AQUA 2015 15 th September 2015, ROME Low velocity areas' influence on contaminant persistence: a field study (Milan, Italy) Alberti Luca 1, Colombo Loris 1, Francani Vincenzo 1 1 Politecnico di Milano, DICA Dipartimento di Ingegneria Civile e Ambientale
AIM OF THE STUDY The city of Milan and its Functional Urban Area (FUA) host many contaminated sites. Soils and aquifer are interested by local (hotspots) and also diffusive (multiple-sources) contamination whose identification is recommended for optimal environmental management. Furthermore, the quality is significantly influenced by many others parameters such as the piezometric fluctuations and change on the regional groundwater flow direction. In particular, chlorinated solvents are showing long time persistence in the aquifers, at least since 1975, when they have been first detected in Milan drinking water. mg/l 25 mg/l 14 TCM_0151460048 TCM_0151460321 20 12 15 10 10 8 6 Serie1 Potenza (Serie1) Serie1 5 4 2 0 0 set-02 set-02 gen-04 gen-04 mag-05 ott-06 mag-05 feb-08ott-06 lug-09 feb-08 nov-10 lug-09 apr-12 ago-13 nov-10 dic-14 apr-12 Recently, Alberti et al. (2014), have underlined the possibility that in some polluted areas the presence of high concentration can be due to lower velocity areas if compared to neighboring zones.
AIM OF THE STUDY The aims of this study are to find possible parameters which can influence the contaminants in urban groundwater : 1. Statistical correlations between groundwater fluctuations and concentration of contaminants 2. Localization of low velocity areas in Milan s FUA and comparison with those areas with persistence concentration of pollutants in time. 3. Study the possible direction change of groundwater flow during time for those areas, and evaluate potential perturbation in concentration trend over time because of such changes.
METHODS AIM OF THE STUDY Possible influence on contaminations (piezometric fluctations, low velocity areas and flow change direction during time) TOOL Implementation of a numerical model in steady (May 2014) and unsteady state (from 2008 to 2014) RESULTS Identification of influencing phenomena on pollutants in urban groundwater Statistical correlations (thanks to a large dataset for the studied area)
CONCEPTUAL MODEL AND NUMERICAL MODEL IMPLEMENTATION (1/4) MODEL DOMAIN The model domain is extended to the north eastern part of Milan FUA and is characterized by areas with high density of productive/industrialized areas Previous studies (DICA-Politecnico di Milano) have underlined the possibility, for some contaminated areas, that the presence of a high contaminant concentration can be due to lower velocity with respect to the proximity zones. This situation could be preventing the renewal of groundwater quality. The preliminary feasibility study here shown could be the first step for individuation of critical areas where contaminants can be persistent during time
CONCEPTUAL MODEL AND NUMERICAL MODEL IMPLEMENTATION (2/4) GRID DEFINITIO N 167 m a.s.l. N 134 m a.s.l. Aquifer A and B reconstructed at hydrogeological basin scale S Horizontal discretization Vertical discretization 132 m a.s.l. 104 m a.s.l. 115 m a.s.l. 85 m a.s.l. 78 m a.s.l. Grid composed by cells (50 m x 50 m) (143,226 active cells) 3 layers to represent Aquifer A*, aquitard and Aquifer B* * ENI-Regione Lombardia aquifer definition -79 m a.s.l. 18 m a.s.l. Source Depth 0-30 m Depth Depth 30-100 m Depth >100 m Total CARG 14 251 163 428 dbarpalog 28 328 171 527 SIF 191 315 100 606 O Km 11 Km
CONCEPTUAL MODEL AND NUMERICAL MODEL IMPLEMENTATION (3/4) (1) Boundary and internal condition (2) Input equivalent hydraulic conductivity interpolation in Layer 1 CH- Piezometric survey map April 2014 Extraction wells Aquifer A Extraction wells Aquifer A-B Extraction wells Aquifer B-C River Lambro Well Villoresi channelinfiltration rate of 9.6*10-6 m/s * Q of wells.:sif and CUI DB Well Martesana channelinfiltration rate of 5.6*10-3 m/s SPECIFIED HEAD SPECIFIED FLOW FLOW DEPENDENT HEAD Constant head -piezometric survey in May 2014 No flow (East-West Boundary) Well (Villoresi and Martesana channel loss estimation and Q of extraction wells estimations*) River (Lambro river) average levels of the end of May 2014 Definition of an equivalent hydraulic conductivity for every layer starting from preexisting information (log stratigraphies of DB ARPA Lombardia - K values assigned for every lithology)
Interpolated zones of recharge PILOT MODEL CONCEPTUAL MODEL AND NUMERICAL MODEL IMPLEMENTATION (4/4) Meteorological station in Cinisello B.mo (January-April 2014) used to interpolate estimated rainfall (Thornthwaite method and ERSAF use of soil data to infiltration). Recharge data and zonation (3) ZONE (-) Red AVERAGE VALUE (m/s) 8.21e-9 TYPE Green and irrigated areas (only infiltration from rainfall) Yellow 8.26e-9 *Urban of Milan Brown 4.00e-9 *Urban of Sesto S. Giovanni Violet 3.96e-9 *Urban of Monza * The value of infiltration in urban areas is estimated by 15% of losses of water supply wells pumping rate
ARPA LOMBARDIA PIEZOMETRIC SURVEY MAY 2014 AND CALIBRATION TARGETS Aquifer A (Layer 1) Aquifer B (Layer 3)
STEADY STATE CALIBRATION Automatic calibration with PEST K matrix (all Layers) with Pilot Points every 2 km Aquifer A Absolute residual mean:16 cm Scaled absolute residual mean: 0,3% K (m/s) Layer 1 Piezometric head and residual (calibrated values) Name X (m) Y(m) K (m/s) Aquifer Kp151 1514480 5046368 0,00110 A Kp150 1519124 5042991 0,00071 A Kp153 1515585 5040048 0,00098 A Kp145 1514493 5046341 0,00110 B Kp146 1514126 5044438 0,00140 B Kp148 1517422 5039390 0,00120 B Kp149 1517765 5039414 0,00070 B Kp152 1520861 5039530 0,00102 B 8 K values from Pumping tests Over-estimated Under-estimated Simulated piezometric head Aquifer A
PILOT MODEL SOME TIME SERIES UNSTEADY VALIDATION Piezometric head (m a.s.l.) Piezometric head (m a.s.l.) The piezometric increase is observed not only within the city but also in some zones of the FUA 121,00 120,00 119,00 118,00 117,00 116,00 115,00 114,00 113,00 112,00 Observed ACQ27 Time 112,50 112,00 111,50 111,00 110,50 110,00 109,50 109,00 108,50 Observed FOG43 Time Simulation horizon: March 2008 September 2014 (6 years) 26 stress periods (1SP= steady state March 2008+25 SP = every 3 month) Variable input: Tri-Monthly variable recharge estimated with Thornthwaite method from rainfall and average monthly T data collected in the meteorological station (effective rainfall infiltration in aquifer) 109,00 108,50 108,00 107,50 107,00 106,50 106,00 105,50 105,00 Piezometric head (m a.s.l.)109,50 Observed Tri-Monthly variable extraction rate from water supply well in Milan Boundary conditions change on the basis of data available ACQ5 Time
Extraction rate (m3) PILOT MODEL UNSTEADY VALIDATION SIMULATION INPUTS Tri-Monthly Recharge estimation An estimation of the rainfall recharge value based on Cinisello B.mo station during 2008-2014 Tri-Monthly Water Supply wells based on data collected by Municipality of Milan (MM) Water supply wells - Crescenzago 1.200.000 Equivalent porosity field estimated from ARPA Lombardia stratigraphic DB in layer 1 Stress Period Recharge (m/s) 1.000.000 1 (April-June 2008) 1,90E-09 2 (July-15 September 2008) 0 3 (16 September -December 2008) 3,80E-09 4 (January-30 March 2009) 1,33E-09 800.000 600.000 400.000 200.000 0 0 5 10 15 20 25 Tri-Monthly Stress Periods Tri-Monthly recharge of the aquifer from rainfall (Thornthwaite method) + CH-Boundary condition change + Tri-Monthly Extraction rate water supply well = Historical series of piezometer REPRODUCTION OF UNSTEADY GROUNDWATER CONDITIONS
Piezometric head (m a.s.l.) Piezometric head (m a.s.l.) PILOT MODEL UNSTEADY VALIDATION WITH TIME SERIES UNSTEADY SIMULATION OF PIEZOMETRIC VARIATIONS IN THE STUDIED AREA 122,00 121,00 120,00 119,00 118,00 117,00 116,00 115,00 114,00 113,00 112,00 ACQ27 Observed Simulated Time In evidence: Good representation of the series trend observed 110,00 109,00 108,00 107,00 106,00 105,00 Piezometric head (m a.s.l.)111,00 ACQ5 112,50 112,00 111,50 111,00 110,50 110,00 109,50 109,00 108,50 Good superposition of piezometric values resulting from transient and steady state FOG43 Observed Simulated Time Observed SimulatedTime Potenza (Observed)
Head mean values (m a.s.l.) Valid Sig. (2-tailed) OBJECTIVES 1 STATISTICAL CORRELATION BETWEEN sogg_mean HYDROLOGICAL AND CHEMICAL PARAMETERS Sig. (2-tailed) Cluster Number of Case Cumulative Frequency Percent Valid Percent Percent 1 277 44,9 45,9 45,9 3 164 26,6 27,2 73,1 4 1,2,2 73,3 5 34 5,5 5,6 78,9 9 127 20,6 21,1 100,0 Total 603 97,7 100,0 Missing System 14 2,3 Total Cluster analysis of piezometric data available 617 100,0 The piezometric increase is observed not only within the city but also in FUA and was proven by statistical trends Temperatura acqua Tetracloroetilene Tricloroetilene Triclorometano (cloroformio) Valore_NUM carico_mean Pearson Correlation 1. a. a N 13 1 1 Pearson Correlation. a. a. a N 1 1 1 Pearson Correlation. a. a. a Sig. (2-tailed) Correlation between all parameters Valore_NUM sogg_mean carico_mean Valore_NUM sogg_mean carico_mean Valore_NUM sogg_mean carico_mean **. Correlation is significant at the 0.01 level (2-tailed). The piezometric trend is correlated with higher concentration of pollutants in the studied area *. Correlation is significant at the 0.05 level (2-tailed). N 1 1 1 Pearson Correlation 1 -.368 **.322 * Sig. (2-tailed),005,016 N 244 56 56 Pearson Correlation -.368 ** 1 -.566 ** Sig. (2-tailed),005,000 N 56 56 56 Pearson Correlation.322 * -.566 ** 1 Sig. (2-tailed),016,000 N 56 56 56 Pearson Correlation 1 -.405 **.461 ** Sig. (2-tailed),002,000 N 239 55 55 Pearson Correlation -.405 ** 1 -.565 ** Sig. (2-tailed),002,000 N 55 55 55 Pearson Correlation.461 ** -.565 ** 1 Sig. (2-tailed),000,000 N 55 55 55 Pearson Correlation 1,049 -,157 Sig. (2-tailed),723,253 N 237 55 55 Pearson Correlation,049 1 -.565 ** Sig. (2-tailed),723,000 N 55 55 55 Pearson Correlation -,157 -.565 ** 1 Sig. (2-tailed),253,000 a. Cannot be computed because at least one of the variables is constant. N 55 55 55
mg/l mg/l OBJECTIVES 2 LOW FLOW VELOCITIES AND CONSTANT CONCENTRATION Legend PCE Concentration microg/l 0,005-1,100 1,110-10,000 10,010-30,000 30,100-70,000 70,400-100,000 102,920-500,000 Time Particle Tracking March2008 - July2008 July2008 - July2009 July2009 - July2010 July2010- July2011 July2011 - July2012 July2012 - July2013 July2013 - July2014 Model Domain Groundwater velocity XYMagnitud 1,00e-008-9,00e-007 9,01e-007-2,00e-005 2,01e-005-3,31e-005 3,32e-005-4,92e-005 4,93e-005-6,97e-005 6,98e-005-2,05e-004 2,06e-004-7,27e-004 7,28e-004-1,62e-003 The persistent concentrations are in the lower velocity area. This can cause a low renewal of groundwater 16 14 12 10 8 6 4 5,5 5 4,5 4 3,5 3 2,5 PCE_152090253 PCE_150770029 If concentrations variations are in a low gap, where max and min are not greater than 3 times average value, the contaminant can be defined as stationary (EPA)
mg/l mg/l mg/l OBJECTIVES 2 LOW FLOW VELOCITIES AND CONSTANT CONCENTRATION Legend PCE Concentration microg/l 0,005-1,100 1,110-10,000 10,010-30,000 30,100-70,000 70,400-100,000 6 5 4 3 2 1 0 PCE_0150770029 102,920-500,000 Time Particle Tracking March2008 - July2008 July2008 - July2009 July2009 - July2010 July2010- July2011 July2011 - July2012 July2012 - July2013 July2013 - July2014 1,2 1 0,8 0,6 0,4 0,2 0 PCE_0150770022 Model Domain Groundwater velocity XYMagnitud 1,00e-008-9,00e-007 9,01e-007-2,00e-005 2,01e-005-3,31e-005 3,32e-005-4,92e-005 4,93e-005-6,97e-005 6,98e-005-2,05e-004 2,06e-004-7,27e-004 7,28e-004-1,62e-003 Particle tracking shows: -Particles arrive in the low velocity area and remain inside (particles in low area run only 200 meters instead of 2000 meters outside in the same interval time) - Particles change direction strongly during time 2,5 2 1,5 1 0,5 0 PCE_0150770023
OBJECTIVES 3 PCE PLUMES CONCENTRATIONS DUE TO DIFFERENT GRADIENT IN TIME 19 Degrees A target flow direction groundwater flow measure the angle of the water flow where observed concentration change quickly during time
OBJECTIVES 3 PCE PLUMES CONCENTRATIONS DUE TO DIFFERENT GRADIENT IN TIME 20 A variation of direction in flow generates also a variation of concentration in monitoring wells
CONCLUSIONS Previous studies (Alberti et al., 2014) have shown the existence of large low velocity areas in the city of Milan. A more detailed steady and transient state model has been implemented for the NE sector of Milano city and its Functional Urban Area, where a huge number of data are available. The numerical model allows with more detail to compare the velocity variation in time in different areas with the pollutants concentration, in order to verify the hypothesis in the previous work. The results show that the low velocity areas are in high correlation with the higher concentrations of contaminants Furthermore, the numerical transient model allows to adequately evaluate the 3D local piezometric gradients, due to the significant changes in piezometric head observed in recent years (2008-2014) in the FUA and the model. The results show that the direction flow can change over time, inducing eventually also a variations of concentrations for monitoring network wells. Proper management plans for groundwater and/or contaminated sites should consider these potential changes in direction of groundwater flow direction in order to avoid potential errors in monitoring wells data interpretation The present work can be improved by an increasing of transient target of groundwater levels and monitoring concentrations wells in proximity to the studied area. Future upgrades may be a possible upgrade of K and porosity values in case of more detailed future piezometric survey.
22 THANKS FOR YOUR ATTENTION! AKNOWLEDGMENT I thank the ARPA LOMBARDIA Environmental department for the database of log-stratigraphies K and porosity values, piezometric survey in May 2014 and concentrations values in the studied area