dropping groundwater relative to settlement? 1

[msucog]

anyone else experiencing settlement issues for no apparent reason. i've noticed on two project where settlement monitoring programs are dragging out longer than usual with odd numbers. one of the programs has been measuring for over a year and experienced flat movement for about 6 months then started dropping again with no additional loading. the drop in elevation tend to be trending somewhat with well water levels dropping. the water levels have dropped only about 3-5 feet in the past six months due to the tough drought being experienced here in the southeast. on one project, there's only 20 feet of mediocre residual soil (7bpf auto average) above rock. there was about 20 feet of new fill placed it the area but we're at least 4 months in to the program which should have lasted only 30-60 days. the underlying and fill soils are sandy silts.
in other words, any else experienced the same thing or know of other suspects for the movement? btw, fill placed to at least 95% std Proctor on top of stable subgrade...nothing funky in soil test borings. consols run during exploration phase too to estimate settlement.

i'm stumped as to how a few feet of water movement would cause 2-4 inches settlement under relatively shallow fills atop shallow soil profiles...especially after flat elevation readings.

 

[fattdad]

From the description, there is some horizon of N=7 residual silts below the water table. These are likley preconsolidated to some degree and respond according to the Recompression Ratio (or index). There is also an unsaturated horizon of residual soil that's likely to respond more quickly according to some elastic modulus value - likely on the order of 80 tsf or so. . .

Added to this is the compression of the newly-placed fill. If this is on-site material (i.e., residual silts) and compacted at or above the optimum moisture content, there could be some compression via the relief of excess pore pressure during the compactive effort.

Considering that there has been 4 months since the last of the 20 ft of fill was placed, there is some likelihood that the saturated zone remains underconsolidated. While lowering of the ground water table can affect effective stresses, when you consider 20 ft of new fill, a drop of 4 ft on the water table would amount to less than 10 percent of the newly-placed loading.

Just some thoughts.

f-d

¡papá gordo ain’t no madre flaca!

 

[DRC1]

Prior to the drop in the water table, the soil had an effective unit weight of 65pcf. after the water table dropped, it remained nearly saturated, so the unit weigt rose to 120 pcf. For a 5 ft drop that would be a uiform surcharge of 275 psf. That may have caused your settlement.

 

[msucog]

on one project (the one with 20' of fill), the fill was placed out over a drainage swale (intermittent creek if you will). the water (originally 2' below the ground surface) was rerouted beginning about 1000-1500' away from the area. the upper 1-3' of mucky silts/clays were undercut. about 10' of rock fill (fist sized good looking shot rock) was placed in 1' lifts on top of stable ground. the remainder of the fill was sandy silt soil fill compacted to at least 95% std Proctor (most tests 98%). we put in monitoring pins the day they hit subgrade and began reading several times a week. the first week the readings were flat then it began to slowly creep and continued and hasn't stopped as of now (4 months later). all together, we're sitting at 2.25" drop. we estimated 10" total settlement with 7" immediately after fill placement (under 25' of new fill). around here, 60 days is usually the longest it takes for it to stop moving with 30 being more typical. however, in this particular geology, i've had another site that took 90 days and moved 12" under 40' of fill. the geology is known be large movements due to the inherent high void ratios.

for the second site, i don't know all the details since it wasn't mine. but the data plots flat for 6 months and then begins to drop and showed 6 inches of drop in the past six months with building construction finished a year ago (so no new loading).

one thing i'm thinking is this: the well readings show the depth to be 80' deep. groundwater here is usually say 10' in these lower lying areas that receive the most fill. so in order for the water table to drop from 80' to 85', the shallow levels would likely experience a much larger drop and effectively adding 60psf per foot of drop. and looking at the historical water levels in the well back about 5 years, the levels have dropped maybe 15' from the max but there was a drop to about the 80' level once before. all in all, i'm thinking there's been much more than a 5' drop in levels which jacked the stresses way up. i wouldn't expect 200-300psf to make a big difference but would provide some minor movement. however 1000psf+ (15-20' drop) likely could generate larger movements.

any other thoughts or similar experiences?
thanks for the input so far.

 

[msucog]

let me add one comment...the well data is only available up to 3 months ago and we didn't receive significant rains until the past month. i hope to see the settlement pins flatten out soon if it really is the ground water causing it.

 

[hokie66]

DRC1 et al,

I am not a geotechnical engineer, so won't try to get technical. But when you say the effective unit weight increases as the water table drops, I can't see it. When the soil is below the water table, the effective soil pressure is reduced by the buoyancy of the water, but to this pressure must be added the hydrostatic pressure. So the loading at the same level would remain about the same as the water table drops. Or am I missing something?

 

[msucog]

you're missing it. if you have a 50' column of soil at 120psf with no groundwater, then you've got 6000psf at the bottom of that column (or top of column below that column). if you've got 50' of soil with groundwater at 10' down then you've got 10'x120psf+40'x(120psf-60psf)=1120+2400=3500psf or almost 1/2 the pressure. and when you look at the consolidation curve for soils around here, once you hit about 1200psf, it begins to drop off non-linearly so that 5ksf might be 5x the strain as 2ksf (all depends on the soil).

 

[msucog]

and to be technical, when the soil is below the water table, the TOTAL pressure is reduced by the pore water pressure to give you the effective pressure. if the pore water pressure is 0 (no water present) then the total pressure is equal to the effective pressure.

effective=total-pore pressure
effective=total-zero (no water)

therefore, the effective stress is always less when water is present versus when water is not present. the effective stress drives this train.

 

[hokie66]

Yep, thanks for that. What my feeble mind wasn't seeing is that the hydrostatic pressure is both up and down, in fact all around.

 

[msucog]

i'm surprised my feeble mind explained it in a way that was understandable...

 

[Mickney]

Just a few thoughts from a Piedmont geotech: If the data seems funky, I would check the benchmarks used to survey the data. Make sure that these locations have not been tampered with. Also, make sure that the settlement plates themselves have not been tampered with. Also, were two separate datum points used to survey the plates?

With the lab data, check the time rate of settlement at the maximum applied earth pressure (i.e. 20 feet of fill). Being an engineer in the Piedmont, 30 to 60 days is a typical recommendation. However, if you have a consolidation plot with good data at each load increment, see what that tells you for time-rate.

For the shot rock fill site, any chance you are loosing the fine-grained soils in the voids of the shot rock fill? Did you cap the shot-rock fill with a filter fabric?

 

[msucog]

the reference point was outside the fill area on a manhole. i didn't set multiple points on this project due to the reference points proximity away from anything and "solidness". the thinner fill areas where pins are located are settling less and the thicker fill areas more. even manhole structure in the fill area settling (no rock fill under them). the thinnest fill is almost flat but it still moved slightly in concert with the other points. the survey data and field data is good based on my past surveying experiences (reading to the nearest 5/1000, closing the loop, balanced backsights/foresights, calibrated equipment, etc). absolutely no activity on the 20+ acre site for the past 4-5 months.

the shot rock fill was excellent. fist sized shot rock and well graded down to fines (they shot the hell out of the rock!--from other area) and placed in 8 inch layers, compacted and proofrolled during each day's placement. and as mentioned, the storm structures out in this area are actually founded in the original ground and they're settling too. i've even got geophysical data across this area and the fill exhibits high velocities.

as i think i previously mentioned, the fill area was a intermittent drainage swale and the water was picked up and piped a 1000-1500' upstream so that water is now gone which might be contributing. the swale was undercut to firm soil prior to filling. all density test results were good. had one of my best tech's on site full time.

i can't find the weak link other than the dropping groundwater levels causing the settlement. and being in amphibolite with dropping groundwater levels seems to throw up red flags to me. i'll see if the numbers flatten with the little bit of rain the past month or two. i'm so far in to the log plot that 30 days (time between readings) doesn't provide much indication as to the movement plot. i haven't personally run the time rate yet since it's a little hoaky in the piedmont anyway but i probably should just to see what it says. i've run it on other projects and the numbers are usually nearly the same with 30-75 days being typical. i've handed all the lab data over to the senior to look in to since my schedule hasn't given me the time to do much else other than survey the pins, plot the data and scratch my head. i'm pretty sure he's already run those numbers and come up with the usual answer.

 

[muuddfun]

The key question for you is to differentiate between settlement of your fill and settlement outside of you fill, you said above that the storm structures in the native ground have acturally settled also? Are they placed in areas completely outside of your fill area, or are they just deep enought that you had fill on top?

If a change in groundwater depth is the culprit it would seem that you may need more than 5 or 10 feet in change in water elevation to make the change in pressure large enough to make some difference. Usually here in california that change is groundwater surface is on the order of 50 feet to really get the subsidence started. But that is not to say that even a small change won't have an effect on your area. At 5 feet you are adding about 300 psf additional stress, is that more additional stress than you had planned on? The other thing to think about is that for the groundwater to have some noticible effect it needs to drop below the past lowest level. You should look into what the historic groundwater levels were in you area. In your earlier post you mentioned that the water level was about 80' in the nearby wells and that it had been at about that elevation sometime in the recent past. So even if the groundwater over the last few years has dropped 15 or so feet, it wouldn't make much difference if it has not gone below the past lowest point, except that maybe now it is feeling the additional weight of fill you have placed and the groundwater drop was enough to push you past the linear recompression portion of your consol?

Check out USGS for information on subsidence. They are doing lots of studies in areas out west were the groundwater withdrawl is causing subsidence in many areas. Usually though it is over a large are, more than the size of one project. You should ckeck if other property nearby has been subsiding.

Here is the url for a recent subidence report by USGS.

http://pubs.usgs.gov/sir/2007/5251/

 

[msucog]

no no...i should have been more specific. the structures that settled are in the fill area but are founded in original ground. the reference structure (where the benchmark is located) is outside the fill area.

i wouldn't expect a few feet (or a couple hundred psf) to make a dramatic difference. i did look at the historic levels back about 5 years...the difference is more than 5 feet and the levels now are the lowest available...and this is the worst drought in history for GA (well actually missed the worst year by like an inch). the lake levels are down 15'-20'. there again, we're only talking about 2 inches in 4 months. 500-1000 extra psf might do it though since the consolidation curves fall off steeper after about 1200-2000psf. i'm convinced it must be groundwater based on the second project's survey data being flat for 6 months then falling off again at the end of summer when no additional loading occurred. i'll keep digging and see what i can turn up...and see what the next readings show since we've had some rains.
thanks for the input.

 

[msucog]

i guess as a caveat, i should note that the second site (with the flat line then dropping) surveying doesn't have the same confidence in the readings as my site since we (our office) weren't in direct control of the project. however, the data does seem to follow the groundwater level trends we've pulled from the nearest well.

i suppose i should double check the time rate before i commit to throwing all my belief behind the groundwater theory since i did estimate 3 inches of settlement after the "instantaneous" consolidation. the 3 inches was to the conservative side based on the lab data.

i'll keep checking to see what i can turn up. thanks again.

 

[muuddfun]

The delayed change in settlement sounds suspiciously like something that does occur with drops in the level of groundwater. You probobly are on the right track. I don't know what the drainage path length of any of the compressible layers you have are. But when the layers of clay are very thick like in my area the time from when the groundwater was drawn down to when the subsidence is first noticed is delayed due to the drainge path length for the porewater out of the clay layers. The downside to that is when they go to recharge the groundwater to stop the subsidence there is also a delay for the same reason, and so they can recharge the groundwater and the subsidence will continue for some time.

You have a very intersting problem. Ill be intersted to see how it turns out.

 

[msucog]

no clays here...all silts...relatively fast draining in many directions. the little bit of clay in the drainage swale was undercut prior to fill placement for the most part.

i will be able to check time rate later this week once i get a chance to come up for air. i think i'll try and dig up more current well data too. it is a pretty interesting scenario...as long as it stops before they get ready to put up the building.

 

[msucog]

checked time rate...it indicated 30-60 days using the more conservative log time method (square root time is about three-quarter to one-half as much). as far as groundwater measurements, i've gotten more recent data (some of which is site specific for the past year or so) and it indicated a severe drop toward the end of '07. the most recent settlement pin readings indicate that the settlement is significantly flattening from last month's reading...perhaps all the rain since the end of december is finally catching up and bringing groundwater levels back up. based on the lab data, the field measurements are generally in agreement with estimated measurements if you assume groundwater dropped 10-15'. also, the more severe settlement appears to be in line with the deeper soil profile areas which tends to agree with my theory.

problem is: what happens in the summer time if the drought continues and worsens? the general trend of the ground water levels seem to continually drop long term even though there are ups and downs from season to season. how the heck are we supposed to estimate "realistic" figures based on what mother nature does or does not do? is anticipating a 5' drop in groundwater enough...how about 15' or even 25'? then once a settlement monitoring program is performed, how do you ever give a straight answer that the settlement has slowed enough to begin construction without having to completely disclaim every opinion? i suppose it's the inevitable cycle of always adding to disclaimers to cover every possible situation...(i sure would have liked to be an engineer in the times when an engineer could be an engineer instead of having to be half lawyer worried about litigation)

 

[muuddfun]

To figure out how much more the groundwater will drop you will need to find out how much pumping is going on and figure out how to stop it, recharge from other sources, or live with the settlements likely to be induced by it. If it is not related to groundwater pumping but soley is a function of site geology, then you will have to figure out how to live with the settlments. Do to the possiblity of having future settlements from additional groundwater drops, I would suggest that you not rely on just monitoring the settlement and then when it is stoped recommending that it is safe to build. You should figure out another method of dealing with the settlements, such a maybe pt-slabs or whaterver is appropriate. You should estimate the maximum settlement you would expect for a proboble worst case senario over the life of the project. Because even if you can stop the settlement now, but in 10 years another drought comes along and drops the goundwater even more than this time you would still have a problem. You should do some hydrogeologic analysis and determine what the likely lowest levels are that the groundwater would reach and start from there. In the central valley of california they have had ground subsidence of close to 30 feet over the years. Look for papers by Poland for more information on that and how to deal with subsidence issues.

I definely aggree that we should get rid of the lawyers. I doesn't matter what disclamers you put, the lawyers will still sue.

 

[msucog]

i think that performing what you suggest may be unreasonable for the most part here in the ga piedmont. on top of that, i'm fairly certain there's not widespread pumping currently going on in the area however that will likely change due to multiple munipalities bringing old pumps back online to substitute taking water from the major storage lakes after the experience with the current drought. but when i do have a lake on a site, i'll check the numbers in the event the lake is drained. but i don't think we can reasonably conclude how far the groundwater will drop due to a drought (other than throwing in a 5' or maybe even 10' drop very near onsite/adjacent lakes). see the attachment (i hope the link works)...it shows the lake level for the major lake here that is many miles away but controls water all the way to the gulf. while a 5' drop in lake level doesn't mean a 5' drop in groundwater a few miles away, the 20' drop appears to be helping it create at least a 10'+ drop in less than 6 months (may be combining with other factors to increase the effects). i can only now imagine what happened to sites near the lake where i'm certian water is/was flowing somewhat significantly under/around the area of the dam.

if we used a similar mentality for building loads, should we (here in ga) bump up the loads provided to us by the structural engineer in case we see a massive ice/snow storm lasting weeks during a drought period (assuming non-drought during exploration/construction) at the exact time an earthquake hits. i suppose you could but that would sort of be pushing the probability of those things happening concurrently based on known occurances. we already estimate to the conservative side...maybe it's more a matter of when portions of the settlement take place (or don't take place). if you've got a lake onsite, at least you can tie some settlement figure to "if the lake is drained effectively lowering the water table 15'".

i think the settlements may be so scattered around town(contained primarily on sites with deeper soil profiles loaded with at least moderate thickness fills) that designing for such events everywhere would be outside the local industry's standard approach (for lack of a better phrase). it's sort of like designing for a 100yr storm event knowing good and well that a bigger event could happen tomorrow. at some point, you've got to chalk it up to "mother nature". we're always careful with how we word settlement related issues and unforeseeable groundwater levels but perhaps that verbage should be stiffened up in light of the current drought.

thanks for the advice. i'll look up the papers you mention.