seepage cut-off and relief wells 3

[jrl802]

Hey all this is my first post so bear with me.

We have this levee that needs underseepage control for about 1980 ft of its length. The levee is set on a fine-grained (CL/ML) layer of soil that varies in thickness and this sits on top of a permeable sandy aquifer that can be anywhere from 60-150 ft thick down to bedrock. The severity of the underseepgae along this length varies. There are certain reaches where our studies have determined that relief wells do the job but there are other reaches where we need a cut-off wall. Essentailly we have reaches were we have a line of relief wells then a cut-off wall and then more wells.

Is there a need to have overlap between the wall reaches and the well reaches. In other words is there a need to have part of the wall extend so that it is "in-front" of the wells to account for seepage that will wrap around the wall.

Thanks!

 

[cvg]

why don't you just run the relief wells for the entire length and then add a cutoff wall in areas where you need additional protection?

 

[jrl802]

Thanks for your reply cvg.

If I udnerstand correctly your suggesting to have the wells run continiously and then have the wall in certain places that are really critical. Therfore offering redundancy when it comes to seepage protection.

The sections where just a wall is being considered have very serious seepage issues and wells would NOT do the trick because they would have to be spaced very close to each other.

It terms of the redundancy that you suggest, I dont know if the cost would be well received.

Thanks though.

 

[fattdad]

I guess this question will be discussed in this forum (noticing you cross-posted in one of the other forums too)?

It's very hard to address this as I can't visualize the problem yet. You have a seemingly impervious foundation soil (i.e., cl/ml) underlain by a pervious natural soil layer (thick). You have the levee, which I'd hope is also fine grained, but then again, it may be the equivalent to sand bags - who knows. You say there is some critical seepage problem, but we don't know whether you've done some field recon and seen boiling or such. Then again, maybe you are addressing this from a "modflow" perspective. Don't know.

We also don't know to what extent you can modify the levee to increase the seepage path, then again, maybe it's not the levee conditions that leads to the design concern.

Seepage wells work. Cut-off trenches work. Helping address needs for overlap, redundancy or consequence of failure is so difficult to understand from your post, I don't know whether I could offer any help.

A sketch attached to your post would provide a thousand words!

I like these sorts of projects!

f-d

¡papá gordo ain’t no madre flaca!

 

[jrl802]

Yea this forum seems more appropriate. Sorry for the cross-post.

You are correct about the way the stratigraphy is set up. From the "bottom" up its bedrock, then pervious natural layer (60-150ft), then impervious layer (varied thickness along levee) and then you have the levee. The levee is on the Mississippi River and for this analysis I am assuming that its impervious.

We had a pretty aggressive and in-depth exploration program where we drilled "matching" borings on the land side and the riverside of the levee every 330ft along its length. We also conducted a seepage analysis to determine where the seepage was worst and we have historical data of boils from past flood events (1993,2008).

I have attached a plan view of the project.

In it there are 6 reaches (330ft sections of levee) with the seepage "fixes".

From left to right:
-3 wells (in red) at 100 ft spacing’s
-6 wells (in orange) at 50 ft spacing’s and the overlap of the wall
-then the two reaches that require a wall
-followed by 7 wells at 50ft plus the overlap and 7 more wells at 50 ft.

The wells that are behind the overlap were designed assuming there is no wall. SO in essence, it is overkill.

My concern is which way would you approach reducing the number of wells based on the fact that you need a wall overlap to prevent seepage from wrapping around teh wall and then pushing up the thinner CL/ML layers.

Hope this helps. If not just keeping asking.

Thanks

J

 

[fattdad]

So, is the distinction between the blue area (i.e., that needs the fix) and the other colors (i.e., to either side) based on modeling or field observations? Does the modeling (i.e., the color regions) account for the wells in service?

Here's where I remain confused: If the near surface natural soils are lower permeability and the levee low permeability, what causes the seepage forces to develop in the sandy soils (i.e., it seems the silt/clay would be a hydraulic barrier. Don't get me wrong, I understand that flow nets (in a homogeneous environment) are independent of permeability, but I also know that the relief wells are in the natural sand, so there is some mechanic whereby you believe a critical gradient is in that layer (i.e., to lead to piping/boiling).

When you say there were two years where boiling were noticed, just what was the relationship to this observation and the tow of the levee?

What model is your attached image based on?

Just wondering, that's all. . .

f-d

¡papá gordo ain’t no madre flaca!

 

[jrl802]

the modeling is done with Geo-Studio 2007 and their Seep/W platform.

The places where fixes are needed and what those fixes are were based on a previous seepage analysis that was performed and back checked and confirmed by the historical observations.

I should’ve explained that the colors are just to point out the difference in Head. SO each color is a contour to visualize the drop in head and each head drop is 5 ft in the model. In the screenshot I sent the blue region means that ther is a much lower head than on the red. Essentially in this screenshot I am showing that the problem is solved by the wells and the wall. but where the overlap exists there is a gross overdesign.

There are no existing wells in service in that particular reach. In the reaches where there are wells existing we have assumed that they dont exist because they will be repalced cuz' they were built in the 40's.

The silt/clay soils are a hydraulic barrier but when there is a flood event a large differential head develops between the flood elevation and the prevailing ground surface on the protected side. This, drives the seepage and then the impervious material causes excess head build up at the landside toe. This build up causes seepage paths through the silt/clay and if there is enough excess head then it can pipe foundation sand out to the surface.

The critical gradient is based on heave or uplift so its just a simple summation of forces in the "y" direction.

And that comes out to be that the critical gradient is equal to the excess head divided by the thickness of the CL/ML layer.

I guess that it is a different "failure" mechanism that if the whole soil layer was homogeneous.

 

[fattdad]

So, are your conclusions based on a steady state hydraulic model or did you you look at the transient time for the flood stage event? It strikes me as awesome that during a flood stage event, sufficient flow can develop to increase the pore water pressures through the foundation silt, through the sand layer, back through (upwards) the foundation silt and then to boil.

Not saying it cant, just amazing!

I'll go back to thinking, but the problem is making more sense.

f-d

¡papá gordo ain’t no madre flaca!

 

[jrl802]

"sufficient flow can develop to increase the pore water pressures through the foundation silt, through the sand layer, back through (upwards) the foundation silt and then to boil."

That’s a good point but the river we deal with is the Mississippi and the seepage entrance condition is considered to be the sand layer (see attached).

Well the modeling is done steady state because I’m just using the model to try and see how many wells on the are protected by the walls are needed. The solutions to the seepage were determined using a steady state analysis as well but its was much more basic than numerical modeling and just included the use of MS Excel.

Attached is another drawing to better clear up any clutter from my previous posts.

Thanks for all the feedback!

 

[GeoPaveTraffic]

First a couple of questions.

Have you ran a 3D model of the problem or just a plan and cross-section model?

Second, why are you running a model every 330 feet? I assume you are going to say "Because that is where the borings are?" My suggestion would be to use your borings, and develope a study reach that incorporates more than one set of boring.

Third, is this a university problem or a real engineering problem?

As a prelminary comment on your cutoff wall relief well concept. I really don't like uysing cutoff walls to control underseepage. They are increadabily expensive to install and ANY defects in installation can cause big performance problems. Additionally, relief wells need frequent and regular maintainence. If they don't get it, they don't work. I suggest the possibility of an underseepage berm be investigated.

As for the possible layout you attached as "Seep_Wplan.pdf", I would not expect any wells to overlap the cutoff wall. I would expect a well at each end of the wall, but believe you have a model problem if the results show that the wells need to overlap the wall.

Mike

 

[GeoPaveTraffic]

fattdad

This is a problem with levees in many areas of the US, and a really big problem with levees along the Missouri and Mississippi Rivers.

If you are interested, I can post a link to the USACE EM for levee design that has a nice discussion of this. The main research was performed by Mansur and others in the 1940's and 1950's.

 

[jrl802]

Thanks for the posts Mike!

This is a real engineering problem and the seepage analysis methods we use are the Maunsr and Kauffman Methods from the 40's and 50's.

We have not run a 3D model because I don't think we don't have the capabilities. The initial "fix" was determined with the M and K methods that have been incorporated into the USACE EM. We use a MS EXcel SS but they can be done by hand if need be.

Why every 330 ft?

You are right. We divide the study areas into 330 ft reaches with a boring in the center of them. The reason being that the closer the borings the more we can narrow in on problem areas. Say, areas with thiner impervious layers that could be more problematic.

As to your comment about wells vs. walls vs. berms. I totally agree with you. It is my personal/professional opinion that seepage berms are the best controllers of underseepage in terms of the way they deal with the problem. But they require a lot of space which also equals money.

Wells work well if they are maintained and also have a limit in terms of very serious seepage issues. Sometimes you might be required to space wells at very close spacings 10-20ft or even less. Cut-off walls are expensive and once you instal them its hard to know if they have been installed correctly and will perform to design standards.

That said some of the areas where a wall has been suggested have very serious seepage issues so wells aren't enough. These levees also protect industrialized areas so land is expensive and berms would require land. A wall in those cases appears to be the most cost effective "fix".

"As for the possible layout you attached as "Seep_Wplan.pdf", I would not expect any wells to overlap the cutoff wall. I would expect a well at each end of the wall, but believe you have a model problem if the results show that the wells need to overlap the wall."

This is a great point. I have the same concern and have expressed this same sentiment to other colleagues. I have modeled the system with just wells at the edges of the wall and no overlap and the model has shown no need for the overlap, as you have stated. On the other hand colleagues have expressed concern over the fact that, if there was no overlap, seepage might wrap around the wall and head towards the are of the land side toe where there is a critical seepage problem and cause boils.

Once again Mike, thanks for your input. I am loving this discussion and Im open to counter arguments and other suggestions. Thanks.

 

[cvg]

don't you need to monitor seepage making it past your cutoff wall? if so, you will be installing piezometers anyways, why not just install a relief well instead and you can also use it to monitor seepage...

As far as seepage "wrapping around" I don't understand the logic of that. That increases the seepage path, reducing the upward gradient. The relief well at the end should be able to handle it.

wells must be maintained to be effective and the Corps requires pump tests every 5 years, however it is not unusual for the pump tests to not be done. Most of the levees are operated by local sponsors who have very little money to spend on O&M. What they do have they spend on mowing, riprap, burrowing animals control, unwanted vegetation removal, etc. and the wells get forgotten. I too would recommend a landside seepage berm. Not only do they provide posative control of seepage, they are easy to maintain and if you do have a boil, it is at least 150 feet from the toe of the levee and does not pose any risk to the embankment

 

[GeoPaveTraffic]

jrl802

I'm not trying to be overly critical, but what I'm about say will likely step on at least a few toes. I'm sorry about that, but I feel it must be said.

Your colleagues that "have expressed concern over the fact that, if there was no overlap, seepage might wrap around the wall and head towards the are of the land side toe where there is a critical seepage problem and cause boils" don't seem to get the way water moves. As cvg pointed out, water takes the path of least resistance. At the ends of a cutoff wall, the water is looking for the shortest path out, that would be through the relief well at the end of the wall.

Also, your statement "We have not run a 3D model because I don't think we don't have the capabilities." really really bothers me. If you don't have the capabilities to do the job, THEN GET SOMEONE ELSE TO DO THE JOB. We are REQUIRED to hold public safety PARAMOUNT. This could not be more true in any situation than a levee. Now with all that said, I don't believe that a 3D model is required in this situation.

It appears that you and possibly your firm has limited experience with levee design. Levees are both very easy and very difficult to design properly. Remember that all levees WILL fail, it is only a mater of when. For us in the goetechnical community, we need to be designing and constructing levees that do not fail before they are overtopped. Underseepage and through seepage leading to piping are the most common geotechnical reasons for levees to fail and we must do a better job of designing these structures.

An industrial area with underseepage problems is not the place to learn how to do these designs. Land costs are nothing compared to the cost of lost lives and property damage that will result from a levee failure. Think about that and how well you wish to sleep each night when doing this type of design.

 

[jrl802]

Thanks cvg,

The seepage problem in that specific reach is so severe that our analysis requires wells spaced at less than 10 ft sometimes less than 5 ft. We have seen that when you begin to space wells at less than 50 ft. their capacity to control the underseepage is reduced.

Its only natural that there will be seepage that wraps around the wall. Drawing flow nets on a plan view will show that. You are right that the seepage path is increased because of that and so will the drop in excess head.

I concur with your thought about the last well being able to handle whatever wraps around as the zone of influence of that well will itself overlap behind the wall.

As per your last paragraph. We are pretty conscious of the hassle wells bring. There are also environmental concerns over pump tests because sometimes groundwater is contaminated and bringing that water up to the surface can be a pain (for the levee districts) to mitigate. Berms effectively reduce your differential head, increase the seepage paths so that there is more head loss and, as you pointed out, push the seepage far from the levee toe this is the ideal solution IMO. However much I LOVE the seepage berm in this case there are several real estate issues that make the berm very costly.

 

[jrl802]

I understand you concerns and I’m not bothered by your comments them because I agree with them and live my life by them.

Maybe I wasn’t clear about where I stand on this issue. Before I posted I was convinced that an overlap was not needed because I have run models with different scenarios (as should be the case). I also KNOW that groundwater will take the path of least resistance. That is why wells work because they are large voids (least resistance).

Additionally this levee already exists and was built in the 1930’s we are simply improving the capabilities of it to perform well during a flood. We can’t control where industry decides to “crop up”. I believe my firm is adequately staffed and has the necessary capabilities to do the job. The capabilities I was referring to rested on software limitations only and as you said no 3D modeling is required.

Thanks for the concerns though and I couldn’t agree more about holding public safety as number one priority.
Thanks!

 

[cvg]

I assume this is a federal levee, built by the USACE.
The controlling design guidance is in the following technical letter:

ETL 1110-2-569 DESIGN GUIDANCE FOR LEVEE UNDERSEEPAGE

That letter states: "A 2-D analysis can be helpful where the substrata system is more complex. A more complicated
3-D type analysis is almost never justified."

I don't feel there is a need to defend the fact that a 3d analysis was not run.

 

[GeoPaveTraffic]

I agree about a 3D analysis almost never being justified. I have only seen it done once on a levee. In that instance, it reduced the level of concern and significantly reduced construction costs. In fact before that change was made, the levee was not going to be constructed because the local sponsor could not meet the match requirements.

The specifics are lost in my mind, but it was a location where the levee had an outside loop, i.e. seepage could reach the protected area from more than one direction. I was only involed a little in the analysis.

jrl802 - this levee wouldn't happen to be in the East St. Louis area would it?

 

[dgillette]

Consider having a small amount of overlap, if for no other reason than that the geology is 3D, not 2D. Stringers of more pervious material may run every which way, and you may not be able to precisely pin down the required limits of the wall and the wells. Mother Nature bats last.

 

[fattdad]

with defined boundary conditions, a 3D model using Modflow would be pretty simple for this case. You wouldn't have all these calibration problems and you are doing steady state analysis. Heck, this is a one-day problem in Visual Modflow (well if you understand how to use the model.

I do find this thread interesting and agree with public safety and the warning not to get too far out of your comfort zone.

f-d

¡papá gordo ain’t no madre flaca!