603 - 5885 1250 (4 lines) info@mes100.com

Case Study: When your Earthwork CutFill result is challenged…

How the story started #

Imagine in a meeting, you just happily present the output of MiTS software for the infrastructure design. You show the earthwork levels you design, the keyplans, the drainage longitudinal cross section, the predevelopment watershed analysis and the 3D view to your client, architects, other fellow engineers and quantity surveyors. It seems that everyone is extremely engaged and towards the end, it looks like you are making the greatest Presentation of your Life Time. You feel that everyone in the room is mesmerized by your charisma, eloquence and you feel like you are the smartest people in the town.

And then a bombshell DROPS; when you present your Earthwork quantity calculation, someone– none other than the client himself– rises from his seat and tells you that your calculation is off by a margin of 25%. What would you do?

This was the situation faced by one of our clients. Of course the details are not as dramatic, but the challenge– and the implication– is nonetheless the same and is very real. So here we would like to narrate what he did afterwards, and how he used MiTS software to justify his results and theirs.

The consultancy involved here is TP Tay, a Civil and Structural engineering firm based in Batu Pahat, Johor. For this particular project, the site involved has an area of 84.6 hectare (846000 m^2). The owner, Ir. Tay was tasked with estimating the cut and fill of the As Built Survey as compared with the Original Ground Level. It wasn’t not really a big area– it’s only about the size of Bukit Jalil Stadium plus National Park, but the cut and fill was on the order of million meter cubes, so a lot of money was at stake here.

Upon receiving the job, Ir. Tay brought up his MiTS software, and imported the two sets of surveyor points– the Original Ground Level and the As Built level–into the software. For this project, he used the DTM method and TIN surfaces.  The software churned out the result in no time: Bank Cut around 2.57 Million m^3, whereas Bank Fill around 1.92 Million m^3.

Sanity check with mesh triangulation #

The very first step when one runs a DTM method, is to verify visually that the triangulation actually conforms to the expectation. Ever a careful man, this was exactly what he did in order to ensure that the mesh triangulation for both surface profiles were sound.

Image 1: Survey Point 1
Image 2: Survey Point 2 (As Built)
Image 3: Survey Point 1 and Survey Point 2 (As Built) overlapped

As shown in the above three screenshots, the meshes for both profiles looked dense enough, which gave him confidence that the result was fairly representative of the actual ground level and hence, the earthwork quantity result was believable.

As a further sanity check, one can mouse over to the surveyor point and click on it, and verify that the actual surveyor point z value and the mesh triangle z value at that particular point actually agrees with each other, as shown in the screenshot below.

Image 4: Verifying survey point at spread input

Everything looked good, so time to submit them to a meeting for discussion!

The Challenge #

But when Ir.Tay presented the result to the client, the client expressed his skepticism. He got an estimation from another software (which shall remain nameless). The result didn’t look too pretty:  Cut was 1.91 Million m^3, and Fill was 1.85 Million m^3. For Fill, the difference is less than 10% and hence is acceptable. But for Cut, the difference was around 25%, way above the commonly accepted range! 0.66 Million m^3 cut soil difference translated into an additional RM 2 million of excavation cost, which is doubly not acceptable in the eyes of a cost-conscious developer.

There are only 3 possibilities here:

  1. The client and Ir. Tay used different surveyor point profiles
  2. MiTS earthwork earthwork quantity is wrong
  3. The other nameless software earthwork quantity is wrong

Point 1) was easy to rule out, after they synchronized the surveyor point profiles, the results between MiTS and the other software still remain different as they were.

Software Internal verification #

Could it be that MiTS earthwork calculation was wrong? After all, software vendors express no liability in the event of inaccurate results; it’s the consulting engineer who has to bear all the consequences of it.

As the software developer of MiTS, we are well aware of this point, so we strive to make sure that engineers can easily cross check or counter check our result. Which is why in addition to providing the DTM method, we also provide the Grid and End Area method for earthwork quantity calculation

So Ir. Tay utilized these tools to counter check his DTM results. He plastered the drawing with enough grid cells, and he also drew cut section lines across the As Built area, and he ran the calculation again. Good news! All the three methods, DTM, Grid and End Area method agreed with one another (as they should)! This gave Ir. Tay extra confidence in his calculation, and also in the accuracy of the software.

Here’s a bonus for those who want more granular validation. Though no human can go through every single grid cell or triangles in his entire lifespan, one can go through at least some of them and verify them to get a feel on the software’s robustness. In MiTS we make the process very easy; here’s a video tutorial on how this can be done!

Image 5: Detailed DTM Calculation

Verification with Civil 3D #

So much about internal consistency check, what about external verification? For this purpose, Ir. Tay turned to us and we decided to use Civil 3D as our external source of truth, because it is the brand that everyone recognizes.

To save us from the hassle of inputting surfaces in Civil 3D, we just used our MCIntegrator plugin to seamlessly synchronize the surfaces in MES to the surfaces in Civil 3D.

Then we used the Civil 3D Volume Surface to compute the earthwork quantity of the original ground level and As Built level in Civil 3D, and we got excellent agreement between MiTS and Civil 3D (Cut: 2.57 vs 2.48 Million meter^3, Fill: 1.92 Million vs 1.92 Million meter^3).

The agreement was excellent; so all the doubts about software accuracy should now be put to rest. It’s already well within the industrial acceptable range– less than 10%. Consultants could merrily present this convincing case to everyone who cares to listen and we could now all call it a day.

Or could we?

But what about the 4.5% difference? #

But the scientists in us were not satisfied; there was still ~4.5% of discrepancy. What could have accounted for those differences?

To answer this question we took a hard look at everything, from the surface profiles used by both software, to steps that we took to operate the software, and to even the MiTS source code (if we had the source code of Civil 3D we would check those too). After a few hours of frustrating search, an epiphany hit us, we suddenly noticed that the As Built surface profile in Civil 3D and MiTS didn’t exactly agree. Here’s the juxtaposition images from both software:

Image 6: Survey Point boundary in Civil 3D
Image 7: Survey Point boundary in MiTS 2

Notice the difference? Can’t? Look harder. Pay attention to the lower part of the As Built surface profile. Note that for Civil 3D, it bends slightly more inwards than MiTS. The difference is very minute and you may need a microscope to spot it.

Image 8: The slightly bend more inwards area in Civil 3D

Now things got easy for us. We can now make an Apple-to-Apple comparison by trying to get the As Built surface profile to be as similar as possible to each other. Civil 3D doesn’t allow you to modify the surface profile ( a flexibility we wish Civil 3D to have, but alas it doesn’t), but we can definitely use MiTS’s Alpha value function to adjust the surface profile as we see fit.

Image 9: Alpha setting set at the middle
Image 10: Alpha setting set to maximum

True enough, by matching the As Built surface profile as closely as we can, we can improve on the result agreement between Civil 3D and MiTS ( Cut: 2.48 Million meter^3 for both MiTS and Civil 3D, Fill: 1.92 Million meter^3 for Civil 3D, 1.90 Million meter^3 for MiTS). 

It’s beyond a shadow of doubt that if we tinker with the Alpha value further, we can get a 100% match for the surface profile and also the earthwork quantity. But we didn’t pursue this further because we were satisfied that we had found the source of the discrepancy and we could make the disagreement disappear if we wanted to.

The duplications of survey points #

One challenging point about the project is that the survey point drawings from the surveyors were full of duplicated points; in fact, only half of the Original Ground Level points are unique! This is a mistake on the surveyors part because it’s not possible to have two points on the same {x,y} coordinate. As such, the duplication poses a challenge to the accuracy of the Earthwork software calculation because there are ambiguities on how you can build the surface profiles.

Civil 3D handles the duplicated survey points by…. just arbitrarily removing all the duplicated points except one. Luckily for this project, all the duplicated points have exactly the same {x,y,z} value, so this approach is entirely justifiable and doesn’t affect the result. In the case where the duplicated points have the same {x,y} value but different z, some result inaccuracy might be introduced because there is no way for Civil 3D to which points are correct and which points are not. 

In MiTS, we handle the situation in a more graceful way; we don’t play God by removing arbitrary points as if the software knows better than the users; we just display a warning message in the verification tab for them to take action.

Image 11: Verification message in MiTS 2

Not only that, we also provide sorting utility in the Spread Input so that the users can sort the survey points, easily identify the duplicated points and delete the points they deem wrong. As a part of our philosophy, we give the tools to our users to do what is right, not to dictate what is right.

Image 12: click button shown by arrow to sort the survey point

In the end, Ir. Tay redid the project by sanitizing the survey points. He did this by going through every group of duplicated points, manually cleansing them (Young engineers take note! THIS IS your example) and then reimporting them into MiTS for another round of calculation. The result is still the same as before, and thus proved that really, the duplicated points in this project didn’t matter at all.

Conclusion #

Ir. Tay was very grateful to us for the support we provided as we together chased down the rabbit hole and resolved the issue. He was very pleased at our mastery of the technical aspects of the software; we didn’t just know how to use the software, we also knew what the result meant, the applicability and limitations ( if any) or the calculation methods, and how one could cross check them. The fact that we know the software inside out is a testament of our technical prowess and commitment to our clients.

The feeling of gratitude is mutual; we also can’t thank Ir. Tay enough for sticking with us for so long and for taking the time to talk to us. Rest assured that we had or will implement his helpful suggestions in our software, so that more consultants can benefit!

Ir. Tay is by no means the only helpful client we have had; there are many more who send us comments, suggestions and bug reports from time to time. It’s no exaggeration to say that the foundation of MiTS is built on top of the hundreds of manyears of cumulative engineering experience. We are really humbled and fortunate to have such a group of engineers who grow with us. 

So! Don’t feel despair if your clients or other consultants pooh-pooh your results, because we got your back. At MiTS, we prepare to stand by you and we can get to the bottom of the matter together, so that you can answer your client in a timely and satisfactory manner. We do this not only because we want to provide excellent customer support, not only because we value our relationship and are always looking for the next idea to improve the software, but because we are insatiably curious ( just read about what I wrote at kindlemalaysia.com to know the kind of topics I’m interested in). 

Oh, what about the nameless earthwork software with the result that we disagreed with? What was their response to this episode? We are sorry to report that we really don’t know, because the last time we heard, the project files couldn’t run in the latest version of the software. 

Powered by BetterDocs