Needing Help in Duncan & Chang (Hyperbolic) Constitutive Model 2

[vvamatin]

Hello people,
I'm reviewing a couple of soil constitutive models including linear elastic, elastic-perfectly plastic, Duncan & Chang (hyperbolic) and Modified Cam-Clay to find the most accurate one that has the closest results of stresses and displacements to the readings of pressure cells and magnetic settlements on a rock-fill dam. The dam has not been finished yet and is still in construction phase, and its core's elevation has reached to about 45 meters from the foundation.

So far, the most accurate results I've got was using Elastic-Plastic model. The linear elastic model was fine until the middle steps of construction, and then its results became far less than what the instruments reported.

Recently, I developed a load/deformation analysis based on Hyperbolic model. I found a paper of Duncan et al, which suggested a typical value range for some types of soils.

The soil I'm reviewing is in the "CL" classification, with a total unit weight of 20.4 KN/m3, poisson's ratio= 0.2 (assumed, works with the elastic plastic model), Young modulus= 15000-33000 (depending on the confining pressure), C= 110 KPa, and Angel of Friction= 10 degrees.

The suggestion in the mentioned paper for the CL-classified soil (of course with C and phi s different from what I have) is K(load)= 60-150, Rf=0.7, K(bulk)=50-140, m=0.2 and n=0.4.

The problem is, using these range for the core, specially
K(load) and K(bulk), results in much higher displacements (about twice bigger than the instruments' reading), and stresses reported. the closest results I can get is with K(load) around 600 and higher and k(bulk)=385.

These values are not so odd, but just if it was a different soil group, such as GW,GP or SW,SP. But for a CL, I don't think it's not a typical value range.

Unfortunately I couldn't find any literature in dams/embankments using this model, to judge if the parameters I'm using are correct or not.

So, do any of you have any ideas if I can go on using these parameters, or simply this means the hyperbolic model cannot describe the dam's behavior?
Thanks in advance.

 

[dgillette]

First of all, I know very little of Duncan-Chang. (I know Mike, but not his model.)

I would be a bit surprised if linear elastic or E-P could give you good results unless it is somehow modified to allow consolidation, i.e., permanent, nonrecoverable volumetric strain. Pretend that the freshly compacted fill is overconsolidated clay. As you place fill over it, it initially consolidates very little, and is roughly linear. Then, it reaches the apparent preconsolidation pressure (a few atmospheres, depending on placement moisture and compaction) and begins to undergo more significant, nonrecoverable volume change. What you may be seeing is that transition from pseudo-recompression to virgin consolidation, causing the initial part of the history to be roughly linear, and the rest to settle much more. Get an undisturbed sample and run a consolidation test on it to see if that will explain things in a MUCH simpler fashion, maybe even with 1-D settlement analysis.

I have never seen a Poisson's rat nearly so low as 0.2 in compacted clay.

 

[vvmatin3]

First of all, thank you for your opinion.
I don't know what is wrong with this forum, it doesn't react when I login to my account. So I had to make another account.

let me tell you that the dam has not experienced its first wetting phase; so in fact there is not much to talk about consolidation. I think it's better to consider the problem as an embankment, with almost dry unit weights and negligible amount of pore water pressure.


And about the "modification" you mentioned, it's kind of true. I assumed a number of construction steps for the dam, in which the Young modulus varies (rises) as the stresses increase (aka the construction continues). Modified models such as hyperbolic (Duncan & Chang) do not need such modification, since they consider Young Modulus and Poisson's ratio as a function of stresses applied in every steps. But in models such as Elastic-Plastic and linear elastic, you need to change these values manually.

(specially Young modulus, since poisson's ratio doesn't change significantly, and the resuluts are very sensitive to the ratio).

The trick, and the non-presence water, makes the model work for this case.

And about the poisson's ratio you mentioned, I have to agree that it's a bit odd. I personally think a value around 0.3 would be better, but I've seen a couple of papers (mostly theoretical of course) with such a relatively low ratio. So theoretically, there's nothing wrong. Besides, I double checked the Young modulus and poisson's range with Bowles' Foundation engineering book and it was a match, Though the value 0.2 for poisson was the least amount of possible rang.

So, all in all, I can consider the Elastic Plastic model working.

The problem still stands, and the Hyperbolic model doesn't work so well, and when it does, the parameters, specially K(load) is much bigger than what it rationally has to be.

As I mentioned, I've seen K around 600-700 for crushed stones or GW-GP or SW-SP materials, but based on what I realized from the lab graphs of USBR classification, the core's material is classified in CL group, not the mentioned groups above.

So the question still is, am I probably doing something wrong? or the model simply doesn't work for this case?

Any helps would be appreciated.
Thank again.

 

[dgillette]

"let me tell you that the dam has not experienced its first wetting phase; so in fact there is not much to talk about consolidation."

I don't agree with this statement. Saturated or not, consolidation occurs during construction as soon as the weight of fill exceeds the "apparent preconsolidation pressure" induced by the compaction - a few atm. Run an oedometer test on compacted fill, and see what happens.

 

[muuddfun]

I agree with dgillete on this one. You are compacting the clay during construction close enough to the ZAV line so that it would not take significantly more pressure to close the gap in density and basically eliminate most all of the air voids, once that has occured then clasical consolidation can begin.

You should just do a remolded oedometer on the fill, and you will get your plot of pressure vs deformation, then compare with your computer and see which is closer to the real thing. The oedometer should be very close to the field if you use for a remold value the average field density and moisture contents you have been achieving during compaction.

If I remember right we got about 1/3 the settlement during construction for an embankment dam I was working on a decade ago. We built 3 feet of camber into about 300 feet of embankment to allow for the settlement. The core was mostly an SM material so it would have behaved a little different than yours.

 

[vvmatin3]

Thank you gentlemen.
I see what I can do about that.
But in the meanwhile, any idea about the validity of the parameters I've used for the model?

 

[muuddfun]

I don't know much about the specific values you are using, but my guess would be that the actual behaviour may be a cross between the elastic-perfectly plastic, and the hyperbolic. You might try to plug the values for one into the other and try to get upper and lower bounds of the settlement in the field then back calculate new parameters and give it an iterative approach until one of the modles comes closest to what you are seeing from the field readings.

 

[vvmatin3]

I've contacted the field supervisor. He told me that the preconsolidation stress is something about 570KPa, and the model I'm reviewing shows about 760 KPa of stress in the last step of construction so far. (the construction is not fully done, it's halfway through).

So It's wise to say that in the most part of the analysis, consolidation doesn't play a significant role. Based on what dgillete said, the consolidation influences the structure, mostly after the preconsolidation stress acquired, AKA the last steps of construction in this case. Maybe that's why the Elastic-Plastic model shows some good results,verifying stresses AND displacements; or simply, maybe I'm wrong!

But about the hyperbolic model,I'm still in doubt about the validity of results based on the parameters used.

Thank you people, anyway.