I've just come across a blog post highlighting a major issue with cut / fill calculations performed by revit on topo surfaces. I'm really blown away by this and slightly concerned having just issued a cut / fill schedule on a job.
I've remodelled the example shown in the blog I read and confirmed the error. And its a biggy.
The pads are all 1m deep. The volumes of the pads themselves report correctly.
All pads are set with a -9m offset to create a 10m excavation.
The volumes only report correctly for 10m3 of cut. All examples above that have random and potentially massive errors.
Come on Autodesk. This needs a urgent fix!
Solved! Go to Solution.
Solved by loboarch. Go to Solution.
checking a simple and small graded topo surface also results in errors
The cut and fill volumes calculated by Revit are approximate. When you test on a very regular surface/shape like this it is very easy to find descripancies in the "correct" vs. the approximate value. On an irregular surface doing the calculation manually you would end up aproximating the calculation as well, and would probably come up with a value closer to what Revit will report.
OK. So a crucial bit of information required. To what level of accuracy are the volumes calculated? Can I write a disclaimed that says "cut and fill volumes are approximate and accurate only to ????? m3" ??
I can't say for sure. This is somewhat dependent on the topographical features, where they fall on the site, size of the site overall. I would say typically the approximation is going to be within +/- 1-2%.
In your example it looks like it is about 1% over.
I'd be happy with that as an answer as long as I can have confidence in the 98-99% accuracy. I just can't help wondering why it can't do the simple numbers in a simple example with 100% accuracy. Its quite unsettling.
russ.green wrote:
I just can't help wondering why it can't do the simple numbers in a simple example with 100% accuracy.
Short answer, the math/algorithim does not work that way.
The long answer, http://en.wikipedia.org/wiki/Monte_Carlo_method
Simple volumes of platonic solids are easy to calculate using memorized formulas. My 7th grader at home can do them. An irrgular shape like a topographical surface is a much harder problem to solve.
Thanks for the this. Very reasuring that its not just a hideous bug. Panic averted!
I would suggest a note about the accuracy is added to the documentation.
..you may want to get an idea to what accuracy the guys are setting out on site (on a rainy day, with shovels and wheelbarrow or big, beefy front loaders). Are they using electronic setting out - is this for a nuclear powerplant of a basic commercial slab?
That would give you a sense of whether there's an issue or not. The last time I cut and filled foundations...it was messy, ball-park work..with a tolerance of around 20%
thanks for the patronising comment. If it were just for a slab it would not be an issue. It is however for a huge volume and as the errors were seemingly random it needed asking to check if it was indeed a bug or other serious underlying fault which meant I could not trust the software.
My sincerest apologies for causing offence Russ. My advice was on the basis of your use of three decmal places in your reporting of earthwork volumes - something I wouldn't do.
Apologies again.
I didn't really. I'm having a bad morning.
I wish the errers were in the decimal places only......
My biggest issue was reading a blog post that said cut/fill calcs are not accurate, then looking hte documentation to see no mentions of the approximations are level of accuracy to be expected and then just hear from ADSK that volumes are approximate only before explaining the 1-2%...erm my head went into a spin and I started to wonder what else revit was deciding to do approximately without documenting it.
As it is I was doing a cut/fill model on a very large hillside to check were were in balacnce so 2% is a lot of material to potentially have to buy in or cart away if discovered later. Granted, its probably the most accurate way of doing it but if you are getting 3 decimal places out of a computer model you might naively expect the computer model to be giving you the answer to within 3 decimal places and doing so accurately. Anexplanation in the help is needed.
At least now I have an answer and can explain likely errors to be expected.
These are the results of making excavations of different sizes, starting from a small 1x1x1 (m), to a very large 1000 x 1000 x 1 (m).
According to these results, I agree with Jeff Hanson's response, above. The variations don't seem to follow a consistent pattern that we can comprehend easily if we see a test as the one shown in that blog that was referenced by Russ in the original post. But we have to keep in mind that the algorithm is made to produce an approximate result not only with simple boxes like in that blog, but also with irregular volumes, which would be the most typical situation.
In this test of mine, with simple boxes of different sizes, the results are always less or just a little more than 1% of the results that you could obtain with simple math. Notice how, for an excavation that is 100 x 100 x 1 m, more than an American football field, the "extra" volume is just 1.6 cubic meters, the 0.02 %, which is less than the actual volume that will result of that excavation because of the "swell factor" of the soil.
Edited by
Discussion_Admin
I found some incorrect entries in the first column of the topography schedule in imperial units that I posted in my previous message, and since I cannot edit that image in that post, I am posting the corrected version in this new post, for future, correct, reference:
Good morning !
Hi,
In my case of 1290173m2 projected area graded surface. It calculates -242038m3 of net cut/fill. After splitting the surface to two; total of two surfaces net cut/fill becomes -236945m3. When I merge the two surfaces again the net cut/fill calculates -230917m3. It looks like with each action on toposurface I get a decrease of 2.2% in calculated value.
Regards,
Evren
Can't find what you're looking for? Ask the community or share your knowledge.