2. Study Area
The water-conveyance tunnel of HPP Pirot (Southeastern Serbia) is used to transport stored water under pressure to the powerhouse. The tunnel is 9093 m long, its diameter is 4.5 m and the gradient 2.51 ‰. It was constructed by standard excavation (first and last parts) and tubing (1/4 and 3/4) with concrete lining (
Figure 1). Three observation wells, PP-1, PP-3 and PP-4, were deep-drilled along the tunnel and the screens installed within the tunnel zone. The HPP has a gravity dam and is usually operated 4–5 h/day. Surface water reservoir is located at an elevation of 612 m, occupies an area of 5.53 km and has a total volume of 170 × 10
6 m
3.
In geological terms, the tunnel was built through the contact of two regionally important tectonic units (Getic and Danubian), which are represented by a system of east-vergent thrusts. The most significant is the Vidlič thrust, which the tunnel intersects (
Figure 2). The study area is mostly built up of carbonate sediments of the Triassic, Jurassic and Cretaceous ages. The Triassic deposits are represented by limestones, dolomites, conglomerates and sandstones, whilst the Jurassic deposits are represented by sandstones, clays, conglomerates and marble limestones. The most frequent deposits of the Cretaceous age are limestones, marls and sandstones. The Mesozoic complex is transgressively overlain by Pliocene sand, and marly clays. (
Figure 2).
The complex geological structure of the terrain has led to the development of a complex hydrogeologic system, where there is a hypogenic karst aquifer in addition to the local hydrogeologic system. Based on the structural type of porosity, the karst aquifer is dominant. Other aquifer types are less developed. The typical contact springs are Berilovac and Izvor, which are the main drainage points of the local aquifer. The epikarst zone is mostly drained through shallow springs within the karst massif (Nišor, Glame, Dobri Do), while the thermal spring Dag Banjica represents the drainage point of a deeper hydrogeologic system characterized by hypogenic conditions.
3. Materials and Methods
Research activities and the selection of methods depended on the operating regime of the water-conveyance tunnel. The campaign lasted for 108 days and included the months of September, October and November 2016. Quantitative and qualitative groundwater data were collected in two primary stages (
Table 1), while the tunnel was in operation (Phases A and C) and out of operation (Phase B). Phase A illustrates the state when the tunnel and the hydrogeologic environment were under pressure, while phase C reflects the state when only the tunnel was under pressure.
During the campaign, the monitoring program included all springs (Berilovac, Izvor, Dag Banjica, Glame, Dobri Do and Nišor), the observation wells along the tunnel (PP-1, PP-3 and PP-4), a surface water reservoir and the Dobrodolska River. Monitoring within the tunnel and at the control gate was conducted while the tunnel was out of operation.
Summary losses of the fully charged tunnel were monitored at the control gate following the standard procedure [
12], by maintaining the water level at the control gate for one hour and then evacuating the water via a bypass for 30 min. Observation of summary losses lasted for seven hours.
Frequent monitoring of the observation wells along the tunnel, as well as at the control gate, was performed using a TD-Diver with a 2 min sampling interval. Vertical probing of water conductivity and temperature was also undertaken during characteristic periods (operating regime, discharging and charging of the tunnel).
While the tunnel was empty, monitoring of quantitative and qualitative parameters included all leakage points in concrete lining fissures and internal boreholes. Quantitative monitoring was performed by continuous gauging of pressure and leakage, and consecutive hydrometric measurements along characteristic sections of the tunnel.
Qualitative prospecting was carried out in situ, using a multi parameter probe (for pH, oxidation-reduction potential—ORP, electrical conductivity—EC, dissolved oxygen—O2, and temperature—T), and included all leakage points where groundwater seeped into the tunnel. On the basis of the conducted prospecting, 10 characteristic sampling points were selected for chemical analyses of the water. Groundwater was sampled on three occasions and a total of 25 samples from the tunnel were analyzed (in some places sampling was not possible because the amount of water decreased over time). The primary chemical composition of the groundwater (HCO3−, Cl−, SO4−, Na+, K+, Ca2+, Mg2+) was determined by standard analytical methods under laboratory conditions.
Potentially threatened/critical zones of the tunnel in terms of groundwater aggressiveness and the tunnel’s structural stability were identified on the basis of the chemical composition of the groundwater sampled from the tunnel. The aggressiveness analyses were based on EU standard EN 206-1 [
13], US Department of Transportation standard (CDT 2003) [
14], and the Larson–Skold (LSkl) index [
15,
16].
4. Results and Discussion
Monitoring of the studied springs did not reveal any distinct changes in quantitative and qualitative parameters. Hence, it was not possible to clearly determine their hydraulic connection with the tunnel. The main changes in the quantitative and qualitative parameters of groundwater were detected at observation well PP-3,
Table 2.
However, despite the fact that the greatest changes were detected at observation well PP-3, a comparative analysis of that well and PP-1 indicated that the groundwater level largely depended on the operating regime of the tunnel. Parallel plots showed different levels of hydraulic interaction between the hydrogeologic setting and the tunnel in the considered zones/sections (
Figure 3). Even though PP-1 was the closest to the reservoir (corresponding to tunnel section 1 + 400 m), the measured data indicated a naturally very high water table, detected above the water level of the storage reservoir. The relatively small changes in groundwater level at PP-1 are primarily a result of specific geologic and hydrogeologic conditions on the location of this observation well (fractured aquifer, low productivity), featuring sandstones, claystones and conglomerates (age J
12+3). However, these changes are also indicative of the hydraulic behavior of the tunnel and the hydrogeologic environment.
The relatively large changes in groundwater level at PP-3 of 43.03 m are a result of the geologic and hydrogeologic conditions on the location of that observation well. The terrain is built up of highly karstified limestones (ages K13 and K13+4), within the zone of the Dobri Do syncline, with the thinnest package of strata that overlie the tunnel (only 70 m). The location corresponds to tunnel section 4 + 200 m.
Particularly pronounced changes in groundwater level at PP-3 were detected when the tunnel was offline (out of operation). The summary water losses in the tunnel were reflected in a continuous decrease in the water level in the observation well, as well as at the control gate (
Figure 4). The decrease in the groundwater level at PP-3 during the measurement process (7 h) was Δh = 7.49 m, which is indicative of intensive interaction of the water in the tunnel and the hydrogeologic environment within the zone of observation well PP-3. Seepage of groundwater of elevated temperature was detected in step IV of groundwater level holding (
Figure 4B). A stable water temperature regime at PP-3 (9.7 °C) indicated that the temperature rise at the control gate did not originate from this zone.
The recession part of the diagram (
Figure 3 and
Figure 4) reflects the state of a discharged hydrogeologic environment in the central zone of the tunnel and the creation of quasi-natural conditions. As a result of rainfall, the groundwater level increased by 1 m at PP-3, which indicated a well karstified zone. In addition to the active hydraulic connection of the transported water and the groundwater, the propagation of precipitation at PP-3 was rapid.
The high reactivity of observation well PP-3 was apparent while the tunnel was being recharged, at which time the pressure in the hydrogeologic environment stabilized. However, there was a particularly high reactivity (instant response of observation well PP-3) while the tunnel was online, when both the tunnel and the hydrogeologic environment were under the influence of the pressure of the transported water (
Figure 5).
In the full operating mode of the HPP (4–5 h/day), at PP-3 the transport of water relieved the pressure not only in the tunnel but within the hydrogeologic environment as well. The pressure decrease caused groundwater to seep into the tunnel. After the HPP was placed offline, the pressure stabilized again and the water from the tunnel seeped into the hydrogeologic environment within that zone. Analyses indicated a potentially unstable zone, which is under constant hydraulic load.
In addition to pronounced groundwater level fluctuations at PP-3, there were also changes in groundwater quality. Vertical probing of EC in observation well PP-3 indicated mixing process of highly-mineralized (Total dissolved solids—TDS) groundwater (detected at sampling point 4 + 230 m in the tunnel, EC: 5000–10,200 μS/cm) and the water transported by the tunnel, whose mineral content was much lower (
Table 3). When the tunnel was in full operating mode and the pressure of the transported water seeping into the hydrogeologic environment was constant, the TDS pattern changed (EC increased with depth). Conversely, while the tunnel and the hydrogeologic environment were being recharged (Phase C), there was EC inversion as a result of tunnel water (low TDS) seeping into the hydrogeologic environment and “pushing” the higher-TDS water towards the surface. Stabilization through diffusion re-established the usual pattern, with water of lower density closer to the surface (
Table 3).
During the course of the investigations in the tunnel, data from consecutive hydrometric measurements were used to identify characteristic tunnel sections (Sections I, II and III) between four observation points (0 + 030, 3 + 160, 6 + 560 and 8 + 530 m). The three sections represent inflows or, more precisely, the differences in inflow between the beginning and end of each tunnel section, and the corresponding observation well (PP-1, PP-3 or PP-4) (
Figure 6).
The rate of inflow into Section I indicates seepage mostly from the fractured aquifer (less than 5 L/s) and a much higher groundwater level (65 m at PP-1) relative to the water level of the reservoir, with a difference of 3.18 m when the tunnel was completely empty.
The rate of inflow into Section II was up to 9 L/s and the groundwater level was generally about 11 m higher than the water level of the reservoir when the tunnel was online. The difference between the online and offline status was greatest (43.03 m) at this observation well. A characteristic of Section II is that it had the highest rate of drainage of the karstified zone above the tunnel.
There was constant inflow into Section III of the tunnel from the hydrogeologic environment, with a rate increase of 10.8 L/s. At tunnel section 7 + 350 m, concentric outflow was registered in the bottom of the tunnel, resulting in an increased overall groundwater inflow into this Section. The groundwater level changes at observation well PP-4 suggested that it was not representative and was therefore excluded from the study. High levels of groundwater and low EC values (130 µS/cm) in PP-4 indicates that it is rainwater and that this observation well is not in interaction with the tunnel.
The physical and chemical parameters of the groundwater in the tunnel varied: EC range 243–10,200 µS/cm, ORP −277 to 458 mV, and pH 7–9.1 (alkaline water). Water temperatures were from 8.35 to 17 °C. The higher range (>12 °C, from 5 + 940 to 7 + 600 m) was measured in the deeper part of the rock mass along the contact of the main tectonic units (Vidlič thrust). Measurements of water losses at the control gate (
Figure 4) showed an elevated water temperature, which may be a result of groundwater inflow from this section. The presence of diverse groundwater characteristics and water types, from Na–Cl through Na–HCO
3 to Ca(Mg)–HCO
3, indicated that the heterogeneity of the bedrock and changes in the tunnel’s operating regime had a considerable influence on the hydrochemical conditions. In the final stages of tunnel prospecting, the highest EC and water temperatures, and the lowest ORP were recorded when the natural groundwater regime was restored.
At tunnel section 4 + 230 m, which corresponds to the location of observation well PP-3, the water samples were of the Na–Cl–SO4 type, with extremely high ionic concentrations: Na: 2100 mg/L, Cl: 2310 mg/L and SO4: 1700 mg/L. When pressurized water was transported down the tunnel, the concentrations near PP-3 increased and altered the natural qualitative and quantitative parameters of the groundwater. High concentrations of these ions were also detected in water samples from tunnel sections 1 + 815 m and 1 + 970 m, which correspond to the location of observation well PP-1. This indicated that there was interaction in the entire section of the tunnel from 1 + 815 to 4 + 230 m and was corroborated by the qualitative and quantitative parameters of the groundwater sampled from PP-1, in which a significant increase in the Cl ion concentration was detected while the tunnel was being charged. This section of the tunnel, nearly 2500 m long, was identified as the zone in which the tunnel hydraulically interacts with the hydrogeologic environment.
Based on the standards used to define the level of groundwater aggressiveness to the concrete lining of the tunnel, the most aggressive water, in terms of high chloride and sulfate ion concentrations (
Table 4), was sampled at 4 + 230 m of the tunnel. In addition, the water samples from 1 + 970 m exhibited a high sulfate aggressiveness to the tunnel lining. Consequently, these two sections of the tunnel are considered to be the most threatened, from a structural stability perspective.