Horiz Subgrade Question in Fininte Element Model for Soldier Pile Wall

[ColonelForbin]

I am Trying to use CWALSSI (non-linear finite element soil-structure interaction software by Corp of Eng.) to model a post and panel wall. Since the program uses a 1 foot trib width, all section properties are in "per foot of wall" basis. I followed the PileBuck (or CalTrans) method to adjust soil pressures above and below the dredge line. I.e. f*Ka and f*Kp where f is a ratio of pile flange width to pile spacing. THE QUESTION IS: Do I need to adjust the horizontal subgrade constant REQUIRED for input, to account for the change in tributary width? I contend that it is a constant and the reduced passive pressure below the dredge line accounts for the reduction in soil stiffness; A coworker of mine (and the boss) contends that the spring constant must also be adjusted to account for the change in width. Does anyone have any design examples or suggestions on whether or not the Horizontal subgrade constant should be adjusted? This is the constant as described by Terzagi in the 1955 paper "EVALUATION OF COEFFICIENT OF SUBGRADE REACTION," Geotechnique, 1955, PP. 297-326

Also, if anyone has a copy of the paper I would love to know how I can get my hands on it.

ColonelForbin

 

[ishvaaag]

If you obtain your spring constant through a lateral modulus of subgrade reaction it will be say (kgf/cm2)/cm, hence to pass to a spring constant, say kgf/cm you will multiply by some tributary area, where the slice thickness has a say.

Other thing is how you get your spring constant in practice. If you get at the depth a passive push per meter thickness of the slice, and read a spring constant, of course this is for such thickness. A table for imperial units would read both the passive push and the spring constant columns multiplied by 0.3. Hence and assuming linearity you would be right if directly reading the spring constant from the same and NOT ADAPTED table, the effect would be contemplated in that you would be entering already a lesser push (by the 0.3 factor) and reading a spring constant also lesser, by the same factor. Alternatively, you would have got a spring constant for the meter thickness, and then accounting the actual thickness would lead to the same result.

However, if the table goes with adimensional input where the thickness is excluded and you read a modulus of subgrade reaction, then the thickness would need to be explicitly accounted for when building the spring constant.

In any case I think to rely on any prefabricated law for the horizontal modulus of subgrade reaction is quite speculative. It may be useful, but reliable I don't think it is. Experimentally testing for it would do better, and maybe just the best we can do about.

 

[Focht3]

Your analysis should be done on the basis of the post (pile) width; this provides the most convenient dimension to work in. The only real transform that has to be made is to account for the continuous nature of the bulkhead - which you have already addressed.

The Winkler springs used to model the soil have units of F*L^-3 so they will be multiplied by the pile width and vertical node spacing to arrive at the spring constant in units of F*L^-1. (Remember to use only 1/2 of a spring at the top and bottom nodes.)

You stated, THE QUESTION IS: Do I need to adjust the horizontal subgrade constant REQUIRED for input, to account for the change in tributary width? I contend that it is a constant and the reduced passive pressure below the dredge line accounts for the reduction in soil stiffness; A coworker of mine (and the boss) contends that the spring constant must also be adjusted to account for the change in width. Your analysis is correct; but you can avoid the argument (and save yourself some effort) by using the post (pile) width instead of a one foot width. That way your boss can't get confused! The bulkhead multiplier becomes the ratio of the post spacing divided by the post width. This is far simpler, and it's easier to check by hand, too.

It is important to note that the k-values of the Winkler springs are roughly zero at the cut level, and increase with depth. Don't use springs of uniform stiffness for the full embedded depth...

[blue]ishvaaag[/blue] correctly expresses some concern over simply using a tabular value of modulus of subgrade reaction for your analyses. After all, soils aren't really linear - and they sure aren't uniform! But the research into non-linear soil springs is fairly well developed, with the research having started in the 1950's. The use of "p-y curves" represents an effort to use non-linear Winkler springs to model the soil, and they do an excellent job of predicting bending moments. They aren't so hot at predicting movement - but no current method is...

I suggest that you contact Ensoft, Inc. in Austin, Texas about their software (PYWALL) for modeling bulkheads. It's a lot better than Pile Buck et al, or guessing values per Terzaghi.

http://www.ensoftinc.com

Please see FAQ731-376 by [blue]VPL[/blue] for tips on how to make the best use of Eng-Tips Fora.

 

[ColonelForbin]

THANK YOU all for your responses. I agree that tabulated values of soil constants are generally not a good idea to rely upon for design. I will have values of the subgrade constant provided to me by the geotechnical engineer for final design.

My question is more related to the computer modeling aspects of the constant. The program I am using (CWALSSI) was produced by The Army Corps. of Engineers for the analysis of sheet pile walls -- which it does exceptionally well. It is a non-linear, soil-structure-interaction analysis program. The program calculates a linearly increasing soil spring stiffness for cohesionless soil and uses a uniform soil spring stiffness over the wall height. I am confined to the 1 ft width by the program architecture, so my problem lies in adjusting the input properly to model a soldier pile wall.

<b>The question put another way:</b> Since I am confined by software to analyze a wall width of one foot, do I need to adjust the constant of horizontal subgrade reaction above and below the dredge line to properly model the change in tributary area of the wall along it's height (above and below the dredge line)?

Based on what <i>Focht3 (Geotechnical)</i> stated above, I would conclude that I should leave the HzSubGrd value constant above and below the dredge line. PLEASE respond back if this is incorrect!

Thanks for your input.