ArticlePDF Available

Parametric and numerical modeling tools to forecast hydrogeological impacts of a tunnel


Abstract and Figures

The project of a hydro electrical diversion tunnel through a crystalline rock massif in the Alps needed a detailed hydrogeological study in order to forecast the magnitude of water inflows inside the tunnel and the possible effects on groundwater flow. The tunnel has a length of 9.5 km and is located on the right side of Toce River at Crevoladossola (Verbania province, Piedmont region, Northern Italy). In the geological framework of the Alps, the tunnel is located inside the Lower Penninic Nappes, in the footwall of the Simplon Normal Fault; the geological succession is mostly represented by Antigorio gneiss (meta-granites) and Baceno metasediments (meta- carbonates). Due to the presence of important mineralized springs used for commercial mineral water, the hydrogeological study focuses both on quantity and quality aspects, by means of: rainfall data analyses, monitoring of major springs flow rates, monitoring of hydraulic heads and pumping rates of existing wells/boreholes, hydrochemical and isotopic analyses on springs and boreholes and hydraulic tests (Lefranc and Lugeon). The resulting conceptual model evidences a dominant low permeability (aquitard behaviour) of gneissic rock masses, except for situations of intense fracturing due to tectonization, and an aquifer behaviour of metasediments, particularly when interested by dissolution. Groundwater flow systems are mainly controlled by gravity. Springs located near Toce river are characterized by higher mineralization and isotopic ratios, indicating long groundwater flow paths. Starting from all the data collected and analyzed, two parametric methods are applied: 1) Dematteis method (Dematteis et al., 2000), slightly adapted to the case study and to the available data, that allows assessing both potential inflows inside the tunnel and potential impact on springs (codified as Drawdown Hazard Index); 2) Cesano method (Cesano et al., 2000) that allows only assessing potential inflows inside the tunnel, discriminating between major and minor inflows. Contemporarily a groundwater flow model is implemented with the EPM (Equivalent Porous Medium) approach, using MODFLOW-2000; it is calibrated in steady state conditions on the available data (groundwater levels inside wells/piezometers, elevation and flow rate of springs). Dematteis method proves to be more reliable and more adequate to the site than Cesano one; it was validated on a tunnel in gneissic rock masses and it takes more into account intrinsic parameters of rock masses than morphological and geomorphological factors. Cesano method relatively overestimates tunnel inflows, taking more into account the variations of topography and overburden above the tunnel. A sensitivity analyses evidences a low sensitivity of parametric methods to parameters values, except for RQD (Rock Quality Designation) used to represent fracturation degree. The numerical model is calibrated in ante-operam conditions and a sensitivity analysis evaluates the influence of uncertainties in hydraulic conductivity (K) values of the different hydrogeological units. Hydraulic head distribution after tunnel excavation is forecasted considering three different scenarios: tunnel only draining; tunnel as a losing source of water; tunnel sealed along its aquifer sectors, using 3 different levels of K reduction. Tunnel impermeabilization results very effective, lowering the drainage rate and the impact on springs. The model defines quantitatively the tunnel inflows and the effects on springs flow at the surface in terms of flow rate decrease. Dematteis method and the numerical model are crossed to obtain a final risk of impact on springs. The study is supposed to overestimate the risk, because all the values assigned to parameters are chosen in a conservative way and numerical simulations at steady state are very conservative too (transient state in such a hydrogeological setting is supposed to last 1-3 years). Monitoring of tunnel and springs during tunnel boring will allow the feedback process.
Content may be subject to copyright.
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
 The project of a hydro electrical diversion tunnel through
a crystalline rock massif in the Alps needed a detailed hydrogeologi-
cal study in order to forecast the magnitude of water inows inside
the tunnel and the possible effects on groundwater ow. The tunnel
has a length of 9.5 km and is located on the right side of Toce River
at Crevoladossola (Verbania province, Piedmont region, Northern
Italy). In the geological framework of the Alps, the tunnel is located
inside the Lower Penninic Nappes, in the footwall of the Simplon
Normal Fault; the geological succession is mostly represented by
Antigorio gneiss (meta-granites) and Baceno metasediments (meta-
carbonates). Due to the presence of important mineralized springs
used for commercial mineral water, the hydrogeological study fo-
cuses both on quantity and quality aspects, by means of: rainfall
data analyses, monitoring of major springs ow rates, monitoring
of hydraulic heads and pumping rates of existing wells/boreholes,
hydrochemical and isotopic analyses on springs and boreholes and
hydraulic tests (Lefranc and Lugeon). The resulting conceptual
model evidences a dominant low permeability (aquitard behaviour)
of gneissic rock masses, except for situations of intense fracturing
due to tectonization, and an aquifer behaviour of metasediments,
particularly when interested by dissolution. Groundwater ow sys-
Received: 7 october 2010 / Accepted: 18 november 2010
Published online: 31 december 2010
© Scribo 2010
Geotema S.r.l., Ferrara,
via Piangipane 141 int.5
44121 Ferrara
Tel. 0532.1862693
Fax 0532.1862767,
Dipartimento di Scienze della Terra
via G. Saragat 1
44121 Ferrara
Dipartimento di Scienze della Terra
via G. Saragat 1
44121 Ferrara
Enel Produzione S.p.A
via Torino 16
10132 Venezia-Mestre
 Il progetto di costruzione di una galleria di derivazio-
ne attraverso rocce cristalline nelle Alpi occidentali ha richiesto un
dettagliato studio idrogeologico nalizzato alla previsione delle ve-
nute d’acqua in galleria e dei possibili effetti sulla circolazione idri-
ca sotterranea. Il tunnel ha una lunghezza di 9,5 km e si trova lungo
il anco destro del Fiume Toce nei pressi di Crevoladossola (Pro-
vincia di Verbania, Regione Piemonte). Dal punto di vista geologico
regionale, il tracciato della galleria attraversa le Falde Pennidiche
Inferiori costituite da potenti falde gneissiche (meta-graniti; Veram-
pio, Antigorio, M. Leone) alternate a più sottili orizzonti metase-
 tunnel, groundwater inow, forecasting, spr ing,
Parametric and numerical modelling tools to forecast hydrogeological
impacts of a tunnel
tems are mainly cont rolled by gravity. Springs located near Toce
river are characterized by higher mineralization and isotopic ratios,
indicating long groundwater ow paths. Starting from all the data
collected and analyzed, two parametric methods are applied: 1) De-
matteis method (Dematteis et al., 2000), slightly adapted to the case
study and to the available data, that allows assessing both potential
inows inside the tunnel and potential impact on springs (codied as
Drawdown Hazard Index); 2) Cesano method (Cesano et al., 2000)
that allows only assessing potential in ows inside the tunnel, dis-
criminating between major and minor inows. Contemporarily a
groundwater ow model is implemented with the EPM (Equivalent
Porous Medium) approach, using MODFLOW-2000; it is calibrated
in steady state conditions on the available data (groundwater levels
inside wells/piezometers, elevation and ow rate of springs). Demat-
teis method proves to be more reliable and more adequate to the site
than Cesano one; it was validated on a tunnel in gneissic rock masses
and it takes more into account intrinsic parameters of rock masses
than morphological and geomorphological factors. Cesano method
relatively overestimates tunnel inows, taking more into account the
variations of topography and overburden above the t unnel. A sensi-
tivity analyses evidences a low sensitivity of parametric methods
to parameters values, except for RQD (Rock Quality Designation)
used to represent fracturation degree. The numerical model is cali-
brated in ante-operam conditions and a sensitivity analysis evaluates
the inuence of uncertainties in hydraulic conductivity (K) values
of the different hydrogeological units. Hydraulic head distribution
after t unnel excavation is forecasted considering three different
scenarios: tunnel only d raining; t unnel as a losing source of water;
tunnel sealed along its aquifer sectors, using 3 different levels of K
reduction. Tunnel impermeabilization results very effective, lower-
ing the drainage rate and the impact on springs. The model denes
quantitatively the tunnel inows and the effects on springs ow at
the surface in terms of ow rate decrease. Dematteis method and
the numerical model are crossed to obtain a nal risk of impact on
springs. The study is supposed to overestimate the r isk, because all
the values assigned to parameters are chosen in a conservative way
and numerical simulations at steady state are ver y conservative too
(transient state in such a hydrogeological setting is supposed to last
1-3 years). Monitoring of tunnel and springs during tunnel boring
will allow the feedback process.
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
In mountain regions, tunnel excavation can be the most threaten-
ing activity on groundwater and this has not ever been considered of
great impor tance in the past, as demonstrated by various Italian case
studies: the huge groundwater level drawdown and springs dewater-
ing as a consequence of the Gran Sasso highway tunnel boring (Pe-
titta & Tallini, 2002), but also in recent times the impact on streams
and springs due to the tun nels of the Bologna-Florence high-speed
railway line (Gargini et al., 2006, 2008). As a consequence, in the
last years people and authorities have become more and more sensi-
tive to the problem and the scientic attention has been drawn on the
possibility to forecast the effects of tunnels and underground exca-
vations concerning groundwater inows and water table drawdown
and to evaluate the risk for springs located in the hydrogeological
basin crossed by the underground opening.
The forecasting of the impacts of a draining tun nel on groundwa-
ter ow systems is one of the most challenging tasks in engineering
geology. Tunnel drainage can produce severe effects on hydrogeo -
logical systems, either at transient or at steady state conditions, such
as springs drying out, hydraulic head drawdown and wells yield
shortage, streams base ow depletion. The resulting environmen-
tal, socio-sanitary and economic damages, if forecasted at the early
stage of the project, should address risk mitigation measures (alter-
native tunnel pathways, tun nel sealing, even the project abandon-
ment) in the framework of a technically correct cost-benet analysis.
In hard rock aquifers it’s extremely difcult to forecast either ma-
jor inows occurrence along the tunnel or the associated effects on
surface waters and groundwater, due to the highly heterogeneous
distribution of hydraulic conductivity (K) and the consequent strong
dependence of major inows on the interception of localized geo-
logical features (i.e. fractured zones, faults, karst conduits). Their
geometry and locations can be achieved by means of a good 3D geo-
logical model (geological mapping, aerial-photos interpretation and
geophysical surveys) with an approximation of some tens or hun-
dreds of meters and, only in a few cases, it’s possible to determine
their hydraulic parameters by means of permeability tests performed
in deep boreholes.
Two different evaluation tools are addressed in order to overcome
these uncertainties and to get a good forecast: parametric methods
and mathematical models.
Parametric studies, also known as matrix methods, identify rel-
evant physical quantities (such as RQD, permeability, overburden,
faults intersecting the tunnel) and rank them and their interactions
by assigning ratings and multipliers to gain nal probability indexes
of impact on each spring. Parametric methods for an impact assess-
ment evaluation can be matrix-based or rating & weight based. In
the rst case a contingency matrix is developed, crossing together
relevant parameters: the classical example is the Leopold matrix for
dimentari (principalmente meta-carbonati; Teggiolo e Baceno) in
origine corrispondenti all’originaria copertura triassico-cretacica
del basamento cristallino. Questa struttura a sandwich ha assunto
l’attuale conformazione a duomo durante il sollevamento a letto del-
la faglia normale del Sempione. La presenza, nei pressi del traccia-
to, di importanti sorgenti mineralizzate utilizzate a scopo potabile e
commerciale ha imposto uno studio idrogeologico basato sia su dati
quantitativi che qualitativi (misure di precipitazione, monitoraggio
delle portate delle sorgenti, monitoraggio dei livelli piezometrici nei
sondaggi e risultati di prove di emungimento in pozzi e sondaggi
esistenti, prove idrauliche Lefranc e Lugeon nei sondaggi e monito-
raggi idrochimici e isotopici sulle sorgenti e sui sondaggi).
Dall’analisi del modello geologico e dei dati idrochimici è stato de-
nito il modello idrogeologico concettuale di riferimento che assegna
un comportamento acquitardo alle rocce gneissiche, ad eccezione
delle fasce tettonizzate e fratturate, ed un comportamento acquifero
ai metacarbonati in particolare dove interessati da fenomeni di dis-
soluzione. Le sorgenti localizzate sul fondovalle del ume Toce, ca-
ratterizzate da una maggiore mineralizzazione e maggiori rapporti
isotopici, costituiscono il recapito di lunghi percorsi sotterranei
all’interno di un sistema di usso idrico sotterraneo essenzialmente
controllato dalla gravità.
Sulla base del modello geologico e di tutti i dati raccolti sono state
effettuate previsioni d’impatto con due metodi parametrici e un mo-
dello numerico. Il primo metodo parametrico utilizzato (Dematteis
et al., 2000), leggermente modicato per adattarlo al caso in esame,
ha consentito di valutare sia le potenziali venute d’acqua in galleria
sia i possibili impatti sulle sorgenti (deniti come Drawdown Ha-
zard Index); il secondo metodo parametrico (Cesano et al., 2000)
ha consentito solamente la determinazione delle potenziali venute
d’acqua in galleria discriminando tra maggiori e minori. Paralle-
lamente è stato costruito un modello numerico alle differenze nite
(MODFLOW-2000) dell’intero massiccio montuoso assumendo un
comportamento poroso equivalente (EPM) e calibrato in condizioni
stazionarie sulla base dei livelli piezometrici nei pozzi e sondaggi e
sulla base della quota e portata delle sorgenti.
Il metodo Dematteis si è dimostrato più afdabile e adeguato al
caso specico rispetto al metodo Cesano; infatti, il primo, validato
su una galleria in rocce gneissiche, assegna maggiore importan-
za ai parametri intrinseci degli ammassi rocciosi piuttosto che ai
fattori morfologici e geomorfologici. Il metodo Cesano sovrastima,
relativamente agli altri metodi, le potenziali venute d’acqua in gal-
leria assegnando più importanza alle variazioni topograche ed
allo spessore della copertura rocciosa della galleria. È stata effet-
tuata un’analisi di sensitività sia per i metodi parametrici che per
il modello numerico; i primi hanno evidenziato variabilità legata
essenzialmente al valore di RQD (Rock Quality Designation, uti-
lizzato come rappresentativo delle condizioni di fratturazione), il
modello numerico è risultato inuenzato dall’incertezza sui valo-
ri della conducibilità idraulica degli ammassi rocciosi. Gli effetti
dello scavo sulla supercie piezometrica, preventivamente simulata
in condizioni ante-operam, sono stati simulati utilizzando, nel mo-
dello numerico, tre possibili condizioni: galleria drenante, galleria
disperdente ed inne galleria impermeabilizzata in corrispondenza
dei settori acquiferi (con assegnazione di crescenti livelli di ridu-
zione della permeabilità). Il trattamento dei settori acquiferi è ri-
sultato particolarmente efcace riducendo il drenaggio e gli impatti
sulle sorgenti; il modello è in grado di denire quantitativamente
le venute d’acqua in galleria e gli effetti sulle sorgenti in termini di
riduzione della portata. L’assegnazione del livello di rischio na-
le per ciascuna sorgente è stato denito incrociando i risultati del
modello numerico con quelli del metodo Dematteis; tali livelli di ri-
schio sono, tuttavia, da considerarsi sovrastimati in quanto i valori
assegnati a ciascun parametro sono stati scelti conservativamente
ed altrettanto conservativamente si possono considerare le condi-
zioni stazionarie generate dal modello numerico dato che, in questo
tipo di ammassi rocciosi, è probabile che le condizioni transitorie si
protraggano da 1 a 3 anni. Il monitoraggio delle venute d’acqua in
galleria durante il futuro scavo e il proseguimento dei monitoraggi
sulle sorgenti consentirà una verica dei modelli e fornirà impor-
tanti informazioni sperimentali.
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
EIA (Environmental Impact Assessment, Leopold, 1971). In the sec-
ond case a parameterization system of ratings and multiplier weights
is applied to variables considered as relevant, in order to obtain
nal quantication indexes: typical examples are given by DR AS-
TIC (Aller et al., 1985) and SINTACS R5 (Civita & De Maio, 2000)
methods, made to assess aquifers intrinsic vulnerability.
On the other hand, mathematical groundwater ow modelling
simulates the actual process causing the impact with a physically
based approach. Adopted simulation codes vary from simple ana-
lytical for mulas, assuming a Darcyan groundwater ow, simple and
stationary boundary conditions and homogenous or simple K dis-
tribution (Zhang & Franklin, 1963; Goodman, 1965; Bar ton, 1974;
Federico, 1984; Lei , 1999, 2000; El Tani, 1999), to complex numeri-
cal codes, more exible and adjustable to real world conditions but
much more exigent in terms of scientic expertise, nancial resourc-
es and input data.
In the here presented case study the occurrence of important min-
eralized springs used for commercial mineral water asked for a solid
risk analysis. The potential impacts on springs due to a draining tun-
nel were so evaluated applying the two different methods: the para-
metric probabilistic and the numerical deterministic. Both methods
started from a detailed hydrogeological study necessary to set up
the conceptual model and the nal drawdown risk of each spring
were obtained from the convergence of the two methods. This paper
presents the main results, together with the effectiveness of the two
approaches in performing such risk analysis.
Geological setting
The diversion tunnel layout crosses the deepest part of the Pennin-
ic Units traditionally assigned to the ancient northern margin over-
thrusted by the oceanic and the southern austroalpine margin during
the continental collision of the alpine orogenesis. The present-day
deep topographic erosion allows a glimpse onto the structure of the
Lepontine nappes, tectonic units formed by at recumbent folds
consisting of orthogneissic cores with discontinuous micaschistic
and paragneissic outer zones overlaying mesozoic metasediments
(Schmidt & Preiswerk, 1905; Castiglioni, 1958; Milnes et al., 1981;
Steck, 2008) (Fig.1). The latter are composed of calcschists, dolo-
mitic and calcitic marbles, quartzites and discontinuous gypsum and
anhydritic horizons. This petrographic and rheologic sandwich is
the result of many deformative phases, the most important of which
sliced and thrusted the former continental, mainly granitic, crust
onto the sedimentary cover with displacements of tens of kilome-
tres (Maxelon & Mancktelow, 2005). The amphibolitic metamor phic
conditions achieved dur ing the main deformation phase allowed the
mineralogical transformations that gave place to the pervasive axial
plane foliation forming the present-day schistosity and most promi-
nent weakness planes of the rock mass. Successive slightly or non
metamor phic folding events gave the nappe pile its present dominant
structure visible along the Antigorio valley (Fig. 2) with a at lay-
ing attitude in its central part, around the so called Verampio win-
dow, passing to a steep southward inclination near Crevoladossola
(Mancktelow, 1985; Grasemann & Manktelow, 1993). The N-S cross
 Geological sketch of the area between the Verampio window (VE) and the Simplon Fault (modied from Grasemann & Mancktelow, 1993). ANT:
Antigorio gneiss; ML: Monte Leone gneiss. The line represents the trace of the cross section of gures.2 and 3.
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
 Geological section along the tunnel; colors according to geological legend in Fig. 2.
 Geological map of the study area, with groundwater monitoring points.
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
section (Fig. 3) shows the geologic and hydrogeologic structure of
the mountain ridge and corresponds to the trace of the diversion tun-
nel, presented on the geological map of Figure 2. The main feature
is represented by the thick late Variscan (ca. 340 My) Antigorio unit,
composed of a monotonous leucocratic orthogneiss that only rarely
shows any structural or petrographic diversity (Bigioggero et al.,
1977). To the north it overlays the deepest alpine unit cor responding
to the as well late Variscan Verampio granitic gneiss and its host
rock represented by the garnet-rich Baceno micaschists. Interlayered
between this two big tectonic units is located the most important
permeable unit (Baceno metasediments), which stratigraphy has
been well dened after the drilling of three deep boreholes. It has
an average thickness around 70 meters and is composed by a lower
gypsum-anhydrite layer, at least t wo cohesionless sugar y carbonate
horizons, dolomitic marbles and calcschists. To the south the geolog-
ical structure is characterised by a late non-metamorphic kilometre-
scale fold linked to the exhumation of the nappe pile along the exten-
sional Simplon Fault (Campani et al., 2010). Near the southern portal
the tun nel trace crosses the second metasedimentar y unit (Teggiolo)
around 70 meters thick and with a composition very similar to the
Baceno metasediments. The tunnel trace ends in the third gneissic
continental unit (Valgrande gneiss) mainly composed of ne grained
quartz-micaschists and g rey biotitic gneiss.
Detailed geological mapping, boreholes and geophysics allowed
both to reconstruct a reliable 3D model and to recognise a huge deep
seated landslide (Deep Seated Gravitational Deformation, DSGD)
in the northern part of the section that might be genetically linked
to the underlying weak metasedimentary horizon. The whole trian-
gular shaped ridge is cross-cut by many brittle faults ar ranged in
two main sets NW-SE and E-W oriented; they are the product of
the brittle last exhumation phase of the Lepontine dome along the
Simplon Fault (Bistacchi & Massironi, 2000; Grosjean et al., 2004;
Zwingmann & Mancktelow, 2004).
Materials and methods
Hydrogeological characterization
The occurrence of important mineralized springs used for com-
mercial mineral water asked for an hydrogeological study focused
both on quantity and on quality of groundwater; the study has been
developed by applying different tools.
A rainfall data analysis, based on 8 meteorological stations, has
been performed in order to estimate the recharge rate to the aqui-
fers. The water sur plus on the modelled area has been calculated by
means of Thor nthwaite & Mather (1957) soil water balance.
A census of the main water points was available, together with one
year of hydrogeological monitoring in ante-operam conditions on
21 major springs and 5 wells in the area. Furthermore, 8 boreholes
bored for the project were available for hydrogeological monitoring
(Table 1). The available data (collected by CESI S.p.A) were: ow
rates at springs and g roundwater levels at wells and boreholes, mea-
sured on a monthly basis. The DVI (Discharge Variability Index)
has been calculated for all the springs as (Qmax-Qmin)/Qav, where
Qmax is the maximum ow rate, Qmin the minimum ow rate and
   
   
2 Bisogno Spring 890 0.02
3 Oira Spring 420 11.17 0.83 1.29E-02 Ca-HCO3-SO4
6 Viceno Spring 850 3.07 0.31 6.50E-04 Ca-Mg-SO4-HCO3
7 La Valle Spring 850 5.38 0.28 3.00E-04 Ca-HCO3
8 Vegno Spring 510 1.55 2.19 2.33E-02 Ca-Mg-SO4-HCO3
10 Flecchio alta Spring 1170 0.23 1.22 1.11E-02 Ca-HCO3
11 Flecchio Spring 1100 14.22 0.73 1.13E-02 Ca-HCO3
11b Flecchio Isolata Spring 1090 3.23 1.47 Ca-HCO3
12 Longio Spring 1140 16.04 0.49 Ca-HCO3
14 Trona Spring 1050 1.37 0.88 3.10E-03 Ca-HCO3
15 Alfenza Nord Spring 1420 12.46 1.05 Ca-HCO3
16 Alfenza sud Spring 874 2.31 0.91 2.95E-03 Ca-HCO3
17 Cavoraga-Faiò Spring 1300 0.08 1.38 9.40E-03 Ca-HCO3-SO4
20 Ronconi Spring 890 0.56 0.96 1.19E-02 Ca-HCO3
21 Cheggio Spring 1390 0.06 1.19 1.91E-02 Ca-HCO3-SO4
22 Cesa Inferiore Spring 545 0.20 0.74 2.10E-03 Ca-HCO3
23 Valle Oro Spring 457 19.97 0.11 6.60E-03 Ca-SO4
24 Lisiel Spring 470 34.96 0.06 4.00E-04 Ca-SO4-HCO3
25 La Conca Spring 470 0.18 1.00 4.55E-03 Ca-HCO3
26 La Rocca Spring 1100 15.00
27 Calantagine Spring 1420 35.00
 Main hydrogeological and hydrochemical data of springs.
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
Qav the average ow rate, relatively to the monitored year; the α
coefcient (recession coefcient according to Maillet, 1905) has
been also calculated for each spring, deriving it from the recession
curves of the available data.
Together with the ow and piezometric measurements, in-situ
physico-chemical parameters have been measured by means of a
multi-parameter probe (temperature, specic elect rical conductiv-
ity, pH, dissolved oxygen and dissolved carbon dioxide).
In the laboratory, ionic concentrations of uoride, chloride, sul-
phate, nitrate, sodium, potassium, magnesium and calcium together
with alkalinity, dissolved silica and xed residue, have been mea-
sured on groundwater samples collected at the monitoring network
on a monthly basis; concentrations of Iron (total), Strontium and
Lithium have been measured twice a year at 5 monitoring points
(11, 12, 22b, 23 and 24; Table 1); isotopic ratios of δ2H, δ18O (stan-
dard VSMOW as in Craig, 1961) and δ13C (standard PDB as in Urey
et al., 1951) have been measured on a monthly basis at the 4 most
impor tant monitoring points (11, 22, 23 and 24, chosen for their
high discharge rates as well as for their drinkable and com mercial
Main data concerning monitoring and classication of water
points (springs, boreholes and wells), coming both from the census
and from data interpretation, are su mmarized in Table 1.
Geological units have been classied with respect to the per-
meabilit y type (porous or fractu red media) and to the expected K
values, which have been derived from the combined use of geo-
logical and geomorphological information available in literature,
perfor med permeability tests (Lefranc and Lugeon tests) i nside 8
boreholes in the area and other tests performed 30 years ago for the
planning of Piedilago-Agaro plant, located 7 km to the north but in
the same geological units (Marti notti per ENEL Produzione, 1993).
Springs have been classi ed according to Civita (1973, 2005) and
the conjunctive analysis of all the data allowed to set up the hydro -
geological conceptual model.
Parametric methods
A parametric method applied to tunnel impacts forecasting is ba-
sically a risk assessment where the risk of a damage occurring to
groundwater as a result of the tunnel drainage is evaluated. Analyz-
ing the general formula (Einstein, 1988) in terms of tunnel impacts:
 (1)
is the risk, e.g. the occurrence probability of an impact against
groundwater ow systems;
is the hazard, e.g. the magnitudo of the impact process, e.g. a ma-
jor inow at the tunnel face;
is the hazard Probability that a major inow could occur along the
 is the receptors Value, that means the environmental, hydrologic
and socio/economic importance of springs, streams and wells (sur-
face emergence of groundwater ow systems) potentially subject to
is the vulnerability, intended as the expected damage in relation-
ship to major inow occurrence due to the hydrogeological connec-
tion between tunnel and receptors.
Two methods, validated by case histories in metamor phic rocks
similar to those of our study, have been chosen: DHI or Demat-
teis method (Dematteis et al., 2001) and Cesano method (Cesano
et al., 2000). These methods consider a multiple set of parameters
increasing the condence in the probabilistic evaluation respect to
single-parameters method (Thapa et al., 2005). They were partially
modied for the intended application, in order to overcome either the
lacking of data measured at the tunnel drilling face (the study was
done in ante-operam condition), available in the original application
of the methods, or some intrinsic limitations of the method in the
nal DHI output.
Whole tunnel trace (about 9000 m) has been divided in 50.1 m
length sectors, with a total of 183 sectors (progressively numbered
from northern to southern portal). The average elevation of each
sector was assigned at the middle, depending on the linear gradient
(9.05 ‰) between two main portals elevation, from 513.3 m (north)
to 505 m a.s.l. (south).
A total of 21 springs and 5 wells were considered as potential
impact receptors for DHI evaluation, according the hydrogeologi-
cal conceptual model. Hydrogeological st ructure, groundwater ow
circuits depth and springs hydrochemistry contributed to dene the
degree of hydrogeological connection between the receptors and the
tunnel sectors.
Application of DHI method
DHI method is a typical rating and weight parametric method for
the evaluation of Risk, expressed by an index called “DHI” (Draw-
down Hazard Index) ranking the probability of spring ow depletion
in relationship to inow probability occurrence and spring vulner-
ability. DHI method is a fully-coupled-model (Jiao & Hudson, 1995)
and takes into account the physical relationships of different var i-
ables. The method is applied through two steps: rstly, variables af-
fecting the probability of PI (Potential Inow) generation inside tun-
nel sectors are rated and weighted; secondly, the obtained PI value
is fur ther weighted, in relationship to local vulnerability, in order to
obtain the nal DHI value.
The original formulas from Dematteis et al. (2001) are:
   
 
1 Sarizzo Well 386 30
4 Molinetto1 Well 424 101
4a Molinetto2 Well 424 54
4b Molinetto3 Well 424 71
22b Cesa Well 573 33
S1 Borehole 508 11.3
S2 Borehole 529 35.5
S3 Borehole 883 376
S4 Borehole 792 304
S5 Borehole 308 30
S6 Borehole 518 15
S7 Borehole 517 15
S8 Borehole 811 412
 : Main hydrogeological and hydrochemical data of wells and
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
where  = rock-mass fracture frequency;  = rock mass hy-
draulic conductivity;  = thick ness of the Plastic Zone around the
bored tunnel;  = tunnel Overburden, including Quaternar y de-
posits;  = occurrence of hydrogeologic connection between spring
and tunnel;  = spring type, related to the depth of groundwater
ow system discharging at the spring;  = geometric distance be-
tween the t unnel sector and the spring.
DHI is ranked in four piezometric drawdown risk classes with a
relative probabilistic output, as shown in Table 2.
With respect to the original formulation (see Dematteis et al.,
2001) the following changes have been introduced.
FF variable: the RQD index (Rock Quality Designation) (Deere et
al., 1969; Deere & Deere, 1989), derived from boreholes S1-S2-S3-S4
and S8, was considered representative of FF, because geomechanical
surveys at the drilling face were not available. An average arithmetic
mean of RQD values has been calculated for each geological unit;
the obtained results have been integrated with data coming from the
geomechanical surveys performed on rock mass outcrops (Astol
& Sapigni, 1999). As a further renement, fractures occurrence at
megascale was derived from photo-aerial lineament traces intersect-
ing the tunnel trace and it has been taken into account as follows:
if a lineament occurs inside a 250 m radius cylindrical buffer zone
coaxial with the tunnel, FF rating for tunnel sectors involved along
±100 m linear distance from the lineament is lowered down by one
or two rating classes, respectively if 1 or more than 1 lineaments are
K: permeability values were assigned to the different rock masses
taking into account both Lugeon tests results and reference data,
available in literature, corresponding to metamor phites in analogous
geological framework (Loew, 2002). Moreover, K values have been
increased of 1 order of magnitude for the tunnel sectors crossed by
a fault or main fracture (with a ±50 m buffer zone) derived by the
longitudinal geologic prole and for sectors crossing the contact be-
tween Antigorio Gneiss and Teggiolo Syncline, due to its important
and recognized aquifer behaviour.
OV: it was derived f rom tunnel longitudinal section, measured at
the midpoint of the corresponding sector and it ranges between 28
and 801 m.
PZ: according to a conservative approach, PZ value was always
considered equal to three times the maximum calculated EDZ (Ex-
cavation Distu rbed Zone) reaching the conser vative value of 14.7 m.
IF: a 100 m and 50 m buffer zone has been dened for each side of,
respectively, primary or secondary structural lineaments (obtained
by eld geological survey or aerial photos interpretation). If a spring
is located inside the buffer zone and, at the same time, the lineament
crosses the tunnel, the spring IF value is 1 (otherwise it is 0). Springs
discharging out from quaternary deposits always received a 0 rating,
because these deposits are never involved by tunnel boring.
ST: three main groundwater ow systems have been identied ac-
cording to springs classication: shallow, deep and mixed, receiving
respectively a 0, 1 and 0.5 rating.
DT: Euclidean distance (e.g. minimum geometrical distance) be-
tween the midpoint of each tunnel sector and the receptors has been
calculated independently of the relative elevation of the receptor re-
spect to the tunnel. DT range is between 8907 and 244 m.
Application of Cesano method
Cesano method actually is not a codied parametric method but a
study showing the results of a multi-regression analysis between tun-
nel potential inows and affecting variables in an analogous hydro-
geological framework (gneissic fractured aquifer below a variable
thickness of Quaternary cover) for a 80 km long tunnel in southern
Sweden (Cesano et al., 2000). It was chosen as a further evalua-
tion tool, proposing a different set of PI factors, in order to better
verify the results. Cesano method does not evaluate DHI but only
PI, discriminating major inows from diffuse dripping. Four major
inows factors, among many others, are recognized and listed here
in order of impor tance: TSW = tunnel proximity to a main surface
water body (i.e. streams, lake, reservoir); BM = bedrock morphol-
ogy underlying Quaternary cover; T = topography; BFF = bedrock
aquifer fracture density. Four diffuse dripping factors, among many
others, are recognized and listed here in order of importance: BM =
bedrock morphology underlying Quaternary cover; QC = thickness
and lithology of Quaternary cover; QA = area of Quaternary out-
crops above the tunnel; PV = peaks and valleys occurrence of BM.
The main change introduced to Cesano method was to transform
the multi-factorial correlation analysis of the original method into a
rating and weighting parametric method, by sum ming the correla-
tion coefcients and normalizing the result to 1. In such a way a
weighted linear combinations of rates allowed a coherent compari-
son with Dematteis method. In Table 3 the weights derived for the
Cesano method are presented.
 
< 0.2 Absence of (or minimal) Drawdown
0.2 ÷ 0.6 Moderate Drawdown
0.6 ÷ 0.7 From moderate to severe Drawdown
> 0.7 Spring drying-up
 DHI classes according Dematteis method.
a) 
 b) 
TSW 0.32 BM 0.30
BM 0.29 QC 0.29
T0.28 QA 0.26
BFF 0.11 PV 0.11
 a) Weights for major inows factors; b) Weights for diffuse dripping
factors (Cesano method).
Relevant variables were parameterized as follows:
TSW: it is expressed by the Euclidean distance between the mid-
point of each tunnel sector and the nearest main stream bed located
above the tunnel. TSW values range between 9930 and 22 m; whole
range was divided in 5 classes of equal amplitude.
BM and T: this two variables have been der ived from Digital Ele-
vation Model compared with tunnel longitudinal geological section,
considering 200 m interval tunnel sectors in order to better represent
the top of the bedrock and topographic surface elevation changes.
Maximum rating (1) is applied to bedrock troughs (areas of potential
concentrated recharge), minimum rating (0) to mounds; a 0.5 rating
was assigned for intermediate BM values. T value range between
505 and 1299 m a.s.l.
BFF: primary and secondary megascale tectonic lineaments have
been taken into account as for FF Dematteis variable. The total num-
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
ber of lineaments crossing the tunnel (divided in 100 m long tunnel
sectors) has been calcu lated: ratings of 1, 0.7, 0.35 and 0 were
assigned to the tunnel if the number of lineaments is, respectively,
≥ 3, 2, 1, 0.
QC: it was derived from tunnel longitudinal geological section.
Maximum thickness value is 260 m, corresponding to the DSGD
mass. Whole QC range has been rated according 5 different classes
of equal amplitude.
QA: the outcropping areal extension (m2) of QC inside a 100 m
long search cylindrical volume coaxial with the tunnel and with
1000 m radius has been calculated, as derived from geomorphologi-
cal map and DEM. Values range from a maximum of 99800 m2 (tun-
nel sector completely covered by Quaternar y deposits; maximum
rating of 1) to a minimum of 0 m2 (0 rating).
PV: peaks and valleys occurrence in the bed rock have been iden-
tied from tunnel longitudinal geological section, according to a
discretization of 100 m long tunnel sectors. A rate of 0 and 1 was as-
signed, respectively, to peaks and valleys; a rate of 0.5 was assigned
to all the other sectors.
Numerical model
Analytical codes, generally useful for limited tunnel sector and
time-scale frames (Loew, 2002), were excluded because too much
simplistic for the complex hydrogeologic setting. A three-dimen-
sional groundwater ow model has been developed by means of the
MODFLOW-2000 code (McDonald & Harbaugh, 1988; Harbaugh et
al., 2000), that solves the ow equation in saturated media according
to the nite difference method.
Assuming that, along the 9.2 km long diversion tunnel, groundwa-
ter ow system at the whole mountain ridge scale follows the Darcy’s
law, the Equivalent Porous Medium (EPM) approach has been used.
It consists in considering the rock matrix together with the fractures
and assigning them bulk hydrodinamic properties, over a rock vol-
ume sufciently wide to be considered statistically representative
(Representative Elementary Volume - REV; Long et al., 1982; Kanit
et al., 2003). Inside the REV it is assumed that fracture distribu-
tion is casual and uniform and that fracture width does not allow
turbulence ow. Geometric and hydrodynamic properties of distinct
fract ures are not requested, small computational efforts are neces-
sary and good results can be obtained working on wide modelling
areas (Mun & Ucrhin, 2004).
A rectangular shaped model domain of 12000 x 17000 m has been
set up, oriented parallel to Gauss Boaga coordinate system and ex-
tending from Torrente Diveria at north to Crevoladossola at south
(Fig. 2). On the horizontal plane it is subdivided into cells of 100 x
100 m, except for the zone including the tunnel and the main springs
in which a grid renement leads to 25 x 25 m cells (Fig. 4). A great
effort has been taken to get a good vertical discretization and 20
variable thickness layers have been used; layer 11 has been used
to represent the tunnel plane and has a thickness that includes the
plastic zone. The domain extends vertically from the topographic
surface, derived from DEM, to an almost horizontal plane with el-
evation of 510 m a.s.l., with a very gentle gradient parallel to the
tunnel slope. The total thick ness of the model varies between 350
and 3000 meters. A section of the domain parallel to the tunnel trace
is shown in Figure 5.
Hydraulic conductivity (K) is always assigned as an isotropic
 Flow domain and K zones at layer 1 of the MODFLOW model. Colors
corresponding to K zones according to table 3; the inactive cells are repre-
sented with the green color; coordinate axis in meters; tunnel trace in black.
Section of the model domain along the tunnel trace (N-S direction), column 85: vertical discretization in 20 layers and assigned K zones (according
to Table 3); vertical exaggeration of 2:1 is used.
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
property, except for the normal faults/fracture zones, where an an-
isotropy factor of 10 resulted necessary along x and z axis during the
calibration process. In Table 4 the nal K values, obtained after cali-
bration, assigned to the 15 K zones are summarized: every geologi-
cal unit is represented by two K values, the rst one for the normally
fract ured rock mass, the second one for the rock mass affected by a
fault/lineament. The K zones distribution can be observed in Figure
4 and Figure 5.
Recharge to aquifer is simulated by a 2nd type boundar y condition
(b.c.) applied to ever y cell of 1st sat urated layer; the low sensitivity
of the model to the recharge allowed to use a uniform value of 270
mm/yr. The regional gradient is represented by a 1st type b.c. on the
northern side of the domain, with a maximum head value below the
mountain ridge (1800 m a.s.l.), decreasing to 1045 m a.s.l. eastward
and to 1250 m a.s.l. westward. The rivers bordering the model on the
western and easter n side, and representing the discharge point of re-
gional groundwater ow system, are represented with the same b.c.:
head values have been assigned equals to riverbed elevations derived
from the DEM. On the eastern side head varies from 1045 to 520 m
a.s.l. along Devero River and from 520 to 300 m a.s.l. along Toce
River; on the wester n side it varies from 1250 to 530 m a.s.l. along
Cairasca River and from 530 to 300 m a.s.l. along Diveria River.
Cells outside the boundary conditions are switched off, so that the
nal shape of the domain is ir regular. The 4 main springs have been
represented by means of the Drain Package (3rd type b.c.), assigning
to the cell a drain elevation equal to ground level at the spring. Con-
ductance values, calibrated on the real ow rate at the springs, are:
139.95 m2/d for Lisiel spring, 55.45 m2/d for Valle Oro spring, 1.11
m2/d for Cesa spring and 7.65 m2/d for Vegno spring.
The calibration process has been performed at steady state condi-
tions on a simulation representing the present situation and solved by
means of the Link-algebraic MultiGrid solver package (Mehl & Hill,
2001). The trial & error process (Zheng & Bennevy, 1995) has main-
ly involved the K values and the uncertain b.c., i.e. the constant head
at the northern boundary. All the 17 piezometric control points have
been used, for which at least one piezometric measure was available;
in Table 5 these data are presented. Flow rates of springs have been
used to calibrate conductance values.
The forecasting simulations have been performed according to
two different behaviors of the tunnel: in the rst one, representa-
tive of the drilling phase, the tunnel is assumed completely draining,
while in the second, representative of the post-operam conditions,
the tunnel is supposed to be dispersant.
The draining tunnel is simulated by means of the Drain Package,
that removes groundwater from the cells in which it is applied as a
function of heads differences (between the aquifer and the t unnel
Zone Color Hydrogeological Unit Kx Ky Kz
1 Antigorio gneiss 1.00E-07 1.00E-07 1.00E-07
2 Quaternary uvial deposits 5.00E-05 5.00E-05 5.00E-05
3 DGPV 7.00E-07 7.00E-07 1.00E-07
4 Metasediments (Baceno syncline) 5.00E-06 5.00E-06 5.00E-06
5 Baceno micaschists 1.00E-07 1.00E-07 1.00E-07
6 Verampio gneiss 1.00E-08 1.00E-08 1.00E-08
7 Metasediments (Teggiolo syncline) 5.00E-06 5.00E-06 5.00E-06
8 Valgrande gneiss 1.00E-07 1.00E-07 1.00E-07
9 Monte Leone gneiss 1.00E-07 1.00E-07 1.00E-07
10 Fault in Antigorio gneiss 7.00E-06 7.00E-06 7.00E-06
11 Fault in Metasediments (Baceno syncline) 1.00E-05 1.00E-05 1.00E-05
12 Fault in Baceno micaschists 5.00E-07 5.00E-07 5.00E-07
13 Fault in DGPV 1.00E-06 1.00E-06 1.00E-06
14 Quaternary glacial deposits 5.00E-06 5.00E-06 1.00E-06
15 Slope debris 5.00E-04 5.00E-04 5.00E-04
   
 
      
 139.95 34.96 3020.5 34.96 3020.2 -0.34 -0.01
 55.45 19.97 1725.4 19.97 1725.7 0.29 0.02
 7.65 1.55 133.9 1.5 130.0 -3.96 -2.96
 1.11 0.2 17.3 0.2 17.2 -0.03 -0.19
 Hydraulic conductivity assigned to different K zones (in m/s).
 Calibration data of the ante-operam simulation.
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
elevation) and of the conductance parameter around the tunnel. The
drain elevation is equal to the tunnel elevation, while the Conduc-
tance (C) values (parameter that represent the resistance opposed to
ow by the rock mass all around the tunnel; Zaadnoordijk, 2009),
were calculated as:
 (4)
where is the hydraulic conductivity of the Plastic Zone, is
the tunnel radius and is the plastic zone thickness.
So, different C values have been assigned to the tunnel reaches
crossing different K zones of the rock mass. A sensitivity analysis
tested 3 different values of KPZ, assuming an increase of 5, 10 and
100 times the original rock mass K.
The simulation with the dispersant tunnel takes into account the
inside ow rate of 18 m3/s and calculates the water exchange with the
rock mass. The dispersant tunnel is represented by the Streamow-
Routing Package STR1 (Prudic, 1989), a 3rd type b.c. usually applied
to simulate river-groundwater interactions, according which the
stream ow rate is propagated starting from the value of the most up-
stream cell (starting point) and calculated for every cell downstream
as the previous ow rate plus or minus the stream ow rate gained
from or lost to the aquifer. The in/out ow is calculated multiplying
the head difference between the stream and the aquifer times the
riverbed conductance. The water level inside the tunnel is 4 meters
above the t unnel bottom, while the conductance is calculated as for
the draining tunnel simulation.
Finally, other simulations have been set up in order to evaluate the
effects of concrete linings on the most critical sectors of the t unnel,
taking into account three different K values for the plastic zone of
the lining sectors: KPZ equal to the K of the drilled rock mass (A),
KPZ = 0.1*K of the drilled rock mass (B), KPZ = 1E-07 m/s (C).
In the sectors without linings the KPZ is 10 times the K of the sur-
rounding rock mass.
Hydrogeological characterization and resulting concep-
tual model
The study area has a temperate continental climate, characterized
by a mean annual temperature of 11.3° C in the city of Domodos-
sola (Federici et al., 1967) and by mean an nual rainfall in the range
of 1250-2600 mm/year on the modelled area (variations are mainly
related to topographic elevations).
Two rainfall maxima occurs every year: the rst one in May (with
values higher than 152 mm) and the second in October/November
(more than 110 mm).
The water surplus on the modelled area has been calculated in the
range of 700-1400 mm/year.
The resulting concept ual model evidences a groundwater ow
mainly controlled by gravity and a general aquitard behaviour of
rock masses (mainly gneissic formations), except for situations of
tectonization, fracturation or dissolution: the main aquifers are rep-
resented by metalimestones.
The spring survey puts in evidence an higher density of springs
with signicant ow rates in the Crodo area and in the Alfenza val-
ley (Fig. 6a).
Springs present ow rates in a wide range: from 0.06 l/s (ID 21) to
35 l/s (ID 24 - Lisiel); the most represented ow rates are in the range
0.05-2.5 l/s. A clear correlation between spring ow rate and spring
elevation does not occur also if three main groups can be spatially
distinguished: springs located at elevations between 400 and 500
m a.s.l., with ow rates higher than 10 l/s (ID3 - Oira, ID23 - Valle
 Location of groundwater monitoring network: a) springs ranked as
a function of the average ow rate in the period Jul-05 – Jun-06; b) springs
and wells ranked as a function of the average EC in the same period.
Oro, ID24 - Lisiel); springs between 1100 and 1500 m a.s.l., with
ow rates in the range between 5 and 10 l/s (ID10 - Flecchio, ID12 -
Longio, ID15 - Alfenza nord); a third group with all the remainders,
located at variable elevations, with ow rates lower than 5 l/s.
Comparing all the available data, it can be assessed that the
springs with the higher ow rates have generally a low DVI, par-
ticularly Valle Oro spring which shows a quite constant ow rate,
independent from the rainfall events, and a low value of α coef-
cient (6.6 10-3 d-1); it is representative of the end discharge of deep
groundwater ow systems and recharged over a huge area.
As in nearby areas (Martinotti, 1993), three hydrochemical water
types occur in the study area (Fig. 7a): Ca-HCO3 facies, representa-
tive of relatively short and shallow ow pathlines circulating inside
gneiss rock masses or quaternary deposits, as for example the case
of Alfenza springs; Ca-SO4 facies, representative of deep groundwa-
ter ow pathlines inside gypsum and/or anhydrites contained inside
metalimestones (as demonstrated by the enrichment in sulphates),
i.e. Valle Oro spring and Molinetto well; a Ca-HCO3-SO4 facies, that
comes from the mixing of the other two.
This classication is conrmed by the groundwater mineraliza-
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
 Piper diagrams: a) all monitoring points in the study area; b) ve groundwater samples collected inside borehole S3 compared with the four main
monitored springs.
 Local meteoric water line derived from all the available data on the study area.
tion degree, that can be inferred from the specic electrical conduc-
tivity (EC), represented in Figure 6b. A further coherent information
derives from the analyses of groundwater samples collected inside
boreholes at different depths: as represented in Figure 7b the two
deepest samples collected inside S3 borehole (respectively CO4 at
300 and CO5 at 370 m b.g.l.) have the same facies of Lisiel spring.
Isotopic composition of groundwater reects quite well the litera-
ture data (Martinotti, 1993; Martinotti et al., 1999; Pastorelli et al.,
2001): the local meteoric line (Fig. 8a) is very similar to the global
meteoric water line of Craig (1961). No signicant ther mal process
occurs because no shifts in δ18O‰ occur. Data fall inside a small
δ18O‰ range, but the distribution is signicant: Cesa well corre-
sponds to the shallower and shorter ow path lines and to the lower
elevation of recharge areas, so characterized by the minor isotopic
deplenishment; Flecchio spring is similar to Cesa well but located
upstream and recharged at higher elevations; Valle Oro spring is
a b
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
  Potential inows inside the tunnel according to Dematteis method
and modied Drawdown Hazard Index (DHIM) at the water point (springs
and wells).
characterized by the most deplenished waters and the highest differ-
ence between theoretical recharge area average elevation and spr ing
elevation (Fig. 8b), as it represents the discharge of deep and long
ow path lines.
Risk evaluation results from parametric methods
The nal result in terms of Potential Inow (PI), expressed as
probability between 0 and 100%, for the entire tunnel is presented in
Figure 9. Most critical sectors are those bet ween 6+062 and 6+864
km, located below DSGD, where PI value (between 40-60% and 60-
80%) is mainly controlled by FF (RQD), with a rating of 0.75. Other
critical sectors are 177 and 178 (km 8+867 – 8+917) where the high
rating of RQD index, OV (around 120 m) and K contribute to in-
crease the PI value. The remainder of the tunnel shows PI values in
the range 0 - 40 %.
The nal results according to Cesano method in terms of major
and minor inows are shown respectively in Figure 10 and Figure 11.
The most critical sectors for major inows, with values between 80
and 100%, are related to lineaments occurrence and to the location
of ground surface and bedrock troughs. Concerning dripping and
minor inows, the most critical sectors are located in the norther n
half of the tunnel where in some sector the maximum value of 100%
probability is attained.
Observing Figure 9 and Figure 10 and tr ying to compare the re-
sults of the two methods, it’s evident that PI value distribution along
the tunnel is quite diversied. On average lower PI values are pro-
duced by Dematteis method. Nevertheless the authors consider more
reliable these results, for two reasons: it is more codied than Cesano
method in the involved geological framework and it is strongly based
on rock mass intrinsic parameters. Cesano method for major inows
 Relation between spring elevation and δ18O‰.
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
Potential minor inows inside the tunnel according to Cesano method.
  Potential major inows inside the tunnel according to Cesano
prediction mainly takes into account morphological and geological
factors, which cannot play a key role with such a thick overburden as
occurring in the investigated site.
Frequency histogram for Dematteis PI (Fig. 12) is typically log-
normal, as should be if we consider the output of a combination of
hydrogeological factors; modal class corresponds to a very low PI
value (15%) as usually observed for actual drainage dist ribution oc-
currence in bored tunnels (Masset & Loew, 2010). In hard rock aqui-
fers big inr ushes are sporadic and scattered drainage events are rare
with respect to diffuse groundwater drops falling down from the
excavated tunnel surface (Gargini et al., 2008). Also for this reason
Dematteis output appears more reliable than Cesano method output.
DHI value for all springs (and for the 5 wells, considered as water-
point potentially subject to impact), according the application of De-
matteis method, is reported in Table 7 as DHID; it is important to
emphasize that DHID never results as “0”, being conditioned by PI
value, always higher than 0.23. So it never happens that there is no
DHI probability occurrence. Springs with higher DHI value are ID
23 (Valle Oro; 0.91 DHI) and ID 8 (Vegno; 0.68 DHI), both located
near a main fault crossing the tunnel and connected to deep ground-
water ow systems.
If DHID values are ran ked in classes based on the limits proposed
originally by Dematteis method (Table 1), then only 1 spring oc-
curs in classes 1 and 2 (respectively Valle Oro and Vegno), whereas
the remainders 19 springs and 5 wells fall in the “partial risk” class
(DHID >0.23). Among these, 10 springs (and 1 well, Sarizzo Well)
resulted with a minimum DHID value equal to 0.23; as already men-
tioned a value of 0.23 derives simply from the fact that PI is always
positive. This consideration shows a limit of the methodology: also a
spring located quite far from the tunnel path, connected to a shallow
groundwater ow system and not related to any tectonic lineament,
e.g. a receptor with very low vulnerability (if any), will receive how-
ever a not fair DHID value, being quite important the control of PI.
If PI value is high, accordingly DHID value will be at least medium
for all springs involved in the evaluation. For this reason Dematteis
method tends to overestimate impact risk for springs with low hy-
drogeological vulnerability to impact; the DHID estimation is more
reliable for medium-high vulnerability receptors.
 Total forecasted tunnel outow (bars, left scale) and specic outow
normalized on distance (line, right scale) in relationship to lithology (DA =
Quaternary uvial deposits, MsB = Baceno Micascists, FMB = Fault zones
in Baceno Micascists, MB = Baceno Metasediments, FMB = Fault zones in
Baceno Metasediments, A = Antigorio Gneiss, FA = Fault zones in Antigo-
rio Gneiss, MT = Teggiolo Metasediments, GVML = Valgrande and Monte
Leone Gneiss, DM = Glacial deposits, TOT = Total).
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
   
1 SARIZZO WELL 0.23 3 0 5
2 BISOGNO 0.23 3 0.11 5
10 FLECCHIO ALTA 0.23 3 0 5
11 FLECCHIO 0.23 3 0.14 5
11b FLECCHIO ISOLATA 0.23 3 0.13 5
15 ALFENZA NORD 0.23 3 0 5
17 CAVORAGA - FAIO' 0.23 3 0.198 5
21 CHEGGIO 0.23 3 0 5
25 LA CONCA 0.23 3 0.11 5
26 LA ROCCA 0.23 3 0 5
27 CALANTAGINE 0.23 3 0 5
24 LISIEL 0.34 3 0 5
3 OIRA 0.46 3 0.07 5
4 MOLINETTO 1 0.46 3 0.09 5
4a MOLINETTO 2 0.46 3 0.13 5
4b MOLINETTO 3 0.46 3 0.14 5
6 VICENO 0.46 3 0.31 4
7 LA VALLE 0.46 3 0.08 5
12 LONGIO 0.46 3 0.16 5
14 TRONA 0.46 3 0.16 5
16 ALFENZA SUD 0.46 3 0.22 4
22 CESA INFERIORE 0.46 3 0.37 4
22b CESA WELL 0.46 3 0.18 5
8 VEGNO 0.68 2 0.75 2
23 VALLE ORO 0.91 1 1 1
20 RONCONI 0.55 3 0.6 3→5
 DHI values for all springs according the application of Dematteis method;
1= highest, 5 = lowest impact probability.
Taking into account these limitations and considering other im-
provements of hydrogeological nature based on previous experienc-
es derived from tunnel drilled in hard rock aquifers, the nal index
of drawdown has been slightly modied by dening a new index
called “modied DHI (DHIM; Table 6). DHIM was evaluated in
the same manner of DHID with the following r ules:
1) If the tunnel-spring Euclidean distance is more than 1000 m and
there is no evidence of hydrogeological connection with the tunnel,
DHID is assumed equal to 0, notwithstanding the PI value; this con-
sideration is based on the results of the hydrogeological monitoring
of the springs and wells along the pathway of Florence-Bologna high
speed railway tunnel connection (Canuti et al., 2009);
2) If the tunnel-spring Euclidean distance is lower than 1000 m
and there is no evidence of hydrogeological connection with the tun-
nel, DHID is assumed to be based on the average PI relative to a
tunnel sector included in a buffer sphere with radius 1000 m and
centred at the spring;
3) If the tunnel-spring Euclidean distance is lower than 1000 m
and there is evidence of hydrogeological connection with the tunnel,
two separate buffer zones are considered: a sphere of 1000 m radius
centred at the spring and a 400 m long tunnel sector across the in-
tersection fault-tunnel; the highest value of PI between these buffer
zones is chosen, conservatively, to calculate DHI at the spring;
4) If the tunnel-spring Euclidean distance is higher than 1000 m
and there is evidence of hydrogeological connection with the tunnel
(with a total distance of connection along faults minor than 2000 m),
a 400 m long t unnel sector buffer zone across the intersection fault-
tunnel is dened for the evaluation of the PI;
5) DHI value so obtained is normalized to 1 and all the entire
value range is ranked according 5 classes regularly spaced between
0 and 1. The normalized DHI value is called modied DHI (DHIM).
The nal result in terms of DHIM for the whole set of receptors
is shown in the map of Fig. 9. The comparison between DHID and
DHIM class ranking is put in evidence in Table 7.
Tunnel inflows and springs dewatering forecasted with
the numerical model
The numerical simulations with the tunnel completely draining
forecast piezometric drawdowns in the range of 200-800 m, depend-
ing on the K values assigned to the plastic zone around the tunnel.
For all the simulations the highest drawdown occurs above the tun-
nel at the big curve close to the northern entrance (around km 1)
and decreases progressively from north to south in the longitudinal
direction of the tunnel; the drawdown stops where the tunnel eleva-
tion is higher than the piezometric surface in natural conditions. In
the transversal direction (W-E) the depressurization effect progres-
sively decreases moving away from the tunnel axis and disappears at
a maximum distance of about 5 km on the west side.
In all the three simulations the piezometric drawdown causes the
drying up of two springs, located in the area of the maximum impact
(Cesa and Vegno). At the other two springs (Lisiel and Valle Oro), lo-
cated in the same area but at lower elevations than the previous ones,
the model forecasts a ow rate decrease in the 4-6% range (Tab. 8).
In the three simulations the total tun nel drainage in steady-state
conditions is respectively 273, 286 and 316 l/s.
The highest contribution comes from the Baceno metasediments,
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
   
 34.96 19.97 1.55 0.2 Q (l/s)
 34.96 19.97 1.5 0.2 Q (l/s)
 33.56 18.92 0 0 Q (l/s)
4.00 5.25 100 100 decit %
 33.47 18.9 0 0 Q (l/s)
4.25 5.39 100 100 decit %
 33.37 18.82 0 0 Q (l/s)
4.53 5.77 100 100 decit %
     decit %
 Impacts on springs forecasted by means of MODFLOW simulations with the draining tunnel and with the 3 dif-
ferent K values for the plastic zones.
which produce ow rates in the range of 100-115 l/s (that is in the
range 31.6-42% of the total drainage). Progressively minor contri-
butions come from the remainder hydrogeological units and related
fault zones: Baceno micaschists, fault zones in Baceno metasedi-
ments, fault zones in Antigorio Gneiss, Antigorio Gneiss and fault
zones in Baceno micaschists (Tab. 9).
Considering the inow normalised by meters of t unnel, the sec-
tors crossing the faults in Baceno metasediments present the high-
est values, from 0.833 to 1.215 l/s*m; these values progressively de-
crease in the following order: Baceno metasediments, fault zones in
Baceno micaschists, Baceno micaschists, fault zones in Antigorio
Gneiss and Antigorio Gneiss. These results mainly derive from the
assigned hydraulic conductivities values but also from the geometric
position of the units with respect to the piezometric surface and the
tunnel alignment.
In the three simulations of the dispersant tunnel (with an inner
ow rate of 18 m3/s), the tunnel still has a draining behavior where
located below the piezometric surface, while it recharges the aquifer
where located above the piezometric surface.
This leads to a small decrease in terms of depressurization effects
in the area of the highest impact; at spr ings Lisiel and Valle Oro
ow rates increase of 0.01 and 0.28 l/s with respect to the simula-
tions with the draining tunnel, while at springs Vegno and Cesa the
impact is the same (Tab. 10).
Finally, the last three simulations, with impermeable linings in
the most permeable tunnel reaches, demonstrate that linings effec-
tively reduce the depressurization in terms of intensity and area of
inuence. Along the tunnel maximum drawdowns decrease of more
than a half with respect to the corresponding values calculated in
the rst simulation with the tunnel completely draining (~350 m of
drawdown), resulting respectively in: 153 m for the A, 140 m for the
B, and 138 m for the C scenario. The ow rates at springs Lisiel and
Valle Oro decrease respectively of 2% and 4.5% respect to the nat u-
ral conditions values. Cesa spring remains completely impacted in
all the scenarios, while Vegno spring gets completely dried up only
in the A scenario, while in the B and C scenar io presents a ow rate
decrease respectively of 82% and 64% with respect to the natural
conditions values (Tab. 11). The total tunnel drainage decreases sig-
nicantly too: 245 l/s (A), 176 l/s (B) and 161 l/s (C).
The general effect of tunnel sealing results quite good, even if it
increases the drainage from the tunnel reaches without linings, due
to the higher hydraulic gradient (Tab. 12 and Fig. 12).
Comparison between the two methods
Before to comment upon a comparison between parametric and
numerical modelling approach it should be noted that both evalua-
tions of tunnel impact are based on a detailed build-up of the geo-
logical model and on a dedicated geological survey. Such a site spe-
cic direct survey is the essential prerequisite for a reliable impact
evaluation, whatever is the chosen approach.
In every approach the conservativity principle was always fol-
lowed, e.g. sizing the thickness of Plastic Zone or modelling at
steady state conditions.
On the other side, the reliability of forecasted scenarios is limited
by the calibration data-set (hydraulic head and springs monitoring
and pumping tests data).
In Table 12 the nal results in terms of tunnel impact forecast
are reported. The impact scenario is relative to the draining tunnel
at steady state conditions. The resulting risk classes represent the
probability and the severity of impact occurrence from, respectively,
the parametric (DHIM evaluation) and the numerical evaluation pro-
Four risk classes are evidenced in colour from red (class I) to blue
(class IV); in the third column of Table 12 the nal output of the inte-
grated tunnel impact evaluation is shown. Final choice was based as
follows: if, for a given spring, parametric and numerical method pro-
duce 2 risk classes separated by an intermediate class, this last one is
assigned as nal output, recognizing an equal degree of uncertainty
to both methods; if parametric and numerical method produce 2 con-
tiguous risk classes for the same spring, the worst one is assigned as
nal output, for the principle of conservativity.
The excavation and consequent drainage of Crevola Toce diver-
sion tunnel will congure the following most expected scenarios of
impact against main hydrogeological receptors:
• medium risk for Valle Oro spring: very high probability (91%) of
a 5% depletion of average annual ow rate (risk is ranked as medium
because probability is high, according DHIM out put, but discharge
depletion is poor);
• medium risk for Vegno spring: high probability (68%) of a 31%
depletion of discharge;
• medium risk for Cesa spr ing: medium probability (46%) of com-
plete drying up;
• medium-low risk for Lisiel spring: low probability (34%) of a 6%
discharge depletion;
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
   
 34.96 19.97 1.55 0.2 Q (l/s)
 34.96 19.97 1.5 0.2 Q (l/s)
 33.75 19.07 0 0 Q (l/s)
3.46 4.54 100 100 decit %
 33.6 19 0 0 Q (l/s)
3.87 4.90 100 100 decit %
33.45 18.83 0 0 Q (l/s)
4.30 5.72 100 100 decit %
3.88 5.05 100 100 average decit %
    
     
6 Quaternary uvial deposits 0 0 0 0 0 0
7 Baceno micaschists 90.83 35.05 105.01 40.52 155.81 60.12
8 Fault in Baceno micaschists 14.44 164.91 14.58 166.44 15.99 182.51
9 Metasediments (Baceno syncline) 114.84 356.15 116.57 361.54 100.08 310.38
10 Fault in Metasediments (Baceno syncline) 27.61 1214.81 23.8 1047.26 18.94 833.31
11 Antigorio gneiss 12.5 4.39 12.85 4.63 13.08 4.76
12 Fault in Antigorio gneiss 12.75 46.17 12.82 46.43 12.33 44.62
13 Metasediments (Teggiolo syncline) 0 0 0 0 0 0
14 Valgrande and Monte Leone gneiss 0 0 0 0 0 0
15 Quaternary glacial deposits 0 0 0 0 0 0
Total and average 7    
   
Observed 34.96 19.97 1.55 0.2 Q (l/s)
Ante-operam 34.96 19.97 1.5 0.2 Q (l/s)
Post 1 order 33.56 18.92 0 0 Q (l/s)
4.00 5.25 100 100 decit %
A scenario 33.73 19.07 0 0 Q (l/s)
3.52 4.52 100 100 decit %
B scenario 34.12 19.39 0.27 0 Q (l/s)
2.38 2.90 82.10 100 decit %
C scenario
34.23 19.48 0.55 0 Q (l/s)
2.09 2.46 63.64 100 decit %
2.66 3.29 81.91 100 average decit %
 Tunnel inows forecasted by MODFLOW simulations, according to the 3 scenarios described in the main text; tunnel reaches crossing the different
hydrogeological units have been distinguished in order to calculate linear ow rates (every m or km of tunnel advancement).
 Impacts on spring ow rates forecasted by model simulations with the draining tunnel (Post) and dispersant tunnel
 Impacts on spring ow rates forecasted by model simulations with the draining tunnel (Post 1 order) and
with the sealed tunnel (linings where the major forecasted groundwater inows, see the main text for more detail).
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
    
     
6 Quaternary uvial deposits 0 0 0 0 0 0
7 Baceno micaschists 113.18 43.67 121.84 47.01 123.4 47.62
8 Fault in Baceno micaschists 4.58 52.28 1.17 13.31 1.2 13.75
9 Metasediments (Baceno syncline) 85.32 264.61 19.33 59.93 4.43 13.74
10 Fault in Metasediments (Baceno syncline) 12.47 548.63 2.75 121.06 0.28 12.37
11 Antigorio gneiss 18.87 6.5 28.22 9.48 30.66 10.22
12 Fault in Antigorio gneiss 10.91 39.51 3.03 10.97 0.56 2.01
13 Metasediments (Teggiolo syncline) 0 0 0 0 0 0
14 Valgrande and Monte Leone gneiss 0 0 0 0 0 0
15 Quaternary glacial deposits 0 0 0 0 0 0
      
 Tunnel inows forecasted by MODFLOW simulations, according to the 3 scenarios with the sealed tunnel (as described in the main text); tunnel
reaches crossing the different hydrogeological units have been distinguished in order to calculate linear ow rates (every km of tunnel advancement).
 Spring forecasted impact scenarios according to the parametric (DHI) and numerical
(MODFLOW) models. See text for the explanation of the different global risk classes denition.
DOI 10.4409/Am-021-10-0017 AQUA mundi (2010) - Am02017: 135 - 154
Aller L., Bennett T., Lehr J.H., Petty R.J. (1985) - DRASTIC - a standa rdized
system for evaluating ground water pollution potential using hydrogeo-
logic settings: U.S. Environmental Protection Agency, Robert S. Kerr
Environmental Research Laborator y, Ofce of Research and Develop-
ment, EPA/600/2–85/018, 163 pp.
Astol G., Sapig ni M. (1999) - Impianto Crevola Toce III - Progetto di Mas-
sima - Relazione geologica. Enel Spa, internal report.
Barton C.M. (1974) - Engineeri ng classication of rock masses for the design
of tunnel suppor t. Rock Mechanics, 6, 4, pp.189-239.
Bigioggero B., Borian i A., Origoni G iobbi E. (1977) - Microstructure and
mineralogy of an Orthogneiss (Antigorio Gneiss - Lepontine Alps). Ren-
diconti Società Italiana di Mineralogia e Petrologia, 33, 99-108.
Bistacch i A., Massiron i M. (2000) - Post-nappe brittle tectonics and kine -
matic evolution of the north-western Alps: an integrated approach. Tec-
tonophysics, 327, 267-292.
Bortolami G.C., Ricci B., Sella G.F., Zuppi G.M. (1979) - Isotope hydrology
of the Val Corsaglia, Maritime Alps, Piedmont, Italy, in Isotope Hydrol-
ogy 1978, Vol.I, IAEA Symposium 228, June 1978, Neu heberg, Ger-
many, pp. 327-350.
Campan i M., Mancktelow N., Seward D., Rolland Y., Müller W., Guerra
I. (2010) - Geochronological ev idence for continuous exhumation
through the ductile-brittle tr ansition along a crustal-scale low-angle
normal fault: Simplon Fault Zone, central Alps. Tectonics, 29, TC3002,
Canuti P., Ermin i L., Gargini A., Martelli L., Piccinini L., Vincenzi V. (2009)
- Le gallerie TAV att raverso l’Appennino toscano: impatto idrogeologico
ed opere di mitigaz ione. Edir-Edizioni Firen ze, Firen ze, pp. 207.
Castiglioni G.B. (1958) - Studio geologico e morfologico del territor io di
Baceno e Premia. Memorie degli Istituti di Geologie e Mineralogie
dell’Università di Padova, vol. XX, 2-82.
Cesano D., Olofsson B., Bag tzoglou (2000) - Parameters regulating ground-
water inows i nto hard rock tunnels-a statistical study of the Dolmen
tun nel in southern Sweden. Tunneling and underground Space Technol-
ogy, V. 15/2, p. 153-165
Civita M. (1973) - Schematizzazione idrogeologica delle sorgenti normali e
delle relative opere d i capta zione. In Memorie e Note Ist. Geol. Aplplic.
Napoli, 12, 1973, pp.1-34.
Civita M., De Maio M. (2000) - Valuta zione e cartogr aa automatica degli
acquiferi all’inquinamento con il sistema parametr ico SINTACS R5.
Pitagora Editore, Bolog na, 275pp.
Civita M. (2005) - Idrogeologia applicata e a mbientale, Casa Editrice Am-
brosiana , pp 331-336.
Civita M., Gargini A., Pran zini G. (1999) - Metodologia di redazione della
cart a della vulnerabilità intrinseca e del r ischio di inquinamento de-
gli acquiferi del Valdarno Medio. Quaderni di Geologia Applicata, 2,
VOL.1, pp. 1.59-1.73.
Clark I., Fr itz P. (1997) - Environmental Isotopes in Hydrogeology, CRC
Press LLC.
Craig H. (1961) - Isotopic variations in meteoric waters. Science, 133, pp.
Deere, D.U., Deere, D.W. (1989). - Rock Qualit y Designat ion (RQD) After
Twenty Years - Technical Report GL -89-1, U.S. Ar my Eng ineer Water-
ways Exper iment Station, Vicksburg, MS 39180.
Deere, D.U., Merr itt, A.H., Coon, R.F. (1969). - Engineering Classi cation
Of In Situ Rock - Air Force Systems Command, Kir tland Air Force
Base, Report AFWL-64-144.
Dematteis A., Kalamar as G., Eusebio A. (2001) - A system approach for
evaluating springs drawdown due to tunneling - AITES-ITA 2001 World
Tunnel Congress: Progress in tunnelling after 2000 p. 257-264. Mi-
lano, 10-13 June 2001
Domenico P., Schwartz F. (1998), - Physical and chemical Hydrogeolog y,
second edition. John Wiley & Sons, Inc., New York, USA
• medium-low risk for Molinetto wells: medium probability (46%)
of well yield depletion;
medium-low risk for the following springs: Oira, Viceno, La
Valle, Trona, Alfenza Sud, Longio and for Cesa well: medium prob-
ability (46%) of impact;
• low risk for the remainder of receptors: low impact probability
(23%) with shallow groundwater ow systems. Also Ronconi spring
was assigned to this last class, independently from the DHIM out-
put, because representative of a shallow groundwater ow system
located at elevations much higher than tunnel.
Conclusions and remarks
An evaluation method of tunnel drainage impact risk against hy-
drogeological receptors has been presented and discussed. In order
to overcome the intrinsic uncertainty of a forecasting process in
ante-operam conditions and to strengthen the evaluation, a rating
and weighting parametric method and a numerical model have been
used contemporarily; their results have been compared and a nal
forecasting resulted from their arrangement with a conservativity
The two methods evidenced their respective limits and advan-
tages. Parametric methods do not take into account the groundwa-
ter ow equation, but allow to assess quite easily a risk probability
for all the springs; another advantage is that they can be adapted to
the case study, as a function of the available input data; anyway the
rough results need a critical evaluation and renement. On the other
side, numerical modelling requires more data (hydraulic heads and
ows) and its reliability depends a lot on data quantity and quality
and on the goodness of calibration process; but the modelling pro-
cess allows to check the concept ual model by applying the ground-
water ow equation.
In the specic case study, the peculiarity and the strength of the
available data set was the use of different eld methods (geologi-
cal surveys, ow and piezometric measurements, hydrochemistr y,
isotopic techniques), helping a lot in the the conceptual model set
up. The data set was sufciently detailed for the application of para-
metric methods, but still scarce for the numerical modelling process.
Furthermore, the model set up required a strong effort due to the
geological complexity and the steep slopes.
The experience made with the here presented case study suggests
to use rstly parametric methods as a “rst level” risk evaluation
on all the water points, in order to assess the most vulnerable water
points; then, on these vulnerable points a “second level” risk evalua-
tion should be developed consequently by means of numerical mod-
elling, based on a more detailed hydrogeological data set.
 Presented at Session “Hydrogeological im-
pacts of tunnels”, GEOITALIA 2009, Rimini, September 2009
(session’s conveners: Alessandro Gargini, Valentina Vincenzi,
Michele Sapigni).
AQUA mundi (2010) - Am02017: 135 - 154 DOI 10.4409/Am-021-10-0017
Einstei n H.H. (1988) - Special lecture: land slide risk assessment proce-
dure” - Atti di “V I nternational Symposium on Landslides, Losanna, 2,
Federici P.C., Saccan i F., Parietti P., 1967 - Le acque salutari della Val
d’Ossola, collana di monograe dell’ateneo par mense, n. 18.
Federico F. (1984) - Il processo di drenaggio da una galleria in avanzamento
- Rivist a Italiana di Geotecnica, V.4, p. 191-208.
Gargini A., Piccinini L., Martelli L., Rosselli S., Bencini A., Messina A.,
Canuti P. (2006) - Idrogeologia delle unitá torbid itiche: un modello
concettuale derivato dal rilevamento geologico dell’Appenn ino Tosco-
Emiliano e dal monitoraggio ambientale per il tun nel alta velocità fer-
roviaria Firenze-Bologna - Bollettino Società Geologica Italiana, 125
(2006), 293-327.
Goodman R.E., Moye D., Schalkwyk A., Javandel I. (1965) - Groundwater
inows dur ing tun nel d rivi ng.- Geol. Soc. America Publication – Eng-
neering Geology, V. 2, p. 39-56.
Grasemann B., Mank telow N. (1993) - Two-dimensional thermal modelling
of normal faulting: the Simplon Fault Zone, Cent ral Alps, Switzerland.
Tectonophysics, 225, 155-165.
Grosjean G., Sue C., Burkhard M. (2004) - Late Neogene extension in the
vicinit y of the Simplon fault zone (central Alps, Switzerland). Eclogae
geol. Helv., 97; 33-46.
Harbaug h A.W., Banta E.R., Hill M.C., Mcdonald G. (2000) - MOD-
FLOW-2000, The U.S. Geological Sur vey modular ground-water model
- User Guide to modula rization concepts and the ground-water ow pro-
cess - U.S. Geological Survey, Open-File Repor t 00-92
Kanit T., Forest S., Galliet I., Mounoury V., Jeulin D. (2003) - Determi nation
of the size of the representative volume element for random composites:
statistical and numerical approach. Inter national Journal of Solids and
Structures, 40, 3647–3679.
Jiao Y., Hudson J.A. (1995) - The Fu lly-Coupled Model for Rock Engineer-
ing Systems. - International Journal of Rock Mechanics and Mining
Sciences & Geomechanics Abstracts, 32, (5), 491-512. Oxford, United
Kingdom: Elsevier Science)
Lei S. (1999). - A n analytical solution for steady state ow into a tunnel.”
Ground Water - V. 37/1, p. 23-26.
Lei S. (2000). - Steady state ow into a tunnel with constant pressure bound-
ary. - Ground Water V.38/5, p. 643-644.
Leopold L. B. Et Al. (1971) - A procedure for evaluating envi ronmental im-
pact, U.S. Geological Survey Circular 645, Washington D.C., U.S. Dep.
Of the Interior.
Loew S. (2002) - In Barla G. & Barla M. (Eds.) - Le opere in sotterraneo e il
rappor to con l’ambiente - IX ciclo M IR, Torino 26-27 novembre 2002,
Patron Editore, p. 201-217.
Long J.C.S., Remer J.S., Wilson C.R., Witherspoon P.A. (1982) - Porous
Media Equivalent for Networks of Discont inuous Fract ures. Water Re-
sources Research, 18(3), 645-658.
Maillet E. (1905) - Essais d’hydraulique souter raine et uviale. Librairie Sci.
Hermann Paris, 218pp.
Mancktelow N. (1985) - The Simplon line: a major displacement zone in the
western Lepontine Alps. Eclogae geol. Helv. 78, 73-96.
Martinott i G. Per Enel Produzione (1993) - Impianto di Piedilago - Studio
geologico strutt urale e idrologico - Rapporto nale del 30 aprile 1993.
Martinott i G., Mar ini L., Hunz iker J C., Perello P., Pastorelli S. (1999) -
Geochemical and geothermal study of springs in the Ossola-Simplon
Region, in Eclogae geol. Helv. 92 (1999), pp. 295-305.
Masset O., Loew S. (2010) - Hydraulic conductivity distr ibution in crystal-
line rocks, derived f rom inows to tunnels and galleries in the Central
Alps, Switzerland. Hydrogeology Journal, 18, pp. 863-891.
Maxelon M., Mancktelow N. (2005) - Three -dimensional geometry and
tectonostratigraphy of the Pen nine zone, Central Alps; Switzerla nd and
Northern Italy. Ear th-Science Reviews, 71, 171-227.
Mcdonald M.G., Harbaugh A.W. (1988) - A Modula r Three-Di mensional
Finite-Difference Ground-Water Flow Model - U.S. Geological Sur vey,
Techniques of Water-Resources Invest igations, Book 6, Chapter A1
Mehl S.W., Hill M.C.,( 2001), - MODFLOW-2000, The U.S. Geological Sur-
vey Modular Ground-Water Model - User Guide to the Link-Amg (LMG)
Package for Solving Matrix Equations Using an Algebraic Multigrid
Solver: U.S. Geological Survey Open-File Report 01-177, 33 p.
Milnes A. G., Greller M., Müller R. (1981) - Sequence and style of major
post-nappe str uctures, Simplon-Pen nine Alps. Journal of Sr uctural Ge-
ology, 3, 411-420.
Mun Y., Uchrin C.G. (2004) - Development and Application of a MODFLOW
Preprocessor Usi ng Percolat ion Theory for Fractu red Media. Jour nal of
the American Water Resources Association, 40(1), 229-239.
Pastorelli S., Mari ni L., Hun ziker J. (2001) - Chemistry, isotop values (δD,
δ18O, δ34SSO4) and temperatures of the water inows in two Gott hard tu n-
nels, Swiss Alps, in Applied Geochemistr y 16 (2001), pp. 633-649.
Prudic D.E. (1989) - Documentat ion of a computer program to simulate
stream-aquifer relations using a modular,  nite- difference, ground-wa-
ter ow model. U.S. Geological Sur vey, Open-File Report 88-729.
Schmidt C., Preiswerk H. (1905) - Geologische Karte der Simplong ruppe.
Beiträge zur Geologischen Karte der Schweiz, [N.F.] 26, Spezialkar te
No. 48.
Steck A. (2008) - Tectonics of the Simplon massif and Lepontine gneiss
dome: deformation structures due to collision between the under thr ust-
ing European plate and the Adriatic indenter. Swiss Journal of Geosci-
ence, 101, 515–546.
Thapa B.B., Nolting R.M., Teske M.J, Mcrae M.T. (2005) - Predicted and
observed g roundwater inows into two rock t unnels - RECT Proceed-
ings, Chapter 50, p. 556-567.
Thornthwaite C.W., Mather J. R. (1957) - Instructions and tables for comput-
ing potential evapotr anspiration and the water balance. Drexel I nstitute
of Technology, Publications i n Climatology, Vol. X, 3, 185-311 p, Cen-
terton, New Jersey (USA).
Urey H. C., Lowenstam H. A., Epstein S., Mckinney C. R. (1951) - Measure-
ment of paleotemperatures and temperat ures of the Upper Cretaceous of
England, Den mark and the Souther n United States. Geological Societ y
of America Bullettin, 62, pp. 399-416.
Zaadnoordijk W.J. (2009) - Simulating Piecewise-Linear Surface Water
and Ground Water Interactions with MODFLOW. Ground Water, 47(5),
Zheng C., Bennevy G. D. (1995) - Applied contam inant t ransport modelling
- Van Nostrand Rei nhold ITP
Zwingmann H., Manck telow N. (2004) - Timing of Alpine fault gauges.
Eart h and Planet ary Science Let ters, 223, 415-425.
... Using the DHI method alongside numerical modeling, Vincenzi et al. (2010) have concluded that the DHI methodology is best to use for a general estimation of the drawdown hazard's risk. It was then suggested to solve a numerical model for the high-risked springs. ...
... Although using numerical methods for groundwater inflow assessment in tunnels is widely common in the literature (Gattinoni et al. 2009;Nikvar-Hassani et al. 2016, 2018Farhadian et al. 2016), examples of using these methods for predicting the impacts of tunneling on the hydrogeologic environment and springs' discharge are scarce (Yang et al. 2009;Vincenzi et al. 2010;Perello et al. 2014;Piccinini and Vincenzi 2010). However, the problem with numerical models is that they require various geological and hydrogeological information in order to run precisely. ...
Water inflow caused by tunneling can have severe impacts on the springs’ discharge rate. If these impacts have not been predicted beforehand, technical, economic, and environmental challenges could occur. While there are a few methods for evaluating the risk of water drawdown, their shortcomings create the need to develop a new one. First, in this research, five main tunneling projects in Iran were studied for evaluating the influence of tunneling on spring’s discharge, and a comprehensive database that contains information on 111 springs located in the vicinity of these tunneling projects was developed. Then, by learning from previously developed methods’ shortcomings and using an appropriate decision analysis method (Analytic Hierarchy Process or AHP), a new model was proposed for evaluating the risk of discharge reduction in springs located in the vicinity of tunneling projects. This new model, named TIS (Tunneling Impacts on Springs), was developed based on four important parameters of a) volume of water inflow toward the tunnel, b) distance between spring and tunnel, c) hydraulic connectivity, and d) aquifer recharge potential. In the next step, using data recorded in the database, TIS values were calculated for each spring, and using suitable statistical methods, the obtained TIS values were classified based on the actual behavior of springs. For using this model in practice, all springs must be checked using a screening process. In this process, according to some limitation criteria (including distance from the tunnel, groundwater condition in tunnel, spring origin), unimportant springs are excluded from the list and only springs with possible influence from tunneling are considered for further assessments. This helps to investigate the in-risk springs more effectively.
... From a hydrogeology point of view, the footprint and negative impacts that the construction of a tunnel can have include groundwater resource depletion, land settlement (Yoo 2016), deterioration of groundwater-dependent ecosystems, groundwater quality degradation, drying of water wells, springs and wetlands, and decline in surface-water supplies (Clements 2006;Golian et al. 2019b). Most tunnel environmental assessment studies over the past two decades have focused on tunnel construction (Molinero et al. 2002;Torri et al. 2007;Yang et al. 2009;Raposo et al. 2010;Vincenzi et al. 2010;Font-Capó et al. 2011;Liu et al. 2015;Golian et al. 2019b). Although tunnel operations may have destructive consequences, few researchers have addressed this problem. ...
... In this regard, the tunnel must first be implemented into the numerical model using the Drain package of MODFLOW (Vincenzi et al. 2010;Golian et al. 2018;Marchese et al. 2020). Through the DRN, the amount of water removed from the aquifer by the tunnel, as long as the groundwater level is above the tunnel elevation, can be determined by the following equation (Golian et al. 2018(Golian et al. , 2019a: ...
Full-text available
Tunneling is often unpopular with local residents and environmentalists, and can cause aquifer damage. Tunnel sealing is sometimes used to avoid groundwater leakage into the tunnel, thereby mitigating the damage. Due to the high cost of sealing operations, a detailed hydrogeological investigation should be conducted as part of the tunneling project to determine the impact of sealing, and groundwater modeling is an accurate method that can aid decision-making. Groundwater-level drawdown induced by the construction of the Headrace water-conveyance tunnel in Sri Lanka dried up 456 wells. Due to resulting socio-environmental problems, tunnel sealing was decided as a remedy solution. However, due to the expectation of significant delays and high costs of sealing, and because the water pressure in the tunnel may prevent groundwater seepage into the tunnel during operation, there was another (counter) decision that the tunnel could remain unsealed. This paper describes groundwater modeling carried out using MODFLOW to determine which option—sealed or unsealed tunnel—is more effective in groundwater level recovery. The Horizontal Flow Barrier and River packages of MODFLOW were used to simulate sealed and unsealed tunnels, respectively. The simulation results showed that only through tunnel sealing can the groundwater level be raised to preexisting levels after 18 years throughout the study area. If the tunnel remains unsealed, about 1 million m3/year of water conveyed by the tunnel will seep into the aquifer, reducing the operational capacity of the tunnel as a transport scheme. In conclusion, partial tunnel sealing in high-impact sections is recommended.
... Al passaggio tra condizioni di deformazione tettonica duttile e fragile era senz'altro ancora presente una circolazione di fluidi caldi che, però, non avrebbe causato fenomeni di alterazione idrotermale, come suggerito dalle analisi di laboratorio. D'altra parte, la circolazione idrica sotterranea recente ha certamente causato la dissoluzione di grandi volumi di rocce evaporitiche e, secondariamente, carbonatiche (Vincenzi et al. 2017). L'alterazione dello gneiss è interpretata come il prodotto della circolazione idrica recente favorita dall'intesa deformazione fragile sia antica che più recente, legata in parte alla dissoluzione dei litotipi solfato-carbonatici e in parte al rilascio tensionale post-glaciale del versante. ...
... Furthermore, if poor parameters and constraints are introduced into the model, poor results will be obtained (Fielding et al. 2011). All this has led most postmining models to explicitly ignore fracture structures by employing the equivalent porous medium (EPM) approach or even both media with lumped parameter models (e.g., see Adams and Younger 2001;Banks 2001;Vincenzi et al. 2010). Notwithstanding that both approaches have been successfully applied in case studies, they exhibit notable drawbacks when describing the fast component of the fractured system (Kim et al. 1997;Rapantova et al. 2007). ...
Full-text available
Underground hard coal mining usually disrupts the mechanical equilibrium of rock sequences, creating fractures within minor permeable rocks. The present study employs a dual-continuum model to assess how both fractured and porous sandstone media influence the percolation process in postmining setups. To test the approach, the software TOUGH2 was employed to simulate laminar fluid flow in the unsaturated zone of the Ibbenbüren Westfield mining area. Compared to other coal mining districts in Germany, this area is delineated by the topography and local geology, leading to a well-defined hydrogeological framework. Results reveal good agreement between the calculated and measured mine water discharge for the years 2008 and 2017. The constructed model was capable of reproducing the bimodal flow behavior of the adit by coupling a permeable fractured continuum with a low-conductivity rock matrix. While flow from the fractured continuum results in intense discharge events during winter months, the rock matrix determines a smooth discharge limb in summer. The study also evaluates the influence of individual and combined model parameters affecting the simulated curve. A detailed sensitivity analysis displayed the absolute and relative permeability function parameters of both continua among the most susceptible variables. However, a strong a priori knowledge of the value ranges for the matrix continuum helps to reduce the model ambiguity. This allowed for calibration of some of the fractured medium parameters for which sparse or variable data were available. However, the inclusion of the transport component and acquisition of more site-specific data is recommended to reduce their uncertainty.
... The capability of numerical simulation of flow as a tool for investigating science in general and hydrogeologic sciences, in particular, has reached a new level in recent days [1,2]. The direction of this type of research represents a promising future aimed at providing useful alternatives to laboratory experiments where realistic input parameters into hydrogeologic simulations are strenuous [3][4][5][6]. ...
Full-text available
Residence time of water flow is an important factor in subsurface media to determine the fate of environmental toxins and the metabolic rates in the ecotone between the surface stream and groundwater. Both numerical and lab-based experimentation can be used to estimate the residence time. However, due to high variability in material composition in subsurface media, a pragmatic model set up in the laboratory to trace particles is strenuous. Nevertheless, the selection and inclusion of input parameters, execution of the simulation, and generation of results as well as post-processing of the outcomes of a simulation take a considerable amount of time. To address these challenges, an automated particle tracing method is developed where the numerical model, i.e., flow and reactive transport code, MIN3P, and MATLAB code for tracing particles in saturated porous media, is used. A rectangular model domain is set up considering a fully saturated subsurface media under steady-state conditions in MIN3P. Streamlines and residence times of the particles are computed with a variety of seeding locations covering the whole model surface. Sensitivity analysis for residence time is performed over the varying spatial discretization and computational time steps. Moreover, a comparative study of the outcomes with Paraview is undertaken to validate the automated model (R2 = 0.997). The outcome of the automated process illustrates that the computed residence times are highly dependent on the accuracy of the integration method, the value of the computational time step, ∆t, spatial discretization, stopping criterion for the integration process of streamlines, location, and amount of seed points. The automated process can be highly beneficial in obtaining insights into subsurface flow dynamics with high variability in the model setup instead of laboratory-based experimentation in a computationally efficient manner.
... As mentioned earlier, the drop in water level makes several problems investigated by researches in different branches. The majority of researches were conducted in ranges over land subsidence (Yoo et al. 2012;Chai et al. 2004;Cheng et al. 2014), surface settlements (Plasencia et al. 2015;Preisig et al. 2014;Loew et al. 2014;Pujades et al. 2017), slope instabilities (Sun et al. 2018;Pinyol et al. 2008;Viratjandr et al. 2006), drying of springs (Dematteis et al. 2001;Vincenzi et al. 2010;Perello et al. 2014;Karimzade et al. 2017), topographic influences (Hajitema et al. 2005), urban damage (Raposo et al. 2010;Paronuzzi et al. 2013;Shourian et al. 2017;Attard et al. 2016), and ecological impacts (Kverner et al. 2008;Yang et al. 2009;Zheng et al. 2017;El Tani et al. 2019). In this paper, water drawdown due to water ingress has been studied. ...
Conference Paper
Full-text available
Water drawdown due to water ingress is a probable issue in tunnelling processes. This makes serious problems including ground settlements, environmental problems, increasing costs, excavation hazards, and several executive difficulties. In this paper, using measured data, data investigation, probabilistic and sensitivity analysis, machine learning methods, curve fitting, and several numerical approaches, a model has been developed using Gene Expression Programming (GEP), which is well-known in machine learning and model generation methods. The proposed model has eight input parameters consists of water ingress, RMR, rainfall, distance and Poisson's ratio. To training the model, 15 km headrace tunnel (HRT) of Uma Oya Multipurpose Development Project (UOMDP) located in Siri-Lanka has been considered as a case study. Based on the analyses, the proposed model has a good performance in predicting the water drawdown amounts and its variations. A graph was generated to consider the simultaneous effects of water ingress and distance on the drop of water level.
... The procedure provided an accurate prediction in comparison with actual measured conditions. Vincenzi et al. (2010) used parametric (DHI method) and numerical (MODFLOW model) methods to forecast impacts of the hydro-electrical diversion tunnel (located in Verbania province, Italy) on adjacent springs and well flow rates. Results of the two approaches were compared and forecasting was carried out. ...
Weather approachable urban planning performs an important part in climate change moderation and revision and permits for maintainable expansion of existing circumstances for forthcoming groups. It has been understood that procedures for example urban greening, established facades and slates or extremely brilliant construction resources are able to diminish additional heat and benefit plummeting energetic budgets. Transporting technical and frequently academic information into real urban planning but essentially includes an interdisciplinary discourse. It proposes to prepare an evaluation of present nonfiction from a climatological viewpoint so as to response the enquiry how consequences from urban climate revisions can be connected to architectural project of forthcoming urban zones. Consequences from advanced investigation are assessed and critically addressed, hereafter preparing a directory for urban organizers and investors which should attend as foundation for a re-evaluation of the term ‘smart urban’.
... The procedure provided an accurate prediction in comparison with actual measured conditions. Vincenzi et al. (2010) used parametric (DHI method) and numerical (MODFLOW model) methods to forecast the impacts of a hydroelectrical diversion tunnel in Verbania province, Italy, on adjacent springs and well flow rates. The results of the two approaches were compared and forecasting was carried out. ...
The probability of the occurrence of porphyry copper ore associated with subduction zone of the Neo-Tethys ocean and volcanic arc of Iran (Urumieh-Dokhtar) justifies the necessity of exploration and prospection of this type of ore in Iran. Zafarghand index and Kahang deposit of porphyry copper-molybdenum located in Isfahan province, center of Iran using Satellite data interpretations were discovered in the years of 2010 and 2003, respectively. Geological studies have indicated the presence of argillic and propylitic alteration halos associated with porphyry copper systems. The 250 rock samples were systematically collected at a sampling distance of 100 m and in the center of the porphyry system by a distance o f 50 m in Zafarghand. Also, 377 samples of lithogeochemicals (185 rock samples and 192 soil samples) were extracted systematically from Kahang deposit. The study of geochemical data of rock and soil samples showed similarity of these two types of mineralization with other mineralization of porphyry copper-molybdenum elsewhere in the world. Finally, the comparison of geochemical anomalies copper with rock units and alteration zones showed that atmospheric waters had washed out the copper in some of these zones and probably the supergene zone was formed in depth as porphyry copper-molybdenum deposit.
Full-text available
This year, World Water Day celebrated Groundwater with the motto “Making the invisible, visible”. Regardless of this particular recurrence, the one expressed by the motto represents the daily mission of every scholar or professional dedicated to groundwater. What is done through hydrogeological maps, through numerical models, through the development and interpretation of data series, tests in boreholes, wells and springs, what else is it, if not a way of visualizing this invisible underground resource? [...]
Questo documento presenta le Linee Guida per la gestione sostenibile delle venute d’acqua e del calore geotermico nelle gallerie, elaborate dal Gruppo di Lavoro GESTAG (GEstione SosTenibile delle Acque nelle Gallerie), istituito il 20/06/2012 dal Comitato Italiano dell’Associazione Internazionale degli Idrogeologi. Il documento affronta e descrive obiettivi, metodi e casi di studio, ciascuno dei quali richiede di adottare misure tecniche di vario tipo, illustrate nei capitoli seguenti, come studi idrogeologici, modelli predittivi degli impatti, soluzioni tecnologiche di contenimento del drenaggio e di captazione, recupero e valorizzazione delle risorse intercettate, monitoraggi. Sono affrontati anche i così detti temi non tecnici, come la comunicazione, poiché è ormai noto che l’accettabilità sociale di una galleria influenza direttamente la sua sostenibilità economica e finanziaria, per gli effetti che può avere sui tempi di realizzazione e sui costi delle compensazioni. Sulla base delle diverse esperienze dei componenti del Gruppo di Lavoro GESTAG, si è optato per una suddivisione in 11 temi principali, in particolare, in ciascuno dei capitoli successivi a quello introduttivo vengono trattati rispettivamente: Capitolo 2: l’importanza del ritorno di esperienza proveniente da gallerie già scavate, le banche dati disponibili e i criteri di analisi dei dati pregressi; Capitolo 3: come realizzare lo studio idrogeologico di una galleria, la relazione con il modello geologico, le metodiche da adottare nelle differenti fasi del progetto, che sono state distinte in ante-operam, in corso d’opera e post-operam; Capitolo 4: il tema degli impatti che il drenaggio in galleria può avere sull’ambiente circostante, le analisi che devono essere svolte e gli strumenti da adottare per la gestione del rischio, infine gli interventi per la mitigazione degli impatti; Capitoli 5 e 6: le metodiche di progettazione per la valorizzazione rispettivamente delle acque drenate in galleria e del calore; Capitolo 7: i prodotti chimici che vengono utilizzati in fase di scavo per il sostegno e il miglioramento del terreno, e la loro compatibilità con l’utilizzo delle acque drenate; Capitolo 8: i criteri di monitoraggio delle acque in un progetto di galleria; Capitolo 9: la comunicazione, il rapporto con i territori e l’analisi di casi pregressi, nonché una guida sulle buone pratiche da adottare; Capitolo 10: una panoramica sulla normativa europea ed italiana in tema di valutazione di impatto ambientale, monitoraggio e scarichi delle acque drenate; Capitolo 11: la bibliografia ragionata inerente l’argomento trattato, che include pubblicazioni tecnico scientifiche, normative di riferimento e risorse web.
This textbook provides an introduction to the study of hydrogeology, and maintains the process oriented approach of the earlier edition. The introduction is followed by chapters on: the origin of porosity and permeability; groundwater movement; equations of flow, boundary conditions and flow nets; groundwater in the basin hydrologic cycle; hydraulic testing; groundwater resources; stress, strain and pore fluids; heat transport in groundwater flow; solute transport; aqueous geochemistry; chemical reactions; colloids and microorganisms; mass transport equations; mass transport in natural groundwater systems and groundwater flow; contaminant hydrology; modelling of dissolved contaminant transport; multiphase fluid systems; remediation; and in situ destruction and risk assesment.
Geological, hydrogeological and geochemical surveys were carried out in the Piedilago area (Ossola-Simplon region) in order to investigate the geothermal resources present in this area. Following these surface exploration efforts an exploratory geothermal well of 248 m was drilled in 1991. It discharges a thermal water with temperatures up to 43°C and calcium (sodium) sulphate composition with a TDS close to 1350 mg/l. Chemical geothermometers suggest a reservoir temperature close to 45°C indicating that the well virtually produces the pure uncooled thermal water. The Piedilago example is here considered as the departure point to establish both general criteria for further geothermal investigations in young mountains chains and taking into consideration all the available data on geology and fluid geochemistry of thermal systems in the Ossola-Simplon region, to constrain a geothermal model for the Lower Pennine Zone.
Water inflows in the Gotthard Highway Tunnel and in the Gotthard Exploration Tunnel are meteoric waters infiltrating at different elevations, on both sides of an important orographic divide. Limited interaction of meteoric waters with gneissic rocks produces Ca–HCO3 and Na–Ca–HCO3 waters, whereas prolonged interaction of meteoric waters with the same rocks generates Na–HCO3 to Na–SO4 waters. Waters circulating in Triassic carbonate-evaporite rocks have a Ca–SO4 composition. Calcium-Na–SO4 waters are also present. They can be produced through interaction of either Na–HCO3 waters with anhydrite or Ca–SO4 waters with a local gneissic rock, as suggested by reaction path modeling. An analogous simulation indicates that Na–HCO3 waters are generated through interaction of Ca–HCO3 waters with a local gneissic rock. The two main SO4-sources present in the Alps are leaching of upper Triassic sulfate minerals and oxidative dissolution of sulfide minerals of crystalline rocks. Values of δ34SSO4 < ∼+9‰ are due to oxidative dissolution of sulfide minerals, whereas δ34SSO4 >∼+9‰ are controlled either by bacterial SO4 reduction or leaching of upper Triassic sulfate minerals. Most waters have temperatures similar to the expected values for a geothermal gradient of 22°C/km and are close to thermal equilibrium with rocks. However relatively large, descending flows of cold waters and ascending flows of warm waters are present in both tunnels and determine substantial cooling and heating, respectively, of the interacting rocks. The most import upflow zone of warm, Na-rich waters is below Guspisbach, in the Gotthard Highway Tunnel, at 6.2–9.0 km from the southern portal. These warm waters have equilibrium temperatures of 65–75°C and therefore constitute an important low-enthalpy geothermal resource.