes per cubic metre, the reported inventory changes by 2,500 tonnes even though no material moves anywhere. Both moisture content and compaction affect density, and Birdi's help documentation states that wet sand is about 20% heavier than dry sand.
A study that has been published demonstrates the extent of the effect in real terms. When a survey of a coal stockpile in Indonesia was carried out using a total station, the result was 71,162 cubic metres and approximately 64,046 tonnes; this figure differed from that obtained by means of weighbridges by 7,916 tonnes, or 12.36 per cent, a deviation that exceeds the tolerance suggested in proper mining practice. The authors point to the moisture content of the coal at the time of weighing and to the different base model employed in the DTM calculation. These are two of the six sources mentioned above, relating to a single site.
So Which One Is Correct? Instead of Voting, Reconcile.
- You should compare the surveys with each other by placing them on the same coordinate system and using the same vertical datum; otherwise, a mismatch in vertical datum can cause the entire set of data to be shifted, a problem we covered in our earlier post on datum mismatches.
- Request that each surveyor provide the same information: the date of the survey, the method used, the results from the control and checkpoint observations, the definition of the base surface, the boundary polygon, the calculation method and settings, and the density employed. We have already explained how to verify an accuracy claim elsewhere, and the same evidence applies in this case.
- Recompute the two datasets using the same base surface and the same boundary. This is exactly what DroneDeploy's case study did when it incorporated the surfaces in order to ensure that the comparison base planes and takeoff areas were identical. If the gap disappears, then the discrepancy was definitional.
- Take one surface away from the other. A difference map indicates the areas where the surveys disagree: at the toe, on steep faces, under vegetation, or over the entire base.
- Compare both surveys with known points taken on solid ground next to the pile; if there is a consistent vertical offset, then this indicates a problem with the control or datum, not with the pile.
- If a gap is still there, then check the dates to see what had been added or taken away between them.
- It is advisable to check the tonnage figure last, and to compare the assumed densities, the moisture content, and the weighbridge or haulage records.
As a general rule, a difference of a few percent can be attributed merely to the method used. If the gap is much larger than that, it usually indicates that the two surveys weren't measuring the same thing.
How to Stop It Happening Again
The majority of these matters can be arranged prior to the first flight. The specification for a stockpile survey should state:
- how the base surface is defined, and who surveys it before the pile is formed
- how the boundary is drawn, and who signs it off
- the survey method, the control and the accuracy checks
- the calculation method and its settings
- the density source and the date it was tested, if tonnes are reported
- the survey date relative to material movement
- the tolerance both parties accept, and how disputes are resolved
That is the step in our pre-survey checklist about confirming what the deliverable is actually for, applied to the case of money.
The Actual Takeaway
If two surveys of the same stockpile differ, the relevant question is not which surveyor is more skilled. Rather, it is what each of them specified—that is, the base, the boundary, the method, the date and the density. A volume figure without such definitions cannot be reproduced or defended by anyone.
If you need stockpile or earthwork volumes that can be reproduced, reconciled and defended, with the base surface, boundary and method agreed before the survey, that is the work our topographic survey and drone imaging and processing teams do.