Bulletin of the Transilvania University of BraşovCIBv 2015 • Vol. 8 (57) Special Issue No. 1 - 2015

THE INFLUENCE OF THE INPUT DATA ONUNSATURATED HYDRAULIC SOILPARAMETERS AND ON UNSTEADY

SEEPAGE RESULTS FOR EARTHFILLDAMS

M.L. BOTOȘ 1 T. MILCHIȘ 2

Abstract: Calculation of seepage on small earth fill dams inevitablyinvolves modelling flow through unsaturated zones. The parameters thatinfluence the flow through unsaturated environments are important for therain water that seeps through the unsaturated slopes, or when we refer tounsteady flow with rapid variations of water level. Obtaining theseparameters involves pricey and time consuming laboratory measurements.Different stages of calculations require different precision of seepagemodelling. These parameters can be estimated using existing databasesobtained over time, the results are influenced by the quality and quantity ofinput data available at the moment. The unsaturated hydraulic parameterscan be predicted from previously estimated soil hydraulic parameters (theforward problem), from observed retention data and quantifying thehydraulic properties by simultaneous analysis of a limited number of soilwater retention data points. The present paper shows differences in seepagecalculations using different parameters obtained for three different inputdata.

Key words: unsaturated soils, parameter estimation, axis translation.

1 Faculty of Civil Engineering, Technical University of Cluj-Napoca, Cluj-Napoca, Romania2 Faculty of Civil Engineering, Technical University of Cluj-Napoca, Cluj-Napoca, Romania.

1. Introduction

Earth fill dams are the most commontypes of dams all over the world.Designing earth dams involves somelogical stages. The first step to identify thecorrect location for an earth fill dam isfinding the right material to build with andon, meaning that we know the type of soilavailable on the site: fat clay, low or highplasticity clay, silty sand, clayey sand orhigh plasticity silt. Estimation of

parameters influencing seepage for thistype of input data is available usingspecialised software like RETC, Hydrus-1D [17]. This software is using existingdatabases created during long period oftime in order to offer satisfying estimatedresults. With the first set of data engineerscan estimate the shape of the cross sectionfor the most economic type of dam. If thesolution is earth fill dam, based on thegeometrical recommendations available[11] we run the first estimations

Bulletin of the Transilvania University of Brasov • Vol. 8 (57) Special Issue No.1 - 2015206

considering the seepage behaviour andstability calculations for the upstream anddownstream slope. If the results are in theeconomic efficiency limits the initial studyfollows its path.

Next step the first geological tests aremade and we can already know thecontent of clay, silt and sand, the bulkdensity, saturated permeability and anumber of other strength relatedparameters. The designers can alreadymake first estimations consideringseepage behaviour, slope stability factorsand shape the filling for the dam.

Moment when the construction starts,earth, which will be used can be tested onthe field and number of compaction testswill be made in order to find the correctmethod of capacitation to obtain theestimated characteristics used in thedesigning phase. Samples can be tested inthe laboratory in order to find the seepagecharacterising parameters [12].

After finishing the construction, the damstarts its operation stage; the behaviour ofthe filling is monitored using specificdevices, placed in sensible sections ofdam. If the results predicted in thedesigning and construction stages are notobtained, more tests are conductedregarding the exact characteristics of theearth fill [13].

The present paper will try to show theinfluence of available input data on theseepage behaviour. After the floodsregistered in Romania in 1970, largenumber of small dams [4], most of themearthfill [4], were raised in order tocontrol and limit flow on small waterstreams. Being small height and with notimportant storage capacity, most of themwere constructed with just a fewcalculations and many times using crosssections recommended [11]. Small damsare characterised by small storage and fastraising levels of water during floods.

The present paper shows some results

obtained for such dam. All thecalculations and tests are made on theCalinesti Oas homogenous earth fill dam.Raised between 1971-1973 the CalinestiOas dam is situated on River Tur, 8 kmupstream of river Talna; Tur, Valea Albaand Valea Rea are the main water sourcesfor the lake. Dam is homogenous earth fillmade of fat clays, 9.50 meters high, crestwidth of 4.40 meters and slopes of 1:3.5upstream and 1:3 downstream. Maximumwater storage capacity is around 20million cubic meters. The dam’s site ischaracterized by the presence continuouslayers of fat clays of 4 to 5 meters laidover a thick layer of river deposits. Samematerial was used in the earth fill.

In this paper are analysing two ofimportant floods recorded in the lake, inthe year 1998 and 2001, trying to show ifany difference on results of seepagecalculations can be observed if theparameters for the conductivity andstorage capacity for the material used areestimated or measured. The calculationsof seepage curves were performed usingMNPNS.exe software [3], and the post-processing of results using GMSH.exe [9].

2. Parameter Estimation

The equation of the unsteady seepagethrough unsaturated porous media is:

yykyxxk

x

t

C

(1)

For the definition of permeabilityfunctions are numerous examples in thescientific literature [1,2,5,6,14,15].

For the estimation of permeabilityfunction is used the Burdine model [6]:

mme

ler SSk /111 (2)

BOTOŞ & MILCHIŞ.: The influence of the input data on unsaturated hydraulic soil ... 207

where:kr - relative permeability;Se - the effective saturation defined as:

rs

reS

(3)

θs is the saturated moisture content andθr is the residual moisture content of thesoil.

The relationship between pressure headand effective saturation derived from thevan Genuchten model [15]:

mneS

1

1(4)

For van Genuchten model the equationof specific moisture capacity becomes:

C

11 1

mnnrs mnC

(5)

In order to run any seepage calculationwith the available measured water levelsin the lake for the major events recordedin the dam’s section, all this parametersare needed. The saturated and unsaturatedpermeability and the unsaturated storagecapacity must be measured or estimatedfor the existing soil in the dam andfoundation.

Measuring in the laboratory theseparameters is time consuming [10],[12],involves skilled personal and the resultscan be obtained with a number of direct orindirect methods. Many times isrecommended the automated orsemiautomated methods for this kind ofslow process over long periods of times[13]. Based on the soil-watercharacteristic curve (SWCC) the indirectmethods are offering satisfying results onthe unsaturated conductivity or on the

storage capacity. The SWCC can be usedor estimated based on the grain sizedistribution, bulk density and void ratio[8], [17].

The parameters for unsaturatedparameters for our study were obtainedusing RETC.exe software [17] for threedifferent types of data:

1. Known type of soil in foundation andinside the dam: fat clay;

Earth fill: Fat clayθr=0.068; θs=0.38; α=0.74;n=1.09; Ks=5.55x10-7 m/s

2. Measured grain size distribution.according to a study conducted by thePolytechnic Institute of Cluj-Napoca ineighties, the dam’s body filling is amixture of clays in a proportion of17.97%, 70.02% clayey silts and claysands at a rate of 12%;

Earth fill: 25.5% Clay; 46.7% Silt;27.8% Sandθr=0.0743; θs=0.4314; α=0.74;n=1.5338; Ks=1.51389x10-6 m/s

3. Measured SWCC for the materialextracted from the dams body.

Two undisturbed probes were extractedfrom our dam in order to performlaboratory measurements. Considering thelow estimated conductivity parameter thematerial was tested using falling columntechnique. The saturated conductivityobtained during the tests was 1.35x10-9

m/s. Axis translation was performed toobtain a number of retention points. Weused pressure cells with HAE plates withthe bubbling pressure of 3 bar.

Axis translation was performed to obtaina number of retention points. We usedpressure cells with HAE plates with thebubbling pressure of 3 bar.

Bulletin of the Transilvania University of Brasov • Vol. 8 (57) Special Issue No.1 - 2015208

Fig. 1. Pressure cell with 3 bar HAE

Table 1Pressure θ Pressure θ

[bar] [bar]0 0.413 0 0.458

0.70 0.300 0.30 0.3450.90 0.289 1.00 0.2641.40 0.266 1.50 0.2492.00 0.246 2.00 0.2252.50 0.209 2.50 0.200

The retention data was used in theRETC [17] software in order to fit theshape parameters m, n and α for vanGenuchten and Burdine retention andconductivity functions.

Fig. 2. Retention data: θ vs. log h

Results:Earth fill:θr=0.068; θs=0.462; α=0.1822;n=1.649; Ks=1.35x10-9 m/s

Fig. 3. Hydraulic properties: C vs. log h

Fig. 4. Hydraulic properties: K vs. log h

3. SEEPAGE Results

Two major floods were analyzedaccording to the available data. The inputwas the water level measured during thefloods recorded in year 1998 and 2001.The floods were considered following along period of time before with levels ofwater close to normal operationscondition. This detail will permit us toconsider that initial seepage conditions aresteady. The measured data at thepiezometers could not be used in our studybecause wrong response caused by lowpermeability and faulty response.

BOTOŞ & MILCHIŞ.: The influence of the input data on unsaturated hydraulic soil ... 209

3.1 Flood Wave Registered Between14.06.-22.06.1998

The period preceding this flood ischaracterized by water levels in the lakeclose to the normal and not major changesin time. The water level in the lakerecorded on 14.06.1998 is considered thelevel at which the seepage is steady.

During the study period the water levelin the lake raised slowly over a period ofsix of days, from 142.43 to 145.05 metersmaximum registered on 21.06.1998. Thistype of flood is recorded in the rainyperiods of the year. June is known interms of rainfall as one of the richestmonths. Boundary conditions used werethe measured water levels in the lake forthe upstream face and the downstream toewas considered drained (given that wehave no water at downstream toe).

If at the downstream toe of the damnothing changes during floods for the

initial conditions, the difference onpermeability shows some importantchanges at upstream slope, differencesthat can change substantially the stabilityfactor for the upstream slope. The seepagecurves obtained during the recorded floodare presented in the following figures andare numbered accordingly to the numbersused for the corresponding hypothesisused to estimate parameters.If at the downstream toe of the damnothing changes during floods for theinitial conditions, the difference onpermeability shows some importantchanges at upstream slope, differencesthat can change substantially the stabilityfactor for the upstream slope. The seepagecurves obtained during the recorded floodare presented in the following figures andare numbered accordingly to the numbersused for the corresponding hypothesisused to estimate parameters.

Fig. 5. Recorded water levels. June 1998

Fig. 6. Seepage curves 16.06.1998

Fig. 7. Seepage curves 18.06.1998

Bulletin of the Transilvania University of Brasov • Vol. 8 (57) Special Issue No.1 - 2015210

Fig. 8. Seepage curves 20.06.1998

Fig. 9. Seepage curves 22.06.1998

3.2 Flood Wave Registered Between02.03.-12.03.2001

The flood is characterized by a suddenincrease during first two days from thenormal water level value of the 146.11meters, well above the warning level;followed by a slow decrease of waterlevels over a week.

Due the important variation of waterlevel at the upstream face of dam,differences at the downstream toe for theseepage curves are also important in caseof this flood. The stability of slopes in thiscase can be significantly influenced for theupstream downstream slope.

Fig. 10. Recorded water levels. March 2001

Fig. 11. Seepage curves 07.03.2001

Fig. 12. Seepage curves 09.03.2001

BOTOŞ & MILCHIŞ.: The influence of the input data on unsaturated hydraulic soil ... 211

Fig. 13. Seepage curves 11.03.2001

Fig. 14. Seepage curves 13.03.2001

4. Conclusions

The constant need of new waterresources, exploiting the hydropowerpotential of river streams, tourism,fisheries, and financial constraints that weface, force us to become more creative,when it comes to find new places to makepermanent reservoirs. In Romania a hugedevelopment potential is represented bydams built over time in order to attenuateflooding in the upper basins, dams withnon permanent reservoirs. These damshave non permanent reservoirs and it is acategory, which due to their small sizesomehow were neglected. Using of optimaldesign solutions according to newoperating conditions will permit increasedefficiency of such investment.

Small earth fill dams with permanent ornon permanent lakes usually weredesigned using charts with recommendedslopes and crest width. Even if luckilythese small dams (constructed mostly afterthe 1970’s floods) were designedconsidering seepage and slope stabilitycalculations, seepage was consideredsteady so they slope stability reserve, andluckily this grows same time with theheight of the dam.

Results are showing that using estimatedparameters and not measured parametersfor seepage analysis will lead to lessfavourable results when it comes to verify

the stability of slopes, meaning that earthfill dams in operation have more safetyreserve.

References

1. Aubertin, M., Mbonimpa, M.,Bussiere, B., Chapuis, R.P.: A modelto predict the water retention curvefrom basic geotechnical properties. In:Canadian Geotechnical Journal,40(6), 1104–1122, 2003.

2. Benson, C.H., Gribb,M.: Measuringunsaturated hydraulic conductivity inthe laboratory and field, 1997.

3. Botos, M.L.: Solving steady andunsteady seepage problems usingcustom programs, C60 InternationalConference, Cluj-Napoca, 2013

4. Botos M.L.: Small dams with non-permanent reservoirs in Romania, C60International Conference, Cluj-Napoca, 2013.

5. Brooks, R.H. and Corey, A.T.:Hydraulic properties of porousmedium. IN: Hydrology Paper3.Colorado State University, Fort, 1964.

6. Burdine, N.T.: Relative permeabilitycalculations from pore-sizedistribution data. Trans. Am. Inst.Min. Metall. Pet. Eng. 198:71.77,1953.

7. Collins. Buckingham, E.: Studies ofthe movement of soil moisture,

Bulletin of the Transilvania University of Brasov • Vol. 8 (57) Special Issue No.1 - 2015212

U.S.D.A. Bur. Of Soils,Bulletin No. 38, 1907.8. Gallage, C.P.K., Uchimura, T., Effects

of dry density and grain sizedistribution on soil–watercharacteristic curves of sandy soils. In:Soils and Foundations 50 (1),161–172,2010.

9. Geuzaine C., Remacle, J.-F.: Gmsh: A3-D finite element mesh generator withbuilt-in pre- and post-processingfacilities, 2009.

10. Hamilton, J.M., Daniel, D.E. andOlson, R.E.: Measurement of hydraulicconductivity of partially saturatedsoils, 1981

11. ICOLD : SMALL DAMS Design,Surveillance and Rehabilitation, 2011

12. Klute, A.: Laboratory measurement ofhydraulic conductivity of unsaturatedsoil. In: Methods of Soil Analysis,Mono.9, Part1. Amer. Soc. ofAgronomy, Madison, WI, pp. 253–261, 1965.

13. Liu, Q., Yasufuku, N., Omine, K.,Hazarika, H.: Automatic soil waterretention test system with volume

change measurement for sandy andsilty soils.In: Soils and Foundations 52(2), 368–380, 2012.

14. Mualem, Y.: A new model forpredicting the hydraulic conductivityof unsaturated porous media. In WaterResources Research, 12(3), 513–522,1976.

15. Van Genuchten, M.T.: A closed-formequation for predicting the hydraulicconductivity of unsaturated soils. In:Soil Science Society of AmericaJournal, 44, 892–898, 1980.

16. Van Genuchten, M.T., Leij, F.J., Wu,L. (Eds.): Characterization andMeasurement of the HydraulicProperties of Unsaturated PorousMedia. University of CaliforniaRiverside, CA, pp.737–748.

17. Van Genuchten, M.Th., F.J. Leij, andS.R. Yates.: The RETC code forquantifying the hydraulic functions ofunsaturated soils. Res. Rep. 600 2.91065. USEPA, Ada, OK, 1991.