Computing Earthworks With Many AOIs and Many Strata Layers

This one comes up a lot with people doing Takeoffs on larger projects where Rock and other strata are a big issue because Rock requires excavating and other excavated materials like sand may be usable for structural fill and site improvement construction if processed correctly. In this post I explore some things that you should take into account when modeling bigger projects with multiple AOIs and multiple strata layers.

To start with I have a couple of pointers here

When you build the boring log strata list from the geology or geotech report, be very aware of which material layers you really care about and which ones can be merged together i.e. Silty Sand and Sandy Silt are going to be very similar in nature - one has more sand content the other has more silt content but they are still both a mix of Sand and Silt. When you look at how these were created - it would be typically by Water or Wind Action that either blew the material to where it ended up or placed it in a flood plain or lake / ocean floor through settling out after a flood event. The bigger the flood event the more sand you will have (faster flows pick up bigger materials) and the smaller the flood event the more silt you will get (slower flows pick up smaller materials but leave bigger materials behind). Also when you have sand and silt in the water body sand drops out faster in still conditions and silt drops out last because it is held in suspension longer - so where a geology report shows “hard bands” they may be different flood events and inside those are the settling out effects (sand at base and silt higher up), so you will get lots and lots of bands over time and various mixes in the column - however unless you need to separate them for some reason treat them all as a single layer called Sand and Silt for example. If there are clear bands of Gravel or Sand that you want to know the quantities of because you can really use them, or if there is a band of a type of clay that you have to waste because it cannot be used for fill for sure treat them as bands / layers in the boring logs. Golden Rule - keep the list of strata as small as possible but as broad as needed. The fewer the strata the faster the results will be computed.

If you have e.g. an Upper Sand and a Lower Sand separated by a silt band - you can either create strata as follows

Top of upper sand
Top of silt
Top of lower sand
Bottom of lower sand

or you could do the following

Top of sand
Bottom of sand

and then say that the Sand is a material type that only 90% is usable for fill (the other 10% or a defined % number based on the thickness of sand and the thickness of silt in the boreholes where the materials show up) then you will get a quantity of usable sand and a waste amount that cannot be used for fill which equates to the silt material found in the sand body. While maybe “not as accurate” we are talking geology here and we are also talking very few boreholes typically across the site so minimal amounts of information to work with and either way you won’t know till you expose the materials during construction. The benefit of this approach is that you just reduced your computations for these specific strata quantities by 50% (read time reduction).

Moving on - the following are tips that you can use to get your Takeoff Report computation times down significantly - we are talking 70% reduction in time at least. The following is from an email analysis I did for a customer this week which you may find helpful. The email analyzes their project - a large site project where there is a risk of a lot of rock excavation on the site. The analysis was to show how you can reduce the time element of computations significantly.

Email Content

  1. Bear in mind that over the whole site area you only have just 17 boreholes to interpret the geology from - that is just 17 points per strata surface over the entire area of the project

  2. The TIN Model of your original ground contains 34376 vertices (points)

  3. If I do a 5’ Grid over the site I have 34764 vertices (so almost the same number)

  4. If I do a 10’ grid over the site I have just 8691 vertices

  5. If I do a 20’ Grid I have 2171 vertices

I use the RPS Surface to Grid command to turn the OG surface into a Grid point cloud and then add the Grid Point Cloud to my OG surface using Add / Remove surface members. The trick here is to not have anything else categorized as OG but you need at least one empty layer (no data on it) in the OG Category for the OG surface to be created (but there is no surface content because there is no data yet). That way I can create the OG and strata and then lock the strata (set rebuild to by user) and then revert my OG back to its original state by recategorizing the linework etc. layers and removing the grid point cloud that was created by surface to grid command.

The bigger the grid size the more noise you get in the OG surface - in the slicer below at exaggeration 1 I have the Full OG and the 20’ Grid surfaces showing in just one slice through the site

At a macro level you can see that there is little difference between the two - so when I take this surface (either one) and use it for strata analysis I am not going to see a huge difference in strata models

At a micro level you can see that over a 20 foot section, the differences can be of the order of a few feet in elevation but they will sometimes be a positive difference and in other areas a negative difference depending on the shape of the terrain - over an entire site - the logic of Grids is that they will cancel each other out on a large data set

On a significantly exaggerated view you can see the positive and negative effects

So when we are interpreting strata surfaces from the OG we can use all 35000 vertices or we can grid the data down to eg 10 feet and reduce the data of the single surface model by 75% and use this to compute the strata surfaces - which on 7 surfaces in this case is a reduction of 7 x 27000 nodes which is very significant in terms of computation time - and you really will not see any significant difference in strata quantities (zero effect on total quantities because you will still be using you OG and FG (or adjusted versions of those) for the full quantities, and only your strata breakouts (which are based on just 17 points of reference anyway) will vary a little as a result (but in my experience that variance will be less than 1-2% max) and the full data analysis is not precise anyway because it is based on the 17 boreholes only - everything else is an interpolation anyhow.

Combine this with the setting in Project Settings - Computations - Surfaces called Volume Computation which by default is set to “Track All Triangles” but can be changed to Do Not Track Breaklines - when changed this can also reduce your surface to surface volume computation times by a further 75% - it is slightly less rigorous in ats approach but for a takeoff if you are looking for speed this can help also without significant impact on accuracy in most scenarios. When you are running your first pass quantities for sure use this method, and only when you are happy that you have everything as you need it would I consider changing that back to the original setting and letting the computations take longer

When you look at what computations the Takeoff Reports require for Earthworks

  1. If you have the adjusted OG and the Adjusted Design and No Overex and no strata and no AOIs - the volume computation is just between two surfaces and you are done

  2. For every AOI that you add that computation is multiplied by the number of AOIs so if you have 4 x AOIs then it will take 4x the base time

  3. For every strata that you add you add another set of surface to surface computations because you have to compute the Volume from OG to Design, from Strata 1 to Design, From Strata 2 to Design etc to get the breakout quantities and if you are doing that to multiple AOIs then that also compounds again soi with 7 strata and 4 AOIs you have to do 28x the computations. So if you can reduce the number of nodes in a strata surface by 75% (and these are a large multiplier here) you can reduce all of those extra computation times by a similar amount - and that will significantly reduce the computation time overall for the whole project. Then take that down by a further 50% to 70% by changing the Project Setting. Not only that but your memory management will be easier and faster and your project files will be a lot smaller. Graphics Regen Time will drastically improve also when you are showing those models.

  4. The only thing that you have to look out for is that in some situations the top most strata (modeled from the 10’ or 20’ grid) may now be higher than the OG because it is not constrained by the same OG surface - however those areas will be small and the associated quantities will also be small and will cancel in 99% of cases so I dont typically worry about that so much - depending on the depth of that strata and the shape of the OG surface and the grid size that you use will dictate if that becomes an issue or not - I typically go with a 10’ grid or maybe a 7.5’ grid because that is going to give you the surface size savings and will mitigate this issue to a large extent.

In the enclosed project I have set all the surfaces to Show Empty as per last night but it jhas the different surfaces computed and the strata computed in this way for you to look at

I hope that we ate Rockpile will get this process streamlined this year before the year end as this is our primary focus area to make the Takeoff process easier and faster

Also be aware that we will always be a little slower using TIN modeling techniques compared to other products out there that use Grid or Section techniques to compute the volumes faster - we can also compute using Grids and using sections if needed and we will be exploring those options as we get deeper into this.TINs will always be the most accurate result but that comes with an increased burden of computation time unless you do these things to streamline your data model.

When I have people working this way they tell me that the times come way down and that the results are very satisfactory

I hope that this helps and that it is useful - I will try and post a video to go with this later