HTIdrotechnicaI Constr~zction. Vol. 3~. No. ~. 2000
MATHEMATICAL G E O F I L T R A T I O N M O D E L OF T H E S Y S T E M "SOIL FOUNDATION - HYDRO INSTALLATIONS" FOR THE PLAVINAS HYDROELECTRIC POWER PLANT
V . V . II'in, V. P. T v e r i t n e v , Yu. S. S h e v l y a g i n a n d A . I. Y u d k e v i c h
T h e Plavinas hydroelectric power plant, which forms the third stage downstream in the cascade of hydro units on tile Daugava River (Latvia), was constructed in 1962-66 under a plan developed by the Russian "'Gidroproekt" Institute. T h e configuration of the hydro unit with a 40-m head includes (Fig. 1) a spillway hydroelectric plant with a headwall length of 200 m, a riverbed earthfill dam with a length of 480 m, a left-abutment earthfill dike extending 2 km into the Dangava watershed and its tributary the Lautse River, a right-abutment earthfill dam (approximately 450 m) and its 1-km continuation, a low-head dike. To lower the heads and prevent seepage soil de~brmations in the foundation of the building structures, the plan calls for various drainage constructions. The thirty-year operation of the hydro unit has been accompanied by a continuous suffosive erosion of unconsolidated rock (riverwash) carrying off fine soil particles. The content of fine earth in the drainage water varies, fluctuating over a wide interval from 10-15 mg/liter to 10 g/liter and even 100 g/liter, depending on the seepage conditions. During the period of operation, according to various estimates, the washout volume has been anywhere from 1.5 to 3.0 thousand cubic meters, resulting in hazardous movements of tile concrete structures and discharge openings. At the present time the heads under the hydroelectric plant buiMing are observed to be rising. Measures to reduce them, for example, by adding more drains, must be carefully evaluated for aftereffects, because even a local drop in the heads can be accompanied by movements of the foundation plate. To investigate tile hydrodynamic regime attending the Plavinas hydroelectric plant and to run prognostic calculations of its modification after plammd remediation measures, the L a t v i a n energy authority Latvenergo has commissioned Gidroproekt to develop a threedimensional (3D) "geofiltration" (groundwater seepage) model of the system soiI ~bundation-hydro installations. G e o l o g i c a l E n g i n e e r i n g C o n d i t i o n s . The geological engineering conditions of the physical buildings of the Plavinas hydro unit are determined mainly by the structural features of the river valley and by the nmltiple effects of the headwaters of several water-bearing horizons. T h e Daugava River valley is set in a Devonian period rock mass overlaid with unconsolidated Quaternary deposits. Its evolution presents a very elaborate history. Within the limits of tile range there is an ancient erosion notch (Fig. 1) gouged out to a depth of 70-80 m relative to the present riverbed and filled primarily with moraine. The buried valley cuts through the bedrock mass to the roof of the Gauja Formation. T h e occurrence of repeated erosion trenching is of considerable significance, because the heavy concrete building of the spillway hydroelectric plant has been erected on the morainic soils filling it, and a drainage channel has been excavated there. Deposits of questionable genesis and variegated composition, primarily unconsolidated rock, have developed nnder the moraine on the right abutment: they form what is called a "trail" and are distinguished by very high permeability. The water-bearing horizons commmficate hydraulically along the trail, fbrming an intricate, 3D flow (Fig. 2). The water-bearing horizons are listed along with their main characteristics in Table t. S o f t w a r e M e d i u m a n d G e o f i l t r a t i o n S c h e m a t i z a t i o n . Tile geofiltration model is constructed in the "Ground W a t e r Flow Simulation" (GWFS) system developed by the Russian-Swedish joint venture Geosoff-Eastlink .IV [1]. G'WFS provides extensive capabilities for modeling steady-state and transient groundwater seepage in stratified and structurally inhomogeneous water-bearing systems in such a way as to execute groundwater flow in three dimensions. T h e software capabilities are utilized to the flfllest in the model of the Plavinas hydroelectric plant. T h e main characteristics of the natural twdrogcologic conditions reflected in the model are simulated by two interactive hydraulic systems with descending and ascending components of the seepage flow. The first system combines the alluvial. Daugava, and Plavinas water-bearing horizons. These horizons have the highest geometrical
Translated from Gidrotekhnicheskoe Stroitel'stvo, No. 4, pp. 40-46, April, 2000.
0018-8220/00/3404-0187525.00 @2000 Kluwer Academic/Plenmn Publishers
187
/
/ !
Fig. 1. Modeling region and schematic layout of tile Plavinas hydroelectric power plant installations: 1) right-abutment dike; 2) right-al)utment earthfill dam: 3) concrete spillway of the hydro plant: 4) riverbed earthfill dam; 5) left-bank dike: 6) drainage tunneh 7") contours of ancient erosion notches, not shown in relief: 8) drainage curtain: 9) boundary of the region of heightened water permeability (trail); 10) boundary of tile "Natural Environment" model, M 1:2000; 11) boundary of the inset mode}, M 1:500: I-I) profile cutaway line. position, a geimrally free surface, an imCiltration supply, and hydraulic connection with the Daugava River. The second system includes }lead and head-headless water-bearing horizons: a glacial water-bearing complex and the Plavinas, Amata, and Gauja water-bearing horizons, which come together along tile ancient erosion notch. The piezometric heads in this system increase with depth, and tile seepage flow together with the horizontal flow is characterized by an ascending vertical component. When the foundation pit was excavated, the horizons were drained off for the most part by the structural lowering of the waterline. Later, once the bull(ling had been completed, the water-bearing horizons not only recovered an abundance of water, but also acquired an additional supply ~}om the reservoir. The geofiltration schematization is based on a 3D geological model of the foundation of tile building installations, adapted to GWFS by means of a built-in "Surfer" program. The main principles of tile geofiltration schematization of the Plavinas hydroelectric plant site are also shown in Table 1. G e o f i l t r a t i o n M o d e l . Nonuniform grid (flownet) discretization with rectangular and nonrectangular (the upper and lower faces being 3D surfaces) mesh cell configurations have been used in constructing the geofiltration model of tile Plavinas hydroelectric plant in the GWFS system. Each cell of the model is a stratigraphic unit whole. Tile cells are joined into computational layers in confi)rmity with tile geofiltration schematization and according to tile availability of infbrmation obtained by fui}-scale, in .situ observations. The model contains 17 computational layers (strata). An exhaustive set of seepage and solid-geometrical characteristics, boundary conditions, vertical, horizontal, ~md areal drains, and systems of relief wells are specified in the computational layers, along with full spatial contours of the buildings and antiseepage elements, forming a 3D seepage flow pattern.
188
T A B L E 1. Geofiltration Schematization of the Plavinas Hydroelectric P l a n t Area Hydrogeologic rock ct~sification
Description
Tech nogenic fbrmations
Sands. sandy Ioams. Ioams, and concret,e
Alluvial deposits
Gravel-pebble deposits with sandstone filler Loams with sandy loam and sand lenses
Sandy loam with sand, loam. and clay seams
8 e-
(3 Daugava strata Sala.spils strata Plavinas strata
Dolomites, fissured, karsted; interstices filled with waste rock Marly clays with dolomite and marl seams Dolonfites, fine-crystalline, fissured, with marl and laminar clay seams
Hydrogeologic unit classification Water-bearing StratiWater-I)earing horizons and graphic complexes. water barriers index Technogenic tit l Alluviowater barrier technogenic Technogenic tt(-' horizon Alluvial aQm horizon Glacial, loamy, gQu (4) Glacial slightly water-bearing permeable lakver complex Glacial gQii(4p) deconsolidated layer Upper gQn (3) subhorizon Middle gQu (2) subhorizon Lower gQll ( 1) subhorizon Daugava Dad horizon
©
$ Amata strata
Gauja strata
Sands, fine-grained, with thin sandstone seams and clay lenses. A persistent cb\v seam is distinguished in the mass Loosely cemented sandstones with sparse clay seams
.i
.~ 8 E0
~2
Salaspils barrier Plavinas horizon Plavinas barrier Amata horizon Amata barrier Bower Amata horizon Gauja barrier Gauja horizon
No.
Thickness, m
Percolation rate, m/day
1.2
2-60
0.0 l-1).001
3
2-20
0.0t-t0
4
t.5-15
2-50
5
2-5
0.001-0.05
6
5-15
0.2-5
I la
2-20
13a
5-28
1 - 1()-:~5. 10-~ 0.1-3.5
15a
2-20
0.00001-0.02
7
10-12
5-100
Daspl
8
12-15
0.00()01
Da pl ~
9
20-22
0.5-100
DapP
10
2.5-3.5
0.001-0.003
Daa ~
11
20-30
0.5-60
Daa t--'
12
2.0
0.12-2. 10-"
D3a ~
13
28.0
0.2-7
Dagj z
14
2.0
0.2-0.001
Dagj
15
> I00
15-25
Tile w o r k proceeded in two stages. Tile first stage entailed construction of tile model " N a t u r a l E n v i r o n m e n t Befi)re the S t a r t of C o n s t r u c t i o n " on a scale of 1:2000 and a modeled area of 230 ha (2.3 kin2). T h e flownet discretization o f tim model in the horizontal p l a n e was a 20 × 20-m square. T h e vertical diseretization was blockstructured. T h e numl)er of cells in one c o m p u t a t i o n a l laver was 71 x 81 = 5751. Nine c o m p u t a t i o n a l layers c o n t a i n e d 51.759 c o m p u t a t i o n a l nodes. T h e " N a t u r a l E n v i r o n m e n t " model was used as a basis for the construction of s u b s e q u e n t large-scale m o d e l s , or insets. I n t h e second stage the scale of the inset models was enlarged to 1:500. T h e modeled area was reduced to 0.8 km ~, b u t t h e (:ell size in p l a n view could be decreased to 2.5 x 2.5 m. T h e n u m b e r of cells in one c o m p u t a t i o n a l layer was 56,680. The total n u m b e r of c o m p u t a t i o n a l ('ells in the model was approximately 800,000. T h e second stage of the m o d e l was developed in steps. A n indeI)en(lent model corresponding to a certain period of c o n s t r u c t i o n or o p e r a t i o n of the i n s t a l l a t i o n was developed in each step. A total of nine retrospective (epignostic) models were form6d, r e c r e a t i n g the successive variation of the seepage regime [?ore the state observed at the tilne of excavation of the f o u n d a t i o n pit at the c o n s t r u c t i o n site. T h e step-by-step m o d e l i n g technique p e r m i t t e d more complete utilization of information accunmlated a b o u t the w a t e r - b e a r i n g horizons in the course of state-of-the-enviromnent observations.
189
m
80 J
L e f t - a b u t m e n t dike
.71J)
C r e s t of d a m
~L.;tl&
70
Riverbed dam
6O 55 50 45
?
....
' ........................
......... , ,
40
35 30 25 20 15 10 5
i
L W a l l LV-5 : ." ~-4,~
9
-; .~.-~, m~..-"~ ~
. . . . . . . . . . . . ~ ~
-
~
JP---
....... ~0
--
!i!i t.o g
.
-4O 45 -50
. Plant building
I
........
-5 -10 -15 -20 -25 -30
Right-abutment dike
!
I---:--r
/
If
15
-55 -60 -30
Concrete s t r u c t u r e s .
2
1.
Stratigraphic
0
Earthfill dam, sand, sandy loam, loam. Technogenic water-bearing horizon [t h( 2)].
~.
Sands, fine-cjl"ained, with thin s a n d s t o n e and clay[D3a(2)]. lenses. ~seams ta horizon
Gravel-pebble deposits with s a n d s t o n e filler. Alluvial horizon (aQ[ll).
~i~
Sandy loam with sand, loam. and clay seams. Upper subhorlzon [gQIl(3)] of Gaula horlzon.
Loam with sandy loam and loam lenses. Glacial '°am layer [gCitl(4)]"
Clays with thin sandstone seams. Antata barrier [D3a( 1-2),.
Sandy loam with sand, loam, and clay seams, Glacial deconsolidated layer [gQIl(4p )].
5 a n d s , fine-grained, with thin s a n d s t o n e
Marly clay with dolomite and marl seams, SalaspilsbarrieriD3spl).
Dolomites, fine-cr:,'staRine, fissured, mad ~!" and laminarclay seams. Plavinas horizon [D3pl(2)]. ~:
Scale i
Technogenic barrier [th(1 )l.
Dolomites, fine-crystalline; fissured, porous, karsted; interstices filled with dolomite w a s t e rock. Daugavahorizon(D3d). ~
Legend
Marls, fissured, with dolomite and laminar clayseams. Plavinashorizon[D3pl(t)].
r
i
i
2000
d
,
•
i
4000 6000 8000 '10000
2. L i t h o l o g i c a l
Sand
~ ., ~,,...:.~
I
~
3. Hydrogeologic .......
Headless
water
level
Sandy loam ,~l"1~1
seams and clay lenses.
1~
L o w e r Amata horizon [D3a(1 )].
~ r / ~ z , ~~
Sandy loam wrlh sand, loam, and clay seams. Middle subhortzon [gQll(2)] of lower glacial horizon.
[
~, Claywiththinsandstoneseams. z-': GaUla bamer [O3Rl(2}]. ~
Headwaters
Loam
i
Clay G r a v e l - p e b b l e soil with sand filler
4. Other C e m e n t a t i o n c u r t a i n in G a u j a a n d plavinas water-bearing horizons
L o o s e l y c e m e n t e d sandstones w e h sparse
,
clay seams. Gauja horizon {D3gJ(1)].
~1~ Sandyloamwithsand, loam, and clay seams. Lowersubhonzon[gQIl(1)]ofGaujahorlzon.
~ I 1 ! Dolomite ' ~ " =" I.I. 1...I "~ [ X-~'~ Sandstone i~ • ~
Stratigraphic boundary Lithographic boundary
Fig. 2. Geofiltration profile along the axis of the head structures of tile Plavinas hydroelectric plant according to the state of the facility in October, 1998. C a l i b r a t i o n o f t h e M o d e l . In each step the model is calibrated from tile results of full-scale observations of tile underground water levels, tile discharges of relief and dewatering wells, and discharge measurements in the drains. Tile calibration is based on the method widely used to solve inverse I~roblems in tile practice of hydrogeologic testing of rocks and simulation. It entails successive approximation of the model levels and discharges to the levels and discharges determined by full-scale measurements with trial-and-error selection of the streamflow parameters of tile strata, wells, and drains and by refinement of the boundary conditions. For steady-state calibration of the model it is assumed that the deviation of the model levels from the measured levels should not exceed 1.0-1.5 m. The reduction of tile error of approximation is limited by the influence of tailrace oscillations, which are ignored in flfll-scale observations. An analysis of tile results of statistical processing of tile deviations between the model and measured levels attained in a calibration of the model in November of 1998 is summarized in Table 2. tt follows from Table 2 that the average deviation of the model levels from the measured tevels is limited to a confidence interwd from -0.44 m to +0.94 Ill. A certain disparity with this interval is noted ill the Gau.}a horizon ( - 1 . 3 9 m) and ill the Daugava horizon (+1.54 m) because samples of three and six vahles, respectively, are nonrepresentative. After the calibration a model of present conditions was obtained from the state ill October of 1998, recording the seepage regime with level marks of 70.6 m for tile reservoir and 34 m fbr /,tie tailrace. The model of present conditions has been certified on a test problem involving retrospective computations with shutdown of the siphon relief wells and a reconstructed situation recorded by full-scale observations in August of 1998. M o d e l i n g R e s u l t s . The model reproduces the aD hydrodynamic regime of seepage flows in interactive water-bearing horizons. Ill tile lower A m a t a water-bearing horizon the piezometric surface ibrms a depression funnel,
190
T A B L E 2. Results of S t a t i s t i c a l Processing of Discrepancies B e t w e e n £lodel arid Measured Levels in O c t o b e r of 19.%
Index
Water-I)earing horizon
Number of piezomet,ers~ n
Mean discrepancy of model and measured levels in piezometers :',f[-
":(h',,,,,,,-h'
Standard deviation
~ :{( s-~,,,,,,~-H.,,- ,.xss] N--
ConIi
Error of standard deviation
V
+
~ H + tp ,/~+~
G. ,/r~
H~AH-
1
..... )
giQH<,
20 21 45
0.2l 0.12 0.04
t.20 0,97 1.15
0.26 0.21 0.17
-0.34 ~< AI7 ~ 0.82 -0.3,t ~< AH ~< 0.58 -0,30 ~< A I t <~ 0.38
glQl~(-al
t7
0.25
1.14
0.27
-0.33 ~< A H E 0.83
34 20 17 3
0.55 0.2 0.42 0.17 -0.5
0.91 0.97 1.0 0.93 0.5
0.37 0. t7 0.22
-0.44 ~< As'/ ~ 151 -0.16 ~< A H ~ 0,56 -0.02 ~< ~ H ~< 0.86 0 ~< A H ~< 0.94 -1.39 ~< A H ~< 0.39
Tech nogenic
thw
Alluvial Glacial sandy loam layer Glacial deconsolidated layer Daugava Plavinas Amata Lower Amata Gau.ja
aQiI-lll
D,~d Dapl D3a
;g
Daa ~ Dagi
m go [
,
0 . '2 '2
,
0.3
_z~
70 |
FB.
":'0.6
< :
60 i,
•
so
"
~0
.
.
ooi:2
t0o
.
: 6.2.0
~
.
.~
~tll
"
Z~~
:
-,
! ~Y4,-
::,.
'~
, : ~
:.
.......
=_.._::7"::r~':"vv:,.."/.~(~#?
'r:: .....
20
-~01
~
i
~
.
! 11
20
10
~0
gO
log
120
140
154}
110
200
~0
241I
260
28(I
]t~
Distance, 120
percolation rate
m
Scale
Legend
--45.0--
equipotential line
0
l0
20
30
40
50
Fig. 3. H y d r o d y n a m i c s e e p a g e flownet along t h e c u t a w a y line I-I.
which st)reads o u t like a contrail in a s o u t h e r l y d i r e c t i o n w i t h a linearly e x t e n d e d center along the drainage curtain. The lowest level m a r k s in the lower A m a t a horizon d r o p to 35 m. In the glacial deconsolidated layer the p i e z o m e t r i c level m a r k s d r o p from 69 ill in tile forebay to 34 m ill the afterl)ay (tailrace). There is a d i s t i n c t l y t r a c e a b l e zone extending a l o n g the axis of the concrete a i d earthfill constructions w i t h a sharp drop in the marks, c r e a t i n g a flow gradient as high as 1.5-2.5. T h e % d r o d y n a m i c seepage grid c o n s t r u c t e d from the s u b l o n g i t u d i n a l 1-1 s e c t i o n of this zone e x h i b i t s a region where the equipotential lines (elevated gradients) bunch together at the m o r a i n e - " t r a i l " i n t e r n e e w i t h a p e r c o l a t i o n r a t e as high as 200 m / d a y (Fig. 3). This c o m b i n a t i o n creates conditions for the c o n t a c t washout of rocks w i t h the subsequent transport of m a t e r i a l into the relief wells and drains. A region of e l e v a t e d g r a d i e n t s is also o b s e r v e d near the relief wells. In addition,
191
similar regions exist near a chister of high-capacity wells located in the immediate vicinity of the hydroelectric plant building. The distribution of elevated-gradient regions and the conditions underlying their formation are have important bearing on the assessnlent of safe installation operating conditions. Consequently, regions of elevated gradients posing a potential hazard for the concrete installations have been established from the modeling results. The constructed model can be nsed to investigate the hydrodynamic regime in these regions and to estimate die efficacy of measures to lower heads {pressure threshold) in the foundation of the hydro installations. Individual engineering decisions have been tested on the model, and their efficacy has been estimated. The following situations have been considered: breaching of the key between the hydroelectric plant building and the front of the spillway, critical deviation of high-capacity relief wells and their compensation by reserve wells, the construction of drainage shafts and new relief wells; etc. One of the principal features of the model is its adaptability to variable hydro unit operating conditions. This particular asset qualifies it as a permanent hydraulic monitoring base, which can be used to analyze installation foundation processes associated with variations of the seepage regime.
Conclusions A three-dimensional geofiltration model of the a r e a of the Plavinas hydro unit has been developed and calibrated from the results of fiill-scale, in sihz observations. The model reproduces complex-interacting groundwater flow-s in the area and contains, for all practical purposes, a coniplete set of in~brmation on the regime-tbrniing structural elements of the installation and on the intervening hydrogeologic environment subjected to technogenic influence during operation of the hydroelectric plant. Regions of elevated gradients posing a potential hazard to the stability of the concrete installations have been established from the n.iodeling results. The model can be used to investigate the hydrodynalnic regime in these regions and to estimate the efficacy of measures aimed at lowering the seepage heads. The model has been used for prognostic computations of changes in the seepage regime both as a result of eniergency situations and due to intentional design nleasures. The model is sensitive to changeable hydrogeologic conditions, offers options for adaptation to new conditions. and it flnlctions permanently. It is therefore a practical tool for the hydraulic monitoring of hydro unit operation.
REFERENCES
1.
192
Soft'ware Package: The Hiqdrogcologi.~t's Workshop [in Russian], Vols. 5 and 15. Geosoft-Eastli~lk JV, Moscow (1996-98).